AUTHOR AVAILABLE from National Academy of Sciences

DOCUMENT RESUME

ED 247 445 CE 039 525

AUTHOR Reskin, Barbara F., Ed. TITLE Sex Segregation in the Workplace. Trends, Explanations, Remedies. INSTITUTION National Academy of Sciences National Research Council)Washington, DC. Committee on Women's Employment and Related Social Issues. SPONS AGENCY Carnegie Corp. of NeW York, N.Y.; Employment and Trai'ning Administration (DOL),, Washington, D.C.; Office of Vocatibnal and Adult Education (ED)., Washington, DC. REPORT NO ISBN-0-309-03445-0 PUB DATE 84 ?CONTRACT 300-81-0282

. NOTE 319p.; Revised versions of papers presented at a workshop sponsored by the U.S. Department of Education, the Employment and Ttaining Administration (DOL), and the Carnegie Corporation, May 1982. AVAILABLE FROMNational Academy Press, 2101.Constitution. Avenue, NW, Washington, DC, 20418. PUB TYPE Collected Works - General (020) Information Analyses (070) -- Reports - Research/Technical (143)

EDRS PRICE MF01 Plus Postage. PC Not Available from EDRS. DESCRIPTORS Adults; Career Choice; Career Education; *Employed. Women; *Employment Practices; *Equal Opportunities (Jobs); Females; Organizational Climate; *PerSonnel Integration; Salary Wage Differentials; *Sex Discrimination; Sex Fairness; Workshops IDENTIFIERS *Occupational Segregation; *Sex Segregation

ABSTRACT This volume includes revised presentations and commentaries from a workship to review evidence for various theoretical explanations for occupational segregation and to report empirical research to enlarge understanding of the topic. An introduction summarizes contents. In partI live chapters on the extent of and trends in segregation document adecline in the segregation index, report an examination of sex segregation within organizations, address change in occupational sex composition experienced with job change and movement by race among occupations with different sex compositions, comment on contradictions among these papers, and project occupational segregation for the1980s. Eight chapters in part II attempt to describe segregation by considering economic approaches to sex segregation, proposing a general theory to explain occupational segregation and wage differentialsr criticizing this theory, reviewing the human capital -explanation attributing segregation to women's preferences,reviewing literature linking sex typirrz in socialization to occupational choice, responding to that icAiew, examininginstitutional barriers to sex integration, and commenting on thatexamination. The three papers in part III on theeffectiveness of interventions to reduce segregation review literature to examine impacts, evaluate occupational desegregation in Comprehensive Employment andTraining Act programs, and comment on the previous paper.Concluding remarks integrate several recurring themes. (YLB) Pi Sex wim Segregation inthe Workplace Tends, Explanations, Remedies

Barbara F. Reskin, Editor

Committee on Women's Employment andRelated Social Issues Commission on Behavioral and Social Sciences andEducation National Research Council

NATIONAL ACADEMY PRESS Washington, D.C. 1984

U.S. DEPARTMENT OF EDUCATION NATIONAL INSTITUTE OF EDUCATION "PERMISSION TO REPRODUCE THIS N EDUCATIONAL RESOURCES INFORMATIO MATERIAL INM'CROFICHE ONLY CENTER (ERIC) This document has been reproduced a, HA; BEEN GRANTED BY received from the person or organization originating 0 /--)' Minor changes have been made to imprive L> L 1/.1)/... reprodur bon quality L "

Points of view or ()Pinions stated in this docu TO EDUCATIONAL RESOURCES' merit 'to not ner essitnlv represent official NIE IhnoliorlI 00th y INFORMATION CENTER (ERIC)." 2 NATIONAL ACADEMY PRESS 101 CONSTITUTION AVENUE,NW WASHINGTON, DC 20418 NOTICE: The project that is the subject of this report wasapproved by the Cover,...ig Board of the National Research Council, whose aembers are drawn from the councils of the NationalAcademy of Sciences, the National Academy of Engineering, and the Institute of Medicine. The membersof the committee responsible for the report were chosen for their special competencesand with regard for appropriate balance. , This report has been reviewed by-a group other than the authorsaccording to procedures approved by a Report Review Committee consisting of members of the NationalAcademy of Sciences, the National Academy of Engi- neering, and the Institute of Medicine.

The National Research Council was established by the National Academyof Sciences in 1916 to associate the broad community of science and technology with theAcademy's purposes of furthering knowledge and of advising the federal government. The Council operates in accordance with general policiesdetermined by the Academy under the authority of its congressional charter of 1863, whichestablishes the Academy as a private, nonprofit, self- governing membership corporation. The Council has becomethe principal operating agency of both the National Academy of Sciences and the National Academy of Engineering in the conduct oftheir services to the government, the public, and the scientific and engineering communities. It is administeredjointly by both Academies and the Institute of Medicine. The National Academy of Engineeringand the Institute of Medicine were established in 1964 and 1970, respectively, under the charter of the NationalAcademy of Sciences.

Library of Congress Cataloging in Publication Data Main entry under title:

Sex segregation in the workplace.

Revised versions of papers, originally presented at a workshop held in May 1982. 1. Sex discrimination in employmentUnited States Congresses. 2. Sex discrimination against women United StatesCongresses. I. Reskin, Barbara F. II. National Research Council (U.S.). Committee on Women's Employment and Related Social Issues. H D6060.5. U5S4751984 331. 4'133'0973 84-8342

ISBN 0-309-03445-0

Printed in the United States of America

3 "kt

Committee on Women's Employment and Related Social Issues

ALICE S. ILCHMAN (Chair), President, Sarah Lawrence College CECILIA BURCIACA, Office of the President, Stanford University CYNTHIA FUCHS EPSTEIN, Department of Sociology, Queens College and the Graduate Center, City University of New York LAWRENCE M. KAHN, Institute of Labor and Industrial Relations, University of Illinois GENE E. KOFKE, American Telephone and Telegraph Co., New York ROBERT E. KRAUT, Bell Communications Research, Murray Hill, N.J. JEAN BAKER MILLER, Stone Center for Developmental Services and Studies, Wellesley College ELEANOR HOLMES NORTON, Georgetown University Law Center GARY ORFIELD, Department of Political Science, University of Chicago NAOMI R. QUINN, Department of Anthropology, Duke University ISABEL V. SAWMILL, The Urban Institute, Washington, D.C. ROBERT M. SOLOW, Department of Economics, Massachusetts Institute of Technology LO VISE A. TILLY, Department of History, University of Michigan DONALD J. TREIMAN, Department of Sociology, University of California, Los Angeles

BARBARA F. RESKIN, Study Director MARIE A. MATTHEWS, Administrative Assistant

iii Contributors

TAMES N. BARON, School of Business, Stanford University ANDREA H. BELLER, Department of Family and Consumer Economics, UniVersity of Illinois `SUE E. BERRYMAN, The Rand Corporation, Santa Monica, California WILLIAM T. BIELBY, Department of Sociology, University of California, Santa Barbara FRANCINE D. BLAU, Department of Economics and Institute of Industrial and Labor Relations, University of Illinois MARY C. BRINTON, Department of Sociology, University of Washington PAMELA S. CAIN, Department of Sociology, Hunter College MARY CORCORAN, Department of POlitical Science and Institute for Social Research, University of Michigan GREG J. DUNCAN, Department of Economics and Institute for Social Research; University of Michigan KEE-OK KIM HAN, Department of Family and Consumer Economics, University of Illinois SHARON L. HARLAN, Center for Research on Women, Wellesley College MARYELLEN R. KELLEY, College of Management, University of Massachusetts, Boston MARGARET MOONEY MARINI, Department of Sociology and Anthropology, Vanderbilt University KAREN OPPENHEIM MASON, Department of Sociology and Population Studies Center, University of Michigan BRIGID O'FARRELL, Center for Research on Women, Wellesley College MICHAEL PONZA, Department of Economics, University of Michigan BARBARA F. RESKIN, Department of Sociology, University of Michigan PATRICIA A. ROOS, Department of Sociology State University of New York, Stony Brook RACHEL A. ROSENFELD, Department of SoPiology and Carolina Population Center, University of North Carolina MYRA H. STROBERi Center for Research on Women and School of Education, Stanford University LINDA J. WAITE, The Rand Corporation, Santa Monica, California WENDY C. WOLF, Public/Private Ventures, Philadelphia

it) Contents

Preface vii

1 Introduction 1 Barbdra F. Reskin

IEXTENT, TRENDS, AND PROJECTIONS FOR THE FUTURE . 9

2 Trends in Occupational Segregation by Sex and Race, 1960-1981 . . 11 Andrea H. Beller

3 A Woman's Place Is With Other Women: Sex Segregation Within Organizations 27 William T. Bielby and James N. Baron

4 Job Changing and Occupational Sex Segregation: Sex and Race Compariscns 56 Rachel A.Rosenfeld

5 Commentary 87 Pamela Stone Cain

6 Occupational Sex Segregation: Prospects for the 1980s . . . . 91 Andrea H. Beller and Kee-ok Kim Han

6 a ,

IIEXPLAINING SEGREGATION: THEORETICAL PERSPECTIVES AND EMPIRICAL EVIDENCE 115

pccupational Segregation and Labor Market Discrimination 117 Francine D. Biau

8 Toward a General Theory of Occupational Sex Segregation: The Case of Public School Teaching 144 Myra H.Strober

9 Commentary: Strober's Theory of Occupational Sex Segregation . . Karen Oppenheim Mason

10 Work Experience, job Segregation, and Wages 171 Mary Corvoran, Greg J. Duncan, and Michael Ponza

11 Sex Typing in Occupational Socialization 192 Margaret Mooney Marini and Mary C. Brinton

12 Commentary 233 Wendy C. Wolf

13 Institutional Factors Contributing to Sex Segregation in the Workplace 235 Patricia A. Roos and Barbara F. Reskin

14 Commentary: The Need to Study the Transformation of Job Structures, 261 Maryellen R. Kelley

IIIREDUCING SEGREGATION: THE EFFECTIVENESS OF INTERVENTIONS 265

15 Job Integration Strategies: Today's Programs and Tomorrow's Needs . 267 Brigid O'Farrell and Sharon L. Harlan

16 Occupational Desegregation in CETA Programs 292 Linda J.Waite and Sue E. Berryman

17 Commentary 308 Wendy C. Wolf

18 Concluding Remarks 310 Francine D. Blau

vi Preface r

The segregation of the sexes into different occupations, industries, and (within &ems) specific jobs is one of the most stable and striking features of the American workplace. Although the sexes have become increasingly similar in their likelihood of employment outside the home, within the workplace women and men differ dramatically in_ the kinds of jobs they hold. Sex segregation is problematic for several reasons. Most importantly, it promotes and sustainsthe wage gap between the sexes. Barring substantial changes in the ways that jobs are evaluated and wages set, women's prospects for economic parity will depend on their migration into mainstream "male" jobs, away from the many low-paying jobs most frequently held by women. In view of the pervasiveness of segregation and, its adverse consequencesfor women, in 1981 several groups sponsored an examinationof sex segregation in the workplace by the Committee on Women's Employment and Related Social Issues of the National Research Council. The sponsors are tLe U.S. Department of Ed- ucation, the Employment and Training Administration of the U.S. Department of Labor, and the Carnegie Corporation of New York. The committee's mandate was twofold: to convene a major interdisciplinary workshop on job segregation and to prepare a state-of-the-art report on the topic. The two-day workshop, held in May 1982, brought together two dozen scholars. This volume includes revised versions of several papers presented there and the remarks of commentators, along with three papers the committee subsequently commis- sioned. These papers served as a resource to the committee in preparing its final report, Women's Work, Men's Work: Segregation on the Job, and stand as a com- panion to that volume. The purposes of the workshop were to bring together scholars from several disciplines to review the evidence for various theoretical explanations for segregation and to report empirical research they were conducting that would enlarge our understanding of its extent, form, and causes. For this reason some of the papers, vii and thus the chapters in this volume, primarily review the literature (Blau, Marini and Brinton, Robs and Reskin, and O'Farrell and Harlan), while others offer up- to-date empirical \findings (Beller, tielby and Baron, Beller and Han, Rosenfeld, and Waite and Ber' man). Two papers combine the presentation of original research with either a critical review of a theoretical perspective (Corcoran, Duncan, and Ponza) or the presentation of a new theoretical approach (Strober). Many of the authors of this volume thank colleagues or assistants for their help. The workshop at which most of these chapters and comments were first presented and this volume also benefited from the work of several people, to whom I express my appreciation. As study director of the committee, Barbara F. Reskin was a valuable intellectual resource and an able manager of our work. Marie A. Matthews, administrative assistant to the committee, was indispensable in organizing the workshop. The members of the Committee on Women's Employment and Related Social Issues and, Heidi I. Hartmann, as associate executive director of the Commission on Behavioral and Social Sciences and Education, helped identify workshop participants, participated in the workshop, and refereed papers for inclusion in this volume. Christine L. McShane, editor for the commission, worked with the authors and the National Academy Press in producing it. This volume would not exist without the behind-the-scene contributions of these people, and I than k them warmly.

ALICE S. ILCHMAN, Chair Committee on Women's Employment and Related Social Issues

viii r

t. ,Sex

Segnation et, inthe Workplace 7ivid; Elciptemation.s Remedies 1Introduction BARBARA F. RESKIN

The concept of segregation was first age, sex, and social class are the most com- brought to public attention in the United mon. Because mostof thjse forms of seg- States to describe the enforced separation regation mirror social norms about the ap- of black and white children in different propriate and "natural" relations between schools. Although strictly speaking segre- groups (just as prior to the 1954Brown de- gation denotes physical separation, it typi- cision many people defined race segregation cally involves an institutionalized form of as natural and appropriate)and because of social distance between dominant and sub- their very pervasiveness, these forms of sep- ordinate groups (Kuper, 1968:144). Cer- aration are not readily thought of as segre- tainly racial segregation in this country en- gation. We take for granted, for example, tailed more than physical separation; not only that children will be separated into age-based did it reflect the belief that black children groups at school and that theywill spend were not fit to associatewith white children, their days apart from most adults. Indeed, but it also made other form& of unequal it is when the accepted patterns of segre treatment possible. gation vary that we noticeforexample, Years of litigation, protests, and busing more than one or twoadults on a school have brought the concept of segregation into playground during recess or children in work the public vocabulary and persuaded most settings. Americans of the, existence of racial segre- The segregation of the sexes in Atome gation in schools and neighborhoods. Atthe spheres is at least as common as that of chil- same time, these activitieshave probably dren from adults. Yet it is often not visible helped to associate the idea of segregation for two reasons. First, cultural expectations, with race discrimination. But our society, which structure our perceptions of the world, like most others, segregates its members on take for granted that most adults live inti- the basis of characteristics other than race; mately with a member of the opposite sex. Because such intimacy is at odds with the model of physical separation implied by the I wish to express my thanks to my friendand col- league, Lowell L. Hargens, for his help in readingand paradigmatic case of racial segregation, it discussing the papers in this volume. masks the existence of sex segregation. Sec-

I BARBARA F. RESKIN ond, the presence of women and men pub- dition, even within integrated industries, licly carrying out a variety of activities to- firms may employ only men or only women gether supports the impression of sex (see Bielby and Baron, in this volume). integration. Superficially these twophe- Clearly, then, a substantial proportion of nomena appear toinvalidate any claim that American workers are physicals segregated the sexes are segregated. from the opposite sex. Our interest in this volume centls onthe If we extend the meaning of segregation segregation of women and men at work, re- beyond physical separation to encompass gardless of whether the sexes are substan- functional separation, the ikorkplace is seg- tially segregated in most partsottheir lives. regated in a third way, with a divisioh of labor by sex the rule. Furthermore, the .In that context, work can becharacterized as sex segregated inthree ways. First, norms practice of employing women and men to that relegate the sexes to separate spheres do different jobs within the same work set- (Welter, 1966; Bloch, 1978) women to the ting is often accompanied by the institu- home and men to the public sectornec- tionalized social distance that segregation essarily imply their physical separation. For frequently entails. This social distance is example, domestic workers in the private marked by differential access to authority sphere, whether theare unpaid or paid, (Wolf and Fligstein, 1979), unequal *wages carry out theirduties in a female environ- (Treiman and Hartmann, 1981), separate job ment, pursuing one ofthe most segregated ladders, and exclusionary practices restrict- jobs. Second, many paid employees work in ing mobility between positions labeled"male" exclusively one-sex settings. Whole indus- and "female" (Roos and Reskin, in this vol- tries are dominated by men;metal and coal ume). Hospitals are a good example. As out- mining, fisheries, horticultural services, siders, we notice female and male employ- logging, construction, and railroads were all ees interacting in variouswaystalking or more than 90 percentmale in 1980 (Bureau joking together in the corridors or wards, of Labor Statistics, 1981:Table 30).Although working side by side over patients in ex, there are nri industries sooverwhelmingly amining and operating rooms, often simi: female, in p, rt because in even the most larly dressed in lab coats or scrub suits. Yet female-intensive industries men hold man- nurses, technicians, clricalworkers, and food agerial posts, women constitute morethan service workers are overwhelminglyfemale, three-quarters of all workers in several in- while doctors, administrators, and orderlies dustries, including direct sales, employ- are predominantlymale. Ironically, it is the ment agencies, convalescentinstitutions, li- functional segregation of the sexes into dif- braries, and apparel and fabricated textile ferent jobs that renders them interdepen- manufacturing. In 1980 over 32million dent and ensures their physical integration. workers were employed in industrieswhose It should be recognized, too, that thephys- work forces were at least 80 percentmale ical integration we observe is preceded, at or female, andslightly more than this num- least for technical and professional staff, by ber 11 million women and 22million separate training programs in which the sexes menworked in detailed census occupa- are physically segregated.This separation tions in which at least 90 percentof the may help preparethem for the unequal sta- incumbents were of their own sex.' Inad- tus and rewards they experiencewhen as

1 The Census.Bureau categorizes Occupations at vary- (1981:Table 30). The number of workers in occupations ing levels of detail. In 1980 theclassification referred that were at least 90 percent metnbers of the incumbent's sex was computed from Bureau of the Census to as -detailed" included 503 occupations.The number of workers in industries that were atleast 80 percent data (1983:Table 1). female was computed from Bureau of LaborStatistics

9 1N7 RODUCTION 3 workers they are physically integrated. overwhelmingly sex typed? What, kinds of Having shown .how the concept of seg- remedies might 'reduce segregation levels? regation applies to women's positions in the It is these questions to which the papers in workplace, we must 'now ask why an inquiry this volume Provide answers. The remain- into sex segregation in the world of work is der of this chapter is an overview of their necessary. Since dividing work on thebasis themes. of sex is customary in the home, why not in the workplace? To answer this question, let's return to the discussion of the consequences EXTENT, TRENDS, AND PROJECTIONS of segregating black and white schoolchil- FOR THE FUTURE dren. Beyond. its stigmatizing effects,- differentiating and separating people are often From its emergence as. a major institution accompanied by differential treatment. Just -in the nineteenth century, the U.S.. ,labor as the segregatedschools to which black force has been highly segregated by' sex. children were sent were inferior to white Most occupations were so dominated by one children'schis (Kluger 1975), the jobs sex that for decades the CensusBureau that wom h 1d provide rewards that 'are changed gender - discrepant' responses for inferior to t irise that "male" jobs offer. certain occupations on the assumption that Foremost is the effect of segregation on they represented coding errors (Conk, 1981). women's wages. The more "female" an oc- Empirical studies assessing the extent of occupation is, the less it typically pays (Rytina, cupational segregation have consistently 1982). Between 35 and 40 percent of the confirmed high leveleof segregation (Gross, well-documented wage gap between female 1968; Blau and Hendricks, 1979; Lloyd and and male full-time workers can he attributed .Niemi, 1979; Williams, 1979; England, 1981). to their segregation into diftetentdetailed Despite dramatic changes in both the com- occupational categories (Treiman and Hui- position of the labor' force and the occupa- mann, 1981). Theadditional segregation of tional structure, segregation levels have been women and men in the same occupations extraordinarily'stable throughout the twen- into different jobs explains even more ofthe tieth century. This raises several questions. differential. The wage lobs associated with First, have social and normative changes in working in female-dominated occupations the 1970s or the existence or enforcement has especially adverse consequences for of antidiscrimination laws led to appreciable women who are the sole supportersof their declines in segregation? What are the pros- families. Ehrenreich and Stallard (1982) pects for.the remainder of this decade? How commented that it is not the ,absence of a much segregation within specific employ- man in a household butthe absence of a ment settings is masked by aggregate esti. male salary that pUshes working women into mates based on data for occupations? What poverty; more precisely, it is the Absenceof ban we learn if we go beyond the static pic- the salary levels that male-dominated jobs tures that occupational distributions yield to pfovide. For women who support families look at workers' job histories? on their -own, segregation may mean pov- The papers on the extent of and trends in erty. segregation in Part I of this volume illumi- rwse factsthe pervasiveness of set nate these questions. In Chapter 2, Andrea segregation and its economic implications H. Beller provides new and encouraging for womenpose important scholarly "and evidence regarding trends in aggregate seg- policy questions. What are the current lev- regation levels since 1970. Using Current els of segregation, and what are the pros- Population Survey data for the period be- pects for the decade ahead? Why iswork so- tween 1971 and 1981, she documents a 10 4 BARBARA F. RESKIN percent decline in the segregation index. atypical occupations and then investigates Unlike most of the previous research, she its determinants. Of considerable interest provides separate analyses for nonwhites and are results brokendown by race that show whites that show more rapid.declines among the proportions of women and men who nonwhites. Especially telling are results that moved between occupations in which mem- reveal particular progress within profes- bers of their sex were a majority and those sional occupations for whites (but not non- in which they were a minority. Rosenfeld's whites). an Outcome that Beller argues is subsequent examination of the wage and linked to desegregation in college majors. prestige consequences of different types of ' It is well known that the more closely one moves points to factorsthat may prompt is able to look into the workplace, the more workers to enter and leave sex-atypical work. segregation one will observe. Thus, segre- Also important are analyses showing (1) how gation indices computed for the 11 major workers' personal characteristics are linked census occupational groupingsshow consid- to an occupation's sex type and (2) whatchar- erably less segregation than do indices com- acteristics are associated with an individual's puted for detailed occupational categories. !'.,..aking an occupation's sex barrier. Spe- However, researchers have not had data sets cific findings, such as the absence of any that permit them to assess segregation levels effect of family responsibilities on the type within firms for a large number of estab- of move a worker makes, bear on theories lishments. William T. Bielby and James N. that seek to explain segregation. Baron's work (Chapter '3) is an important In commenting on the first three chapters exception. They examined U.S. Employ- .in Part 1, Pamela S. Cain notes in Chapter ment Service data for'aimost 400California 5 some apparent contradictions between firms employing more than 60,000 workers them and offers a resolution. She also re- to address several issues previouslyb4ond minds the reader of the inherent limitations the reach of scholars. The result is a set of that available tools and data place on study- striking and disturbing findings. For ex- ing sex segregation. ample, over half the firms were totally sex In the final chapter in Part I, Andrea H. segregated: not a single job title was held Beller and Kee-ok Kim Han use trend data by both men and women. Furthermore, to project the level of occupational segre- across all firms, the proportionof workers gation at the end of the decade (Chapter 6). who held nominally integrated jobs (i.e., jobs They use several models to generate a set held by both men and women in a fir in) was of projections. Of particular relevance to only 10 percent. An analysis of the small policy makers are the results for models based number of firms that were minimally inte- on optimistic, intermediate,and conserva- grated permitted the authors to identify tive assumptions about the rate of decline, mechanisms that support segregation in dif- which could reflect such factors as whether ferent types of establishments. These find- affirmative action regulations are enforced. ings contribute to our understanding ofthe Under the most optimistic assumption, seg- organizational bases of sex segregation. regation. would decline markedly, but the It is also possible to get beyond aggre- models that Beller and Han judge to be more gated occupational data by tracking workers' realistic predict only modest declines. Social patterns of movement betweensegregated policy must be guided by what is likely to and integrated jobs. Rachel A. Rosenfeld's happen in both the presence and absence research employs such an approach.In of deliberate interventions to reduce seg- Chapter 4, Rosenfeld estimates the amount regation. Their paper provides such infor- of such mobility between sex-typical and sex- mation and draws its implications for policy.

14 INTRODUCTION 5

maintain patriarchal privilege in the home EXPLAINING SEGREGATION and that pursuing this goal gives rise to both The chapters in Part II grapple with the segregation and lower wages for women. controversial, and difficult question of why Historical data on shifts in the gender label gender is linked to the work people do. In- of public school teaching illustrate the the- dividually,,each summarizes and weighs the ory. empirical evidence associated with a partic- In a close analysis of Strober's theory ular explanatory orientation. Jointly, they (Chapter 9), Karen Oppenheim Mason takes provide both a sound foundation and an issue with certain assumptions as empiri- agenda for needed research. cally unsupported. Mason disputes Strob- Francine D. Blau's paper on,labor market er's claim that existing ideas cannot ade- discrimination and occupational segregation quately explain segregation and offers a set (Chapter 7) is one of three that consider of theoretical approaches that she contends economic approaches to sex segregation. The account for the persistence of segregation. economics literature on discrimination has It has been suggested that the concentra- concentrated on the role of discrimination tion of women in certain occupationsreflects in the wage differential betweenthe sexes their own preferences, which in turn stem (see Blinder, 1973; Osterman, 1979; Cabral either from beliefs that these occupations et al., 1981), but very littlehas been pub- are compatible wit`women's domestic roles lished specifically on the role of labor market or from a socialization processthAt predis- discrimination in maintaining sex segrega- poses tkiem. toward certainkinds of work. tion. Focusing on this question, Blau criti- Each alternative has stimulated large bodies cally appraises the utility of several general of research. Mary Co:coran. Greg J. Dun- theories of discrimination, including those can, and Michael Ponza review Chapter invoking taste, overcrowding, monopsony, 10 the human capital explanation that attri- statistical discrimination, and dual labor butes segregation-to wompn's desire to find markets as well as the human capital alter- jobs that do not conflictwiththeir domestic native. °Having laid out the theoretical al- obligations. They put this explanation to a ternatives, Blau evaluates the empirical evi- test with evidence from other researchand dence on the extent to which discrimination their own current work. The authors pre- contributes to 2egregation. In doing so she sent results from their analysis ofdata from details the difficulties in trying to measure the Panel Study of Income Dynamics on the discrimination and emphasizes the need for duration of work experience, part-time work, research that can distinguish between the and occupational sex composition that chal- various alternatives. lenge predictions based on the human cap- 1n Chapter 8, economist Myra H. Strober ital approach. Of special interest are anal- rejects existing theories of discrimination as yses that cast doubt on thehuman capital inadequate to explain how occupations get assumption that skill depreciation and con- assigned to one sex or the other and what comicant wage losses associated with time contributes to stability or change in these out of the labor force prompt women to es- gender designations. Exploiting ideas from chew certain occupations. Their findings existing theories, she proposes a provocative represent an important contribution tothe new "general theory" toexplain both oc- development of a body of knowledge re- cupational segregation and wage differen- garding how familial roles influence wom- tials. The argument claims that the labor en's occupational outcomes. market behavior of menemployees and Margaret Mooney Marini and Mary C. workersis governed by their desire to Brinton provide in Chapter 11 a compre- 6 BARBARA F RESKIN hensive synthetic review of the massive lit- tions in the research that Roos and Reskin erature that links sex typing in socialization review and questions the omission of the to occupational choice. Their review covers effects of such factors as job design and eval- research on (1) the existence of sex differ- uation. Noting that little is known about how ences in occupationalpreferences, knowl- women are channelled into sex-typed career., edge, skills, and traits and (2) whether ob- paths, she calls for research on this topic. served differences result from sex-role socialization within families, schools, and the REDUCING SEGREGATION mass media. Because of the direct link between education and occupational options, Policy makers will find the chapters in they pay special attention to education in Part III on the effectiveness of interventions general and mathematics and science edu- to reduce segregation especially useful. In cation in particular. This chapter, which ul- Chapter 15, Brigid O'Farrell and Sharon L. timately draws conclusions about the effects Harlan examine the impacts of various in- of sex typing on segregation, is an important terventions on the basis of an extensive resource for researchers. leading of case studies. From these data they In response to Marini and Brinton, Wendy draw some general conclusions about what C. Wolf cautions that, in view of the mul- kinds of intervention succeed and the con- titude of factors implicated by the occupa- ditions under which they work best. They tional socialization literature, the outcomes point out, for example, that, to increase of any particular intervention attempts are women's representation in male-labeled jobs, unpredictable (Chapter 12). She reminds the companies had to modify certain personnel reader that most of the literature reviewed practices, such as recruitment procedures, by Marini and Brinton deals with differ- seniority systems, required qualifications, ences between the sexes before they enter and job training. the labor market. She points out that the In contrast to O'Farrell and Harlan's sur- constraining effects of such factors may de- vey of workplace-based remedies, Linda J. cline for adult women who face the eco- Waite and Sue E. Berryman evaluate the nomic realities of earning adequate wages. effectiveness of a single program, the Com- In Chapter 13, Patricia A. Roos and Bar- prehensive Employment and Training Act bara F. Reskin draw on labor market the- (CETA), for several employment outcomes ories to develop a framework in which a of black, white, and Hispanic women and variety of institutional barriers to sex inte- men (Chapter 16). Their statistical analyses gration are examined. They focus on formal fail to show effects of race or Hispanic eth- procedures within establishments and the nicity but do show sex differences in pro- organization of labor markets that discour- gram assignment consistent with sex seg- age or exclude workers from enteringjobs regation. Two especially interesting analyses that have been defined as belonging to the address CETA's ability to foster desegre- other sex. They consider, in turn, harriers gation. The first examines the link between to job training (including apprenticeships), the sex label of participants' pre-CETA jobs barriers to entry-level positions, and struc- and their CETA placements, and the second tural barriers that limit women's promotion looks at CETA's record in meeting partici- into and retention in sex-atypical jobs. They pants' preferences for sex-atypical assign- cite a wide variety of studies that show how ments. However, the data Waite .and Ber- these barriers perpetuate the segregation of ryman use were collected prior to 1978, when the sexes. CETA reauthorization legislation made sex Ili commenting on this paper (Chapter equity an explicit program goal, as Wendy 14), Maryellen R. Kelley points to liiiiita- Wolf notes in her commentary (Chapter 17).

1 6 INTRODUCTION 7

Post-1978 evaluations might yield a different view of Economics and Statistics 63(Nov- picture. ember):573-80. Conk, Maego A. 1981 Accuracy, efficiency and bias: the interpretation of women's work in the U.S. Census of CONCLUSION Occupations, 1890-1940. Historical Methods In recapping the papers in thisvolUme, 1(Spring):65-72. Duncan, Otis Dudley, and Beverly Duncan Francine D. Blau integrates se--.tral recur- 1955 A methodological analysis of segregation in- ring themes (Chapter 18). She points to the dices. American Sociological Review 20:200- variety of ways that federal activities may 17. help reduce or sustain sex segregation. Blau Ehrenreich, Barbara, and Karin Stallard 1982 The nouveau poor. Ms. Magazine 11(Au- reminds readers that economic parity is not gust):217-24. a necessary consequenceof occupational de- England, Paula segregation. On the basis of the papers in 1981 Assessing trends in occupational sex segregation, 190011976. Pp. 273-95 in Ivar Berg (ed.), this volume, howet it seems unlikely that Sociological Perspectives on Labor Markets. we shall have to copewith that concern in New York: Academic press. the near future. It is to be hoped that the Gross, Edward publication of these papers will help move 1968 Plus ea change ...? The sexual structure of occupations over time. Social Problems 16:198- us closer towardthat goal. 208. Kluger, Richard 1975 Simple Justice. New York: Random House. r Kuper, Leo 1968 Segregation. Pp. 144-150 in David L Sills (ed.), REFERENCES International Encyclopedia of the Social Sci- ences 14. New York: Macmillan. Blau. Francine D., and Wallace E. Hendricks Lloyd, Cynthia B., and Beth T. Niemi 1979 Occupational segregation by sex: trcnds and prospects. Journal of Human Resources 1979 The Economics of Sex Differentials. New York: Columbia University Press. 14(21:197-210. Osterman, Paul Blinder, Alan S. Wage discrimination: reduced form and struc- 1979 Sex discrimination in professional employ- 1973 ment: a case study. Industrial and Labor Re- tural estimates. Journal of Human Resources lations Review 32(July):451 -64. 8(Fall):436-55. Rytina, Nancy F. Bloch, Ruth H. American feminine ideals in transition: the rise 1982 Earnings of mer and women: a look at specific 1978 Monthly Labor Review105 of the moral mother, 1785 -1815. Feminist occupations. (April):25-31. Studies 4(June):101 -26. Treiman, Donald J., and Heidi Hartmann (eds.) Bureau of the Census Detailed Occupations and Years of School 1981 Women, Work, and Wages: Equal Pay for Jobs 1983 of Equal Value. Washington, D.C.: National Completed by Age, for the Civilian Labor Force by Sex, Race, and Spanish Origin: 1980. Wash- Academy Press. ington, D.C.: U.S. Department of Ce.n- Welter, Barbara merce. 1966 The cult of true womanhood. American Quar- terly 18(Summer):151-74. Bureau of Labor Statistics 1981 Employment and Unemployment: A Report on Williams, Gregory 1980. Washington, D.C.: U.S. Department of 1979 The changing U.S. labor force and occupa- Labor. tional differentiation by sex. Demography 16:73- Cabral, Robert, Marianne A. Ferber, and Carole A. 88. Green Wolf, Wendy C., and Neil D. Fligstein Sex and authority in the workplace. American 1981 Men and women in fiduciary institutions: a study 1979 of sex differences in career development. Re- Sociological Review 44(April):235 -52.

17 Part I Extent, Trends, and Projections for the Future

is Trends in Occupational. Segregation Zi by Sex and Race, 1960-1981

ANDREA H. BELLER

Interest among economists in occupa- 1973), and 1950-1970 among professional oc- tional sex segregation stems from the fairly cupations (Fuchs, 1975). Using the index of well established relationship between the segregation from the Duncan Index (Dun- sex differential., in earningsand women's can and Duncan, 1955), thesestudies con- concentration in a small number of occu- cur in the relative lack of changenoted in pations. It also stems from a family-based occupational segregation through 1960 and analysis of women's roles, although this con- the small decline during the 1960s. (The nection continues to be controversial. Such decline of sex segregation in the professional an analysis says that because oftheir family occupations during the 1960s was somewhat roles, women invest less in market-oriented greater than that for all occupations.) human capital than men do (Becker, 1981), With the strengthening of equal employ- and this includes choosing traditionally fe- ment opportunity (EEO) legislation and the male occupations (Polachek, 1979). Recent promulgation of equal educational oppor- empirical studies tend to refute this expla- tunity legislation in 1972, one might have nation of sex differences in occupational anticipated an accelerated decline in occu- choices (Beller, 1982b; Corcoran and Dun- pational segregation during the 1970s. can, 1979; England, 1982). While untan- Moreover, there is a general perception that gling the causes of occupational sex segre- many women are becoming increasingly ori- gation has proved an ambitious challenge, ented toward nontroditional family roles and measuring its trends is no less difficult. nontraditional jobs in the workplace.' Sur- This paper assesses the trends in occu- prisingly, the index of segregation remained pational segregation of the sexes during the unchanged through 1976 ca 1977, according 1970s and compares them with those of the to two recent works (Lloyd and Niemi, 1979; 1060s. A number of studies have examined changes in occupational segregation between census years: 1900 1960 (Gross, 1968), 1950-1970 (Blau and Hendricks, 1979), 1960- See, for example, Cherlin and Walters (1981) and 1970 (Economic Report of the President, Mason et al. (1976).

11 12 ANDREA H. BELLER

U.S. Commission on Civil Rights, 1978). The t, and f t = the percentage 'of the female segregation indexes computed by there labor force employed in occupation i in year studies (as well as by Blau and Hendricks t. The index may take on a value between and by Fuchs) are presented in Appendix 0 and 100, where zero represents perfect B, Table B-1. But findings from other studies integration and 100 represents complete seem to conflict with these reports of no segregation. The number tells the propor- change. Beller (1982b) showed that EEO tion of women (or men) that would have to laws reduced occupational segregation by be redistributed among occupations for the 1974 z...nd that EEO laws combined with equal occupational distribution to reach complete educational opportunity legislation inequality between the sexes. creased the effects of years of college com- The index of segregation has two completed on women's entry into nontraditional ponents, labeled the mix effect and the com- occupations between 19'11 and 1977, espe- position effect by Blau and Hendricks (1979). cially among new entrants (Beller, 1982a). The value of the index depends on both the The reason these studies detected no change _relative size of various occupations and the is a lack of comparability between the two sex composition within occupations .2 Changes data sets they used to compute the segre- in the index thus cinriv. e from two sources: gation indexes. In fact, I have found that the changes in the occupational distribution and index of segregation declined from 68.32 in changes in the entry of the sexes into various 1972 to 64.65 in 1977 and 61.66 in 1981, a occupations. (It also depends on the inter- rate of decline almost three times as large action of the two.) These changes may be in as that during the 1960s. reinforcing or opposing directions. Signs of In the next section, the trends in occu- progress within occupations, for example, pational segregation from 1971 to 1981 are could be masked by unusual growth in oc- documented and compared with those of the cupations that are predominantly single sex. 1960s. Trends in segregation among all Oc- A standardization procedure can he used to cupations, among professional occupations, determine the influence of each of these two and among college majors are discussed. An effects. For example, to determine the ef- analysis of cohort differences in occupational fect of changes in the sex composition within segregation during the 1970s follows. I then occupations on the changes in the segre- compare and contrast changes in the sex gation index from year t 1 to year t, the composition of detailed occupations during index of segregation for year t can be com- the 1960s and the 1970s. Finally, race dif- puted standardizing the size of occupations ferences in trends in occupational segrega- to year t 1. Thus the employment stand- tion are presented. ardized index of segregation holds constant the distribution of employment across occupations (occupational mix) and enables one MEASUREMENT AND DATA Trends in occupational segregation are commonly measured by the index of segregation (Duncan and Duncan, 1955). The index is defined as follows: The value of the index may also depend on the degree of aggregation of the occupations. Typically, the greater the degree of aggregation, i.e.. the 'fewer the St ='12Ymu-fit occupations, the lower the level of measured segregation. For this reason, in comparing indexes over time, one should use the same number of occupations at the where m =,the percentage of the male same degree of aggregation. This methodological issue labor force employed in occupation i in year is discussed in England (1981).

20 TRENDS IN OCCUPATIONAL SEGREGATION BY SEX AND RACE 13 to observe the effects of changes in the sex ulated by the BLS from the monthly CPS.4 composition within occupations alone.3 More detail on these sources and on issues Similarly, to observe the effects of changes of comparability and the choice of occupa- in the occupational structure alone, the pro- tions included in the sample are discussed portion standardized index of segregation can in Appendix A. As discussed there, the CPS be computeri holding constant the sex com- occupational datacollected during the 1970s positidn at year t 1 and using the em- are not comparable to the 1970 census data ployment distribution of year t. (Standard- even though the same occupation codes are izing by the size of occupations in year t used, because the Census Bureau changed 1 arbitrarily assigns the interaction term in its method of assigning individuals to oc- one direction. Standardizing by the size of cupations in December 1971. Hence, sta- occupations 1:, year t would assign the in- tistics based on these two sources should not teraction tern in the other direction.) These be compared. Although their reliability dif- standardization procedures can also be ap- fers (see Appendix A), I make some com- plied to a given year to determine how the parisons between the two different sources index of subgroup j differs from that of the of CPS data in 'order to include 1981 data population as a whole. This allows us to de- in the analysis. The 1960 and 1970 census compose the index of the subgroup into the occupational data are used to show trends effects of occupational mix and sex compo- during the 1960s; these data were made sition. For example, by standardizing the comparable by the Census Eureau's recod- segregation index of the youngest cohort to ing of the 1960 data according to the 1970 the occupational mix of the whole labor force, occupation codes (U.S. Bureau of the Cen- it can be seen how the sex composition within sus, 1972). occupations for the youngest cohort differs from that .3f the rest of the labor force. TRENDS IN OCCUPATIONAL To assess trends in occupational segre- SEGREGATION, 1960-1981 gation during the 1970s, I used data from the Current Population Survey (CPS) con- Occupational segregation of the sexes de- ducted monthly by the U.S. Bureau of the clined continuously during the 1970s at a Census. Data for the years 1971-1974 and 1977 used for the detailed analyses presented in this paper come from the March Data on occupations for 1971-1974 and 1977 come from the 1972-1975 and 1978 Annual Demographic Files, Annual Demographic Files (ADF) of the CPS. which are available on public-use computer tapes. These The ADF data are supplemented here with files contain considerable demographic detail, making more recent data from the Bureau of Labor it possible to cross-classify occupation by such char- Statistics' (BLS) annual averages (AA), tab- acteristics as labor market experience, which is done later in this paper. These were the only years for which I had these data at the time of this writing. To incorporate more recent data than 1977, I obtained from The employment standardized index of segregation the BLS unpublished tabulations of annual averages is defined as follows: (AA) for 1981; to ascertain comparability between the S = 'IRE AA and the ADF data, I also obtained these tabulations for 1977 and for 1972, the earliest year for which th"y where In: = (MIT) (T111) (100)/E (MulT)(L-1). are available. (The cooperation of Elizabeth Waldman and Jack Bregger of the BLS, who made these data = (Fa) (T _,)(100)/E (FIT) (T(._1), F = the available expeditiously, is gratefully acknowledged.) number of females in occupation i in year 1. M = the These data are not cross-classified by demographic number of males in occupation i in year t, and T = characteristics. Thus, while overall trends can be as- F + M = total employment in occupation i in year sessed through 1981, cohort trends can be assessed t. only through 1977.

21 ANDREA H. BELLER 14 rate that exceeded thedecline during the trends in one data set should he comparable long as compara- 1960s. The index of segregationdeclined from to trends in the other as 68.32 in 1972 to 61.66 in 1981,according to bility has been established within each data the BLS AA data, According tothe ADF, it set. Between 1972 and 1981the index of declined from 68.14 in 1971 to 64.15 in1977, segregation declined at an averageannual high as during the and the decline occurredcontinuously over rate nearly three times as the intervening years. Theseindexes, which 1960s, i.e., - 0.74 compared with - 0.28. are computed over a commongroup of 262 These figures appear in the bottom part of three-digit census occupations, appear in Table 2-1. The annual rate of decline in the Table 2-1, lines 2 and 3. For comparison segregation index appears to haveacceler- have purposes, indexesof segregation for 1960 ated slightly in the mid-1970s and to and 1970 computed from the decennial cen- remained steady through 1981. sus over the same262 occupations are in- To decompose the change in the segre- cluded. According to Census Bureaudata, gation index during the 1970s, Ihave stand- the index declined from 68.69 in 1960 to ardized it to the employment mix atthe 65.90 in 1970. beginning of the decade. As mentioned As pointed out earlier, theindex levels above, we want to make comparisons only are notcomparable across data sets, but within a data set, so we standardize the Cen-

TABLE 2-1Segregation Indexes, 1960-1981,All Occupations Census CPS 1977 1981 1960 1970 1971 1972 1973 1974 Index Level Unstandardized Census 68.69 65.90 Annual Demographic 67.09 66.39 64.15 File (ADF) 68.14 67.36 64.65 61.66 Annual Averages (AA) 68.32 Employment Standardized Census(1970) 68.06 65.90 67.36 66.97 66.64 64.49 ADF (1972) 67.99 68.32 65.18 62.88 AA (1972) Change in Index Total Average Annual Rate of Change Change 1972-1981 1972-1981 tinstandardized 1960-1970 1971-1974 1974-1977 1977-1981 Census -0.28 -0.58 -0.75 -0.75 -0.74 -6.66 AA Employment Standardized Census (1970) - 0.22 0.45 -0.72 ADF(1972) -0.56 -0.60 -5.44 AA (1972) NOTE: These indexes are based on a common groupof 262 three-digit census occupations. 1970, Detailed SOURCE: U.S. CommerceDepartment, Bureau ofthe Census, U.S. Census of Population Printing Office, Characteristics. Final Report PC(1)-D1, U.S.Summary(Washington, D.C.: U.S. Government and 1978,computer tapes; 1973), Table 221; Annual DemographicFiles of Current Population Survey, 1972-1975 Surveys, 1972,1977, and 1981, and Bureau of Labor Statistics,annual averages of monthly Current Population unpublished tabulations. TRENDS IN OCCUPATIONAL SEGREGATION BY SEX AND RACE 15 sus Bureau data to 1910 andthe ADF and from 46.08 in 1969 to 35.62 in 1978. The AA data to 1972. Employment standardized average annual rate of decline in thisindex indexes reveal the amount of the change in is 1.16 per year.5 Thus, segregation by field segregation that is due to changes in sex of study among bachelor's degree recipients composition within occupations of afixed declined rapidly during the 1970s, followed size, the -composition effects. As shown in by the professional occupations, and, finally, Table 2-1, the employment standardized the work force as a whole. segregation index declined from 68.32 in 1972 to 62.88 in 1981, or by almost as much as COHORT DIFFERENCES IN the unstandardized index declined. Thus, OCCUPATIONAL SEGREGATION, index most of the decline in the segregation 1971-IR77 during the 1970s was due to changes in the sex composition within occupations,but the Was the decline in occupational segre- employment distribution also shifted slightly gation by sex during the 1970s distributed toward a less segregated work force. throughout the labor force or concentrated Professional occupations continued dur- in groups most able to take advantage of ing the 1970s to be less segregatedthan the improved access to nontraditional jobs and work force as a whole and to experience a opportunities for advancement? Beller (1982a) somewhat larger decline in segregation. The found the effect of equal employment op- segregation index for 59 professional occu- portunity legislation between 1971 and 1977 pations declined from 59.44 in 1972 to 50.55 to be largest among college-educated new in 1981, according to the annual averages and recent entrants into the labor market. data. This yields an average annual rate of That is, compared to 1971, the chances of decline of nearly 1 percentage point, 0.99. women with 1-10 years ofpotential labor Since these occupations are composed pri- market experience (new entrants) in 1977 or marily of individuals with a college degree, with 7-16 years (recent entrants) finding emit is instructive to examine an index of seg- ployment in a nontraditional occupation in- regation for earned bachelor's degrees con- creased more than for older cohorts in 1977. ferred on men and women by field of study. Recent entrants in 1977, who were new en- These data, published annually by the Na- trants in 1971, found' ncreased opportuni- tional Center for Education Statistics (NCES), ties to move into nontraditional occupations are based on thedistribution of all degrees as they aged over this period. granted by all accredited degree-granting To examine changes in occupational seg- institutions in the United States during a regation by cohort, I stratified the labor force specific academic year.5 The segregation in- by potential work experience, defined as dex computed over college majors declined AgeEducation6, as in this previous study. Using the ADF data for 1971 and 1977, I stratified women and men into groups 5 These data are taken from National Center for Ed- with the following years of experience (EX- ucation Statistics, Projections of Education Statistics to PER): 1-10, 7-16, and 11-40 + .I believe /988-89, by Martin M. Frankel and Debra E. Gerald, that new and recent labor market entrants April 1980. The original sources for the major portion of these data are the annual NCES reports on Earned are best able to benefitfrom improved op- Degrees Conferred. Further information came from portunities, and I hypothesize that young

". . education and professional associations, experts in other academic areas, and other agencies in thefed- eral government .." (as cited in NCES, April 1980, p. 49). The numbers in this reportdiffer slightly from 6 The index of segregation did not decline for post- the ones for the same year published in Earned Degrees graduate degrees, however, it is at a lower level than Conferred. for bachelor's degrees (Beller and Han, in press). 16 ANDREA H. BELLER cohorts will show greater changes thanolder TABLE 2-2Segregation Indexes by cohorts. If access to nontraditional occupa- Experience Cohort tions increases, new entrants willhave more Change opportunities to enter the occupational Experience Cohort 1971 1977 1911-1977 structure at preferred pointsthan older co- Unstandardized horts with the same education. Sincead- 1-10 67.47 62.51 -4.96 justments in education can only occurwith 7.16 69.94 64.03 - 5.91 66.31 -3.05 some iag, new entrantsalso have the great- 11-40+ 69.36 education Standardized to Employment est opportunities to acquire more of Whole Labor Force and to alter their field of study in response 1-10 67.44 61.96 - 5.48 to perceptions ofimproved opportunities in 7-16 69.07 64.89 - 4.18 the labor market. In general, theeduca- 11-40+ 69.13 66.65 -2.48 tional attainment of younger cohorts of Standardized to Proportions older cohorts, and of Whole Labor Force women is higher than of 1-10 67.78 64.35 3.43 women are increasinglylikely to obtain ad- 7-16 68.54 63.93 4.61 ditional degrees at all degree levels (Beller 11-40+ 67.84 63.69 - 4.15 and Han, 1984). Recent entrants in theearly NOTE: The indexes are computed oil the basisof stages of their careers cantake advantage of 258 occupations. Occupations with no employment in new opportunitiesfor advancement.' Thus, any experience cohort weredropped from all groups. SOURCE: Current Population Survey, Annual De- I compared new entrants intothe labor mar- mographic Files, computer tapes. ket (EXPER = 1-10) with the restof the labor force (EXPER = 11-40+), new entrants in 1971 with new entrants in1977, and new entrants in 1971 (EXPER =1-10) groups over this period.The differential be- with themselves 6 years later in 1977(EX- tween the youngest cohort andthe rest of PER = 7-16). We can see howsegregated the labor force widened over time, how- the youngest cohort is compared withthe ever, as segregationdecreased more rapidly rest of the labor force;how segregated the in the youngest cohort ashypothesized. The enter ng cohort is at thebeginning com- segregation index for this group (EXPER = pared with the mid-1970s, and howmuch 1-10) declined from 67.47 in 1971 to 62.51 change in occupational segregation the1971 in 1977, or by 0.83 percentage points per entering cohort experienced as itaged. year, while the indexfor the remainder of The segregation indexes for these expe- the labor force (EXPER = 11-40+)de- rience cohorts appear in Table2-2. The clined from 69.36 to 66.31, or by 0.51 points youngest cohort is lesssegregated than the per year. Duringthis period the decline is remainder of the labor force in both 1971 greatest not for the youngestcohort but for and 1977, and segregation declinedfor all the group with 7-16 years of potential work experience. For this group (EXPER =7- 16) the segregation index declined from 69.94 in 1971 to 64.03 in 1977, orby 0.99 points 7 While these arguments and data strongly suggest per year. If wefollow the entering cohort that the results should be stronger for yopngcohorts, 7- a potential bias in ourresults ev.ists in that the sex in 1971 for 6 years to 1977 (EXPER = difference in actual experience probably widens with 16), we find that the segregation indexde- potential experience. Thus, stronger resultsfor younger clined within this cohort by 3.44 percentage cohorts might be related to the actual versuspotential points from 67.47 to 64.03, or by 0.57 points experience issue. In the absence of comparablecohort cohort in 1971 be- the effect per year. .The entering data prior to 1971, it wasn't possible to assess aged through 1977 of such a potential bias. came less segregated as it

2 4 TRENDS IN OCCUPATIONAL SEGREGATIONBY SEX AND RACE 17' , when the entering cohort is less segregated trants) are in different occupations than their than in 1971. This implies that each entering older couiterparts. cohort is less segregated than in the past and experiences a greater decline in seg- MAJOR COMPONENTS OF CHANGE IN regation as it ages. THE OVERALL INDEX, 1972-1981 To decompose these cohort changes in occupational segregation, I standardized each The specific detailed occupations contrib- subgroup to the whole labor force in each uting the largest amounts to the decline in year. To determine how segregated this group the index of segregation between 1972 and would be if it had the same occupational mix 1981 based on the AA data are the following: as the labor force as a whole but maintained accountants; elementary school teachers; its owkifex composition within occupations, bank officers and tinancial managers; sales I standardized the segregation index to the clerks, retail trade; secretaries, not else- occupational mix of the whole (employment whe.classified (n.e.c.); telephone opera- standardized). To determine how segre- tors; typists; sewers and stitchers; delivery gated this group would be if it had the same and route workers; janitors and sextons; sex composition within occupationslia the cooks, except private household; child care whole but its own occupational mix,' I stand- workers, private household; and maids and ardized to the sex composition within oc- servants, private household. cupations of the whole (proportion stand- Any differencein contribution can be due ardized). to either a change in the size of a segregated As it turns out, the occupational mix is occupation or a change in the sex compo- quite similar across cohorts; the employ- sition within an occupation. Two of the tra- ment standardized indexes are nearly iden- ditionally female occupations, private tical to 'the unstandardized indexes, What household maids and servants and sewers this implies is that, while the occupational and stitchers, showed a large decrease in distribution does not differ between older size over the period. Each of these declines and younger generations as a whole, the sex took over 1 percentage point off of the seg- composition within occupations differs sub- regation index in 1981. Other traditionally stantially'between recent and older cohorts. female occupations that decreased in size Thus, for example, while approximately the are telephone operators -and private house- same proportion of the youngest and the hold child care workers. One traditionally older cohorts are accountants, a higher pro- male occupation, delivery and route work- portion of youthful than of older accountants ers, also decreased in size over the period. are women. The proportion standardized in- Although the numbers of secretaries, nee. c., dexes for 1977 show th. t the youngest co- and elementary school teachers increased hort (EXPER = 1-10) would be much more between 1972 and 1981, the proportion of segregated if it had the same sex composi- the female labor force that crowded into these tion within occupations as the labor force as traditionally female occupations declined a whole (64.35 compared to the actual 62.51); from 9.2 to 8.7 percent and from 3.6 to 2.8 symmetrically, the remainder of the labor percent, respectively. A smaller proportion force would be less segregated if it had the of the female labor force crowded into the sex composition of the whole (63.69 com- constant-sized female occupations of retail pared to the actual 66.31). Although as a sales clerk and typist in 1981 than in 1972, group new entrants have the same occu- 4.2 percent as opposed to 5.4 percent, and pational distribution as everyone else, new 2.4 percent as opposed to 3.3 percent, re- female entrants (as well as new male en- spectively. But the female share #4 these

25 18 ANDREA H. BELLER occupations did not decline. Womenalso many different nontraditionalplaces in the entered three rapidly growing traditionally labor force, changes which bear a closer look. male occupations: accountants, bank officers and financial managers, and janitors and sex- CHANGES IN THE SEX COMPOSITION tons. The female share in these occupations OF OCCUPATIONS, 1960-1977 increased from 21.7 to 38.5 percent among accountants, from 18.7 to 37.4 amongbank This section examines changes in the sex officers, and from 10.5 to 19.0 among jani- composition of size-standardized occupa- tors. Cooks is a rapidly growing occupation tions, assuming that all are of equal size. It that men are entering in greater numbers also summarizes material presented in greater than previously; the percentage of females detail in Beller (1981). Occupations are cat- in this occupation declined from 62.4 to 52.3 egorized according to their sex label and percent over this period. broad occupational group. Changes in the The segregation index declined despite sex composition ofdetailed occupations dur- the fact that some occupations contributed ing the 1960s are contrasted withchanges more to segregation in 1981than in 1972'. between 1971 and 1977. These analyses are The occupations that contributed more to based on data for 262 occupations from the the index are primarily rapidly growing fe- decennial censuses and the ADF and focus male occupations. The largest increases came exclusively on changes in the sex composi- from registered nurses; office managers, tion within occupations, a variableamenable n.e.c.; hank tellers; computerand periph- to alteration through public policy. eral equipment operators; and miscella- Each detailed occupation is assigned a sex neous clericalworkers. Nurses and bank label defined by deviations in its sex com- tellers are both rapidly growing predomi- position of ± .05 from that of the labor force nantly female occupations. The field of mis- as a whole.According to this definition, oc- cellaneous clerical workers is both growing cupations are categorized as male if in 1960 and becoming increasingly female as are the the percentage of males equaled or ex- fields of office managers and computer and ceeded .722; in 1970, .669; in 1971, .668; peripheral equipment operators. and in 1977, .640. Table 2-3 shows the num- Changes in segregation during the 1970s ber and percentage of occupations that are may be summarized asfollows. While women male, female, and integrated in each year.' continued to enter some of the traditionally Although a majority of occupations continue female occupatior s in large numbers, such to be male dominated, the percentagede- as registered nurses,they decreased their clined during the 1970s, though it had in- rate of entry into others,such as secretaries. creased during the 1960s; a number of oc- While many nontraditional occupations be- cupations changed from male to integrated, came slightly lessmale dominated, large de- while the percentage that was female re- clines in segregation occurred in only a few, mained unchanged. e.g., accountants.Also contributing to a de- A comparison of changes in women'sshare cline in segregation were the dramaticde- of employment by occupation from 1971 to clines in the size of the traditionally female occupations of sewers and stitchersand telephone operators, presumably the first due 8 Although the choice of the Yalu- ± .05js somewhat tdva declining industry and the second to arbitrary, it has little effect on substantive conclusions rapid mechanization, eliminating the need in this paper. It simply affects how wide a segmentof for as many telephone operators. These the occupational distribution we choose to call inte- changes suggest that women are working in grated. of.

TRENDS IN 06CUPATIONALJECREGATION BY SEX AND RACE 19

TABLE 2-3Sex Label of Detailed Occupations Census CPS Sex Label° 1960 1970 1971 1977 Male 159 165 157 140 Integrated 17 19 15 32

Female 86 78 . 90 90 Total 262 262 262 262 Percentage

Male 60.7 63.0 59.9 , 53.4 Integrated 6.5 7.3 5.7 12.2 Female 32.8 29.8 34.4 34.4 Total 100.0 100.0 100.0 100.0 ° Defined relativeto the sex composition of the labor force in the year given. SOURCE: U.S. Bureau of the census, U.S. Census ofPopulation:1960. Final Report PC(2)-7A. Subject Reports. Occupational Characteristics. (Washington, D.C.: U.S. Government Printing Office,1963), Table 1; U.S.Bureau of the Census,U .S . Census of Population: 1970. Final Report PC(2)-7A.Subject Reports. Occupational Charac- teristics. (Washington, D.C.: U.S. Government Printing Office, 1973), Table 1; 1972 and 1978 Current Population Survey. Annual Demographic Files, computer tapes.

1977 with those from 1960 to 1970 reveals 1977 it became 0.6 percent relatively less the following Women's share of employ- male. Thus, while women had become more ment increased absolutely in a majority of occupationally concentrated during the 1960s, occupations iri both periods: in 77 percent they began entering nontraditional occu- of occupations between 1960 and 1970 and pations at a greater rate than the labor force in 71 percent between 1971 and 1977. But as a whole during the 1970s. Male occupa- women's share of employment relative to tions also became relatively more male on their share of the labor force increased in average during the 1960s in every broad oc- many more occupations during the 1970s cupatiOnal category with the exception of than during the 1960s: in 45 percent of oc; clerica! In the 1970s, with the exception of upations between 1971 and 1977 as com- crafts and operativein every broad occu- pared to 26 percent between 1960 and 1970. pational category m e occupations became Women's relative share in male occupations relatively less male.or example, the av- also increased much more widely during the erage male managerial occupation, which 1970s: in 46 percent of male occupations became 2.5 percent more male during the compared to 25 percent during the 1960s. 1960s, became 4.9 percent less male be- These changes were most pronounced among tween 1971 and 1977. the white-collar occupations, especially In summary, the contrast in changes in the professional and managerial, and little or no sex composition of occupations between the change occurred among the blue-collar oc- 1960s and 1970s shows that a new pattern of cupations. Women's share continued to grow female entry has emerged. Rather than con- both akolutely and relatively in the already tinue to crowd into a limited subset of occu- predominantly female clerical occupations. pations, v mien are entering a wide variety I have also examined the magnitude of of nontraditional occupations. These changes change in the sex composition of the average are most prominent at the white-collar level, occupation. Between 1960 and 1970 the av- especially among professional and managerial erage occupation became 2.8 percent rela- occupations. But little such change appeared tively more male, while between 1971 and fir the blue-collar occupations.

417 20 ANDREA H. _RULER

TXpLE 2-4Sex Segregation Indexes for All Occupations and Professional Occupations by friace All Occupations Professional qccupations 1972 1977 1981 1972 1977 1981 Index Level Unstandardized White 68.39 64.96 62.08 60.05 54.68 50.75 Nonwhite 68.00 63.29 59.39 51.58 49.95 48.88 Employment standardized to 1972 White 68.39 65.35 63.07 60.0.5 55.20 50.89 Nonwhite 68.00 65.43 63.52 51.58 48.58 50.59 Change in Index, 1972-1981

Unstandardized Annual Average Total Annual Average Total .. White -0.70 -6.31 -1.03 -9.30 Nonwhite -0.96 -8.61 -0.30 -2.70 Employment Standardized to 1972 White - 0.59 - 5.32 -1.02 -9.16 Nonwhite - 0.50 -4.48 -0.11 -0.99 NOTE: These indexes are based on a common group of 262 three-digit census occupations of which 59 are professional, occupations. SOURCE: U.S. Bureau of Labor Statistics, annual averages of monthly Current Population Survey, 1972, 1977, and 1981, unpublished tabulati, ins,

ition within occupations, the indexes were TRENDS IN THE SEX SEGREGATION OF standardized to the occupational distribu- OCCUPATIONS BY RACE, 1972-1981 tion of employment in the initial year, 1972. Occupational segregation by sex declined By c..ntrast to the unstandardized indexes, continuously for both whites and nonwhites the standardized indexes, shown in Table between 1972 and 1981. While the index of 2-1, declined slightly more for whites than segregation was approximately the same for for nonwhites, from 68.39 to 63.07 for whites both races in 1972-68.38 for whites and and from 68.00 to 63.52 for nonwhites. Thus, 68.00 for nonwhites -it declined relatively the greater decline for nonwhites than whites faster for nonwhites during the 1970s. For in the unstandardized index can be attrib-, whites it declined to 62.08 in 1981, while uted to shifts in the nonwhite occupational for nonwhites it declined to 59.39. These distribution from heavily single -sex occu- indexes appear in Table 2-4. The figures in pations toward less segregated ones. As a the bottom part of this table show the av- matter of fact, nearly one-half of the decline erage annual rate of decline in the index of in sex segregation among nonwhites was due. segregation between 1972 and 1981, 0.96 to such changes in their occupational dis- for nonwhites compared to 0.70 for whites.° tribution (4.13) toward less segregated oc- To identify the portion of the overall change cupations as compared with changes in their attributable to changes in the sex cornpo- sex cotnposition within size-standardized occupations (4.48). I conclude that changes in the sex composition within occupations was about the same for both races over the a Although in this sample many occupations contain only a few nonwhites, the level of the segregation in- decade, but the occupational distribution of dexes changes very little when occupations with fewer nonwhites also shifted toward less sex seg- than 10 nonwhites are excluded from the computations. regation.

28 TRENDS IN OCCUPATIONAL SEGREGATION BY SEX AND RACE 21

The picture for professional occupations rates .of professional employment of 90 per- contrasts dramatically. On the one hand, cent for nonwhite women,,69 percent for while the levels of the segregation indexes nonwhite men, 54 percent for white women, are lower for both racesfor professional than and 26 percent for white men. Neverthe- for all occupations 1972, these indexes less, profession:' occupations still Com- are well lower fornonwhites. As shOwn in prised a smaller proportion of nonwhite than the second panel of Table 2-4, the 1972 seg- white employment in 1981. regation index was 60.05 for white profes- If sex segregation declined as much for sionals and 51.58 for nonwhite professionals nonwhites as for whites over all occupations in contrast to 68.39 and 68.00, respectively, ut not among professional occupations, then for all occupations. On the other hand, non- ost change among nonwhites must have whites experienced little decline in sex seg- occurred at other levels of the occupational regation among professional occupations distribution. during the 1970s, while whites experienced larger declines than among all occupations. MAJOR COMPONENTS OF CHANGE IN By the end of the decade, white profession- THE OVERALL INDEX BY RACE als had become slightly less sex segregated., than nonwhite professionals had been at the The occupations contributing the most to beginning of the decade:---an index value of decreasing the segregation index among 50.75 compared to 51.58---whild' nonwhites nonwhites between 1972 and 1981 differ had become somewhat less segregatedan considerably from the ones for whites, al- index value of 48.88. The annual average though some similarities exist. The occu- rate of decline in the segregation indexes for pations that contributed the most to de- professional occupations over the decade wa.s clines in the index of sex segregation among 1.03 for whites and 0.30 for nonwhites, in nonwhites only are primarily laborer and contrast to 0.70 and 0.96, respectively, for service worker occupations. Frequently, they all occupltions. According to the employ- are typically male occupationsthat declined ment stadardized indexes, this entirede- in size. over this period and in which.non- cline for whites resulted from changes in whites are represented disproportionately. their sex composition within occupations The larger decline for nonwhites than for (1.02), while for nonwhites most of this de- whites in the index of segregation over all cline came from shifts in their occupational occupations can be traced to this source. distribution toward.less (nonwhite) sex-seg- The specific detailed occupations taking over regate& professional occupations (0. I 1 ). one-half a percentage point off the segre- The question naturally arises as to what gation index for nonwhites only between 1972 proportion of each racial group constitutes and 1981 are the following: storekeepers and the professional occupations. In 1972, 14 stock clerks; clothing ironers and pressers; percent of white men and 15 percent of white construction laborers, except carpenters' women were inprofessional occupations, helpers; freight and material handlers; gar- while only 8 percent of nonwhite men and deners and groundskeepers, 'txcept farm; 11 percent of nonwhite women were. By miscellaneous laborers; unspecified labor- 1981 the proportions had risen for all groups, ers; farm laborers, wageworkers; chamber- with the largest increase occurring for non- maids and maids, except private household; white women to 15 percent, the next largest cleaners and charwomen; nursing aids, or- increase for nonwhite men to 11 percent, derlies, and attendants; and practical nurses. and identical increases forwhites of both Among these, nearly all laborer occupations sexes to 16 percentfor men and 17 percent declined in size, while nearly all service for women. These percentages reflect growth worker occupations increased in size. Non-

29 22 ANDREA H. BELLER white women increased their share of store- among whites also declined among the fol- keeper jobs from 20 to 41 perc'ent between lowing white-collar occupations: account- 1972 and 1981, ants; elementary school teachers; bank of- The segregation index declined for both ficers and financial managers; sales clerks, races because of the dramaticdecline in the retail trade; bookkeepers; secretaries, n.e.c.;. size of the private household maids and and typists. Two of these occupations servants occupation, but the decline was bookkeepers and secretaries, n.e.c.ac- much greater for nonwhites. The decline in tually became more segregated among non- the number of nonwhite females in this oc- whites because women but not men entered cupation took an exceptional 8.04 percent- these fields. age points off the segregation index for non- The occupations that contributed more in whites. 1981 than in 1972 to the segregation index The specific detailed occupations contrib- for whites are identical to those for the pop- uting the largest amounts to the decline in ulation as a whole reported earlier, Occu- the index of segregation among whites be- pations that became more segregated among Iween 1972 and 1981 are mostly the same nonwhites only include bookkeepers; sec- as for the whole population presented ear- retaries, n.e.c.; investigators and esti- lier. Nevertheless, some occupations con- mators, n.e.c.; statistical clerks; electricians; tributed to the decline in the index for whites and assemblers. These weft all sex-segre- only: bookkeepers, garage workers and gas gated occupations in which the numbers of station attendants, waiters and waitresses, nonwhites employed grew. The typically fe- and hairdressers and cosmetologists. Al- male occupations among thealso became though the number of whites employed as increasingly female: While the number of bookkeepers increased between 1972 and electricians grew rapidly, the female share 1981, the proportion of,the white female of nonwhite employment increased from 0 labor force that crowded into this tradition- in 1972 to 3.9 percent in 1981. Contributing ally female occupation declined from 5.2 to toward increasing the segregation index by 4.6 percent. While men entered the two comparable amounts for both ra-i-is are reg- expanding traditionally female occupa- istered nurses, hank tellers, computer and tionswaiters and waitresses, and hair- peripheral equipment operators, and mis- dressers and cosmetologistsat an increas- cellaneous clerical workers. ing rate over this period, the male share of In summary, much of the decline in oc- white employment increased from 7.2 per- cupational segregation by sex during the cent to 9.2 percent in the former and front 1970s occurred for both races; however, ma- 9.2 to 11.1 percent in the latter. The tra- jor differences exist. The major exodus of ditionally male occupation of garage workers nonwhite females from the occupation of and gas station attendants showed a decline private household maids and servants and in size over the period. the decline in size of a number of laborer Comparable declines- in segregation for occupations in which nonwhite males dom- whites and nonwhites occurred for the fol- inated shifted the nonwhite occupational lowing occupations: the clerical occupation, distribution toward a greater reduction in telephone operators; the operative occupa- occupational sex segregatibn during the 1970s tion, sewers and stitehers; the laborer oc- than fir whites. On the other hand, white cupation, delivery end rout,. corkers; and women reduced their rate of entry into a the three service worker occupations, jani- number of traditionally female white-collar tors and sextons, cooks (except private occupations that nonwhite women contin- household), and child care workers (private ued to enter, and white women increased household). By contrast, sex segregation their entry into a number of traditionally

30 TRENDS IN OCCUPATIONAL SEGREGATION BY SEX AND RACE 23 male white-collar occupations more than than the decennial census must be selected. nonwhite women did. The monthly Current Population Survey (CPS) collects detailed three-digit census occupation data from a random cluster sam- CONCLUSION ple of (initially around 50,000) around 60,000 Occupational segregation of the sexes di households (1/1500)- designed to represent minished significantly during the 1970s, as the civilian noninstitutional population of the measured by the index of segregation. Most United States. Two sources provide detailed Of the decline was due to changes in the sex occupational data from this survey. The first composition of traditionally male occupa- is the Bureau of the Census's March Annual tions, particularly at the professional and Demographic Files (ADF), available on managerial levels. Declines in segregation public-use tapes since 1968. The question among new and recent job market entrants on "longest job held last year" should pro- were greater than for the rest of thelabor vide reasonably reliable estimates of the force. While nonwhites experienced a greater Previous year's occupat'onal distribution. The decline in occupational sex segregation than second is the monthly statistics compiled by whites over the decade, about the same the Census Bureau for the Bureau of Labor amount was due to changes in the sex com- Statistics (BLS) from which the latter composition of traditionally male occupations. pute, and since 1974 publish in Employment The marked declines in sex segregation in and Earnings, the annual averages (AA). AA professional 'occupations apparent among data were used in the two studies referred whites did not hold for nonwhites, but non- to in the text (Lloyd and Niemi and U.S. white professionals were much less segre- Commission on Civil Rights), which com- gated than white professionals at the start puted segregation indexes for more recent of the decade. Continued declines in oc- data than the 1970 census. Unpublished tab- cupational segregation by sex depend on the ulations of AA, which include data for smaller apparent momentum for change continuing occupations than the published data (50,000 into the next decade. incumbents), are available directly from the BLS for 1972 on. In comparing the CPS with the decennial ACKNOWLEDGMENTS census, the primary disadvantage of the CPS This research was supported in part by is its smaller sample size. The AA data are funds from the University of Illinoisat somewhat more reliable on these grounds Urbana-Champaign, Research Board. Help- than the ADVata. For total labor force data ful comments on an earlier draft were pro- other than a'cultural employment and un- vided by Marianne Ferber, Victor Fuchs, employment, the sampling error of the an- Barbara Reskin, and several reviewers. Ex- nual averages is 0.67 times the sampling er- cellent research assistance was provided by ror of the monthly data(Employment and Kee-ok Kim Han. Computational assistance Earnings, May 1982, Table J). To improve by John Boyd and Alex Kwok is gratefully reliability, the smallest occupations should acknowledged. be excluded. For purposes of this paper, all tabulations excluded occupations with fewer than 25 survey respondents in either the APPENDIX A 1975 or. the 1978 ADF data set (representing occupations with fewer than approkimately Data So(rces 40,000 incumbents). Out of the 441 detailed To assess trends in occupational segre- three-digit 1970 census occupations, this left gation for intercensual years, a data set other 267 in 1974 and 280 in 1977, 262 of which

31 24 ANDREA BELLER are common to both years.All tabulations December 1971 a question eliciting infor- in this paper include only those 262 occu- mation on major activities or duties was added pations, or fewer where noted. to ,the monthly CPS in order to determine more precisely the occupationalclassifica- tion of individuals. According to the BLS Comparability of Data (Employment and Earnings, January 1979, p. 207), "this changeresulted in several dra- In attempting to assess ':rends in occu- matic occupational shifts, particularly from pational data over time, two comparability managers and administrators to other groups. problems arise, depending on the period of Thus, meaningful comparisons of occupa- interest. The first is changes in the Census tional levels cannot always be made for 1972 Bureau's occupation codes with each decen- and subsequent years with earlier periods." nial census. A variety of techniques for deal- For this reason, the 1970 census data are ing with this problem are discussed in Eng- not comparable with the CPS data after 1971. land (1981). The 1960 census data were The two studies that found no change in recoded according to the 1970 census codes segregation in the 1970s relied on such a by John A. Priebe (U.S. Bureau of the Cen- comparison. The earliest comparable data sui, 1972). These data are published in U.S. would be from the 1972 ADF on the longest Bureau of the Census (1973, Table 221); the job held last year, 1971. For these reasons, 1960 and 1970 census data in this paper, as presented here are census data for 1960 and well as in Blau and Hendricks (1979), come 1970 comparisons; data from the ADF for from this source. The 1980 census used a 1971-1974 and 1977, the years for which we substantially revised set of occupation codes, have the data tapes; and the unpublished and thus its occupation data will not be com- annual averages data for 1972, 1977, and parable to earlier census occupation data un- 1981 for more recent data. It was found that less the Census Bureau double codes them the computations based on the ADF and the with the 1970 and 1980 occupation codes. AA data sets are quite similar, although in- Another comparability problem arose in dividual occupations can differ. To include that the segregation indexes computed us- data as current as 1981, comparisons across ing the CPS data were inconsistentwith the these two data sources were sometimes made, one computed using the1970 census data although their reliability differs. Additional despite the fact that both used 1970 census comparability problems will arise when the codes. In an attempt to find out why, the CPS converts to the new 1980 occupation following information was discovered. In codes, beginning with the 1982 data.

APPENDIX B

TABLE B-1Indexes of Segregation From Other Studies Census CPS' Projected 1960 1970 1976 1977 1985

AU Occupations Blau and Hendricks (1979) (N = 280) 68.33 65.77 60.10 Lloyd and Niemi (1979) (N = 236) 64.5 64.3 U.S. Commission on Civil Rights (1978) (N = 441) 65.8 66.1 Professional Occupations Fuchs (1975) (N = 33) Unstandardized 66.2 59.2 Standardized to 1960 66.2 62.7 ° Computed from the BLS's AA data.

32 TRENDS IN OCCUPATIONAL SEGREGATION BY SEX AND RACE 25

TABLE B-2Percentage of Occupations With Changes in Sex Label by Initial Sex Label and Period

1960-1970 1971-1977 1971-1974 1974-1977 All 9.5 11.1 11.5 11.1 Male 3.1 10.8 8.9 7.5 Integrated 53.0 33.3 53.3 37.5 Female 12.8 7.8 8.9 9.9 SOURCE: Same as for Table 2.3.

REFERENCES Becker, Gary S. 1982 "The Failure of Human Capital Theory to Ex- 1981 ATreatise on theFamily. Cambridge, Mass.: plain Occupational Sex Segregation."The Harvard University Press. Journal ofHumanResources17 (Summer): Beller, Andrea H. 358-70. 1981 "Changes in the Sex Composition of U.S. Oc- Fuchs, Victor R. cupations, 1960-77."-Unpublished paper. 1975 "A Note on Sex Segregation in Professional 1982a"The Impact of Equal Opportunity Policy on Occupations." Explorations inEconomic Re- Sex Differentials in Earnings and Occupa- search2 (Win ter): 105-11. tions."The American Economic Review, Pa- Gross, Edward pers and Proceedings(May):171-75. 1968 "Plus ca change...? The Sexual Structure of 1982b"Occupational Segregation by Sex: Determi- Occupations Over Time."Social Problems16 nants and Changes."The Journal of Human (Fall):198-208. Resources17 (Summer):371-92. Lloyd, Cynthia, and Beth Niemi Beller, Andrea H., and Kee-ok Kim Han 1979 ThesEconomicsof Sex Differentials. New York: 1984 "Trends in Major Fields of Study Among Columbia University Press. Women in Higher Education." In Kathryn Mason, Karen D., et al. Rettig and Mohamed Abdel-Ghany, eds.,Eco- 1976 "Changes in U.S. Women's Sex-Role Atti- nomic Decisions of Families: Security for the Elderly and Labor Force Participation of tudes, 1964-1974." AmericanSociological Re- view41 (August):573-96. Women.Washington, D.C.: American Home Economics Association. Polachek, Solomon W. Blau, Francine D., and Wallace E. Hendricks 1979 "Occupational Segregation Among Women: 1979 "Occupational Segregation by Sex: Trends and Theory, Evidence and a Prognosis." Pp. 137- Prospects." The Journal of Human Resources 157 ir. Cynthia B. Lloyd et al., eds.,Women 14 (Spring):197-210. inthe Labor Market.New York: Columbia University Press. Cherlin, Andrew, and Pamela B. Walters 1981 "Trends in United States Men's and Women's U.S. Bureau of the Census Sex-Role Attitudes: 1972 to 1978." American 1963 Census of Population: 1960.Subject Reports. Sociological Review46 (August):453-60. Final Report PC(2)-7A. Occupational Charac- Corcoran, Mary, and Creamy Duncan teristics. Washington, D.C.: U.S. Govern- ment Printing Office. 1979 "Work History, Lahor Force Attachment, and Earnings Differences Between the Races and 1972 "1970 Occupation and Industry Classification Sexes."The Journal of Human Resources14 Systems in Terms of their 1960 Occupation and (Winter):3-20. Industry Elements," by John A. Priebe. Tech- Duncan, Gus Dudley, and Beverly Duncan nical Papers, No. 26. 1955 "A Methodological Analysis of Segregation In- 1973a Census of Population: 1970, DetailedChar- dexes."American SociologicalReview 20 acteristics. Final Report PC(1)-D1. U.S. Sum- (April):210-17. mary. Washington, D.C.: U.S Government Printing Office, Table 221. Economic Report of the President 1973 Washington. D.C.: U.S. Government Printing 197313 Census of Population: 1970.Subject Reports. Office, pp. 155-59. Final Report PC(2)-7A. Occupational Charac- England, Paula tjristics. Washington, D.C.: U.S. Govern- ment Printing Office. 1981 'Assessing Trends in Occupational Sex Seg- regation, 1900-1976." Pp. 273-95 in lvar Berg, U.S. Commission on Civil Rights ed.,Sociological Perspectives on Labor Mar- 1978 Social Indicators of Equality forMinorities and kets.New York: Academic Press. Women. August.

33 ANDREA H. BELLER 26

the Cur- U.S. Department of Labor, Bureau ofLabor Statistics 1982b Labor Force Statistics Derived from rent Population Survey: ADataboolc 1 (Bul- 1979 Employment and Earnings 26(1):207. letin 2096):Table C-23. 1980 Employment and Training Report of the Pres- ident. Employment and TrainingAdministra- U.S. National Center for Education Statistics Projections of Education Statistics to 1988-89. tion. 1980 April. 1982aEmployment and Earnings29(5):Tabre J.

34 A Woman's Place Is With Other 3Women: Sex Segregation Within Organizations

WILLIAM T. BIELBYand JAMES N. BARON

Sex segregation in the workplace is one us personally. Social structures that gener- of the most visible signs of social inequality. ate gender segregation are of great concern In almost every work setting, it is unusual to social scientists, and the inequities that to see men and women working at the same segregation engenders' are obvioully rele- job. When they do, they typically perform vant to social policy. Yet sociologists know different tasks, with unequal levels of re- surprisingly little about job segregation by sponsibility and authority. Even when job sex. Most of what we have learned concerns tasks are virtually identical, it is not uncom- segregation among occupations. For exam- mon to find men and women allocated to ple, we know that equalizing the detailed distinct job classifications within an orga- (census 3-digit) occupational distribution for nization. men versus women would require moving Even women working full time, year round roughly 60 percent of women working out- are 'paid less than men. While the earnings side the home across occupational cate- gap is partly due to unequal access to high gories, and this has changed very little since wage firms and to unequal pay for compa- 1000 (Gross, 1968; Blau and Hendricks, 1978; rable work, a substantial portion is due to Williams, 1979; England, 1981a). We are differences in pay scales for job classifica- also learning more about how men and tions filled by men and those filled by women women make occupational choices (Bielby, (Bridges and Berk,1974; Halaby, 1979; 1978; Marini, 1978, 1980). Empirical re- Treiman and Hartmann, 1981). Sex segre- search on job segregation across organiza- gation has social-psychological conse- tional settings, however, is quite sparse. quences as well. For example, groups with Accordingly, this paper examines sex seg- limited opportunities for advancement may regation in the workplace, utilizing data de- respond with psychological disengagement scribing work arrangements in nearly 400 from the firm, lowered career aspirations, establishments across a wide range of in- and an increasingly narrow, instrumental dustrial and institutional settings. We dis- orientation toward work (Kanter, 1977b). tinguish situations in which employers place In short, sex-segregated workplaces affect men and women in the same job classifi-

3 28 WILLIAM T. BIELBY AND JAMES N. BARON cation from those in which jobtitles seg- (e.g., Blau and Jusenius, 1976; England, regate the sexes withinestablishments. 1981b). Much more has been written, how- Our study could be viewed as astraight- ever, about whyemployers treat men and forward job-level disaggregation of findings women differentlythan about the extent to regarding occupational sex segregation. Those which they do so. The sparse literature ad- studies acknowledge that considerable seg- dressing why some firms are more segre- regation may exist even within detailed oc- gated than others falls into three categories: cupational categories. Our measures and institutional accounts, explanations based on methods parallel such studies, and our find- tasted for discrimination, and humancapital ings confirm their speculationsabout per- market models. vasive segregation within occupations. Institutional accounts stress how statisti- However, our aitn is not merely to reveal cal discrimination in hiring and allocating hidden segregation among jobs within firms. employees places men and women in dis- Rather, since sex segregationis accom- tinct career trajectories., Men tend to enter plished in organizations and is affected by internal labor markets in which they can technical, administrative, and social exigen- expect an orderly progressionthrough suc- cies of the workplace, it is important to ex- cessively more attractive jobs, insulated from amine how organizational structuresand competition outside the firm. This increases processes produce sex segregation. organizational loyalty, decreases costly Our research does not consider howmen's worker turnover, and allows employers to and women's occupational choices, laborforce recoup investments infirm-specific training participation, and human capital invest- (Doeringer and Fiore, 1.971). Women are ments affect the sex compositionof the perceived to have weaker commitments both workplace (for contrasting interpretations, to specific firms and to paidemployment in see England, 1982,and Polachek, 1979). Nor general and are thus allocated to jobs with are we investigatingthe demand side in the low turnover costs and limited opportunity economist's sense of the term, since wehave for security and advancement (Bielby and no information onthe productivities of dif- Baron, 1982). Not all firms, however, re- ferent classes of workers and the wages em- quire specifically trained workers orhave ployers are willing to offer them (cf. Blau, internal labor markets. Therefore, if sex biases 1977). Rather, the intersection of labor sup- in allocating workers to job ladders werethe ply and demand enters into our analysis inonly basis for segregating men and women, directly, since occupational compositionand one would expectless segregation in firms skill mix of the firm are examined as deter- lacking institutionalized employment ar- minants of sex segregation. However,if jobs rangements -- particularly small,labor-in- in most establishments arehighly segre- tensive, entrepreneurial firms in theso-called gated by sex even across firmsdiffering economic periphery (Averitt, 1968). dramatically in their production functions This is certainly not the only mechanism and cost structures then it seems unlikely placing men and women in distinct job clas- that marginal adjustments of supplyand de- sifications, and perhaps a more reasonable mand account for distinct job assignments hypothesis is that the process of segregation of men and women (for a similar view, see differs according to an organization's admin- Blau and Jusenius, 1976). istrative arrangements and locationwithin the economy. For example, small manufacturing, service, and retail establishments WHY ARE SOME FIRMS MORE typically rely on an unskilled secondary la- SEGREGATED THAN OTHERS? bor market and use simple hierarchy or en- Diverse explanations of sex segregation trepreneurial despotism to control workers have been reviewed 'thoroughly byothers (Edwards, 1979).

. 36 SEX SEGREGATIOPWITHIN ORGANIZATIONS 29

These enterprises mightapply one ofsev- mizing behavior by bothworkers andem- era, strategies in allocating men andwomen to jobs. Highly trained ployers. That is, firms willassign men and line workers with women to the same job titles only job- and firm-specificskills typically under spe- are not cific, andrare, circumstances: (a) when there employed in such establishments,nor are is an available labor poolcomposed ofmen highly rationalized personneland job clas- and women and (b) sification procedures when employersperceive utilized. Thus, these that the costs of firms might provide employing men andwomen precisely the workcon- roughly are thesame. texts in whichmen and women who lack To summarize,certain analysts credentials formore desirable employment argue that gender segregationat work is caused by ad- work together withinbroadly defined job ministrative arrangements forhiring, allocat- categories. Furthermore,if employersmust ing, and controlling employees.Others em- sacrifice profits in orderto discriminate, they phasize the impact oftastes or prejudices, while must be able to afford thecosts of their pal- still others claim thatsex segregation reflects ivies. Marginal firms withweak competitive rational decisions regardinghuman capital in- positions can least afford thesecosts and have vestments on the part of workersand em- an economic incentive to ignoresex in hiring ployers. Perhaps because and allocating workers segregation is such (Arrow, 1973). a natural attribute of most work In the absence, however,of institution- situations, little has beenwritten about the conditions alized procedures forhiring and allocating under which it does workers, male employers) not occur. in the economic Our empirical analysis peripherymay have mor4- discretion to is guided b4sev- im- eral general hypotheses.First, institutional plement tastes for discrimination,which can reflect their accounts suggest that lesssegregated firms own preferences or those of their lack the administi'ative employeesor even their clients. In the apparatus to differ- most entiate workers bysex and cannot afford the extreme case, patriarchal controlstrategies costs of implementing employers' would exclude tastes for women from the workplace a segregated work force. Second, entirely. Such neoclass- arrangements should be most ical accounts, grounded prevalent in organizational in notions of tech- niches thatare nical efficiency,suggest that desegregated protected from competitivepressures (e.g., organizations do not relyheavily on firm- through satellite linkageswith larger firms) specific skills but employworkers in or where preferences fora segregated work occupations that are attractiveto both men and force are so widely heldwithin an industry women and for which bothsexes are eligi- or area that they have the force ofcustomary ble. Ofcourse, each of the mechanismssum- law constraining marketforces (Doeringer mariied above might and Piore, 1971:22-27). operate but within specific organizational If labor supply and settings. Conse- technical requisites de- quently,we examine the heterogeneity termine the distribution ofmen and women among highly segregated establishmentsto across job categories, thena firm's inix of oc- see if there are alternate s by which cupations and skills should largelyaccount for employers achieve thesai,jrial. It: distinct its tendencies tosegregate men from women. job assignments formen aiaworrien. According to such humancapital models, workers expectingintermittent labor force participation (primarily women)choose toen- DATA AND METIIODF, ter occupations in which job skills do not atro- We analyzed dataon work arrangements phy fromnonuse (Polachek, 1979). Indeed, if in hundreds ofeconomic establishments stud- jobs with the highestturnover costs are also ied in California between1959,and 1979 by those m which skillsatrophy most quickly, the CaliforniaOccupational Analysis Field then extremesegregation can reflect maxi- Center of the U.S. EmploymentService. These

37 30 WILLIAM T. BIELBY AND JAMES N. BARON

data, used primarily in preparing the Dic- are overrepresented in the sample. Major tionary of Occupational Titles (DOT), and California industries not represented in our procedures for collecting them are described sample include construction trades, truck- in U.S. Department of Labor (1972). Our unit ing, apparel and general merchandise retail of analysis is the establishment, the "physical trade (department stores), and insurance location where business is conducted or where carriers. The first two industries are male services or industrial operations are per- dominated and highly segregated; the latter formed" (U.S. Bureau of the Census, 1976:iv). two employ many women and may be less Industrial characteristics serve as contextual segregated. While these data do not char- factors, and suborganizational information about acterize a distinct pbpulatior), they do re- workers.and jobs has been aggregated, to char- flect a diversity of work arrangements across acterize each enterprise. The inaj'ority othe a broad range of industrial and organiza- establishments are firms; others are branches, tional contexts. In our view they provide regional divisions, subsidiaries, and produc- invaluable comparative evidence regarding tion sites. Since we focused on work sites rather how administrative, technical, and environ- than firms, corporate headquarters of multi - mental contingencies in organizations affect plant organizations are typically not included the structuring of work. in our data. Corporations often direct' initial The data collected and coded for our proj- desegregation efforts at headquarter !liana- ect include 742 observations in over 500 dis- Aerial and office work (Staeffer and 1,yliton, tinct enterprises. About one-fifth of the es-

1979), and progress toward equal employ-o tablishments were visited inure than once ment opportunity (EEO) goals in these areas by Employment Service analysts. The most will not be reflected in our results. recent analysis was used for firms with follow-up data. Since some of the information The Sample used to characterize organizatioual attri- No well-defined sampling frame guides butes, however, was derived from narrative the Employment Service's selection of en- reports (described below), precedence was terprises to study, but they try to represent given to complete observations' that 'also the diversity of activities carried out within possessed a contemporaneous narrative re- any industry (Miller et al., 1980). The Cal- port, even if a more recent follow-up anal- ifornia Field Center tended to study those ysis, lacking a narrative, had occurred. industries that are regionally concentrated To ensure comparability, analyses rein the state, so our sample of establishments stricted' to the firm's productive component includes, for example, firms engaged in ag- or some other subset of jobs or departments riculture, aircraft manufacturing. banking, were omitted, since they do not accurately fishing, and motion picture production but characterize an entire work site. This re- nut automobile or furniture manufacturing. striction reduces the sample of establish- While our sample provides a reasonable ments to 415. The sex comptIsition of jobs representation of the composition of estab- was not reported for 22 of these firms, re- lislunents within industries,' the actual in- ducing the sample size fbr analyses reported dustries studied are not fully representative in this paper to 393..2 Of these, about 26 of economic activities in California. Most importantly, manufacturing establishments JOI) composition was not enumerated by sex after 1977-197h. apparently because of increasing resistance Kewighting our obsratiims according to pl.lb from establishments approached by the Employinnt Ilshtli data on the siw distribution of establishments ServiceUnfortunately. this occurred when the within California's industries has virtually no effcton fornia Field (:etittr wasstudyingAgricultural estab- thy (listributam of mgani tat uinalattrikrtMln our gam lishmnts. therflae. 7 of the 22 ubserations ply information on sex composititom arc in agriculture

38 GT

SEX SEGREGATION WITHIN ORGANIZATIONS 31 percent are plants, franchises, or production ulec supply, in essence, a complete organi- sites within some larger entity (though not zational division of labor for the plant or firm necessarily corpOrations); 10 percent are ad- in which job titles are analyzed. Face Sheets ministrative divisions, regions, or branches provide identifying information about the es- of larger companies; 3 percent are subsidi- tablishment and analysis. After assigning the aries (distinct firms owned by another firm); firm to one or more categories of the Standard and 61 percent are independent busi- Industrial Classification (SIC), the Employ- nesses.3 Over half, 54 percent of these es- ment Service classifies the enterprise by its tablishments were studied between 1968 and primary product(s) and supplies information 1971, and 76 percent were visited by the about any unique or noteworthy characteris- Employment Service between 1965 and 1973. tics of the firm, such as its jobs and processes. The 393 establishments in our sample em- Narrative reports prepared for many estab- ploy nearly 47,000 men and over 14,000 lishments include information on some or all women.' of the following: history and purpose of establishment; environmental conditions; operations and activities (departmentation, The Documents workflow, processes or services); personnel policies and practices; recent job restructur- This paper uses two types of data obtained ing; effects of automation on personnel and from 'the records of the California Occupa- operations; and optional sections dealing with tional Analysis. Field Center. Staffing Sched- such topics as the product market and relations with government, the community, or other firms.

3 In practice, it sometimes was difficult to determine precisely if establishments studied by the Employment Service were autonomous firms or productive or ad- Operationalization ministrative units within larger companies. When our materials indicated an owner or president, we assumed Staffing schedules, face sheets, and nar- the enterprise was autonomous, owner-operated, un- rative reports were used to measure various less other information indicated to the Contrary When environmental, organizational, and techni- the top position had such titles as plant superintendent, cal attributes of establishments as well as plant manager, general manager we assumed the en- the composition of occupations and skills terprise was a subdivision of a larger firm, unless background information suggested otherwise. Anomalous employed in each enterprise. Operational- cases were referred for clarification to the Employment izations are summarized in Table 3-1, which Service analysts who conducted the original studies. also reports descriptive statistics for varia- Confidentiality restrictions precluded access to estab- bles used in our analysis. Organizational scale lishments' identities, preventing us from resolving such is measured by the natural logarithm of the ambiguities directly. 41, °The disproportionate share of manufacturing estab.. number of employees, and positional spe- lishments in our sample accounts fcr the underrepre- cialization is measured by the logarithm of sentation of women workers. Nevertheless, the range median job size. The latter measure is com- of industries covered represents nearly every work.con- puted across workers, so a median of 10 in- text in which women labor. One important exception: dicates that one-half of the workers are in The Employment Service tends to analyze branch plants and to overlook corporate headquarters. Therefore, vir- establishment job titles with 10 or more in- tually every kind of nonmanual work performed by cumbentsas opposed to half the jobs con- women is represented in our study, but, unfortunately, taining 10 or more workers. This measure we have no instances of such work done at the head- indexes the degree to which establishments quarter offices of large corporations. Evidence from the Imassify" the work force by assigning many early 1970s suggests that efforts to desegregate nonmanual work occurred first in such contexts (Shaeffer workers to the same job title. Consequently, and Ly,itori. 1979). low scores correspond to high levels of spe-

39 32 WILLIAM T BIELBY AND JAMES N. BARON

TABLE 3- scriptive Statistics and Operationalizationsof Organizational Attributes (N = 393) Standard Variable (range) MeanDeviation N . Description Establishment Scale and Specialization inestab-' Log establishment size (.69-8.97) 3.67 1.48 393 Natural log of number of employees lishment. Log specialization (0-6.49) 1.12 1,03 368 Natural log of median job size (see text). Manufacturing Industry (0-1) .63 393 One or more of establishment's standard industrial classification designations is in the manufacturing industry. Economic Sector Core (0-1) - 393 See text. Ambiguous (0-1) .41 - 393 See text. Periphery (0-1) .21 +- 393 See text. Social Organization Log fragmentation ( -1.10-1.39) .14% .28 360 Natural log of ratio of establishment job tales to unique DOT titles. Union or bidding arrangements .25 393 1 = some or all employees unionized or covered (0-1) by formal bidding arrangements. Proportion women (.00-1.00) .32 .26 393 Proportion of female employees/ Occupational and Skill Composition Proportion production workers .56 .33 380 Proportion of workers with DOT codes denoting (.00-1.00) production occupations. Proportion clerical and sales .20 .23 379 Proportion of workers with DOT codes denoting workers (.00-1.00) clerical and sales occupations. Proportion service workers .08 .21 379 Proportion of workers with DOT codes denoting (.00 -1.00) service occupations. Proportion professional, mana- .16 .20 379 Proportion of workers with DOT codes denoting gerial. technical workers professional, maleagerial, or technical occu- (.00-1.00) pations. Average complexity: data 1.92' 1.09 336 Mean of ratings indicating complexity of work- (0-5.71) ers' involvement with data. Average complexity: people 1.27 1.11 ' 336 Mean of ratings indicating complexity of work- (0-8.00) ers involvement with people. Average technical skills 0.01 1.00 379 Mean of standardized ratings of 'production ( -2.19 -3.08) workers' technical skill (see text). Skill,specificity (.00-.99) .51 .27 173 Proportion of workers in nonentry-level jobs. Sex Segregation (12.5-100) 93.4 13.8 393 Index of dissimilarity computed acrosS'job titles. 100 = all male, all female, or completely segregated.

cialization.5 One way to segregate workers regati'on in more specialized establish- is to place each worker in a uniquejob cat- ments. We failed to detect any netdifferences egory; therefore, we expectgreater sex seg- in relationships between sex segregation

scale; This measure correlated .87 with Gibbs and Pos- is almost completely determined by organizational ton's (1975) "M4" index of distributive differentiation. the correlation between the number of employees and describing the evenness of the distribution of workers the number of job titles is .92 when both are measured dif- across positions. Functional(horizontal) differentiation in a logarithmic metric. Consequently, while such of departments is reflected in our criteria for assigning ferentiation may mediate the effects of scale, it is un- establishments to economic sectors (see below). Struc- likely to affect work arrangements independently. tural differentiation - the proliferation of workroles - 40 SEX SEGREGATION WITHIN ORGANIZATIONS 33 across 6 industrial sectors: manufacturing, 6-digit DOT titles assigned by the Employ- state, services, social overhead capital, ag- ment Service analyst.fi It measures the de- riculture, and trade. We distinguish man- gree to which the organization differentiates- ufacturing from nonmanufacturing, how- its work force administratively beyond what ever, to assess indirectly whether physical might he expected from a breakdown of de- demands of work account for patterns of sex tailed occupational functions (Braverman, segregation. 1974:70-83; Edwards, 1979).. Fragmentation We argued aka we that organizations con- is' one strategy for segregating male and fe- front different incentives to segregate de- male workers who perform similar job tasks; pending on their niche within the economy that is, separate names are attached to men's and their size, structure, and technology. and women's work. Other facets of the social These differences in organizational forms and organization of workplaces are measured by environments capture distinctive locations a dumin'y variab14'that denotes the presence within the economy what some institu- of unions or formalized bidding arrange- tionalists and Marxists have called sectors ments, covering some or all workers, and or se gments. In certain, respects, these are the sex ratio, the percentage of workers who the organizational equivalent of ';classes "; are female. Unions and formal bidding ar- that is, actors presumed to share certain in- rangements are an institutional arrange- terests and attributes by virtue of their comment that may constrain employers' (and male mon market positions. Core firms are typ- employees') ability to indulge tastes for dis- ically large, differentiated, use automated crimination. Sonie argue, however,that technologies, pnxluce multiple products, are unions can exacerbate gender inequalities unionized, and are linked to larger organi- (e.g., Milkman, 1980; Baron and Bielby, zational entities. Their environments arc 1982). Finally, tokenism a highly unbal- characterized I.; interorganizational de- anced sex ratio can facilitate the segre- pendence within kz.y industrial sectors, and gation of women or men into one or two these establishments tend to be dominant separate job titles, while a more balanced actors in their milieux. In contrast, the eco- work force may be more difficult to egre- nomic periphery is composed of small, un- gate (Kanter, 1977a). differentiated enterprises, typically operat- The occupatignal composition of each en- ing in highly competitive markets in terprise was cdmputed from the DOT oc- industries other than manufacturing. This cupational codes corresponding to each job congruence of organizational fOrmand en- title.' We measured the distributions of vironment does not characterize all firms; workers across the following broad cate- the 161 establishments allocated to the am- gories: professional, managerial, and tech- biguous sector include those. in small-wale manufacturing and many nonmanufacturing firms in less vulnerable situations with re- h In many firms. not every job title., was mapped to gard to their environments. The measures a DOT title by the Employment Service analyst. The fragmentation measure was computed only for firms in and proceduiles underlying this sectoral which both 90 percent of the jobs and 90 percent of classification scheme are discussed in detail the workers were assigned a 6-digit DOT code (see by Baron (1982:Chapter Baron. 1982: Chapter VI). Six-digit DOT codes do not Badica! accounts 'of workplace relations correspond to unique titles in the DOT. but more de- suggest that three aspects of an establish- tailed classification is possible only for those jobs as- ment's social organization should he asso- signed to occupational categories according to the fourth edition classification scheme. ciated %Nall its level of sex segregation. Frag- Establishments in which less than 75 percent of the mentation is measured by the logged ratio workers could be assigned a DOT code were treated of lob titles in the establishment to unique as missing data on this outcome.

41 .11 34 WILLIAM T. BIELBY AND JAMES N. BARON nical; clerical and sales; service domestic, partments who were not in entry -leveljobs, business, personal and production,in- as indicated on thestaffing schedule.") cluding extractive and transportation occu- The index of dissimilarity measures the pations. Levels of informational and inter- percentage of workers of one sex thatwould personal skill were measured by the mean have to be moved to new job classifications of ratings indicating the complexity of workin order to equalize the jobdistribution of ers' involvements with data and people, re- workers by sex (Duncan and Duncan, 1955). spectively." It equals zero when the percentage distri- The measures of technical skills used de- butions of men and women across job cat- pended on the edition of the DOT used in egories are identical, and it equals 100when the analysis of an establishment. Production no men and womenwork in the same job." occupations were classified as skilled, semiskilled, or unskilled in the second edition, The Analysis and we assigned values of 2, 1, and 0 to these categories, respectively. Ratings ofworkers' We first describe the distribution of es- involvements with "things" began with the tablishments by level of sex segregation. Then third edition. Average ratings were com- we examine theorganizational attributes that puted for each establishment's labor force di: aguish propensities to segregate. The using either the measure based onsecond relative contribution of social, administra- edition DOT codes or the one based on sub- tive, and technical attributes to patternsof sequent versions of the DOT.Each estab: segregation may suggest mechanisms that account for thoSe results, but wedo oot ex- lishment's score was then normalized rela-. live to all others sharing the same version pect conclusive results fromcross-sectional of the skill measure. That is, each enter- findings. Accordingly, these analyses are prise's level of technical complexity is meas- supplemented in two ways. First, we ex- ured relative to other organizations incor- amine specific eases for whichqualitative porating the same version ofthe DOT information exists on the hiring and alloca- occupational classification.9 Finally, the level of firm-specific skills was assessed indirectly scheme. Mean ratings were computed from second edi- by the proportion of workers in "line"de- tion codes only if.(a) detailed occupational codes existed for at least 75 percent of the employees in the enterprise and (b) at least 25 percent of the workers This information. when avivlable, was coded from were in production jobs. ,,Preferene was given tothird job analyses pertaining to each position in thefirm. edition data for establishments that contained jobs an- Otherwise, it was obtained from 6-digit DOT iwcu- alyzed in terms of both second and third edition DOT pational codes listed oil staffing schedules. Third edi- procedures. tion DOT ratings of 7 or S for relations withdata were ") Workers in nonproduction departments were ex- recoded to 6 to conform to the fourth edition rating cluded because the Employment Service did not al- scheme. Mean ratings for each establishment were ways collect informationdenoting entry-level jobs in computed only if (a) at least half the jobs and workers those departments. The measure was not computed for establishments with less than 15 percent of their labor rnhne depart roe',s could be characterized on thedata and people dimensions, (1)1 at least half of the jobsand force in production-related departments. We also elim- workers in other departments could he characterized; inated observations in which any department had no and (c) no more than 10 percent of the establishment entry-level workers, or in which certain traditionally labor force was missing data on these variables.Finally, entry-level iwcupations occurrede.g., janitor, re- scale saloes were inverted so that large values corre- eptionistbut none was coded as such. Given these spond to high levels of involvement with data and peo- restrictions, a measure of skill specificity is available ple for less than half of our observations. The TIVAn level of technical skills was computed " We consider all-male and all-female establish- value of from third or fourth edition ratings (from jobanalyses ments to be perfectly segregated and assign a or DOT codes]. subject tothe same restrietions de- 100 to the index of dissimdarity. However, in analx ses sribed in footnote 7. Third edition codes of h were reported be -loss. these enterprises are considered sep- recoiled to 7 to conform to the fourth edition rating arately

12 SEX SEGREGATION WITHIN ORGANIZATIONS 35

TABLE 3-2Distribution of Establishments and Workers by Level of Segregation (N = 393)

(4) (3) Cumulative(5) (2) Percentage PercentageMedian (8) (1) Number ofof of Establish- (6) (7) Cumulative Segregation Establish- `Establish- Establish- ment Number ofPercentagePercentage Index ments& ments ments Size Workers of Workersof Workers 0-9 0 0.0 0.0 0 0.0 0.0 10-19 1 0.3 0.3 30 30 0.0 0.0 20-29 2 0:5 0.8 94.5 189 0.3 0.4 30-39 4 1.0 1.8 64 457 0.7 1.1 r. 40-49 2 0.5 2.3 25.5 51 0.1 1.2 50-59 7 1.8 4. 1 11 186 0.3 1.5 60-69 13 3.3 7,4 28 540 0.9 2.4 70-79 18 4.6 12.0 29.5 3,746 6.1 8.5 80-89 27 6.9 18.8 57 2,281 3.7 12.2 90-96 35 8.9 27.7 72 7.260 11.9 24.2 96-99.99 52 13.2 40.9 195 26,587 43.6 67.8 100 202° 51.4 92.3 25 19,250 31.6 99.4 All male 21 5.3 97.7 7 322 '0.5 99.9 All female 9 2.3 100 5 51 0.1 100 TOTAL 393 100 36 60,950 100 ° Subsequent tables, report 201 completely segregated establishments. After completing our analyses, however, we discovered 1 establishment for which the sole integrated job was in fact due to a coding error. tion of women. Second, we examine the only 10 percent Of the nearly 61,000 workers subset of organizations for which longitu- are in establishment job titles that have both dinal infOrmation is available in orriir to learn men and women assigned to them. Even (1) the extent to which patterns of sex seg- among the 162 establishments having some regation change over time and (2) the or- men and women in the same job titles, the ganizational circumstances. under which mean segregation index is 84.1. In short, gender segregation increases, decreases, or the workplace is substantially more segre- remains constant. The concluding section gated by sex than is suggested by studies addresses implications of' our findings for that aggregate work force composition across policies aimed at equalizing job experiences establishments and into 3-digit occupational and attainments of men and woineji. categories.12 One way to segregate male and female workers is to employ either men or women RESULTS exclusively in an establishment. The 21 es- Descriptive Statistics tablishments without female workers, listed by establishment size in Table 3-3, are al- Table 3-2 presents a remarkable story: Most establishments are either completely segregated or nearly so. Less than one-fifth of 12 Ofcourse, statistics on the dish ihution of workers the establishments have segregation indices in our sample are not representative of the California lower than 90, and they employ less than labor force, since establishments, not workers, were one-eighth of the workers in our sample. sampled. Nevertheless, these results show that there are very few work contexts in which men and women Over one-half of the establishments are are assigned to the same job titles, and results reported completely segregated, and over three- below suggest that even the least segregated enter- quarters of' the workers are in organizations prises are seldom examples of workplace equity behaving indices between 96 and 1(X). Indeed, tween the sexes.

,43 TABLE 3-3Organizational Attributes of Establishments That Employ Only Males orOnly Females Occupational Composition: PercentagePhysical in Job Dem odds 'Product!, or Employ- Job Titles Median Produc-Clerical Exclude Women"Comments Activities Sector°Industry!' ees Titles per DOTJob Sizetion & Sales p-rm, Service Establishments With No Female Workers 0 100 0 NA Flight instruction P SOC 2 1 5 2.0 0 0 (1 Y Food products A MFG 3 3 1.0 1.0 80 20 A 3 2 .67 1.2 100 0 II 0 NA Class products ..._MFC 0 0 NA Stonework A MEG 3 3 1.0 10 100 0 NA Ste rework A MR; 4 4 1.0 1.0 75 0 25 0 0 NA Stonework P TRADE 4 2 1.0 2.0 100 0 0 0 NA Sports equipment A MEG 4 3 1.0 1.0 75 0 25 0 50 0 N Wallpaper A MEG 4 2 0.5 2.0 50 u N Canning A MEG 6 4 0.8 1.0 67 0 33 0 0 0 Y Subsidiary of a large is Plastics A MEG 7 3 1.0 2.3 1(X) chemical company. 0 50 0 N Crop dusting P M; 7 6 NA 1.0 50 0 25 0 NA Pens and pencils A MEG 8 5 0.8 2.0 75 25 0 NA Sports equipment A MEG 8 3 0.5 2.7 75 0 0 0 0 Y Fish hatchery A .AG 10 2 1.0 4.2 1(X) 9 0 Y Branch of large dairy. Dairy products A M EC 11 9 0.9 1.0 77 13 NA N Laundrs P SERI' 17 10 NA 2.0 NA NA NA 6 0 NA Fertilizer A MFC 17 7 1.0 3.0 71 23 94 :3 3 0 Y Branch plantof larger Brick & the A MEG 36 14 1.1 3.0 compam.. 0 48 0 NA Subsidiary 01 larger %Voter transportationA MX: 46 7 1.0 10.0 52 company. 54 (1 4 NA City-owned. Golf course A STATE, 56 9 to 18.4 43 SERI' 0 Y Production plant of larger Wire C MEG 66 21 1.3 4.0 92 0 8 company. 44- Establishments With No Male Workers

Domestic service P SERV 2 2 1.0 1.0 0 0 50 50 Licensed foster mother who employs a housekeeper. Real estate rentals P SOC 2 2 1.0 1.0. 0 100 0 0 Real estate escrow P SOC 2 2 1.0 1.0 25 75 0 Business services P SERV 3 2 NA 12 NA NA NA NA Female labor contractor of office services. Confection A MFG 5 5 1.0 1.0 100 0 0 0 Female-owned. Hairpieces A MFG 5 4 1.0 1.0 60 20 0 20 Female-owned. Real estate escrow P SOC 6 4 1.0 1.5 0 50 0 50 Female owner-manager. Library P STATE, 10 5 1.7 2.3 0 20 0 80 Library branch of veterans' SOC hospital; civil service. Business services P SERV 16 2 1.0 7.5 0 t20 0 80 Female-owned telephone answering service. P = periphery; A = ambiguous; C = core. bSOC = social overhead capital; MFG = manufacturing; AG = agriculture; SERV = service. = professional, technical, and managerial occupations. dNA = information not available; Y = yes; N = no.

45 38 WILLIAM T. BIELBY AND JAMES N. BARON

most entirely small manufacturing opera- pletely segregated. Moderately segregated tions in which most workers are in produc- enterprises are larger still, followed by highly tion jobs. Only 1 of the 9 all-female and almost perfectly segregated ones (antil- establishments listed in Table 3-3 has more ogs of the means are 4, 9, 27, 30, 67, and than 10 employees, and only 2 employ women 159 employees, respectively). Among the in manufacturing. The typical all-female es- 384 establishments employing men, 30 per- tablishment provides services and does not cent of the variance in log size occurs across utilize the technical skills of blue-collar the 5 segregation categories. workers. 13 This strong association between organi- While extreme sex segregation is perva- zational scale and segregation persists in sive, hiring policies that exclude one sex multivariate analyses (see below) and seems entirely are utilized only by establishments to involve the changing mix of employer dis- engaged in marginal economic activities on cretion versus the impersonal rule of bur- a very small scale. Only 5 of the all-male eacratic procedures as organizations grow. establishments are branches of larger com- For example, employers operating on a very panies, and 4 of those cite physical demands small scale may exercise tastes for discrim- of work as the reason for excluding women. , ination by excluding women altogether.

. It scents that autonomous employers oper- However, as tasks become increasingly dif- ating small firms need no explicit rationale ferentiated and. specialized clerical roles are for excluding female workers; they can uni- introduced, inexpensive female labor can be laterally exercise their preferences for an all- utilized in segregated job classifications. male work force. However, not all small establishments dif- The strong association between organi- ferentiate job tasks to the same degree, and zational scale and segregation is docu- some allocate both men and women to mented in the first line of Table 3-4. The broadly defined job classifications. Contin- table reports mean levels of various orga- ued expansion leads to the implementation nizational attributes across 6 categories of of rationalized, bureaucratic personnel pro- sex segregation: moderate (A 75); high (75 cedures in neatly all firms. Mechanisms seg- < A 90); very high (90 < A < 100); com- regating the sexes become institutionalized, plete segregation (A = 100); exclusive em- and in large establishments men and women ployment of males; and exclusive employ- are almost always assigned to separate job ment of females.Excluding the 9 families. establishments employing Only women, col- Other organizational attributes listed in umn 7 reports the proportion of variance in Table 3-4 are moderately associated with each organizational attribute occurring across segregation levels. Means for each variable categories of segregation." Establishments listed in Table 3-4 differ monotonically across that exclude men or women are the smallest categories of moderate, high, and very high on average, followed by those that are corn- segregation. Indeed, the characteristics most strongly associated with segregation core sectoral location, specialization, and frag- " There are 13 establishments with just 1 male worker, mentation are also highly correlated with and these are quite similar to those listed in Table 3- 3. The largest has 22 employees and 9 have 5 workers or less. The 60 firms with just I female employee are seems to reflect processes qualitatively different from concentrated in small-scale manufacturing and social those that exclude or segregate women. Accordingly, overhead capital industries. Only 1 has more than 30 we exclude the 9 all female establishments from the employees. multivariate analyses reported below and from the var- 14 Total exclusion of males is not only rare but also iance explained computations reported in Table 3-4.

46 SEX SEGREGATION WITHIN ORGANIZATIONS 39

TABLE 3-4Means of Organizational Attributes by Level of Sex Segregation Level of Sex Segregation (A) i (1) (2) (3) (4) (5) (6) 12 All All Columns A s 7575 < A E 90 90 < A < 100 A = 100 Male Female(1)-(5) Variable (N = 42) (N = 40) (N = 80) (N = 201) (N = 21) (N = 9)(N = 384) Log size 3.40 4.20 5.07 3.31 2.17 1.47 ,30 Log specialization 1.34 1.40 1.72 .87 .74 .39 .12 (1.52)a (1.12) (.93) (1.14) (1.64) (.03)E Manufacturing industry .45 .68 .76 .62 .67 .22 .03 (.46) (.66) (.71) (.63) (.71) (,02) Core sector .17 .so .78 .29 .05 .00 .21 (.23) (.40) (.49) (.37) (..`..5) (.02) Ambiguous sector .52 .42 .21 .45 .11 .22 ,06 (.50) (.47) (.33) (.41) (.58) (.01) Periphery sector .31 .08* .01 .26 .24 .88 .08 (.27) (.13) (.18) (.21) (.05) (,02) Log fragmentation .09 .21 .33 .10 -.14 .06 .16 (.12) (.16) (.20) (.13) (.00) (.02) Union/bidding .05 .30 .46 .22 .14 .00 .08 arrangements (.09) (.23) (.28) (.28) (.35) (.02) Proportion women .44 ..42' .34 .28 .00 1.00 - Proportion production .44 .55 .61 .56 .74 .20 .04 workers (.45) (.54) (.56) (.56) (.80) (.04) Proportion clerical & sales .28 .20 .17 .20 .02 .39 .05 workers (.27) (.21) (.21) (.19) (- .02) (.06) Proportion service workers .06 .08 .09 .07 .03 .33 .00 (.07) (.08) (.09) (.08) (.03) (.00) Proportion PTMc workers .22 .17 .11 .16 .21 .33 .03 (.21) (.18) (.14) (.16) (.19) (`.01) Average complexity: dafa 2.24 1.88 1.73 1.95 1.66 2.36 .02 (2.23) (1.89) (1.77) (1.94) (1.62) (.02) Average complexity: people 1.64 1.05 .91 1.36 1.41 1.50 .04 (1.58) (1.14) (1.17) (1.28) (1.15) (.07) Average technical skills .37 - .04 .12 .02 .30 - .96 .02 ( - .34) ( .09) ( .02) (.05) (.48) (.04) Skill specificity .43 .48 .52 .50 .53 .31 .01 (.57) (.52) (.48) (.50) (.47) (.0(1) ° Entries in parentheses for columns (1)-(5) are means adjusted for linear associationwith size. b Entries in parentheses in right-most column are the increments to explained variance due to segregation categories, net of size. Professional, technical, and managerial. log size, suggesting that the pattern of means in the table may simply reflect concomitants of organizational scale. Therefore, we also 15 Specifically, each attribute was regressed on log control for each attribute's linear relation- size and binary variables representing segregation level. ship with log size (coefficients in parenthe- and adjusted means were computed at the average value of log size. The value reported in parentheses below ses). 15 the zero-order eta-squared for each attribute "y" is the While controlling for size weakens most increase in explained variance when variables denoting relationships in Table 3-4, the general pat- segregation group are added to the regression of y on tern of coefficients across categories of seg- log size.

47 40 WILLIAM T. BIELBY AND JAMES N. BARON regation is unchanged in many instances. in the first two columns include the 21 all- Net of size, segregation is most strongly as- male establishments; accordingly, the vari- sociated with the level of interpersonal skills able measuring organizational sex ratios and the proportion of clerical and sales (proportion women) is excluded from that workers in the enterprise. The former var- discriminant analysis. The other results are iable seems to differentiate moderately in- based on the 363 establishments that em- tegrated establishments engaged in per- ploy both men and women. Standardized sonal services from all others, and the latter weights are computed by scaling both the differentiates all-male establishments, with linear composite function and the organi- virtually no clerical component, from those zational attributes to unit variances. Coe. with a significant fraction of the work force ficients were computed from a canonical engaged in clerical duties. correlation analysis applied to a pairwise- In sum, work settings in our sample ap- deletion correlation matrix-with binary var- proach complete gender segregation, and iables denoting group membership. Cate- descriptive statistics suggest two mecha- gory means on the discriminant function are nisms that may contribute to patterns of seg- metric coefficients for those binary varia- regation: one reflecting the impact of ad- bles. ministrative structures and personnel Organizational scale clearly dominates the procedures that vary with organizational scale results. Log size has a standardized loading and the other pertaining to the occupational of .79 on the first function and is correlated composition and skill level of establishments ..89 with the linear composite/Consequently, that rely primarily on nonmanual tasks typ- the group means convey the same message ically done by women. The aoalyses reas Table 3-4: Scale accounts fir most of the ported below examine 'those mechanisms in association between organizational attributes greater detail. and segregation. The second discriminant function differentiates the 42 moderately integrated en- a'Lfu1firuriate Analyses terprises and the 21 all-male ones from other Standardized coefficients-for discriminant establishments, the overwhelming majority analyses reported in Table 3-5 indicate the of' which have segregation indices between organizational attributes that best differen- 90 and 1(X). The function apparently reflects tiate establishments according to levels of tho impact of institutionalized personnel sex segregation. The first function defines drocedures and occupational composition. the linear composite of organizational attri- Formal bidding procedures for job advance- butes that diflers most among segregation ment, combined with a work force occu- categories, relative to variation within cat- pying specialized, nonentry jobs, appar- egories. The second function extracts an ad- ently inhibit the assignment of men and ditional dimension differentiating among women to the same job classification, in some segregation categories and is uncorrelated cases :acilitating the exclusion of women with the first function. The standardized empl iyees altogether. This can occur when weights index the relative importance of each men and women are assigned to separate attribute in distinguishing among categories of segregation, and the group means locate the segregation categories along etih com- nut met in our data. We are, however, not making posite dimension. 61 Theanalyses reported inferences to a larger population. so significance tests are not appropriate for our analyses. We are reporting descriptive statistics about group differences, waling 1' Discriminant analysis rests on distributional as- levels of segregation to maximize the correlation with sumptions of withingroup multmormality that are clearly a linear combination of organizational attributes.

48 SEX SEGREGATIQN WITHIN ORGANIZATIONS 41

TABLE 3-5Discriminant Analyses Relating Level of Sex Segregation to Organizational Attributes Including Establishments With Excluding Establishments With Only Male Workers Only Male Workers (N = 384) (N = 363) Standardized Weights Standardiied Weights 1st 2nd 1st 2nd Variable. Function Function Function Function Log size .79 .02 .74 .20 (.89)° ( - .01) (.87) (- .02) Log specialization -.33 .36 -.22 -.68 (.48) (- .23) (.52) ( -.37) Core sector .37 -.03 .33 -.06 (.72) (.32) (.72) (.26) Ambiguous sector .07 -.23 .07 -.29 (-.40) ( -.00) (-.34) ( - .20) Log fragmentation .24 -.15 .21 .00 (.66) ( -.04) (.59) (.09) Union/bidding arrangements .00 .53 .14 .32 (.40) (.45) (.42) (.37) Proportion women .35 -.55 (.14) (- .67) Proportion prodUction workers .27 .21 - .14 -.06 (.02) (.59) (.16) (.34) Proportion clerical and sales .15 -.36 .16 - .20 workers (.08) ( .85) (- .11) ( -.30) Proportion PThib workers -.09 - .01 .05 -.34 (- .21) (- .24) (-18) ( - .31) Average complexity: data -.11 -.49 - .27 -.27 ( -.09) (,- .39) (- .18) (- .23) Average complexity: people -.07 .25 - .05 .21 ( .29) ( .30) (- .34) ( .20) Average technical skills .05 .03 .12 -.03 (.03) (.47) (.11) (.35) Skill specificity .06 .49 .08 .23 (- .04) (.35) (.0k) (.28) Segregation Group Means on Discriminant Function Moderate (N = 42) 0.01) 0.00 0.00 0.00 High (N = 40) 1.43 2.23 1.36 1.36 Very hibil = 80) 2.19 2.98 2.37 2.71 Complete (N = 201) 0.19 2.64 - 0.01 2.98 All male (N = 21) -1.76 4.52 Canonical correlation .610 .309 .584 .345 Correlations between composite function and organizational attributes appear in parentheses. Professional, technical, and managerial.

49 42 WILLIAM T. BIELBY AND JAMES N. BARON entry classifications and department-spe- of solidarity and power (Simmel, [1923] 1950; cific seniority systems keep women from Kanter, 1977a), but our analyses provide only transferring into male career lines (Shaeffer partial,support for the relative numbers hy- and Lynton, 1979), Net of these tendencies, pothesis. Since sex composition' defined similar consequences occur when the cler- membership in 1 of the 5 segregation groups ical and sales component is small and few (all-male),it was inappropriate to include workers have complex informational tasks. 17 that item in the discriminant analysis. In other words, establishments in which Nevertheless, the second discriminant func- most workers perform nonmanual taskstend tion should be highly correlated with pro- to have the occupational composition,task portion women if the measures loading highly requirements, nonunion environment, and on it are simply proxiesfor organizational unspecialized job structures that facilitate a sex ratios. Thecorrelation, however, ranges modicum of parity in work arrangements by from .27 and .32 depending on how sex. Indeed, 9 of the 16least segregated missing values are treated, about half the firms have no workers in production occu- size of the correlations between the second pations and 2 others have less than 10 per- function and occupational composition. cent in such roles (see Table 3-6). In con- Therefore, it appears that technical and ad- trast, most all-male establishments are ministrative concomitants of production work engaged in manufacturing activities and em- are more important thanrelative numbers ploy most of their workers in production in differentiating levels of segregation. roles (see Table 3-3). Note that economic On the other hand, sex ratios do figure sector dominates neither discriminantfunc- more prominently in thediscriminant anal- tion. Thus, organizational arrangements as- ysis restricted to the 363 establishmentsthat sociated with sectoral location not sector employ men and womea, According to Table per se affect sex segregation. 3-5, the first dimension remains dominated While the discriminant analyses seem to by organizational scale, but specialization and support institutionalist accounts of sex seg- sex ratios clearly define thesecoi'd factor: regation, a more parsimonious explanation Segregation increases monotonically as or- may account for theresults on the second ganizations become more specialized and less function: It is easier to segregate women dependent on female personnel. Unfortu- when they are a minority of the work force, nately, the 2 discriminant analyses are not regardless of administrative arrangements. directly comparable, since the latter differ- Relative group size is often a crucial basis entiates among segregation levels condi- tional upon a mixed work force, while the distinction between all-male establishments " Since the 4 variables characterizing occupational wand others is prominent in the former anal- composition sum to 100 percent. coefficientsdepend ysis. on which of the 4 is omitted.When the proportion of In short, there apparently is some strength workers in production occupations is omitted, the coef in relative numbers, but that strength can ficients are - .49, .13, and .14 for clerical and be offset by countervailing organizational ar- sales: service; and professional, technical, and mina gerial, respectively. Of course, zero-order correlations ran,;ements: Union contracts and formal between the discriminant function and the measures bidding procedures, positional speciali?a- of occupational composition are not affected by alter- tion of the work force, reliance on firm-spe- native choices for the omitted variable. Percent service cific skills, and manual job tasks facilitate correlates .01 with the second discriminant function. employer strategies that either keep women The second function correlates most highly with the percentage of clerical and sales workers (.65) and the out of the establishment completely or con- proportion of production workers (.59). fine them in segregated job classifications.

50 SEX SEGREGATION WITHIN ORGANIZATIONS 43

segregation index among the other 61 is 75. Deviant Cases: Moderately Desegregated Ten of the 16 establishments in Table 3-6 Establishments have relatively low segregation indices sim- The foregoing analyses discriminate very ply by virtue of having one job title in which, segregated organizations from segregated a few men and women are employed. ones, but we must not lose sight of the fact Two real estate firms listed in Table 3-6 that there is very little variance to explain. have men and women in integrated job titles Less than one-eighth of the establishments but segregate them locationally. The third in our study are even moderately integrated least segregated enterprise in our sample is (using a generous definition of moderation), a real estate management firm whose 23 male and the remaining, highly segregated or- and 126 female apartment managers work ganizations display virtually every possible and live in 149 different buildings, while the configuration of organizational form and en- managers and officers of the escrow service vironment. The completely segregated es- are dispersed across field offices throughout tablishments include huge bureaucracies that a large metropolitan area. In both instances, dominate their industrial environments, as men and women have the same rank and well as small entrepreneurial firms at the may have similar responsibilities, but within economic margins. These enterprises are each of the individual workplaces there is public and private, in manufacturing and perfect or near-perfect segregation. nonmanufacturing; some employ women al- Another source of integrated job classifi- most exclusively, while women have only a cations is the sex-linked practices in dealing token presence in others. Indeed, an un- with clients, accounting for low segre segregated work force is so rare that it is indices in 4 other establishments. In 2 res- worthwhile examining commonalities among idential children's camps and an institution the few aberrant enterprises that do assign providing educational therapy, male coun- men and women to the same job classifi- selors supervise boys and femalt.aunselors cations. supt...vise girls. Another example is a lan- The top panel of Table 3-6 lists organi- guage school, the second least segregated zational attributes of the 16 establishments establishment in our sample, in which "it is with indices of dissimilarity less than 60. deemed necessary that [students] be taught Examining them on a case by case basis re- by both men and women as they are likely vealed some regularities not detected in the to need to communicate with each sex when statistical analyses. First, we discovered that using their language" (Narrative report 1712, small enterprises with just one mixed job 1970). Gender role ideologies have histori- classification typically have indices substan- cally played an important part in creating tially less than 100. That is, small establish- and sustaining inequities against women in ments that employ men and women can ap- the teaching professions (Tyack and Strober, pear moderately desegregated,but that 1981). These educational establishments show desegregation can be nominal. The most ex- that cultural definitions of mea's and wom- treme instances in Table 3-6 are the retail en's work and responsibilities for social- bookstore and pet store, each employing just izing the next generation can demarcate one woman a sales clerk: Theirindices are responsibilities by sex even within detailed 37.5 and 50, respectively. Accordingly, sex organizational poF.tions. segregation appears bimodally distributed Integrated work forces are utilized in sev- among small firms. Of the 234 establish- eral establishments under circumstances that mews with fewer than 50 employees, 173 corroborate neoclassical accounts of sex seg- are completely segregated. But themedian regation (Polachek, 1979). Real estate sales 44 WILLIAM T. B1ELBY AND JAMES N. BARON

TABLE 3-6Occupational Attributes of the Least Segregated and LargestEstablishments

Segregation Job Entry Products or Total Female Employees Index Titles No. Activities Sector° lndustryb Employees Sixteen Least-Segregated Establishments 30 14 12.5 3 1 Citrus P Agriculture 23.2 6 .2 Language schools A SOC 20 9

137 28.1 9 3 Real estate management A SOC 169 6 33.3 5 4 Real estate ^ P SOC 23 36.3 12 5 Real estate escrow A SOC 105 95 37.0 63 6 Real estate A SOC 320 128 9 1 37.5 4 7 Retail books P Trade 5 40.0 5 8 Children's camps P Services 10 42.3 6 9 Tomatoes P Agriculture 41 23 7 50.0 2 10 Real estate P .SOC 9 3 1 50.0 2 11 Retail pets P Trade 8 2 50.0 5 12 Scientific instruments A Manufacturing 57 53.6 35 13 Educational therapy A SOC 98

2 55.6 14 Plumbing supplies A Manufacturing 11 7 57. 1 8 15 Garments A Manufacturing 14 '43 17 57.7 24 16 Children's camps A Services

Eleven Largest Establishments 18 99.5 204 17 Mining and quarrying C Manufacturing 825

268 96.3 60 18 Bakery products C Manufacturing 888 236 100.0 148 19 Printing and publishing C Manufacturing 985 82 99.0 337 20 Sugar refining C Manufacturing 1277

1464 49 96.1 83 21 Thoroughbred racing C Services 967 99.2 129 22 Ordnance C Manufacturing 1727 1384 79.2 346 23 Banking C SOC 2340

59 99.5 83 24 Thoroughbred racing C Services 2845 2987 661 90.0 252 2.5 Airline C sot

369 99.2 78 26 Telephone C SOC 6874

334 100.0 615 27 Naval shipyards C State, 7b25 Manufacturing = periphery; A = ambiguous; C = core. b SOC = social overhead capital.

is a vocation for which reentry costs to em- nation. If market forces, however, account ployees and turnover costs to employers are for the disproportionate share of real estate minimal. Consequently, one would expect firms among the moderately integrated es- an abundantsupply of and demand for qual- tablishments, they cannot account for 4 of ified female workers for these positions. At the other 5 real estate enterprises that com- the same time, real estate sales canbe suf- pletely segregate men and women in other- . ficiently lucrativeespecially in Califor- wise similar market and organizational cir- niato attract males as well. Further,be- cumstances. cause salespersonswork primarily outside Similarly, it is not surprising to find in- the office, there should be fewer costs as- tegrated work forces engaged in harvesting sociated with employee tastes for discrimi- of fruits and vegetables. There is an ample

52 SEX SEGREGATION WITHINORGANIZATIONS. 45

TABLE 3-6(Contir,u?d) Occupational Composition; Percentage in Entry Job TitlesMedian Clerical No, per DOTJob SizeProduction Saks VFW' Service Comments Sixteen Least- Segregated Establishments

1 1.0 13.5 100 0 0 0 Picker-1 iM, 14F. 2 0.9 6.2 0 8 92 0 Language teachers-6M, 6F; department head-3M, 2F. 3 1.1 67.9 0 7 93 0 Apt. maniger-23M, I26F. 4 0.5 6.8 0 76 24 0 Salesmen-13M, 4F. 5 1.8 25.5 0 22 78 0 Escrow manager-4M, 24F; escrow officer-3M, 32F. 6 1.4 59.1 8 82 10 1 Salesman-130M, 80F. 7 1.0 2.2 0 78 22 0 Sales clerk-5M, 1F. 8 1.0 1.8 0 0 70 30 Counselor-3M, 3F. 9 1.0 11.4 100 9 0 0 Worker-I1M, 12F. 10 1 0 4.1 0 94 5 0 Salesman1M, 7F. 11 1.0 1.2 0 67 0 33 Salesperson-1M, 1F. 12 10 10 62 12 25 0 Instrument maker-3M, IF. 13 1.3 2.0 0 17 80 3 Psychologist-3M, 2F; therapist-13M, 25F; camp counselor-4M, 3F. 14 0.8 1.0 57 25 18 0 Assembler -4M, 1Y. 15 1.3 2.0 86 0 14 C Four women in 3 integrated production jobs. 16 1.4 2.0 9 5 70 16 Counselor-11M, 11F.

Eleven Largest Establishments

17 1.9 8.0 79 6 14 1 No women placed in production jobs due to nature of work. ' 18 1.4 61.2 35 54 3 8 19 1.4 18.0 79 13 5 3 Women mostly in clerical and bindery classifications. 20 2.4 NA 75 10 11 4 Eleven women in 3 integrated quality control classifications; none in production. 21 1.3 111.5 15 60 7 18 Women mostly in clerical classifications. 22 1 4 102.6 84 6 8 2 See text. 23 3.2 15.6 0 68 30 2 Company policy to place women in other than routine clerical positions. Several integrated officer and manager ;,,bs. 24 1 3 268 3 8 33 25 34 Women mostly in clerical classifications. 25 NA 111.7 NA NA NA NA Women employed only in clerical, stewardess, and ticket agent classifications. Women employed without restriction except in jobs re- 26 1.6 157.5 74 26 1 quiring lifting 25 pounds or more. 27 2 1 69.2 77 8 15 No women assigned to production classifications due to vigorous requirements of various crafts.

PIN = professional.technical, and managerial occupations.

supply of male and female workers who are of his or our findings, since the sex com- ill prepared for most other types of em- position of jobs was not compiled for the ployment, and employers bear none of the Employment Service's more recent analyses training costs (Thomas, 1980). Neverthe- of agricultural work. less, Thomas's study suggests that the two In sum, no single dimension of desegre- agricultural establishments in our sample may gation emerges 'from our analysis of "de- be atypical. He found that men in lettuce viant" organizations. Instead, we found harvesting are typically assigned to higher qualitatively different but sometimes over- paid piece-rate jobs, while women are con- lapping sets of circumstances that contribute centrated in hourly crews. Unfortunately, to a desegregated workforce: (1) nominal de- we have no evidence ofthe generalizability segregation of a single job title in a very AO.

46 WILLIAM T. BIELBY AND JAMES N. BARON small firm; (2) spatial segregation across work peatedly in both narrative reports and job sites of male and female workers assigned analyses. For example, the Naval shipyard to the same job classification; (3) sex-linked in Table 3.6 assigns no women to production desegregation of jobs like camp counselor, jobs due to "vigorous requirements of var- which employ men and women, but toward ious crafts" (Narrative report 3, 1965), and different ends.; and (4) market desegregation the mining enterprise followed the same that occurs when a mixed labor pool is avail- policy due to the "nature of work" (Narrative able to employers who perCeive that train- report 1800, 1971):' A company providing ing and turnover _costs are identical for male telephone service to a large metropolitan and female workers. Moderate desegrega- area states that it was company policy to tion occurs rarely in the work contexts we employ women "without restriction" except have examined, and, when it does occur, it in jobs requiring lifting 25 pounds or more;... is typically of the nominal variety. yet the segregation index was 99.2 (Narra- tive report 100, 1965). The ordnanc.6 plant, studied in 1970, em- lArge Establishments and Bureaucratic ployed 555 females as assemblers and 243 Sel;regation males as production workers. Each is an en- Almost all large establishments are highly try job, and both were mapped to the same segregated, and most have written rules detailed DOT occupational category by the governing the hiring and allacation of work- Employment Service analyst.18 According ers. Consequently, if bureaucratic control to the job analysis, these jobs differ pri- strategies segregate men from women (Ed- marily in that male "workers" lift 25 to 40 wards, 1979), this should be most apparent pounds, but female "assemblers" lift 5 to 20 in the largest establishments. The bottom pounds. panel of Table 3-6 lists attributes of the 11 Weight restrictions are not mentioned ex- establishments in our sample employing more plicitly in the narrative for the printing es- than 800 workers. Case materials provide tablishment in Table 3-6 (one of the largest insight into administrative roles and pro- in the western U.S. in 1968), but as in nearly cedures that support sex segregation. all other manufacturing plants, the only pro- All but one of these establishments seg- duction activities assigned to women are light regate employees almost perfectly by sex, assembly tasks. Most production tasks in this and narrative information available for 7 of establishment are done by skilled crafts- them suggests that this total segregation is men. The union contract establishes pro- accomplished largely through bureaucratic cedures for hiring and apprenticeship, and rules and procedures. Possibly the most im- it seems reasonable to conclude that the union portant factor, particularly in manufacturing plays a substantial role in enforcing sex set; establishments. is the existence of legal re- regation in this plant. A narrative report strictions on the weight that women may lift. prepared in 1966 one of the two race California law specified until 1970 that "no tracks provides evidence of that role in an- female employee should be requested or other organizational context: Union dis- permitted tolift any object weighing 50 patching policies explicitly exclude women pounds or over," and regulations enforced from the job of parimutuel clerk (Narrative by the state'.i Industrial Welfare Commis- report 1536). No report was prepared for sion further restricted the maximum to 25 the other track in the same area, but pre- pounds. In 1970 a federal court ruled that this law conflicted with Title VII of the Civil Rights Act (Utility Workers' Union of Amer- '8 in other words, at the occupational level, entry- ica vs. Southern California Edison [69-543]). level production work has a relatively balanced sex ratio References to these restrictions occur rein this firm.

54 SEX SEGREGATIONWITHINORGANIZATIONS 47

sumably it falls under the jurisdiction of the in a short period of time. This is especially same union. Of course,these industries might true of large bureaucracies employing many have been sexL,;egregated long before they workers in nonentry jobs. As long as sen- were unionized. That is,unions may be per- iority and accrued skills remain important petuating gender-based inequalitiesrather bases for advancement in such contexts, than creating them. workplace equity cannot happen overnight: Lifting restrictions and union contracts The existing stratification regime favoring cannot account fbi all sex segregation inlarge males essentially guarantees more work- establishments, because segregation is place inequality in the short run, just as equally pervasive in large, nonunion estab- reduction of childbearing to below replace- lishments outside manufacturing. Indeed, ment levels would notimmediately elimi- in most firms including all but 1 of the nate population growth in a society. ,11 in the bottom panel of Table 3-6 0 non- This particular bank's egalitarian policies production jobs are as segregated as. pro- toward women seem to have overcome bu- duction jobs, if not more so. While this sit- reaucraticinertia because they were imple- uation primarily reflects distinctionsbf.tween mented during a time of extraordinary or- male managers and female clerical workers, ganizational growth and change. When it is also true that the few male clericalwork- analyzed in 1968, the bank was described ers and female managers inour sample of as one of the "largestand fastest growing establishments are hardly ever assigned to business concerns in the nation" (Narrative job classifications with workers of the op- report 415, 1968). Employmentincreased posite sex. nearly 50 percent between 1961 and 1968, One notable exception is the bankstudied when the bank was automating its data pro- in 1968, which employs 'tore womenthan cessing operation and establishingregional any other establishment in oursample. The offices and branch banks throughout south- bank claimed it had recently initiated a pro- ern California.Growth and technological gram to hire and promotewomen into ofchange appear to be directly responsible for ficer classifications (Narrative report 415, the desegregation of' several management, 1968). The staffing schedule supportsthis administrative, and data processing job clas- claim: In .1968, women comprised 7 of 81 sifications in this firm. Its atypical experi- vice presidents, 18 of 108 assistant vice pres-- ence demonstrates thatthe segregation ob- idents, 16 of 49 management trainees, and served in other large establishments is not 29 of 118 operations officers. Whilefemales inevitable. 19 These jobs, however, may have were used exclusively in routinedata pro - cessing jobs like keypunch operator, 10 of Other data gathered in California by the Employ- 23 systems analysts were women. ment Service corroborate this. They studiedabout 30 In one sense the bank's efforts areonly other establishments with more than MOO employees noteworthy when contrasted againstthe' not included in our analysis because of incomplete ov- uniformly high levels of segregation in other rag of sonic aspect of their operations.Of these, only comparable large establishments: Fully :3 were moderately desegregated. a university campus (A = 70 5), a unified school district (A = 76.5), and equalizing the job distribution by sex would an insurance company (A 77.2) Among large bu- still require reclassifying 80 percent of this reaucracies, assignment of both sexes to the samejob I)ank'!; female employees. Most managerial titles occurs most often in social overhead capital or- and professional positions remained exclu- ganizations --firms in health, education, and welfare sively male, while few men were employed srrViCeS. transpoaation, utilities, finance, insurance, and real estate industries. Hut the banking and finance in routine' clerical duties. Anorganization's industry is not unifOrml!, desegregated Of the 5 other demography. history, technology, and labor establishlts in our sample engaged in such activi- ipply. however, constrain the degree to ties. 3 were compltelsegregated, I was nearly so (A whit h its work fin-c composition canchange 03 3). and 1 was modrateb, integrated (A 73 71

r- L.J J 48 WILLIAM T. BIELBY AgD JAMES N. BARON

TABLE 3-7Descriptive Statistics for Longitudinal Sample (N = 75) Time 1 Time 2

;1 Standard Standard Variable Mean Deviation Mean Deviation Log size 4.26 1.21 4.38 1.21 g specialization 1.37 0.99 1.48 1.03 Manufacturing industry .79 Core sector .61 . Ambiguous sector .32 Pe: iphery sector .07 Log fragmentation .18 .20 .17 .19 Union/bidding arrangements .37 .39 Proportion women .29 .25 .29 .24 Proportion production workers .64 .30 .62 .29 Proportion clerical and sales workers .16 .17 .18 .18 Proportion service workers .08 .22 ,.22 Proportion VFW workers .10 .12 " .11 .11 Professional, technical, and managerial. been integrated precisely because they were lishments selected for restudy were slightly new; it may take several years for the sex larger on average and more concentrated in label of a new line' of work to become es- manufactuting; other differences between tablished. descriptive statistics for these observations and those for the entire sample reflect con- Longitudinal Analyses: The Permanence of comitants of organizational scale (cf. Tables Sex Segregation 3-1 and 3-7). In most instances, the temporal Seventy-five of the 393 establishments in data describe changes between the mid-1960s our sample were studied more than once. and the early 1970s the period immedi- The average interval between visits was about ately following passage of Title VII of the 5 years, with a range of 2 to 12 years. The Civil Rights Act of 1964. Most establish- size composition of the follow-up sample is ments expanded employment between anal- very consistent with the age and size-spe- yses; the labor force was stable in just 6 cases cific establishment mortality rates reported and was reduced in 25 establishments. by Birch (1979). While moderately large, As Table 3-8 shows, neither legislation older enterprises were slightly more likely nor organizational change effected much to be revisited, we detected no other sys- change in sex segregation in the late 1960s tematic biases in the Employment Service's and early 1970s. Two-thirds of the estab- choice of establishments for follow-up anal- lishments remained all-male, completely yses (fir details see Baron, 1982, Chapter segregated,r almost fully segregated (see VII). the 9 cells ine bottom-right corner of the Descriptive statistics for the 75 establish- table). Indeemuch of the change in Table ments are reported in Table 3-7.20 Estab- 3-8 reflects very small differences in segregation indices. Table 3-9 lists characteristics of the 18 2" Several estabitsh.nents were studied more than twice establishments for which the segregation in- by the Employment Service. In those instances, we dex changed by at least 5 points. Only a few selected the pair for which the interval between analyses Was closest to 5 years. Consequently, analyses for attributes differed systematically between 16 of the establishments in the longitudinal analyses the 11 organizations that became less seg- do not include the one selected for the cross-sectional regated, between the 7 that became more sample of 393 observations. segregated, and between the 57 in which

56 SEX SEGREGATION WITHIN ORGANIZATIONS 49

TABLE Segreption Levels Over Time (N = 75) Segregation: Time 2 le Vety All Segregation: Time. 1 Mod High High Complete Male Total-

Moderate(A s 75) 1 1 (1%) High(75 < A <90) 3 2 2 2 9 (12%) Veryhigh(90

All male 1 1 2 (3%) TOTAL 5 10 23 34 3 75 (7%) (13%) (29%) (47%) (4%) (100%) sex segregation remained virtually constant. reels of paper" in the other production work Unionized establishments and those with (Narrative report 1476, 1970). Only a man- formal bidding arrangements tended not to ufacturer of kitchen ranges displays a delib- change from their high levels of segregation, erate effort to desegregate its work force. nor did those in Mich women comprised a Only 2 women were employed in produc- small minority of the work force. Organi- tion departments in 1966. As in the case of zations that partially desegregated were typ- the bank described earlier, organizational ically in small-scale manufacturing or were growth appears to have facilitated gender providers of personal and social services. parity. By 1970 the work force of this man- Most declines in segregation are attrib- ufacturing firm had more than doubled, and utable to a nominal change in the compo- women worked alongside men in 14 entry- sition of 1 or 2 jobs rather than a major level jobs in the assembly department, re- change in personnel practices. For example, ducing the segregation index from 99.5 to 2 male checkers were hired by a large-chain 78.8. Nevertheless, as in the bank, employ- supermarket, a male operator was hired by ment in several other noriadministrative de- a firm providing mobile telephone service, partments remained restricted to males in and a woman was hired to work as a lens 1970. finisher in a firm manufacturing contact len- The 7 establishments that became more ses, The "integrated" title of thip girl in the segregated over time were all h4hly seg- poker cardroom is completely segregated by regated initially, and typically the, segrega- shift; apparently, only males work in the tion of only 1 job produced most or all of early morning hours.21 A company t*..aan- the change. In several instances, increased ufitctured silkscreened wall cove) hired segregation took place in the context of or- 4 female inspectors, but continued to dis- ganizational retrenchment or consolidation. criminate statistically, hiring women only as But in no case was there any evidence that inspectors, paint mixers, and clericals "due increased segregation resulted from delib- to otx.asional job requirements of lifting heavy erate employer strategy to manipulate the sex composition of jobs. Given the high initial levels of segregation, the small increases 21 The apparent integration of the chip girl position nia sinipl reflect ,t pographical error on the 1970 in the index could easily be attributable to staffing scheduleIf so. the establishment Was perfect lv random perturbations in labor supply and segregated in 196ii and 1970. demand. 50 WILLIAM T. BIELBY AND JAMES N. BARON

TABLE 3-9Attributes of Establishments With Increasing or DecreasingLevels of Sex Segreiaon

Number of Number of Segregation Employees Females Index Entry Year Tl-T2 T1 -T2 T1 -T2 No. Products or Activities T1 -T2 Sector° Industry" Establishments With Decreasing Segregation Trade 56 -49 16-12 100.0-94:6 1 Supermarktts 1961-1967 A

Manufacturing 42-44 18-21 77.8-69.6 2 Optical goods 1965-1968 C SOC 149 -145 124-118 96.0 -86.4 3 Mental health facilities 1966-1970 A

Manufacturing 139-142 12-26 96.1-85.8 4 Wall coverings 1965-1970 C 15-30 13-21 100.0-88.9 5 Mobile telephone service 1966-1971 P SOC 20-35 3-11 100.0-87.5 6 Clay products 1964-1971 A Manufacturing 149-190 97-66 100,0-87.1 7 Poker cardrooms 1966-1970 A Services 9-10 3-3 1tn0-85.7 S Optical goods 1965-1968 A Manufacturing 198-438 8-32 99.5-78.8 9 kitchen ranges 1966-1970 C Manufacturing

11-11 10-6 90.0-63.3 10 Wigs 1965-1971 A Manufacturing 27-54 6-17 95.2-67.9 11 Wall coverings 1965-1970 C Manufacturing

Establishments With increasing Segregation 75-58 29-33 86.6-92.0 12 Truphies 1961-1971 C Manufactaing

Manufacturing 95-49 20-11 84.0-89.5 13 Electronic instruments 1967-1970 C Manufacturing 309-535 137-294 92.7-99.2 14 Potter) 1960-1968 C Manufacturing 164-173 10-4 89.0-95.9 15 Musical instruments 1960-1970 C Manufacturing , 21 -42 9-18 91.7 -100,0 16 Ceramic tiles 1971-1977 A SOC 60-55 42-21 80.1-100.0 17 Securities exchange 1960-1968 A 25 26 16-17 77,8 -100.0 18 Garments 1960-1970 A Manufacturing

" P peripheo. A -z ambiguous, C = core. bas.n.SOC six.3a1 overhead capital.

In sum, changes in organizational limns political and legal climate nor an influx of and environments and shifting labor supply women workers in the late1960s affected and demand had little effect on sex segre- definitions of women's work within this lam- gation in the late 1960s and early1970s.22 * of work organizations, especially in large, In most of the establishments weexamined, unionized establishments with institution- the consistently high levels of segregation alized procedures governing hiring and ad- are probablydue to long-standing policies vancement. for hiring and allocating workers,perhaps reflecting industrywide practices predating DISC 1[5,S1ON the establishment itself. Neither achanging In most establishments, kw jobclassifi- cations are staffed by both men and women. 22 tiintteen establishments were visited 3 or more Indeed, complete segregation was the norm times the' Emplo)ment Service'Seven were per- in establishments studied inCalifornia by fetb segregated at each visit, and segregation de creased between analyses in only I establishment. In the U.S. Employment Service between 1959 short, there is no evidence of a secular trend injob and 1919, and segregation levels were vir- segregation by sex between 1959 and 1979. tually constant in these organizations during

58 SEX SEGREGATION WITHIN ORGANIZATIONS 51

TABLE 3-9(Continued) It segregation due mostly to one title:

T1 T2 Number Number Entry Numberof Numberof No. Title of MalesFemalesof MalesFemales Comments Establishments With Decreasing Segregation

1 Checker 0 10 2 5 Managerial promotion track for box boy but not for checker. 2 Several Integration of 2 lens grinding Jobs. 3 Several Integration of kitchen helper (initially male) and head nurse (initially female) titles. 4 Several _ - - - Integration of 2 inspector titles (initially male). 5 Operator 0 10 1 9 6 Press operator 2 0 3 2 7 Chip girl 0 35 16 18' Chip girl title segregated by shifts., 8 Finisher 2 0 1 1 9 Several _ - - - Desegregated 14 of 19 jobs in assembly department; 7 of 8 other departments completely segregated. 10 Ventilator 0 3 1 3 11 Several Integration of I nonproduction and 3 production Jobs; 23 of 28 jobs still segregated in 1970.

Establishments With Increasing Segregation

12 Several - - Discontinued jewelry manufacturing, eliminating 2 integrated jobs. 13 Assembler 10 13 3 8 Decline in production between 1967 and 1970. 14 Packer 5 5 0 ,8 15 Assembler 3 3 5 0 16 Kiln loader 1 1 8 0 Kiln loader only integrated job in 1971. 17 Tabulation clerk 6 5 0 0 Data processing operation consolidated as separate subsidiary between 1960 and 1968.

18 Sewing machine 1 3 0 3 Administrative jobs not included in staffing schedule for operator 1970.

the late 1960s' and early 1970s. When bal- Institutional arrangements that prevail in anced sex ratios did occur, they almost al- core firms rather than sectoral location per ways reflected just 1 or 2 integrated job titles se shape sex segregation, aid they are within an establishment. most visible in the largest establishments in Multivariate analyses revealed that or- our sample. Many had sex-specificpolicies ganizational scale'is strongly associated with for allocating workers, oftentimes sanc- levels of sex segregation. Small enterprises tioned by collective bargaining agreements. are either completelysegregated or trivially In manufacturing establishments, legal re- desegregated; larger enterprises tend to have strictions on physical demands of women's almost all workers in segregated jobs. Union work typically rationalized complete sex seg- contracts and formal bidding procedures, regation among production jobs, even when positional specialization, reliance on trained lifting was only required infrequently. Estab- nonentry personnel, and manual job tasks lishments in the economic core also specialize facilitate a division of labor segregated by skills and fragment job tasks more than other sex, while women are lessextensively seg- enterprises, and these aspects of bureaucracy regated when they comprise more than a are not gender-neutral. small minority of the work force. Of course, most women are not employed

59 52 \ WILLIAM T.BIELBY AND JAMES N. BARON in such manufacturing contexts. Less than were no less segregated than others (see also one-quarter of all women workers are in Salancik, 1979). manufacturing enterprises with 50 or more Nevertheless, we can point to aim; ex- employees, and one-half the females in non- . traordinary circumstances under which de- manufacturing industries work in establish- segregation did occur, As noted above, one ments with fewer than 50 employees.° Even bank and a manufacturing enterprise inte- small organizations outside of manufactur- grated a number of nonclerical jobs during ing, however, are highly segregated. Of the a time of rapid growth or technological change 100 nonmanufacturing establishments in our or both, before there was any substantial sample having fewer than 50 employees, 73 government pressure to change personnel were perfectly segregated, and women em- policies regarding minorities. Short of these ployed in those establishments are more likely fortuitous and idiosyncratic circumstances, to be segregated through the patriarchical large and systematic reductions in gender actions of an entrepreneur or supervisor than segregation seem unlikely to occur in the by institutionalized buieaucratic rules and absence of fundamental shocks to the social procedures. In short, there is equifinality in system. For example, during World War II, sex segregation of the workplace. Personal employers faced extreme labor shortages. and patriarchal control prevails in small es- California's Industrial Welfare Commission tablishments, bureaucratic structures seg- granted 60 permits that exempted women regate men and women in large nonmanu- from its 25-pound restriction on lifting, al- facturing enterprises, and technical control lowing 4,539 women workers to enter pro- excludes women from many production jobs duction jobs formerly closed to them.25 Thus, within the manufacturing sector. unusual extraorganizational circumstances Do these findings imply that sex segre- forced employers to reject long-standing gation is immutable? We detected little practices based ostensibly on physiological change in segregation indices, even among differences between the sexes. Neverthe- firms analyzed after court decisions in 1971 less, after the war these women were de- struck down California's restrictions on mobilized as rapidly as they had been in- women's lifting.'`' In 1968, affirmative action tegrated into the work force. The procedures goals and timetables were required of firms for obtaining exemptions from the state la- holding federal contracts, yet establish- bor code remained in effect but were barely ments in our sample covered by this order utilized, and until 1971 the code rationalized sex segregation of production jobs in manufacturing. To the extent that our results are gener- 23 Figures are based on the 1967 Technological Ad- alizable, two policy implications of our study vance in an Expanding Economy survey by the Uni- versity of Michigan's Institute forSocial Research are clear.First, sex segregation will un- (Mueller et al.. 1969). Of 766 women working outside doubtedly persist if policy makers adopt a the home, 560 were employed in nonnianufacturing laissez faire stance. Neither demographic industries, and 282 of those were in establishments trends, nor technological changes, nor bu- with fewer than 50 employees. Since then, the female work force has become even less concentrated in man- reaucratic imperatives are natural forces that ufacturing, although there may have been a shift to- lead to balanced sex ratios within jobs or ward employment in larger establishments. firms. Second, policy intervention is un- 24 Of 46 establishments analyzed between 1972 and 1979, 26 were completely segregated. Eleven had segregation indices less than 75, but 5 were real estate enterprises studied by the Employment Service in 1973. 2' Personal communication, Ni ar gar et Miller, Exec- Levels of segregation remained uniformly high in man- utive Officer, California Industrial Welfare Commis- ufacturing establishments studied after 1971. sion. SEX SEGREGATION WITHIN ORGANIZATIONS 53 likely to make matters worse most estab- tingent of minority employees within an or- lishments are about as segregated as they ganization even at the lowest ranks may can possibly be. Whilethere may be polit- facilitate\cfforts, to desegregate work. ical or economic incentives for focusing on Efforts tc impose workplace equity in the certain kinds of work contexts, policyefforts absence of such a constituency might ac- could justifiably be directed at almost any tually backfire. "Tokees" or "solos" can so- area of economic activity, sincealmost all lidify resistance by male workers and de- establishments are equally segregated. meralize those who should champion and Our findings also suggest some strategies benefit from equal employment opportunity for attacking sex segregation. First, policies programs (Kanter,1977a; Northcraft and that segregate men and women are most Martin, 1981). This underscores once again visible in large organizations. They are doc- the limits of laissez faire approaches, since umented in written contracts, rules, job def- nearly all of the natural desegregation we initions, and procedures; they do not exist observed in our sample involved tokenism. solely in subjective tastes of employers, em- Our recommendations are based on statis- ployees, and clients. Large firms are often tical associations rather than examinations of subject to public scrutiny, and their greater specific policy interventions. Nevertheless, dependence on government contracts and recent surveys of attempts to reduce sex seg- regulatio.is makes them susceptible to pol- regation do document the effectiveness of the icy interventions (Salancik, 1979). They are organizational strategies we have suggested also more likely to have slack resources with Shaeffer and Lynton, 1979; O'Farrell and which to absorb the costs of social change. Harlan, 1982). One conclusion cannot be dis- In short, it seems reasonable first to redress puted: Doing nothing guarantees persistent visible and easily identifiable mechanisms sex segregation in allareal of economic activ- of sex segregation in organizations that are ity. Although not the focus of this study, we vulnerable to outside pressures, even if only repeatedly encountered instances of sex seg- a fraction of all womenwork in such estab- regation of jobs leading to gender-specific pro- lishments. motion lines an orderly progression through Second, interventions seem more likely jobs of successively greater authority and re- to succeed in organizations thatalready have sponsibility for men and dead-end careers for a sizable female workforce. Sex ratios affect women. The pervasive sex segregation across the balance of power among organizational organizational and institutional contexts that constituents (Kanter, 1977a), and our results our study hasdocumented almost certainly show that, as women comprise a larger per- accounts for the substantial sex differences re- centage of an organization's labor force, em- vealed by individual-level analyses of work ployers seem less likely to segregate them. inequalities (e.g., Baron and Bielby, 1982; Changing the sex composition of jobs will Bielby and Baron, 1982; Wolf and Fligstein, require modifying organizations'rules for 1979). advancement through internal labor mar- The degree and persistence of sex seg- kets. Such changes are easier to accomplish regation leave us somewhat pessimistic about when female workers command firm-spe- prospects for rapid or extensive change. On cific experience. In short, segregation seems the other hand, by documenting the impact likely to persist in the absence of severe of organizational structures and dynamics on external pressures on the organization. Fur- the sexual division of labor and by un- thermore, the technical and political viabil- derscoring the pervasiveness of itender seg- ity of efforts to abate the sexual division of regation we hope to facilitate more in- labor depends on the existence of a constit- formed researchefforts and policyinter- uency inside. The presenceof a visible conventions in the future.

61 WILLIAM T. BIELBY AND AMES N. BARON

prospects." Journal of Human Resources 14 ACKNOWLEDGMENTS (2):197-210. Blau, Francine, and Carol Jusenius The authors were supported in part by a 1976 "Economists' approaches to sex segregation in grant from the National Science Foundation the labor market: an appraisal." Pp. 181-99 in (SES 79-24905). We gratefully acknowledge Martha Blaxall and Barbara Reagan (eds.), Women and the Workplace: The Implications the cooperation an .1 assistance of the Cali- of Occupational Segregation. Chicago: Uni- fornia Occupational Analysis Field Center. versity of Chicago Press. Our thanks go to Isabelle M. Al lain, Hanle Braverman, Harry Anderson, and Kelly Moreno who assisted 1974 Labor and Monopoly Capital. New York: Monthly Review Press. in this research, and to Trina Miller who Bridges, William, and Richard Berk processed many versions of this manuscript. 1974 "Determinants of white-collar income: an eval- We are indebted to our many colleagues uation of equal pay for equal work." Social Sci- who have infiLenced our thinking on sex ence Research 3:211-34. Doeringer, Peter, and Michael Piore segregation and have commented on earlier 1971 Internal Labor Markets and Manpower Anal- versions of this report, especially to Donald ysis. Lexington, Mass.: D. C. Heath. J. Treiman and Karen 0. Mason for their .Duncan, Otis, and Beverly Duncan 1955 "A methodological analysis of segregation in- detailed suggestions. dexes." American Sociological Review 20:210- 17. Edwards, Richard REFERENCES 1979 Contested Terrain. The Transformation of the Workplace in America, New York: Basic Books. Arrow, Kenneth 1973 The theory of discrimination," Pp. 3-33 in Or- England, Paula ley Ashenfelter and Albert Rees (eds.), Dis- 1981a "Assessing trends in occupational sex segre- crimination in Labor Markets. Princeton, N.J.: gation, 1900-1976." Pp. 273-95 in Ivar Berg Princeton University Press. (ed.), Sociological Perspectives on Labor Mar- kets. New York: Academic Press. Averitt, Robert 1968 The Dual Economy. New York: Hot: 1981b"Explanations of occupational sex segregation: an interdisciplinary review."Unpublished Baron, James manuscript, University of Texas at Dallas,(Jan- 1982 Economic Segmentation and the Organization of Work. Unpublished Ph.D.. dissertation, uary). University of California, Santa Barbara. 1982 "The failure of human capital theory to explain occupational sex segregation." Journal of Hu- Baron, James. and William Bielby man Resources 17:358-70. 1982 "Workers and machines: dimensions and determinants of technical relations in the work- Gibbs, Jack, and Dudley Poston, Jr. place." American Sociological Review 47:175 - 1975 "The division of labor: conceptualization and 188. related measuresSocial Forces 53:468-76. Bielby, Denise Gross, Edward 1978 "(:artier sex-atypicality and career involvement 1968 "Plus ca change . . ? The sexual structure of of college educated women: baseline evidence twciipations over time." Social Problems 16:198- from the 1960's," Sociology of Education 51 208. (January):7 -28. Halaby, Charles Bilby, William, and James Baron 1979 "Job-specific sex differences in er.fanizational 1982 "Organizations, technology, and worker attach- reward attainment: wage thscrir.imati.in vs. rank ment to the firm." in Donald J. Treimati and segregation." Social Forte', 58 (Stl.tember; Robert V. Robinson (eds.), Research in Social 108-127. Stratification and Mobility, Vol. II. Greenwich, Kanter, Rosabeth Conn.. JAI Press. 1977a "Some effects of proportions on group life: Birch, David skewed sex ratios and responses to token 1979 The Job Generation Process. Cambeidge, Mass.: women." American Journal of Sociology 82:965- MIT Program on Neighborhood and Regional 90 Change. I977bMen and kVornen of the Colimation. New York Blau. Francine Harper & Row. 1977 Equal Pay in the Office. Lexington, Mass. Marini, Margaret I,exington Books. 1980 "Sex differences in the process of occupational Blau. Francine. and Wallace Hendrick, attainment a closer look." Social Science Re- 1978 -Occupational segregation by sex. trends and search 9:307-61.

62 SEX SEGREGATION WITHIN ORGANIZATIONS 55

ft 1978 "Sex differences in the determination of ado- Job Opportui'lities. New York: The Conference lescent aspirations. a review of research." Sex Board. Roles 4:723-53. Simmel, George Milkman, Ruth 1950 "Quantitative aspects of the group." Pp. 87- 1980 "Organizing the sexual division of labor: his- 177 in Kurt H. Wolff (ed.), The Sociology of torical perspectives on women's work and the George Simmel. New York: Free Press. Orig- American labor movement." Socialist Review inally published in 1923. 49:95-150. Thomas, Robert Miller, Ann K., Donald J. Treiman, Pamela S. Cain, 1980 "The social organization of industrial agricul- and Patricia A. Roos ture." Paper presented at the 1980 meeting of 1980 Work, Jobs, and Occupations: A Critical Re- the American Sociological Association, New view of the Dictionary of Occupational Titles.7 York. Washington D.C.: National Academy Press, Treiman, Donald J., and Heidi I. Hartmann (eds.) Mueller, Eva, Judith Hybels, Jay Schmiedeslcamp, John 1981 Women, Work and Wages: Equal Pay for Jobs Sonquist, and Charles Staelin of Equal Value. Washington, D.C.: National 1969 Technological Advance in an Expanding Econ- Academy Press. omy. Ann Arbor, Mich.: Institute for Social Tyack, David, and Myra Strober Research, University of MiChigan. 1981 "Jobs and gender: a history of the structuring Northcraft, Gregory, and Joanne Martin of educational employment by sex." Pp. 131- 19?' "Double Jeopardy: why some women and mi- 52 in Patricia Schmuck and W. W. Charters norities objet to affirmative action programs." (eds.), Educational Policy and Management: _Research paper 599, Graduate School of Busi- Sex Differentials. New York: Academic Press. ness, Stanford University. U.S. Bureau of the Census O'Farrell, Brigid, and Sharon Harlan 1976 County Business Patterns, United States Sum- 1982 "Job integration strategies: today's programs mary. Washington, D.C.: U.S. Government and tomorrow's needs." Paper presented at the Printing Office. National Research Council's Workshop on Job U.S. Department of Labor Segregation by Sex, May, Washington, D.C. 1972 Handbook for Analyzing Jobs. Manpower Polachek, Solomon Admihistration.°Washington, D.C.: U.S. Gov- 1979 "Occupational segregation among women: the- ernment Printing Office. ory, evidence, and a prognosis." Pp. 137-57 in Williams, Gregory Cynthia B. Lloyd, Emily S. Andrews, and Cur- 1979 "The changing U.S. labor force and occupa- tis L. Gilroy (eds.), Women in the Labor Mar- tional differentiation by ." Demography ket. New York: Columbia University Press. 16:73-88. Salancik, Gerald Wolf, Wendy C., and Neil D. Fligstein 1979 "Interorganizational dependence and respon- 1979 "Sex and authority in the workplace: the causes siveness to affirmative action: the case of women of sexual inequality." American Sociological and defense contractors." Academy of Man- Review 44:232-52. agement Journal 22:375-94. Shaeffer, Ruth Gilbert, and Edith F. Lynton 1979 Corporate Experiences in Improving Women's

63 Olo

A Job Changingand Occupational Sex Segregation: Sex and Race Comparisons

RACHEL A. ROSENFELD

The U.S. occupational structure is and occupations with different sex compositions has been extremely sex segregated (see and will predict movement to or from sex- Beller, in this volume). The extent and sta- atypical occupations using a range of indi- bility of such sex segregation prevents most vidual and job history variables. individuals from considering possible mobility between sex-typical and atypical oc- EXPLANATIONS OF SEX SEGREGATION cupations. Such mobility does, however, occur. World War II was adramatic example Explanations for why women end up in of a situation where many women changed typically and predominantly female occu- from typically female to typically male jobs. pations vary depending on whether the ex- Even under less extreme circumstances, the planations focus on labor supply or labor sex composition of aperson's occupation is demand. Both types of explanations usually not a constant throughout one's worklife, fail to consider that a person might break as will be shown hereand as others have through the sex segregation barrier. demonstrated (Wolf and Rosenfeld, 1978; England, 1982b; Corcoran et al., in this vol- Labor Supply Explanations ume; Jusenius, 1975;Sociology of Work and Occupations, vol. 9, number 3, 1982). Using Labor supply arguments about sex segre- 1973 data on job changers, this paper will gation often give the impression that a person focus on change in occupational sex com- makes a once-in-a-lifetime and usually sex- position that people experience with a change typical occupational choice. Socialization ex- in employer. It will describe the movement planations, one type of supply side argument, by black and white women and men' among suggest that women are socialized to plan for

There are, of course, other ethnic groups whose ever, identifies race/ethnicity only as white,black, and occupational sex segregation would he interesting and other. The other group is extremely small and probably useful to study, One might like, for example, to follow quite heterogeneous, In what follows, those identified Malveaux's (1982) lead and look at the position of His- as other are dropped from the analysis, leaving win- panics and their mobility. The data set used here,how- parisons between blacks and whites. 56

64 OB CHANGING ANDOCCUPATIONAL SEX SEGREGATION 57 and enter occupations that society considers who does not take clerical courses in high appropriate for women, while men are so- school or vocational school may not be able cialized to choose from a wider range of suit- to 'get a secretarial job. An early choice of ably male occupations. Consistent with this training thus determines a person's later oc- explanation is the evidence of students' early cupational career. expectations of and aspirations for sex-typical occupations (see Marini and Brinton, in this Labor Demand Explanations volume, for a review of this literature). Hu- man capital explanations of sexsegregation, Labor demand arguments se that the ex- another supply side argument, see the choice clusion of women from traditionally male jobs, and preparation for a sex-typical occupation especially the ones that supposedly require as part of economically rational planning. Since continuous commitment, is largely a.result of women expect to take time out from thelabor employers' and male workers' preferences, not force for full-tine work at home, they train women's choices. The internal labor markets' for and later enter occupations that offer easy literature hypothesizes that on some career reentry and low depreciation of their skills and ladders that are protected from outside mar- training while they are out of the labor force. kets and for which workers receive on-the- Such occupationr, of course, become predom- job training, employers are unwilling to take inantly female (Polachek, 1979, 1981a; Op- a chance on losing their training investment penheimer, 1970). (See the detailed discus- by hiring members of high-risk groups; that sion of this approach and evidence contrary is, members of groups known to be unstable to its assumptions and hypotheses in Corcoran workers (e.g., Doeringer and Piore, 1971; Ed- et al., in this volume.) wards, 1979). Because of their propensity to Suppl! side arguments carry with them as- marry and bear and raise children, women sumptions about people's education and train- are considered one of these groups. Onthe ing. Much formal training for work careers basis of their group characteristics, then, all takes place early in a person's life, often before women might be screened from certain male one starts full-time workbecause in part, as jobs by what is known as statistical discrimi- the human capitalist explains, early training nation (Phelps, 1972).2 Employers may also provides a longer time over which to receive the returns to this training. Women's occupations are not necessarily low-skilled, but 2 It is not necessary for the beliefs about certain groups they are described as occupations in which a to be true for statistical discrimination to occur (Spence, woman must bring her training with her to 1974; see also England, 1982a). If the supply of appropriate labor is great enough, and in the absence of other the job because her expected short tenure pressures, the employer will not be hurt even if the does not allow enough time to receive returns stereotypes about groups are objectively incorrect. At on her training there(Oppenheimer, 1970). times, employers nave even created the link between An early choice of a typically female occu- women's domestic roles and intermittent labor force pation, therefore, must be to some extent a participation, reinforcing stereotypes about women as workers. It is not that women have always chosen to choice about the type of training to get or the leave the labor market when they marry or have chil- type of major to take in college (Polacheck, dren; it has also been the case that employers required 1978). Training for a typically female job might that women who marry or have children leave their preclude training for a typically male job; for jobs. For example, secretaries, teachers, and flight at- example, when a woman decides or is advised tendantsthe latter as late as 1972have been re- to attend nursing school rather than medical quired to leave their jobs when they marry (Davies 1975; Cohn, 1982; see also Cook and Hayashi. 1980, school. Thus prepared, the woman lacks the on forces in contemporary Japanese firms pushing or credentials to enter a medical occupation encouraging women to leave the labor force at marriage atypical for her sex. Likewise, a young man or childbirth).

C5 58 RACHEL A. ROSENFELD

feel that on-the-job training is easier if the change over a person's work life, and 3) work force is homogeneous. Even for low- consider changes in demand for certain Icir,is skilled jobs for whicha long-term commit- of labor. ment is not expected, however, employers may hire on the basis of the sex appropriate- Occupations and Sex Stereotyping ness of the applicant to the sex type of the .job (Levinson, 1975). The contrast often made is betweentyp- In addition, male workers in competition ically female white-collar occupations and with female workers have excludedwomen typically male professions and crafts. One is from predominantly male training programs led to forget the variationamong typically and jobs (see Baker, 1964; Hartmann, 1976). male jobs and among typically female jobs. Such exclusion may be hidden in seemingly Some typically male jobs do not require much universal entrance requirements and hiring skill or continuity, although theymay pay and promotion procedure (Roos and Res- more than a predominantly female job (Eng- kin, in this volume). land and McLaughlin, 1979; Englandet al., 1982). Early decisions about futureoc-a- Reinforcement of Sex Segregation pations and occupational training alonecan- not account for the level of sex segregation Once women have selected out of, or are one observes. _ excluded from, male occupations,some ex- Not all female jobs are lower in theiroc- planations go on to say, their tenure in typ- cupational rewards when compared with all ically female occupations further reduces their male jobs, either. While female occupations ability to change to a typically maleoccu- pay less, on the average, than male occupation. In their typically female jobs, they pations pay to either male or female incum- get returns from the skills in which they bents, certain white-collar femaleoccupa- have already invested; they do not have the tions have relatively high status and relatively chance to learn new skills necessary for male good working conditions. And while female jobs; they may be isolated from information occupations are described as giving little networks about typically male job. openings chance for advancement, some upwardmo- (e.g., Roos and Reskin, in this volume); and bility may be possible. Men in theseoccu- therefore they do not have access to the pations are often the ones who take advan- male career ladders that provide better ad- tage of these possibilities. Oppenheimer vancement than do female job ladders. As (1970) has suggested that forsome men of a result of few advancement opportunities lower socioeconomic status, such female jobs and the demands of both a family andan may be seen as a move up, perhaps a step outside job, women may indeed lackcom- on the way to a managerial position. For mitment to their jobs and decide to drop example, men in teaching tend to endup out of the labor force. disproportionately as principals andsuper- intendents (Schmuck et al., 1981). Ingen- eral, one finds thatmen in the female semi- EXPLANATIONS OF MOBILITY professions are overrepresented in administrative jobs (Grimm and Stern, 1974). (See To begin to study mobility to and froma also discussion in Sokoloff, 1980: 55-63.) In sex-typical occupation, rather than stopping a case study of one firm that had encouraged with the conclusion of immobility,one must people to move to jobs atypical for their (1) examine more carefully the stereotypes gender, Schreiber (1979) found thatmen in of typically female and male jobs, (2)con- clerical positions, in contrast with thewomen sider how supply side characteristics might in those jobs, felt that these were jobs that 108 CHANGING AND OCCUPATIONAL SEX SEGREGATION 59 would increase their chances to move into Changes in Labor Demand management. What is interesting about their perception is that it was not supported by Another source of individuals' mobility the promotion records of men who had been between sex-typical and atypical jobs is from clerical workers within the firm. changes in labor demand. A person may be able to fill an occupation usually held by the opposite sex, if the employer is experiencing a demand for labor in that occupation. For Individual Life Changes example, the feminization of clerical work Individuals' needs, employment behav- and school teaching seems to have resulted ior, and human capital can change over their from a scarcity of willing and suitably edu- lives. Not all men, for example, are highly cated men (Davies, 1975; Strober ana Tyack, committed to the work force over all their 1980; Oppenheimer, 1970), although the lives. Some are employed while still in school, women who entered these jobs usually were although often in sex-typical occupations not moving from another job but from out- (Greenberger and Steinberg, 1983; Lewin- side the labor force. World War II has al- Epstein, 1981). Still, at this stage, a young readyleen mentioned as an example where man might end up in a typically female oc- women were suddenly brought into' men's cupation that is convenient for hime.g., jobs. After World War II, of course, women the graduate student who is also a depart- were just as brusquely forced out of the la- mental secretaryand then change to a more bor force or into the usual female jobs (An- typically male occupation after graduation. derson,1981). Dramatic changes in de- Many women have fairly long periods of mand, then, can bring about equally dramatic their lives when they are not involved with changes in the sex distribution of occupa- childrearing and when they would be suit- tions, which strongly suggests the impor- able candidates for jobs that require a rel- tance of demand for maintaining and chang- atively long commitment (Kreps and Leaper, ing occupational sex segregation. 1976). Schooling and training necessary to Legislation can also affect labor demand make an occupational change are possible and, as a result, movement from sex-typical later in life, as evidenced by the increasing to sex-atypical jobs. The enactment of the numbers of women attending college at older Equal Employment Opportunity (EEO) ti- ages (Ileyns and Bird, 1982). Other changes tle of the 1964 Civil Rights Act, its expansion in a woman's life circumstances could lead and strengthening in 1972, and Affirmative her to seek a job considered atypical for her. Action legislation may have permitted and ht any of the descriptions of women'slcareers encouraged at least some women to move implicitly place. women 'in families where Co jobs that previously were closed to them. the husband provides the main economic Beller (1982, and in this volume) did find support; in such a setting the woman can dramatic decreases in occupational sex seg- make her job decisions using criteria other regation in the 1970s and some links of this than income maximization. With increasing decrease to legislative enforcement of EEO. divorce, inflation, and unemployment rates, At the same time, some movement back however, more women find themselves to sexLtypical occupations may occur after without husbands or with husbands who do individuals experience harassment by co- not earn enough, or anything, to support workers or difficulties in adjusting to work the family. Under such circumstances, that was designed for the opposite sex women may decide they can no longer afford (Schreiber, 1979; Gruber and Bjorn, 1982; to remain in low-paying, though sex appro- Harlan and O'Farrell, 1982; ;Kanter, 1977; priate, jobs. Roos and Beskin, in this volume).

G7 60 RACHEL A. ROSENFELD

The meaning of a move to a typical or thors make some initial analyses of the de- atypical occupation might be quite different terminants of such change for Avoknen. depending on whether the person is of a minority race. In the past, because of race ORGANIZATION OF THE PAPER discrimination in lower-level, male occupations, a black woman or tan moving to a This paper will examine both patterns and typically female occupation might find not determinants of changes of occupational only an increase in status, but also an in- gender typicality by sex and race. The next crease in pay and ajob more in line with section describes the data for the analysis. her or his educational credentials. Many of The third section shows the types of occu- the typical occupations for blacks, however, pational mobility undertaken by women and were sex typed aswell as race typed; -for men, black andwhite. The fourth section example, private household workers were examines the individual and job-levelde- black women and porters were black men. terminants of a move to or out of a sex- Among blacks, therefore, one might still ex- atypical occupation. The final section sum- pect to see differences in occupational.lo-.. marizes the results and discusses their im- cation and mobility by sex. It hasbeen,sug- plications. gested that one result of the civil rights movement of the 1960s was the movement DATA of some black males into higher level positions typically occupied by white males and The data for this paper come from the the movement of black women into the lower- January 1973 Current Population Survey paying and lower authority positions typi- (CPS) and its supplemental questionnaire cally filled- by white women (Lewis, 1977; that surveyed persons with new jobs since Jusenius, 1975; Malveaux, 1982); Analysis. January 1972. Data were selected on outof mobility across sex-type boundaries needs of-school whites and blacks 20 to 50 years to include differences related to race. of age whq had new employers at the time Why a person has either a sex-typiCal or of the CPS and who had been employed at sex-atypical occupation has received re- some time in the preceding 5years.3 The markably little investigation (see Polacheck, data set includes information on the re- 1979, 1981b; England, 19826; Jusenius, 1975; spondents' previous jobs, their 1973 jobs, Daymont and Statham, 1981; Corcoran et the way in which they spent time between al., in this volume; and Beller, 1982 forstrome jobs (if there was such a period), the reasons of the exceptions to this statement). Even less has been studied regarding the extent and determinants of changes in occupational In January 1973, respondents were asked regarding

sex-type during the adult years. Some case each person in the CPS, "Was . . .doing the same kind studies (e.g., Schreiber, 1979; Kanter, 1977; of work a year ago, in January 1972?" The supplemental Mcllwee, 1982; Dressel and Petersen, 1982) questionnaire was distributed to those over 16 who had new employers since January i*,zand who were not provide interesting hints about what hap- self-employed or working without pay in tefamily busi- pens, but they do notgeneralize their find- ness in January 1973; the questionnaireasked about ings to the occupational structure as awhole. the previous job and about the job search.Although Using national data, Jusenius (1975) and the question was meant to includel persons who changed Corcoran et al. (in this volume) have shown occupation or job with the same employer, almost311 respondents seemed to interpret the question as re- that some black and white women move ferring to employer shifts. Of the 102,374 people about among male occupations;Corcoran and co- whom. the January 7973 CPS inquired, only 326 were. workers show that at least some black and reported as (-hanging jobs with the same employer, too white men change, as well. Both sets of au- few to include in the analysis here. 108 CHANGING AND OCCUPATIO.NAL SEX SEGREGATION 61 that they changed jobs, and such personal white women aged 30 to 44 in 1967, about characteristics as marital status, age, and ed- 50 percent of those who moved from an oc- uoation. These data enable one to examine cupation predominantly of one sex to one the joh-changing process and the extent and predominantly of the other sex between 1967 determinants of changes in the sex compo- and 1971 did so within a given firm, For sition of occvations held by individuals. black women, however, she found "chang- The timing of this survey on employer ing type of occupationeither to or from a 1:hangers was fortunate, since January 1973 typical onewas much more likely to be is a particularly interesting period in which accompanied by an employer change." Fur- to examine gender differences in job shift- ther, "these results are ;n part explained by ing.It was just after EEO legislation en- the listing of occupational changes. . . . forcement became stricter. It was also a time Among white women, there were those of recovery in the business cycle, when jobs whose movement into an atypical occupa- may have been more available and the ef- tion appeared to be an intrafirm promotion, fectiveness of EE() legislation enhanced. from operative to foreman, for example. The availability of data.011 a relatively large Among the black women, however, the sample of persons that changed employers atypical jobs into which women moved were during a particular period is also fortunate. constderably different from their previous, Selecting only employer changes avoids typical employment, from private house- confounding general inertia with the bar- hold worker to operative, for example" (Ju- riers to changing occupational type faced by senius, 1975:28). movers. Of course, people canchange jobs There is considerable sex segregation within the same employOr. (Data on job shif- within firms (Blau, 1977). Occupations that ters within a firm were not available from are atypical for a given sex in thelabor mar- the (PS, see footnote 3.) One might expect, ket as a whole may be predominantly of that how's . that the' chances for changing from sex within a particular firm. Some of the se-typical occudation to a sex-atypical oc- intrafirm joh changes that Jiisenius labelled cupation would be greatest acrossemploy- as changes in type of occupation may have ers. Those changing jobswithin the same actually been changes from one predomi- firm might be more likely to continue' along natly female (male) job to another predom- some lob ladder rather than changingjob inantly female (mak) job in that particular halite, jusenius 11Y 75) found that, among firm. While using data on employer changers does not solve the problemof possible inconsistency between the sex composition of an occupation ati a whole and the sex com- 100.1 %%1111 .1 111 \11.1111/110.1'1 11.11 1 110%11 brell position for a given job, it does increase the had nil? benttipimed th 1 1111111Id litc.t Ill C'halice's that a change in occupational type 1, , Lik,rCurb et:thitits tills. ti dehiti tp.nIt"t h.p.t had a 11n1..11)11N I(11) .111(1.(1 art. slit walk represents a real change on the job. More ,hiltts rillt In, 11)111.11 ul 11)1 MIA% 'ONF11/ %yin be said about this in the conclusions. Ho 1"1N11 1/1,1f1111 II1.11(111('',1 11/1111,1111' .1%kr(1 1111 While theme' are advantages to using this (111.111rd 111.11 'Mt .11 10111 III' 11111 111'1(1 1/1 t\ 11)11% 111 1111 particular sample, sonic important statistical 1t)"Jill it .1s Mit' rnt \1 III( 11111t it J.1111;a/ problems are raised in limiting the study to 1111111111 \1111,,1 111 146.'1 ;if1.11e1 1111 11.11,1 1111 1111

.111.11 \ '.1' 111 1'',1'1111'd 1111.111 11.11)11 .110 11111% 111I 11110e \Olt, tilos,. persons changing employers. ln aria-

11.1,1 11.0111.t11t 111111,01'1 S111111' 111111'III1111' i Wa1% 11it' /.ing the outcomes of job chages across l ,111s41.11,11,11% 1.1; Ins t\ VA/ 11.1t INIII111.11 1111, employers, 011e would like to be Ale to ge- 11.11,te11.11tI\ h/tnn 1111' 11111 111 J11111,11 1(17 En! I1111,1 trah/e these results to all potential job-shift- 111 11111 11W, \kJ, .11111 11,111 (1111 111; 11)7.!1'111 14111'1s. 11 crs. nickeling tII those persons whir) re- " h" " l''" it"1" 011 JAM1,11 \1'17 i lnII IA a 13111,R1 ti. 1 till ot tilltttph.d Illailled With then currentemployers from 62 RACHEL, A. ROSE?,'FELD

January 1972 until January 1973 but who in the last 5 years, perhaps thus selecting might, if observed longer, change employ- on those with more continuouslabor force ers after January 1973; (2)those persons pre- participation. At the same time, since this viously employed who were not in the labor previous job could be. one held at any time force or who were unemployed and between between 1968 and January 1973, the selec- jobsas of January 1973; and (3) those persons tivity bias is probably less than if the re- who did not have any previous job. Select- quirement for inclusion in the analysis had ing only those person's for study who changed been employment on two particular dates employers over a given period could bias (Corcoran et al., in this volume). coefficients estimated on the selected sample (Berk and Ray, 1982; Heckman, 1979; DESCRIPTION. OF OCCUPATIONAL SEX BarnoW et al., 1980; Olsen, 1980). The au- COMPOSITION AND EMPLOYER thor was not able to investigate all forms of CHANGING selectivity in these data. The author did, however, use procedures described by Ray Table' 4-1 describes the current and pre- et al(1981) to cre4tte a selectivity measure vious occupations of the sample in terms of fin- moving from, as compared to staying with, the average percent male.'' kVhite males we're' an employer, 1972 through 1973.This meas- most segregated on this measure: On the' ure was highly collinear with tenure on pre- average, they held occupations composed of vious job fOr those who shifted jobs, not a over 80 percent of the' same sex. Black men surprising result given the decline of job were' in somewhat less sex-segregated oc- mobility with tenure (hall, 1980). Tenure' cupations, Women, both white and black, on the' job, then, differentiatesthose who at the mean tended to be in occupations that changed employers from those who stayed were only 26 to 28 percent malt' (or, con- with the saint' employer be'twe'en January versely, 74 to 72 percent female). As a result 1972 and January 1973. Since the correction of changing employers, there was little chin ige proposed by Ray et al. (1981) is fer an or- in average' sex composition of occupations dinary least squares model and this study for any group.5 used logits for much of the' analysis, a direct The overall low average change in percent measure of tenure on previous job tocontrol col selectivity bias isused here rather than the derived selectivity measure. In general. percent male was calculated for each .3- The selection of those who change em- digit occupational code from the 1970 U. S census. Ten ,plovers over those who remained with an largemiscellaneous groups of occupations iinluding )f.rnployer is probably most significant fOr adult cli.ruld workers not specific mill iniwellantims, man espocially %%hite adult males. For agers and administrators not elsewhere classilieci. and inspectors c 1 Here further broken down by Indus- women, especially white women,the' dis- tr ,Se('Slimmer. 1977, for further det.uh. Using the tinction between being in the labor force at 1971) data to categorize occupations is problematical in all and not in the' inrce call be equalb thatat least some occupations changed their sex cony important. The W011W11 in the sample Wl position betsveeti 1971 and 1974 iltllr. in this %i1- nine)Itis not clear that such changesalTrcta large employed at a given time. January 197:3. moldier of people one shoald keep in mind, bosses er, Only about 50 percent of all adult women as disciissed a1)051. thatjust as 11O(115rclu,tlsVal) 01,111Ve are employed at a particular time. However, to 0«11),01011Swith diflerent sex types, so whole o selecting on employment at any given time copatlims mer tone can change then see does not seem to bias cross-sectional anal- and se's label. Since the(:1'was .1multistage rather than simple ysis of %voinens occupational rewards and random sample. the significance les els in the tables ate characteristics (Corcoran et al.. in this yob 01115 approunatc l'sing the weights poi% tiled ss ith the Fligstill and \von, 197)ii. Mc, hilwa5ei. did notIlimILT the ihNtrIbilt11)111)1 tit AS() tt'litHrtti to }lave had sonic prrriott, job tit.pndrlit of intlplidrilt 5,111,11111.5

p OB CHANGING AND OCCUPATIONAL SEX SEGREGATION 63

TABLE 4-1Description of Job Shifts and Occupational Sex Composition by Race and Sex: 1973 CPS Employer Changers Black White Black White Women Women Men Men Mean percent male of previous occupation 26.4, 27.3 76.0 81.8 Mean percent male of 1973 occupation 28.9 27.3 76.6 82.1 Average change in percent male (percentage points)° 2.23 .44 ..86 .52 Correlation between percent male of previous and 1973 occupations" .34 .32 .26 .29 N 185 , 2009 176 2358 ° None of the changes is statistically significantly different from 0 at the .05 or .1 level. " All correlations significantly different from 0 at the .05 level. male might be interpreted as indicating lit- would like, therefore, to go from a contin- tle change of occupational sex type 4vith a uous measure of occupational sex composi- change of employer. This would be consist- tion to a categorical or ordinal measure of ent with the representation of the labor mar- occupational sex-type or typicality. ket as sex-segregated and preventing any While most occupations can be identified change to less sex-typical jobs by members as male or female, the particularcoding of of a given sex. Averages, though, can hide occupations into these categories as some- large changes in both directions, i.e., to and what arbitrary. In the remainder of this pa- away from occupations with high propor- per, occupations in which men are the ma- tions of the respondents' sex. As Table 4-1 jority (at least 51 percent of those in the also shows, there are surprisingly low cor- occupation) will be labelled "male," "atyp- relations between the sex composition of the ical for women," or "male dominant," while previous and the January 1973 occupations those with less than 51 percent male will be of the CPS employer changers, from a high called "female," "atypical for men," or "not of .34 for black women to a low of .26 for male dominated."6 black men. Such low correlations suggest Table 4-2 describes the employer changes considerable change in occupational sex type of the 1973 CPS sample in terms of these with a change of employer, perhaps some- categories. While there is considerable sex what less so for women (black and white) segregation, some people do move from or than for men. One hypothesis to account fin- to sex-atypical jobs. Ten to 15 percent move these results is that something is wrong with from a sex- typical to a sex-atypical occupa- the data. Others (e.g., England, 1982b), lion with a change of employer; over 60 per- though, have found similarly low correla- cent move back to a sex-typical one. While tions. While lower than might have been expected, the correlations are positive. One that This categorization is similar to that used by Cor- can imagine that what is happening is coran et al. (in this volume). The distribution of race people are changing occupations within and sex groups over the occupational percent male did ranges of sex composition. Forexample, not show any natural breaking points, but it was ex- women could be shifting easily among oc- tremely skewed. The median percent male of the 1973 cupations with anywhere from 0 to 30 per- occupation was 16 for black women, 16 for white women, 87 for black men, and 93 fir white men. Since most cent male but be stuck at a hypothetical 30 people are in the tails of the distributions, the exact percent male barrier. The image of the labor points chosen to separate male from female occupations market is of barrieis across which it is dif- should have relatively little importance as long as they ficult to move (e.g., Sokoloff, 1980). One are somewhere in the middle of the distribution. 71 64 RACHEL A. ROSENFELD

TABLE 4-2Sex-type of 1973 Occupation by Sex-type of Previous Occupationby Race and Sex: 1973 CPS Empldyer Changers Black Women White Women Previous Occupation Previous Occupation Male, Not Male- Male- Not Male- DominatedDominatedTotal DominatedDominatedTotal 1973 Occupation° (%) (%) (%) (%) ( %)

Male-dominated 39.0 15.3 . 20.5 36.2 15.9 21.0 Not male-dominated 61.0 84.7 79.5 63.8 84.1 79.0 Total 100.0 100.0 100.0 100.0 100.0 100.0 N 41 144 185 503 1506 2009 22.2 # 77.8 100.0 25,0 75.0 100.0

Black Men White Men Previous Occppation Previous Occupation Male- Not Male- Male- Not Male- DominatedDominatedTotal DominatedDominatedTotal 1973 Occupation (%) (%) (%) (%) (%) (%) Male - dominated 86.5 62.9 81.8 90.2 71.6 88.4 Not inale-dominated 13.5 37.1 18.2 9.8 28.4 11.6 Total 100.0 100.0 100.0 100.0 100.0 100.0 N 141 35 176 2136 222 '2358 80.1 19.9 100.0 90.6 9.4 100.0 a Male-dominated occupations are those that are at least 51 percentmale.

the flow is stronger iii the direction of sex- somewhat more likely to be in female oc- typical occupations, at least at the level of cupations, which is what Malveaux (1982) census occupational codes,the barriers be- reports. Here, though, one does not see a tween male and female occupations are not greater tendency for black womenthan white impermeable.? In Table 4-2, results are gen- women to move to female occupatio ss.The erally consistent with other descriptions of mobility patterns of women by rte are sim- occupational sex type by race. Black women ilar. Black men, on the other hand, show as compared with white women,and black greater movement than white men out of men as comparedwith white men, are male-dominated occupations and less movement to male occupations. As Malveaux (1982) and others have emphasized, in making comparisons of sex type The movement across occupational sex types as de of occupations (and mobility between types) Feed here usually involves a relatively large change in by race, one needs to keep in mind the race the sex composition of a person's occupation. see Ap- differences in occupation within the cate- pendixridix A. (Results for blacks, not shown, are similar to those in Appendix A.) The majority of thosewho gories of male and female occupations. One staff with an occupation labelled male. or femalehold way of understanding race' and sexdiffer- occupations after their employer shift that are within ences in mobility across en-within occupa- 10 percentage points of the percent male of their pre- tional sex-type categories is by examining vums occupation A majority of those crossing the sex- type boundaries move to an occupation with a sex com- the outcomes of such mobility-i.e the. position that differs by. more than 54) percentage points types of occupations where' people work after from that of their previmis occupation. Further, the a change. Table 4 -:3 showsthese outcomes direction of the change is as one would expect Those by race, sex, and sex type of 1973 occupa- fur woolen going froth typical to atypical eiccupations, tion. example. are going to occupations that are consuierabl . Typical occupations for the white women more male than their previous occupations.

72 IOR CHANGING AND OCCUPATIONAL SEX SEGREGATION 65 job shifters are clerical occupations, al- status, despite the greater proportion of though 13 percent end their employer white-collar jobs held by white women and changes as health workers or teachers and black and white men who were in female 19 percent as service workers. Atypical oc- occupations. Going to or even staying in male- cupational outcomes for white women tend dominated occupations, however, does not to be within the professional and managerial necessarily result in the greatest increases categories, though 16 percent are durable in wages. For white men, moving to a fe- and nondurable goods operatives. While al- male occupation may not increase status, most 40 percent of the black women who but it does increase wages more than any end up with sex-typical 1973 occupations are other type of move. For black women, those clerical workers, more of them as compared going to typically female occupations ex- with white women have sex-typical blue- perience the greatest gains. Moving to a male- collar jobsas operatives, nonhousehold 1ominated occupation does provide greater service workers, and household workers. wage increases for white women than other Black women's atypical occupations are types of moves, but white women do not roughly comparable to white women's atyp- significantly gain in wages by going from one ical occupations. male occupation to another. The failure of Among white men, sex-typical occupa- male occupations to be wage growth occu- tional destinations are most often skilled and pations for women has been found else- semiskilled blue collar. Almost one-quarter where (Corcoran et al., in this volume; Ju- of those with male-dominated 1973 occu- senius, 1975; Rosenfeld, 1983). At the same pations have crafts occupations, while an- time, those who stayed in male-dominated other one-fifth are operatives. Atypical oc- occupations as compared with others in th- :r cupations for white men after a job shift are race/sex category have the highest 1973 wage in lower white-collar retail sales and clerical levels. occupations, as well as in durable and non- Even among people changing employers, durable operatives and service categories. then, there is a considerable amount of sex For black men, both sex-typical and sexsegregation in both the origin and the des- atypical occupational outcomes tend to be tination occupations. At the same time, be- lower skilled, as compared with the out- tween 10 and 15 percent, depending on race copes for white men. More of those with and sex category, go from a sex-typical to a sex-typical destinations, as compared with sex-atypical occupation with their employer white men (or women), are laborers or serv- move. A considerably larger proportion, 60 ice workers. While 18 percent of the black to over 70 percent, of those who previously 111(.11 with sex-atypical outcomes are clerical had a sex-atypical oct upation had a sex-typ- workers and 12 percent are health care ical one in 1973. The sex-type distribution workers and teachers, another 35 percent and mobility across occupations differ be- are service workers, tween the sexes and between the races. Black Another approach to understanding race women resemble white women in their dis- and sex differences in mobility across and tributions and mobility more than black men withi sex-type boundaries is look at the resemble white men. Yet even among women outcomes of the mobility in terms of relative the' characteristics of jobs labelled male or job rewards. Table 4-4 shows changes in sta- female and the consequences of mobility tus and wages by race and sex for difkrent within and across categories differed by race. t,pes of moves. Such differences need to be kept in mind Consistent with Wolf and Rosenfeld's when interpreting sex and race differences (1978) findings. all sex and race groups mov- and similarities in individuals moving to or int; to a male-dominated occupation gained between sex-atypical occupations, the sub- the most, or close to the most. in terms of ject of the next section. 66 RACHEL A. ROSENFELD

TABLE 4-3Occupational Destinations by Sex, 'lace, and Sex-typeof 1973 Occupation: 1973 CPS Employer Changers Black Women Occupational Destinations Typical' Atypical Total 1973 Occupation (%) (%) Engineer, physician, dentist 0 456 .54 Health worker, teacher (except college) 2 04 10.26 3.773 Engineering and science technicians ,0 0 0 4.84 0th er professional, salaried 1.36 17.95 Manager, salaried, manufacturing 0 0 0 Manager, other, salaried 0 7.69 1.61 Sales-retail 3.40 0 2.69 2.56r- 4.08 Sales-other, .68 34.41 Clerical 38.10 20.51 7.69 1.62 Crafts 0 Durable and nondurable goods operative 15.64 17.95 16.13 3.76 Other operative 2.04 10.25 0 Nonfarm labor 0 0 8.60 Private household worker 10.88 0 t.56 20.97 Service worker 25.85 0 Farmer, farm manager 0 0 0 0 Farm laborer, foreman 0 99.99 Total" 99.99 99.98. 39 186 N 147 Percent white collar' 46 62

White Women

Engineer, physician, dentist 0 .71 .15 12.61 Health worker, teacher (except college) 12.82 11.85 Engineering and science technicians 0 1.42 .30 4.72 Other professional, salaried 1.32 17.54 .15 Manager, salaried, manufacturing 0 .71 0 13.03 2.73 Manager, other, salaried . Saks-retail 5.40 1.18 4.52 ., Saks-other 0 6.64 1.39. 40.02. Clerical 47.36 12.32 3.09 .85 Crafts .25 11.07 Durable and nondurable goods operative 9.74 16.12 5.45 2.54 Other operative 1.76 4.26 .95 Nonfarm labor .06 0 1.84 Private household worker 2.32 4.97 16.04 Service worker 18.98 0 Farmer, farm manager 0 0 .15 Farm laborer, foreman 0 .71 100.03 Total° 100.01 100.00 422 2014 N 1596 Percent white (mile' 67 65 " Totals differ from 100 due torounding. than 50 percent male) for men and one that is not Atypical occupation is one that is male dominated (more male dominated for women. Whitecollar occupations are those in the major occupationalcategories of professional and technical, managerial and administrative, clerical. and sales.

74 JOB CHANGING AND OCCUPATIONAL SEX SEGREGATION 67

TABLE 4-3(Continued) Black Men Occupational Destinations Typicalb Atypical 1973 Occupation (%) (%) Total Engineer, physician, dentist .68 0 .55 Health worker, teacher (except college) 1.35 11.76 3.30 Engineering and science technicians 0 0 0 Other professional, salaried 7.43 2.94 6.59 Manager, salaried, manufacturing 1.35 0 1.10 Manager, other, salaried 5.41 0 4.40 Sales-retail 0 5.68 1.10 Sales-other 1.35 0 1.10 Clerical 4.05 17.65 6.59 Crafts 14,86 0 12.09 Durable and nondurable goods operative 10.81 23.52 13.18 Other operative 12.84 2.94 10.99 Nonfarm labor 22.30 0 18.14 Private household worker 0 0 0 Service worker 14.86 35.28 18.5r., Farmer, farm manager 0 0 0 Farm laborer, foreman 2.70 0 2.20 Total° 99.99 99.97 100.02 N 148 34 182 Percent white collar' 22 38

White Men Engineer, physician, dentist 3.09 0 2.73 Health worker, teacher (except college) 1.97 9.22 2.82 Engineering and science technicians 2.34 0 2.07 Other professionals, salaried 8.25 3.55 7.70 Manager, salaried, manufacturing 2.48 0 2.19 Manager, other salaried 914 0 8.07 Sales-retail 1.73 15.60 3.35 Sales-other 5.90 0 5.22 Clerical 3.98 23.05 6.21 Crafts 24.46 .71 21.69 Durable and nondurable goods operative 8.10 27 30 10.35 Other operative 12.93 3.19 11.80 Nonfarm labor 8.95 0 7.90 Private household worker 0 (1 0 Service worker 5.01) 17.38 6.48 Farmer. farm manager' .19 0 .17 Farm laborer, foreman 1.45 (I 1.28 Total" 99 96 100.(k) 100.03 2416 N 2134 '182 Percent white collar' 39 51 a Totals diner In n» 1(X) due to rimming. h A tpial occupation is one that is male thlinmated (more than 50 percent male)for men and one that is not male dominated for women. ' White collar 011'Ilpations are those in the major occupational categoriesof proless101ial and technical, managerial e.fld alfillitliStUdel%e. eleru al, and sales 68 RACHEL A. ROSENFELD

TABLE 4-4Mean Status and Wage Changes With Employer Changes by Occupational Sex-type, Sex and Race: 1973 CPS Employer Changers Black Women White Women Previous Occupation Previous Occupation 1973 6ccupation Atypical Typical Atypical Typical Atypical Previous wage° 4.25 2.97 3.30 2.68 Current wage 3.84 3.70 3.54 3.10 Wage change -.42 .73 .24 .41* Previous SEI 53.7 32.3 48.0- 39.6 Current SEI 60.4 39.3 48.9 46.4 SEI change 6.7t 7.0 .9 6.8* Typical Previous wage 2.29 2.62 2.71 2.50 Current wage 3.30 2.59 2.54 2.83 Wage change 1.01* -.03 -.16 .33* Previous SEI 33.5 30.5 44.2 40.3 Current SEI 37.4 30.2 40.5 41.8 SEI change 3.9 -.3 -3.7* 1.6*

Black Men IA'hite Men Previous Occupation Previous Occupation 1973 Occupation Typical Atypical Typical Atypical Typical Previous Wage 3.54 2.98 3.98 2.75 Current wage 4.73 3.56 4.35 3.30 Wage change I . 18t .59 .37* .55* Previous SE( 26.5 28.9 36.0 28.8 Current SEI 2,5.3 32.1 38.3 37.7 SEI change 1.2 3. 2 2.2* 8.9* Atypical Previous wage 3.42 2.81 3. 16 3.11 Current wage 3.39 2.77 4.01 3.40 Wage change .53 .25 .85 .30 Previous SEI 27.7 22.7 32.7 27.8 Current SEI 30.0 25.9 33.9 30.0 S E I change 2.2 3.2 1.2 2.2 " All wages are exPressed in 1972 dollars. * Change significantly different from 0 at the .05 level. t Change significantly different from 0 at the .1 level.

PREDICTING MOBILITY TO AND for different race and sex categories was BETWEEN SEX-ATYPICAL ( CCUPATIONS shown in Table 4-2 as the probabilities that either (1) persons who begin their employer In this paper occupations have been di- change from a sex-typical occupation will chotonazed into sex-typical (for men, oc- end up in a sex-atypical one or (2) persons cupations with a male majority; for women, who start from a sex-atypical occupation will occupations with ---550 percent male) and sex- go to a sex-typical one. In this section the atypical jobs. Mobility between occupations two kinds of probabilities arc taken as the JOB CHANGING AND OCCUPATIONAL SEXSEGREGATION 69

dependent variables, to be explained by a phasize sex differences in continuity, exten- set of individual and jobcharacteristics.8 siveness, and intensity of employment.If Since the dependent variables aredichot- such explanations are valid, then both women omous, ordinary least squaresis not an ap- and men who have less continuous, exten- propriate method of analysis. Logistic sive, or intense employment should be more regression is used instead. Thedependent likely to go to or stay in female occupations. variable with logit analysis is actually the The variable marital status (presently logged odds of having one kind of occupa- married or not) used here indicates the ex- tional destination versus the other kind. This tent of a woman's familyresponsibilities, since paper, however,will sometimes refer to ef- responsibilities accompanying marriage are fects of independent variables on theprob- often the reason for women's more inter- ability of being in an atypical occupat:an, mittent and less intense employment(e.g., since a simple transformation of thecoeffi- Sweet, 1973; Shaw, 1981). For womenthis cients that does not change their signsdoes variable would be expected, if anything, to give the effect of the independentvariable decrease the probability of going to a sex- on the probability(see Hanushek and Jack- atypical occupation and to increase the son, 1977). probability of moving to a sex-typical one. For men, on the other hand, one would expect effects in the oppositedirection, since Independent Variables responsibility to provide for a family has been While the various explanations of sex seg- described as part of men's motivation to re- regation usually focus on its extentand per- main continuously employed in goodjobs sistence rather than on individuals'mobility (see Duncan et al,, 1972; Rumbergerand from or to sex-typed occupations, theydo Carnoy, 1980; and Rosenfeld, 1980, for some suggest what might affectmobility across evidence consistent with this view).9 occupational sex-type boundaries. For ex- For women, continuity of employmentand ample, many labor supply explanations of extent of family responsibilities arealso why women are in female occupations em- measured by whether a woman was out of the labor force caring for her family between her previous and 1973job. Of course, it is possible that a woman reenteringthe 5 An alternative to analyzing mobility across occu- labor force after a period of caring for afam- dependent pational categories would be to use as the ily is embarking on a fairly continuous work variable the distance moved, measured by thedifference in percent malebetween the previous and the life; the data, however, do not give sufficient 1973 occupation. Percent male on the previous occu- information on numbers or ages of children pation could be included as oneof the independent to predict this possibility. Tothe extent that variables to control for the degree of sex segregation there is statistically significant discrimina- Such equations, however, are on the origin occupation. tion by sex and age, one would expect women likely to violate the assumptions of ordinary least squares (01,5) regression. At least one of the independent var- who leave employment for family care to be iables (the lagged dependent variable) isusually correlated with the error term. ()LS regressionthen gives biased and inconsistent estimates. The degreeof bias 9 An alternative explanation of this evidence is that may sary over groups,making comparisons risky. If one wants to keep the emphasis onmobility, and es- men who are more motivated,who remain employed, pecially if the values on independent variablesdo not and who achieve higher occupational rewards, are also change (see Corcoran et al., in this viklunte), one can those who arc motivated toward and successful at get- do little to correct this problem with data at only two ting and remaining married. A preferablevariable would times. (See Rosenfeld and Nielsen, 1984.for a discus- be changes in marital status rather than simply marital 511)11 Of this problem and ofsolutions when one has data status as of 1973. The CPS, however, did not give inure than two times./ information on previous marital status. 70 RACHEL A. ROSENFELD less likely to move out of female occupations eluded, but its interpretation as an indicator and more likely to move to them. This of job attachment is complicated by its role variable is not included in models for men, as a measure of selectivity into the sample. since almost none of them had taken occu- Intensity of employment is measured by pational leaves for care for a family. whether the previous and the 1973 job were Another kind of break in employment is full-time (at least 35 hours a week). Women's unemployment, searching for a joban- occupations have been described as offering other of the independent variables. Such a more flexibility in hours'than typically male period between jobs does not necessarily jobs. This flexibility, however, benefits em- lead people to jobs that allow or encourage ployers by enabling them to pay lower wages, low labor force attachment; a period of give feVver fringe benefits, and more easily searching fOr a job may be necessary to find hire and fire (see, for example, Sokoloff, a better job or one that is unusual for a 1980:106-107). Women are overrepresented person of a given gender." Extent of at- among part-time workers, but they are also tachment to a particular job is measured by overrepresented among the involuntarily whether a person had more than two jobs part-time(Barrett,1979).To theex- in 1972. (One generally would expect those tent that typically female occupations offer with any jobs in 1972 to have at least two less than full-time employment, one might one held previously and one held as of Jan- expect thosemale or femalewho seek uary 1973.) Tenure on previous job is in- or find full-time work after being employed part-time to be those who are also moving from female tq male occupations. Con- versely, those taking part-time employment Daymont and Statham (1981), who included num- may be going to a more typically female ber and various age groups of children in an analysis occupation. With cross-sectional data, Beller of middle-aged black and white womens occupational (1980, 1982) found some evidence to sup- atypicality, did not find significant negative effects of children on sex-atypical occupation. Beller (1980, 1982), port this speculation. In general, though, using data on a less restricted age range, likewise found the evidence for the effects of home re- that a large number of children did not significantly sponsibility, continuity, and extent of em- decrease the probability of a woman being. in a male ployment on the sex composition of wom- occupation. en'sand in the case of Beller's research, ttLength of time unemployed between jobs was di- chotomized into no time and some time unemployed on men'soccupations is weak (Beller, 1980, because the distribution was extremely skewed away 1982; Corcoran et al., in this volume; Po- from 0. The data contained a general measure of time lachek,19811); England, 1982b; Daymont nut in the labor force between jobs, which wits also and Statham, 1981). One might expect sim- positively skewed. This information was used to con- ilarly weak effects of these variables on mo- struct a variable representing whether persons had a period out of the labor force befih their 197:3 jobs (net bility among occupations. of a 13(.00 out for caring for a family by women). A Cohort, age, and stage in the work career third variable was lreated, indicating whether a job may all affect the nature of one's occupation. search had begun because a person left a previous job Older women and men may have sex-role involuntarily. something that might be expected to attituues that make them less willing than modify the effects of having a period of unemployment. These three variablesNome unemploynwnt between younger people to enter atypical occupa- jobs, some time Out of the labor force,and an int ot- tions, they may also be less acceptable to untury job search wvre highly intercorrelated. In- employers as candidates for such occupa- clusion of them in various combinations showed no tions.Beller (in this volume) shows that significant effects of an ins oluntary search or a period younger age cohorts as compared with older out of the labor force on subsequent job attachment. nor modifie mum oldie efh.t of a period of unemplo,.- ones are less sex-segregated. Your;., persons went. who arc new to the labor niarket, however,

78 011 CHANGING AND OCCUPATIONAL SEX SEGREGATION 71 may not yet have learned about the possi- diplomamight be consideredqualified to bility of atypical jobs (Malveaux, 1982). Age move into more traditionally malemana- is included as a measure of both cohort and gerial or administrativepositions or may have age per se, while whether a respondent was taken majors that preparedthem explicitly in school in 1972 indicates stage in the work for atypicalcareers. career. Many persons are employed while Getting additional educationor training attending school (Lewin-Epstein, 1981); more could also influence whethera person moves persons have at least one period of employ- to an atypical occupation. The CPSsupple- ment or other activity between leaving high mental questionnaire askedrespondents, school and entering college. Thus, attend- "Did you takeany occupational or educa- ance in school the previous year andem- tional trainingcourses or programs during ployment and out-of-school in 1973do not the time youwere looking for work?" The necessarily mean that thosepersons have response to this question is included with just entered the work force for the first time. the independent variables.Relatively few At the same time, for men who have had a people did haveany training, which pre- sex-atypical job while attending school, cluded coding theresponses into general leaving school could mean acquiring a more types of training. Training isa change in sex-typical occupation. human capital; it allowsone to change sex Education has been shownto be a major type of occupation only ifone is trained for determinant of occupational achievement, a sex-atypical occupation and if thetraining and is among the independent variables here. that the perion took is acceptedby em- In general, though, occupations withmore ployers. Job trainingprograms often offer women do not have a less educated work sex-stereotyped training that isnot useful in force and do not require less education than getting a private sector job (seeRoos and those occupations with more men (England Reskin, in this volume; Waite andBerry- et al., 1982); this might leadone to hypoth- man, in this volume). The CPS employer esize that education does not affect mobility changers could have gotten theiradditional between types of occupations. Research to training from other than jobprograms (the date, however, hints at curvilineareffects of source of the training is not specified), but education: Both low and high levels of ed- if training was not aimedat sex-atypical work, ucation increase the chances having an atyp- it would not necessarily increasemovement ical occupation (Beller, 1980; Polachek, to or between sex-atypical jobs. 1981b; Daymont and Statham, 1981). Those Since skills required by particular jobslimit with the least education would "tend to ap- mobility and since the relatively highstatus ply for and obtain jobs that requiredless of some female jobsencourages mobility, skill or training and among which transfer- Specific Vocational Preparation(SVP) re- ability is relatively great. The atypicaljobs quired for the previousoccupation and for which they qualify would haveas few whether this (iccupationwas white collar were skill requirementsas the typical jobs they also includedamong the independent var- already held" (Jusenius, 1975:24).Jusenius iables. Especiallyamong men, those who fbund support for this ideaamong white have white-collar jobs thatrequire more skill middle-aged women: Those with fewerthan could be those less likelyto change their 12 years of schoolingas compared with those occupation, let alone thesex type of their who had at least 12years were more likely occupation, across firms. to go from typical to atypical occupations. Region (South or not) is includedas a con- At the Cher end oldie spectrum, those with trol variable. In cross-sectionalanalysis of oc- the credential of at least a college degree cupational atypicality, Dayinont andStatham as compared with those without a college (1981) found black middle-agedwomen in the

\79 72 RACHEL A. ROSENFELD

TABLE 4-5Logistic Regressions for Moving From a Sex-typical Occupation to aSex- atypical Occupation by Sex and Race: 1973 CPS EmployerChangers. Black Women White Women (N = 138) (N = 1485) Intercept .45 .96 Independent Variables Highest year of education completed" 0-11 -1.08 .32 1-3 years of college .002 .06 >3 years of college 2.80* 1.19** Training while job seeking .69 .14 Age 20-29 -1.29 -.30 30-39 -1.11 .28 Currently married .98 .06 In school in 1972 -3.70* -.43 Taking care of a family before 1973 job -1.94t .02 >2 different Jobs in 1972 .64 .43t Some time unemployed before 1973 job .26 .11 Going from part-time to full-time work 1.92** .18 part-time to part-time work 2.08t -.15 full-time to part-time work .83 -.53* SVP` of previous job .28 -.05 Previous job white collar - .19 -.06 Tenure on previous job's -.34t - .07 Region = South .95 -.06 32.0 51.4 X2 df 18 18 .02 .00 p D .21 .03 .01 < p .05. ** p .01. t .05 < p < .10. o A sex-typical occupation is one that is at least 51%male (for men) or less than 51% male (for women). Only those who were previously in a sex-typical occupation are included. The outcomevariable = 0 if the 1973 occupation is sex-typical, = 1 if it is not. Dummy variables are coded 1 = Yes, 0 = No.

South less likely to be in sex-atypical occu- to a sex-atypical one. The coefficients show pations. the effects of the independent variables on The independent variables thus include the logged odds of ending up in a sex-atyp- measures of employment continuityand in- ical rather than a sex-typical occupation. The tensity, changes in life cycle stage, age,ed- chi-square statistic compares the estimated ucation, training, and level of previousjob. model with one including only the inter- They do not, however, include labor demand cept, thus enabling one to calculate whether variables, which will be discussed in the con- the estimated made] as a whole is significant clusions. in predicting the outcome. The D statistic is equivalent to R2-and measures the overall fit of the model (Harrell, 1980). Means and Results standard deviations for the variables are in Table 4-5 gives the results by race and Appendix R. sex for moving from asex-typical occupation For all groups except white males, having

80 IOB CHANGING AND OCCUPATIONAL SEX SEGREGATION 73

TABLE 4-5(Continued) Black Men White Men (N = 139) (N = 2105) Intercept -.73 -.88 Independent Variables Highest year of education completed"

0-11 . .20 -.22 1-3 years of college 1.05 .18 >3 years of college 3.75* .11 Training while job seeking 1.13 .21 Age 20-29 2.09t .41 30-39 -.90 .33 Currently married -.36 Ass In school in 1972 .93 .16 Taking care of a family before 1972 job' >2 different jobs in 1972 .49 .10 Some time unemployed before 1973 job .24 .16 Going from part-time to full-time work .17 .30 part-time to part-time work' full-time to part-time work 1.37 .60' SVP' of previous job - .14 .13" Previous job white collar -1.34 26 Tenure on prievious jobs .17 -.09t Region = South .16 - .13 X2 30.02 71.7 df 16 16 p .02 .00 ll .20 .03 SVP = Specific Vocational Training. From Temme's (1975) aggregation from the Dictionary of Occupational Titles (3rd ed.). 1 = short demonstration only, 10 = over 10 years of specific training required. " Tenure on previous job: 1 = less than 4 months; 2 = 4-6 months; 3 = 7-11 months; 4 = at least 1 yearbut less than 3; 5 = at least 3 years but less than 5; 6 = at least 5 years but less than 10; 7 =at least 10 years but less than 15; 8 = 15 or mare years. ' Not included for men because there were too few cases in this category to give a stable estimate.

more than 3 years of college (in contrast with occupation, the results here suggest that some other level of educational attainment) higher, rather than lower, educational levels increases the chances of going from a sex- enable one to break the sex-type barrier, typical to a sex-atypical occupation. Remem- unless one is a white male. For white men, ber that for white women, their male-dom- education has no effect, although it is also inated and other occupations were about true for them that those who hold typically equally likely to be white collar, and for female occupations are more likely to be in black women and men, sex-atypical occu- white-collar occupations. pations, as compared with sex-typical ones,. For white men, skill level of their pre- were more likely to be white collar, and vious job is significant. The higher the skill perhaps more likely to require college cre- level on their preyious job, the less likely dentials. In contrast, then, with what Ju- they are to go to an atypical job when ,they scrims (1975) found in her cross-classifica- change employers. This result fits with the tion of education categories by type of image of white men going to atypical jobs

81 ti

74 RACHEL A. ROSENFELD only if such moves are to better jobs, an from part-time to full-time employment thus image suggested by other authors (e.g., Op- has no effect on the sex type of their occu- penheimer, 1970), as well as by the average pation. wage increase white men earn with ashift For black men, the move from full-time from male-dominated to other occupations to part-time work or part-time to full-time (Table 4-4). Getting additional training or work has no effect on the sex type of their education while lodking for a job has no ef- 1973 occupations. Black men tended to have fect for any race and sex group on chances both typical and atypical occupations that for moving to an atypical occupation, per- were lower in status andskill than other haps because of the sex typing of such train- groups. For black men, even morethan for ing. black women, male and female jobs may be For women and white men previously in equally likely to require or provide les#, than sex- typical occupations,changes in amount full-time employment. of labor supplied had an effect on the type In trying to interpret the effect of a change of occupational destination. Whitesmen in hours, one needs to keep in mind that as well as womenwho wentfrom full-time part-time work is not simply redundant with to part-time work were also likely tobe going female occupations: The correlation be- to typically female occupations. Black women tween percentage male of 1973 occupation who went from part-time to full-time work and whether the job is full-time is only .14 (as compared with those with other patterns for white men, .12 TOr black men, .05 for of hours) had an increased probability of going white women, and .06 for black women. to occupations in which men were a major- Human capital explanations for why women ity. are in typically femalejobs predict that being It is not clear whether part-time employ- presently married and having taken time out ment represents involuntaryunderemploy- for family care reduce the chances that a ment or a choice to work less thanfull-time. woman will move from asex-typical to a sex- For both white women and the relatively atypical occupation. If employers exercise kw white men who go to part-time em- statistical discrimination on the basis of more ployment. the move to part-time (and fe- than sex alonethat is, to the extent that male) work could be due to (1) other op- they see all married women or all women portunities closing (although whether the job who have been out of the labor force as less search was involuntary did not change the committed to employmentone would also size of the work intensity variables)that predict negative effects of these variables on is, the choice might be between part-time going to an atypical occupation. The only work and no workor (2) white men and effect of family responsibility, however, is women might choose to take part-time em- for white men: They are less likely to change ployment, which is more readily available from a sex-typical to a sex-atypical occupa- in typically female occupations, inorder to tion if they are married. combine employment with other activities. In this analysis of movement from a sex- One can speculate that black women are tWcal occupation, significant effects ciao motivated by underemployment when they or work career stage are presentonly for seek full-time work that is more often as- blacks, and only strongly for black women. sociated with typically male occupations than Black women who were in school the pre- with typically female ones. White women vious year--and who tended to be younger going from part-time to full-time workcould were less likely thanothers to leave female be those who have taken time out from a occupations (see also Malveaux, 1982). If op- hill-time work career to raise their families portunities in traditionally female occupa- and now are reentering the lal)or -force in a tions were just opening up to black women, typically female occupation. The movement one would not besurprised to find that those

82 JOB CHANGING ANO OCCUPATIONAL SEX SEGREGATION 75

TABLE 4-6Logistic Regressions for Moving From a Sex-atypical Occupation to a Sex - typical Occupation by Sex: 1973 White CPS Employer Changers! AVhite Women VIliteMen (N = 493) = 222) Inter( rpt 2 09t 2.20 Wiepeildeil ilrlablesh %ear of education ompkted (1- I 1 .66 .11

1 I %ears of olle,4e 26 --1.11* II Vears ofcollege 1 66 .53 Framing while job seeking 27 .64) Ake 20 29 57t .17 30,39 31 1.14 airrcrit married .39 .32 In school in 11)72 .55 1.84" akink caw of a famil% before 1973 }cif 2 -2 (16.11.nt ink in1972 .61t 6') 'Nome nine liefm 197.1 )01) .04 .32 Going from part tune to 11111 time work Ik -.22 part tune to part-time work' .5 full time. to part hint' work 1.04* .40 SVP of pre% imic job .27" _ ($3 Peres ious lob whitc collar 60 .82t .1 enure, in, pre )ious job IN .05 Region South .1.1 28 ).in 7 29 3

df It lti 00 .02 I) 1St 13

(15 p Ill

(11 p A. p *A.01 " ()ilk th(Ist %%host pies ious ia.iiipation was se\ atspic il arc in( lulled The outcomesariabl .0 if the 1973

I itit it11Pa11,ill .11%P1(.41 Se laid(I 1 fit «ding I11111' tin) ft'%%, tdSt'S Iu tilt, Catgnr. ti Itlrludr 1111.11

lust Ia% lift', stlinol wcre taking, adv-ntage of previom occupations that are sex-atypical. female in copittions and keeping diem. There The coefficient estimates give the effects of is alsoa nu.trgiltul11t giltlVt effect of'wing the independent variables on the logged odds under '10 on :he chances that black men will of ending the employer change in a sex- sta \vitt) 1111illy-(101160:itt'd (1(.111pat1011.Since typical rather than a sex-atypical occupation. predi)illittaIltl\ !tulle (X ..upations are dn. ones Table 4-6 does not include results for blacks, both black II) and Nita wonnen because of the small number of blacks tlI 11()%t t(' gain terms of wagnjevel, begin their job shifts from sex-atypical oc- the lifilli-;(. blacks Yer at a disadvantage upations. Means and standard deviations in the ()II'( nine loation of their employer for the variables are in Appendix 13. Again, education helps predict the cc- tVhat about going the other \ya,fruni eupational outcome Here one finds the ur- st\ II) d \ (X1111)iitR)117' vilincar tqlect of education reported front 4-f) pr"sunts results fin tnst \vim in((. ross-sectional analyses. Women \yid' less

LS 3 76 RACHEL, A. ROSENFELD

than 12 years of schoolingor more than 3 Some life cycle and age effectsappear. years of college are more likely to remain Men who Lad sex-atypical occupations and with a sex-atypical occupation than thosewith were previously in school have greater odds intermediate amounts of education.Fur- of going to sex-typicaloccupations than those ther, the higher the specific skillprepera- who were out of school in both1972 and to required on the previous job, the lower 1973. Younger women,as compared with the odds that a woman willgo to a sex-typical older women,may have a somewhat greater occupation with a change of employer. tendency to stay witha sex-atypical occu- As Table 4-5 showed, higher previousMT pation even across firms, although here the also decreases the probabili'j that white men effect is only marginally significant. will leave sex-typical occupations. Maleoc- While marital status and family related cupations requiring higher skillsor educa- interruptions in labor force participation again tion may be especially likely to keep their have' no significant effectson women's oc- incumbents evenacross employers, al- cupation-type mobility, changes in hours do. though women with lower educationstay The woine:i whogo from full-time to part- with their atypical ocvupations, too. For men time work are again more likely tomove to who began their employer change froma typically female occupations. typically female occupation, it is having only Thus, as in the analysis of mobility from a few years of college-, as compared with sex-typical occupations in Table 4-5, edu- other levels of education, thatinhibits cation, skill requirements, life: cyclestage, movement to a sex-typical occupatie The and hours employed playa part in predict- interpretation may again have to with ing who will move across sex-type bound- credentials. If :aime women's occupations aries, although the way in which these'var- offer relatively good positionsto men, one iables explain mobility is somewhatdifferent have expected that thosemen who here. hats filled higher-skilled and white-collar jobs and who had more' than 3years of education would also be less likelyto move from a SUMMARY AN I) DISCUSSION female to a male occupation. This doesnot The labor market is extremelysex seg- seem to he the case.. For white men, having regated. Some' people, however, domove held a White'- collar oc,pation actuallyin- across the harriers built upon sex-typicality creased the chances of leavinga sex-atypical of occupations. Thispaper used data on 1972 occupation, although this effect was ofmar- to 1973 black and white employer changers ginal statistical significanceIf' the mana- to examine ttns phenomenon. At the level gerial and administrative womnsoccupa- of .3 -digit occupational code's, this studyfound tions do indeed offer titan chances her that of those persons beginninga job shift promotion, perhaps men are moving to them in increasing numbers, resulting inoccu- pations that are now more male, thoughcon- of multicollinttrits. The correlation betweenwttlicr sidered female. Forwomen, having a pre- the. pry% ions jolt was white collar and its t'I' is 11111- vious atypical occupation that was white eratel!,high about .48for white. women. For black women and men it WAS apprazittnitely .43 and fr white* collar, net of everything else., increasesmo- Men, Vhti %Nflit-collar occupation for theIt bility to a sex-typical occupation.'` %ions job was dropped. thy effet of SVP continued to be significant') tigatist her white SP was tilt' t'fftl't of whether thi. previous job was white collar was positis v. but not agnifiant For h, Tahle i 6. the' opposite s1,411s for white, svoniil other grotips and other kinds of niohditinludity4 it cht,tliyi the Ines ions job w.ts ina white-collar oc olds SVP titmil!.%Allot collar 04.1 not change there (tipattctit And of its 'IV I' look stoqw.-too.,l.. like. the result stilts

84 /pH GilANGING AND (x SEX sEGREGATION 77 from a sex-typical occupation (defined in than men and about why they are in wom- terms of whether the occupation had a male en's occupations focuses on women's roles majority), 15 percent of both black and white within the home. For men, on the other women, 13 percent of black men and 10 hand, responsibilities fbr a family are usually percent of white men moved to an occu- ignored. Finding effects of family status for pation atypical for their sex. There was also men but not for women suggests that our a strong flow from atypical to typical occu- stereotypes about the interface of family and pations: Somewhat over 60 percent of women employment need to be reexamined with and black men made such a move, as well respect to both women and men. as over 70 percent of the white men. In Other variables, though, that indicate general, the mobility patterns across sex types changing commitment to the labor force did of occupations fbr black and white women have effects on the nature of women's and resembled each othc r closely. Indeed, sex men's mobility between occupations with differentiated occupational locations and different sex types. Moving to part-time work types of mobility to a much greater extent was associated with moving to or staying than did race. At the same time, there were with a typically female occupation for white race differences. Even among the women, women and men, while moving to a full- the black women's female occupations were time job from one that had been part-time much less likely to be white collar ones than increased the probability of going to a typ- those held by white women, in part because ically male occupation for black women. of the overrepresentation of black women Having been in school the previous year also in typically female service occupations. Black increased the chances of moving to a sex- women who were able to remain in pre- typical job for black women and white men dominantly male occupations actuallii had previously in typically female occupations. higher wage levels and status gains than white As discussed in the preceding section, it is women in male occupations. Among men, not possible to determine whether part-time blacks were more likely to be in or move to employment represents a choice about how female occupations, although those who re- many hours to spend on market work versus mained in predominantly male occupations other activities or whether it represents in- received the greatest wage gains of any of voluntary underemployment. For those the four race by sex groups. For black men, persons who have demands on their time too, both the male and female occupations beyond employment, it would be possible', they held we're relatively unlikely to be white- in some cases, to extend the range ofjobs collar ones. open to them through such options as flex- For neither white nor b) ik women was time.. there much support for the idea that extent , Level of education infltaired the types of family responsibilities influences the of occupation changes that persons made. chance to move from or to a sex-typical oc- !laving more than 3 years of college, which cupation. One could argue that this is be': in most cases would indicate having a col - cause the variables measuring family re- lege degree, made it more likely that women speaisibilities are not detailed enough. These and black men would go from sex-typical to results, however, are ci insistent with a num- sex-atypical occupations and that ,_White ber of other studies that show the effects of women would go from one atypical occu- marriage and children on women's labor force pation to another. UnfOrtunAtely, the (;I'S participation but not on the status, income, did not give infbrmatifm on college major. or sex-type of the occupation womenhold It is not clear, therefore, whether it is e;ie. once they are in the lab n- fierce.Much of degree as a credential or the substaike-e of the lore about why women get lower wages the degre that enables someone to go to

vJr r RACIII:1, A. ROSENFELD and stay with an atypical occupation.If de- being in a male-typed rather than a female- mand is sex-segregated, then even having typed job. the training for an atypical Occupation may The analysis did not present strong age not help the individual getthe desired job; effects, such as one might have expected by virtue of gender, he or she willbe con- from changes in the degree of sex segrega- sidered inappropriate. If employers are tion over cohorts (Beller, inthis volume). It pressured to desegregate the workplace, they may be that both the strong movementback may he more willing tohire people from to sex-typical occupationsand the failure to other fields and train them on the job. With find age effects are the result of using data the decreasing sex segregation ofcollege from a period when changes in the climate majors (Beller, in this volume;Heyns and facilitating sex desegregation were just' Bird, 1982), it will he interesting to measure underway. whether persons use such majors to follow These analyses provide some insights into atypical careers. 13 the nature of male and female typical jobs White women and men who held male by race and sex and into the mechanisms by occupations that required inureskill were which persons change from an occupation more likely to staywith a typically male oc- of .one sex type to one of another sex type. cupation when they changedemployers. This The explanatory power, however, of the in- effect, together with the effect ofhaving a dependent variables taken together is low, college education, suggests the attractionand as indicated by the Dstatistic.It is espe- retentive power of the higher-levelmale jobs. cially difficult to predict, fiir whites, who The effects of having less than a highschool will leave a sex-typical for a sex-atypical oc- dvgree on a woman remaining withatypical cupation, at least using theseindividual and occupations emphasize again the rangeof job-shifting characteristics as the independ- occupations that are predominantlymale. ent variables. One reasonfhr this could be It is not skill and training alonethat keep that the individual and job-shifting variables won- en out of male occupations.It could be were not detailedand extensive enough that the less-educated women who stay in sex- Another reason could he that the factors that atypical occupations are trapped inlow-level affect whether a person is in or moves to a jobs; they remain in these jobs becausethe sex-atypical occupation are outside the per- pay is higher thanthat which they could earn son. As alreadydiscussed, individual har- elsewhere. Daymont and Stratham (1981)fiaind acteristics can seem to produceeffects that, forces. The de- that itis precisely among theblue-collar (c- in reality, reflect outride the cupations, which often hav,..lower educational mand for certain kinds of labor, e.g., requirements, that one finds anadvantage to demand for women in atypical occupations, can work against eventhe strong sex role ;ocialization that most of us receive. Kanter " 1:sing data on female college graduates ut 1961 097, for example, recounts how within a 'ileitis (197h) showed that those who Imosexatyptcal given firm, managementpersuaded women college maims were more likely than other womento against their initial resistance to move to star with sex-atypical careers.Conversely, Hearn and more typically male positions.Within and that )1/ak !Wit! reported data un 1976 collette senior~ among firms, employers vary intheir en- showed women were actually noire likely than Liento studs soationally specific majors Thejobs for which couragement of persons of the wrong sex to these typically female majors were preparingthem, apply and be hired for jobs; this variation boweser, were lower in status than thoseAnticipated could e.:plain whether an individual ends lip bs tspicalls male vocationally specific majorsHearn in a sex-typical or atypical jolk, Once a per- and Oltak were nut able to follow these seniors to see sex-atypical job, whether of ( son is hired into a how t lusls the %MIMI', majors predicted types that person stays with the joy, maydepen:1 reers 10B PIANGING AND OCCUPATIONAL SEX SEGREGATION 79 on management's Aims to curtail harass- sex. One would hope that sensitivity to this ment and to adapt work conditions so that level of measurement problem would lead both men and women can continue em- more of those who design large surveys, es- ployment there (Roos and Reskin, in this pecially longitudinal ones, to include ques- volume). tions on the sex composition of the respond- Such efforts at desegregating the work- ents' specific jobs. The conditions on a place usually depend on whether it is prof- person's job can he important in determin- itable to make them. Legislation can make ing that person's rewards and motivation to it unprofitable not to hire and retain a sex- stay with the job. O'Farrell and Harlan (in integrated labor force by enacting direct this volume) have urged that serious efforts monetary sanctions and by creating a el, be made to study sex segregation within or- mate in which expensive sex discrimination ganizations, where advancement and work suits are possible. An unfavorable economy conditions can be traced more precisely. The such as we have now, however, can redirect rewards of a given occupation over time may emphasis toward providing jobs and away indeed depend on its overall sex composi- from eflOrts to integrate jobs. While women tion, with those occupations dominated by may have a greater need to support their women offering lower rewards than other families now and rrif.!y be more attracted to aspects of the occupation would predict typically male jobs, competition for jobs in- (England and McLaughlin, 1979). Future creases the posibility of reverse discrimi- research needs to include not only the ques- nation charges by male and white workers, tion of why individuals enter and stay with as well as informal efforts to exclude women sex-typical as compared with sex-atypical oc- from male occupational territory. Rapid cupations, but also the larger questions of changes in decreasing occupational sex seg- why and how women, whose work is un- regation and improving sex role attitudes dervalued, have been excluded from the work occurred in the 1970s; social scientists and that society values highly. policy makers need to .nonitor carefully what' happens in the unfavorable economic climate of the early 1980s if these gains are to ACKNOWLEDGMENTS persist. The dependent variables in this analysis The support of the Carolina Population 1. were movements across occupation sex types Center and NSF grant DAR-79-1585 made that were created from the sex composition this paper possible. I am grateful to referees of 3-digit occupational codes. Information at and participants from the Workshop on Sex the 3-digit census code level i!P gerally the Segregation of Jobs (National Research most detailed available for netional samples. Council, Washington, D.C., May 24-25, Sexsegregationoccurswithin3-digit 1982) for comthents on an earlier paper that axles. Firms are sex segregated (Blau, 1977), informed the work on this one: to David and within firms, women rarely work in the Mau me and Roberto Fernandez for merging same jobs as men ( Bielby and Baron, in this the data, to Kenneth Spenner for providing volume). It is not clear, then, that the women the sex composition data, to Francois Niel- who move to male occupations or the men sen and Kathy Ward for their advice, and who move to female ones are really going especially to Barbara Reskin lint her exten- to jobs shared with members of the opposite sive and helpful suggestions. X30 RACHEL A. ROSENFELD

APPENDIX AExtent of Change in Percentage Male by Type ofOccupation Change: White 1973 CPS Employer Changers White Women Type of Move° Typical to Typical to Atypical to Atypical to Typical Atypical Typical Atypical (%) (%) (%) (%) 1973 percentage male-previous occupation percentagemale = (1973 occupation greater % male) 91 to 100 0 0 1.2 0 81 to 90 0 0 3.4 0 71 to 80 0 0 8.1 0 61 to 70 0 0 9.7 0 51 to 60 0 0 19.3 0 41 to 50 .9 0 22.7 .5 31 to 40 2.2 0 14.3 2.7 21 to 30 2.9 0 14.0 4.9 11 to 20 8.9 0 5.6 8.8 1 to 10 11.8 0 1.6 12.1 (No change) 0 42.4 0 0 34.1 (1973 occupation smaller % male) -1 to -10 13.1 2.1 0 13.2 -11 to -20 9.6 3.8 0 8.2 -21 to -30 4.7 5.0 0 10.4 -31 to -40 2.8 12.1 0 3.8 -41 to -50 .7 20.5 0 1.1 -51 to -60 0 20,9 0 0 -61 to -70 0 16.3 0 0 -71 to -80 0 12.1 0 0 -81 to -90 0 4.6 0 0 -91 to -99 0 2.5 0 0 99.8 Totalb 100.0 99.9 99.9 .321 182 N 1267 239

38 JOB CHANGING AND OCCUPATIONAL SEX SEGREGATION 81

APPENDIX A(Continued) White Men Type of Move° Typical to Typical to Atypical to Atypical to Typical Atypical Typical Atypical ( %) (%) (%) (%)

(1973 occupation greater % male) 91 to 100 0 1.9 0 0 81 to 90 0 5.2 0 0 71 to 80 0 5.7 0 0 61 to 70 0 17.1 0 0 51 to 60 0 18.1 0 0 41 to 50 1.5 25.7 0 0

31 to 40 . 2.2 10.5 0 4.8 21 to 30 3.5 12.9 0 7.9 11 to 20 5.9 2.9 0 3.2 1 to I() 18.6 0 0 11.1 (No change) 0 31.1 0 0 47.6 (1973 occupation smaller % male) - 1 to - 10 18.6 0 .6 15.9 11 to - 20 7.4 C 2.5 1.6 -21 to -30 5.1 0 5.0 4.8 -31 to -40 4.9 0 10.7 3.2 -41 to -50 1.2 0 23.3 0 -51 to -60 0 0 21.4 0 -61 to - 70 0 0 15.7 0 -71 to -80 0 0 11.3 0 -81 to -90 0 0 6.9 0 91 to - 99 0 0 2.5 0 Total' 100.0 100.0 99.9 100.1 N 1926 210 159 63 ° a sextypical occupation is one that is at least 51 percent male for men or less than 51 percent male for women. b Totals differ from 100 due to rounding. Characteristics by !dace, Sex, andTypicality of APPENDIX BMeans (and Standard Deviations)of Individual and job- Shifting Previous Occupation. 1973 CPSEmployer Changers" White Men Black Women White. Women Blackc Men Previous Occupation Previous 0±xlipation Previous Occupation Previous Omupation Typical Atypical Typical Atypical Tn.:Leal Atypical Typical Atypical

Highest year el education completed?' .20 .23 .20 .29 .17 .16 .18 .44 0-11 (.36) (.39) (.50) (.41) (.42) (.40) (.46) 1,18) .11 20 .26 .16 .17 21 .17 .14 1-3 years of college (.35) (.321 (.40) (.44) (.37) (.38) ( 40) (.38) .14 .21 .25 .06 .20 .15 .22 .11 >3 ye:a.% of college (.361 (.41) (.43) (.40) (.351 .41) (Al) ( 23) .05 .04 .05 .06 .07 .11 .12 .17 (.20) -11 Training while job searching (.26) (.32) ( 2!) (.32) (.38) (.22) (.24)

Age .56 .74 .60 .77 .63 .61 .63 .57 20-29 (.50) (.441 (.49) (.42) (.48) (.49) (.48) (.50) .14 24 .13 .26 27 .21 .25 .26 30-39 (.36) (.43) (.34) (.44) (.45) (.41) (.43) ( 44) .74 .77 .63 .66 .59 .70 71 .71 Currently married (.42) (.48) (.46) (.46) (46) (.44) (.43) (.50) .11 .24 .12 .13 .07 09 school in 1972 12 .12 (.31) (.43) In (.32) (.34) (_26) (.281 ( 32) (.33) .002 0 .28 .28 0 0 job .23 .17 (0) Taking care 01 a Lindy liore 1973 (0) (0) (0'.4) (.42) (.38) (.45) (.45) u

111 .18 *-12 different jobs in 1972 .06 .07 .12 .13 .15 .20 .21 (.24) (.26) (.32) (.34) (.36) (.41) (.40) (.39) Some time unemployed betbre 1973 job .20 .15 .08 .10 .14 .23 .16 .16 (.40) (.36) (.27) (.30) (.35) (.43) (.36) (.37) Going from .17 part-time to full-time wprk .19 .10 .14 .16 .05 .17 .07 (.40) (.30) (.35) (.37) (.22) (.38) (.26) (.38) .03 part-time to part-time work .04 .07 .12 .12 .03 0 .02 (.20) (.26) (.33) (.33) (.17) (0) (.13) (.16) .08 full-time to part-time work .12 .10 .14 .11 .06 0 .04 (.32) (.30) (.34) (.31) (.25) (0) (.19) (.27) 5.35 4.08 SVP of previous job 3.71 4.88 4.52 5.13 4.44 4.05 (1.64) (1.45) (1.63) (1.69) (1.40) (1.77) (1.45) (1.27) ,..-- .,., .40 Previous job white collar .40 .63 .63 .69 .25 .26 (.49) (.49) (.48) (.46) (.44) (.45) (.48) (.49) Tenure on previous job 3.34 3.56 3.04 3.13 3.32 2.88 3.19 3.07 (1.86) (1.84) (1.66) (1.74) (1.92) ;1.87) (1.82) (1.78) Region = South .59 .56 .33 .33 .55 .57 .32 .27 (.45) (.49) (.50) (.47) (.47) (.50) (.50) (.47) 35 2136 223 NI° 144 41 1506 503 141 ° An atypical occupation is ma- held by a woman that is more than 50 percentmale or one held by a man that is 50 percent male or less. b See Tables 4-5 and 4-6 for coding of variables. Ns on which means are based vary somewhat across variablesdue to missing data.

91 RACHEL A. ROSENFELD 84

Doeringer, Peter, and Michael Piore REFERENCES 1971 Internal Labor Markets and ManpowerAnal- ysis. Lexington, Mass.: 1). C.Heath. Anderson. Karen Dressel, Paula, and David Petersen 1981 Wartime Women: Sex Roles,Family Rela- tions, and the Status of Womenduring World 1982 "Becoming a male stripper:, Recruitment, so- Work War II. Westport, Conn.:Greenwood Press. cialization, and ideological development." and Occupations 9:387-406. Baker, Elizabeth York: Duncan, Otis Dudley, David L. Featherman,and Bev- 1964 Technology and Women's Work. New Columbia University Press. erly Duncan Goldberger 1972 Socioeconomic Background and Achievement. Barnow, Burt, Glen Cain, and Arthur New York: Seminar Press. 1980 "Issues in the analysis of selectivitybias." In Ernst Stromsdorfer and GeorgeFarkas (eds.), Edwards, Richard Evaluation Studies Review Annual,Vol. 5. 1979 Contested Terrain. New York: Basic Books. Beverly Hills, Calif.: Sage, pp. 93-59. England, Paula 1382a"Explanations of occupational sex segregation: Barrett, Nancy S. Unpublished 1979 "Women in the job market: Unemployment An inte disciplinary revi?w." and work schedules." In Ralph E. Smith(ed.), manuscript, University of Texas atDallas, De- The Subtle Revolution: Women atWork. partmer4 of Sociology and Political Economy. Washington, D.C.: The Urban Institute, pp. 1982b "The failure of human capitaltheory to explain 63-98. occupational sex segregation." Journal of Hu-. Beller. Andrea man Resources 17:358-70. Linda Mc- 1980 "Occupational segregation by sex: Determi- England, Paula, Marilyn Chassie, and nants and changes." Paperpresented at the Cormack Population Association of America annual 1982 "Skill demands and earnings in female andmale meeting, Denver, April 12. occupations." Sociology and Sccial Research 66:147-68. 1982 "Occupational segregation by sex: Determi- nants and changes." Journalof Human Re- Ekland, Paula, and Steven D. McLaughlin in- sources 17:371-92. 1979 "Sex segregation of jobs and male-female Rodolfo Alvarez, Ken- Perk, Richard, and Subhash Ray come differentials." In Social neth Lutterman, and Associates (eds.),Dis- 1982 "'Selection biases in sociological data." Science Research 11:352-98. crimination in Organizations. San Francisco: Jossey-Bass, pp. 189-213. Bielby, Denise Fligstein, Neil, and Wendy Wolf 1978 "Career sex-atypicality and careerinvolvement evidence 1978 "Sex similarities in occupational status attain- of college educated women: Baseline the restriction of from the 1960s." Sociology of Education51:7 - ment: Are the results due to the sample to employed women?" Social Sci- 28. ence Research 7:197-212. Blau, Francine Equal Pay in the Office. Lexington,Mass.: Greenberger, Ellen, :aid Lawrence Steinberg 1977 "Sex differences in early labor force experi- Lexington Books. 1983 ence." Social Forces 62:467-86. Cohn, Samuel Grimm, James, and Robert Stern 1982 "Synthetic turnover: A reinterpretationof hu- occupational sex-typ- 1974 "Sex roles and internal labor market structures: man capital models of Social Pi ob- ing:. Unpublished manuscript.Department of The 'female' semi-professions." Sociology, University of Wi-,consin-Madison. ferns 21:690-705. Cook, Alice, and ifiroko Hayashi Gruber, James, and Lars Bjorn "Blue-collar blues: The sexual harassmentcif 1980 Working Women in Japan. Ithaca, N.Y.. Cor- 1982 nell International Industrial andLabor Rela- women auto workers."Work and Occupations tions Report #10. 9:271-98. Hall, Robert E. Da% ies, !Margery U.S. "Women's place is at the typewriter.The fem- 1980 "The importance of lifetime jobs in the 1975 economy." Cambridge, Mass.: National Bu- inization of the clerical laborforce." In Richard Edwards, Michael Reich, and DavidGordon reau of EmnomicResearch Working Paper (eds.), Labor Market Segmentation.Lexing- #560. ton. Mass.: I). C.Heath. pp. 279-95 Hanushek. Eric A.. and John E. Jackson Statistical Methods for Social Scientists. New Daymont, Thomas, and AnneStatham 1977 and York: Academic Press. 1981 "Occupational atypicality: Changes, causes. consequences." In Lois Shaw (ed.), Dual Ca- Harlan, Sharon, and Brigid O'Farrell reers, Vol. 5.Columbus, Ohio: Center for Hu- 1982 "After the pioneers: Prospects foromen m nontraditional blue collar jobs." Work and Oc- m= man ResourcesResearch, Ohio State Uniser- sity. pp. 107.139. ctip.aions 9.363-86

P p OD CHANGING AND OCCUPATIONAL SEXSEGREGATION 85

Harrell, Frank ing Its Growth and Changing Composition. Population Monograph Series, No. 5. Berke- 1980 "The LOGIST procedure." In SAS Supple- mental Library User's Guide: 1980 Edition, ley: University of California. Cary, N.C.: SAS Institute, Inc., pp. 83-102. Phelps, Edmund Hartmann, Heidi 1972 "The statistical theory of racism and sexism." American Economic Review 62:659-61. 1976 "Capitalism, patriarchy, and job segregation by sex." SIGNS 1:137-69. Polachek, Solomon Hearn, James, and Susan Olzak 1978 "Sex differences in college major." Industrial and Labor Relations Review 31:498-508, 1981 "The role of college major departments in the reproduction of sexual inequality." Sociology 1979 "Occupational segregation among women: of Education 54:195-205. Theory, evidence, and a prognosis." In Cyn- Heckman, James thia Lloyd, Emily Andrews, and Curtis Gilroy (eds.), Women in the Labor Market. New York: 1979 "Sample selection bias as a specification error." Econometrica 47:153-61. Columbia University Press, pp. 137-57. Heyns, Barbara, and Joyce A. Bird 1981a "Occupational self-selection: A human capital 1982 "Recent trends in the higher education of approach to sex differences in occupational women." In Pamela l'erun (ed.), The Under- structure." Review of Economics and Statis- graduate Woman: Issues in Educational Equity. tics, pp. 60-69. Lexington, Mass.: Lexington Books, pp 43- 1981b"Occupational segregation: A 69. man capital predictions,"Unpublishlliman- Jusenius, Carol uscript, Universit*bNorth Carolina-Chapel 1975 "Occupational change, 1967-71." In Dual Ca- Hill, Department of Economics. reers. Vol. 3 Columbus,Ohio.: Center for Hu- Ray, Suhhash, Richard Berk, and William Bielby man Resource Research,Ohio State Univer- 1981 "Correcting sample selection bias for bivariate sity, pp. 21-35. logistic distribution of disturbances." Unpub- Kanter, Rosabeth Moss lished manuscript, University of California -Santa Barbara. Revision of a paper presented at the 1977 Men and Women of the Corporation. New York: Basic Books. 1980 meeting of the American Statistical As- sociation. Kreps, Juanita, and R. John Leaper 1976 "Home work, market work, and the allocation Rosenfeld, Rachel A. of time." In Juanita Kreps (ed.), Women and 1977 "Breaks in women's employment and labor the American Economy. Englewood Cliffs, N.J.: market re-entry." Paper presented at the an- PrenticeHall, pp. 61-81. nual meetings of the Canadian Sociology and Anthropology Association, Fredericton, New Levinson, Richard M. Brunswick. 1975 "Sex discrimination and employment practices: An experiment with unconventionaljob 1980 "Race and sex differences in career dynamics." inquiries." Social Problems 22:533-43. American Sociological Review 45:583-609. Lewin-Epstein, Noah 1983 "Sex segregation and sectors: An analysis of gender differences in returns from employer 1981 Youth Employment During High School. Re- port to National Center forEducation Statis- changes." American Sociological Review 48:637- tics, U.S.(Vepartment of Education. Chicago: 55. National Opinion Research Center. Rosenfeld, Rachel A., and Francois Nielsen Lewis. Diane 1984 "Inequality and careers: A dynamic model of socioeconomic achievement." Sociological Meth- 1977 "A response to inequality: Black women, racism and sexism.' SIGNS 3:339-61. , ods and Research 12:279-321. Malveaux. Julianne Rumberger, Russell, and Martin Carnoy "Segmentation in the U.S. labour market." 1982 "Recent trends in occupational segregation by 1980 race and sex." Paperpresented at the' Work- Cambridge Journal of Economics 4:117-32. shop on job Segregation by Sex. National Schmuck, Patricia, W. W, Charters, Jr., and Richard Academy of Sciences, Washington, D.C., May- 0. CAtion (eds.) 24-25. 1981 Educational Policy and Managewnt: Sex Dif- NIcIlwee, Judith ferentials. New York: Academic. Press. 1982 -Work satisfaction any ng women in nontra- Schreiber, Carol Tropp ditional occupations." Work and Occupations 1979 Changing Places: Men fiend Women in Tran 9.299-336. sitional Occupations. Cambridge. Mass.: MIT Olsen, Randall Press. 1980 A least squares correction for selecti% itbias. Shaw, Lois Econometrica 48.1815-20. 1981 "Causes of irregular employment patternsIn Lois Shaw (ed.). Dual Camrs. A Decade of oppenheimer. Valerie The Female Labor Force in the United States:L) Changes in the Lives of Mature Women, Vol. 1970 5. Columbus, 011;o: Center Human Re- Demogi And EC01101111 Factors Gea ern 93 86 RACHEL A. ROSENFELD

source Research, Ohio State University, pp. Strober, Myra, and David Tyack 79-106. 1980 "Why do women teach and men manager Sokoloff, Natalie SIGNS 5;494-503. 1980 Between Money and Love: The Dialectics of Sweet, James A. Woman's Home and Market Work. New York: 1973 Women in the Labor Force. New York: Sem-

Praeger. , inar Press. Spence, Michael Temme, Lloyd 1974 Market Signaling: Information Transfer in Hir- 1975 Occupation: Meankgs and Measurement. ing and Related Screening Processes, Cam- Washington, D.C.:ureau of Social Science bridge, Mass.: Harvard University Press. Research. Spenner, Kenneth Wolf, Wendy, andRachel A. Rosenfeld 1977 From Generation to Generation: The Trans- 1978 "Sex structure of occupations and job mobil- mission of Occupation. Unpublished Ph. D. ity." Social Forces 56:823-44. thesis, University of Wjsconsin- Madison.

94 5Commentary tts PAMELA STONE CAIN

Chapters 2, 3, and 4 view segregation from 60 percent of workers of either sex would three different vantage points. Beller pre- have had to change occupations in order to sents an analysis of aggregate occupational achieve identical male and female distri- trends. Bielby and Baron explore the or- butions. Indeed, over a 20-year period, the ganization of jobs within firms. Rosenfeld index dropped by only 5 percentage points. looks at the occupational shifts of job chang- Rosenfeld finds that over a 1- to 5-year ers. The resulting picture is a confusing one. period, 10 to 15 percent of the individuals Beller emphasizes the decline in segrega- in her sample of job changers moved from tion from 1960 to 1981, Rosenfeld the ex- an occupation typical of their sex to one that istence of significant (i.e., nontrivial) move- was atypical; 60 to 70 percent moved from ment across the boundaries of sex-typed anatypical to a sex-typical occupation. Thus, occupations. In contrast, Bielby and Baron over a relatively short period of time, only highlight the finding that segregation is vir- about 15 to 30 percent of job changers failed tually complete across jobs and firms of widely to cross the sex-typed boundary. These fig- varying characteristics. From their results, ures indicate a movement of individuals across it appears that only about 10 percent of segregated occupations that is not imme- workers. are employed in integrated job ti- diately obvious given the level and con- tles, and even in those jobs they often work stancy of segregation indices for the aggre- at different sites or with different clients. gate distribution. These results, however, Moreover, Bielby and Baron found no de- like those of Beller, are less cause for op- cline in job segregation over a period roughly timism when one realizes that fewer than 5 comparable to the one studied by Beller. percent of the 1973 Current Population Sur- How to reconcile these results." In part, vey sample changed jobs. Their behavior, the contradictions can 1w attributed to inter- theT1, represents a flux at the margin. Most ',roil; kin. Although the segregaion index workers were stayers not movers, and, judg- declined at an ever-increasing rate, Beller ing from the other papers in Part 1, they finds Out the resulting level of segregation stayed in sex-typical jobs. was still very high: in 1981 approximately The inconsistency in results between Beller

87 n r- 88. PAMELA STONE CAIN and Rosenfeld, on one hand, awl Bie lby and Act early in the period under study, Putting Baron, on the other, is also attributable to aside questions about using time - ordering different units of analysis. Be ller's and Ro- as a basis for attributing cause, the credi- senfeld's use of relatively aggregated occu- bility of this interpretation is undermined pations underestimates the degree of seg- by what we know about the slowness and regation,given the considerableheter- inefficiency of the enforcement process and ogeneity of jobs within even fairly detailed by the limited scope of some federal en- three-digit census categories. Beller and Ro- forcement efforts, especially contract com- senfeld recognize this but nonetheless adopt pliance. a more optimistic, "half full" interpretation Bielby and Baron's cross-sectional and of their results. In light of Bielby and Bar- longitudinal analyses suggest that more im- on's findings and the considerat, ms cited portant than regulatory effbrts are larger above, a -half empty" interpretation might structural changes in the economy that lead be more in order. to the creation of new types of work, to new This is not to deny that Beller and Rosen- forms of work organization, and/or to changes feld fbund evidence of change, however in in employee-employer relationships. They cremental and slow-moving. Moreover, each find less pronounced segregation in estab- found that sex-typed patterns were most lishments engaged in nonmanual service yielding among younger workers. Both also work, with unspecialized job structures and flame' that the decline in segregation was cem- an absence of formal bargaining agreements. centrated among certain occupations, espe- Larger firms with well- structured internal cially in the professional, managerial, labor markets as well as very small firms service, and health care sectors. Other sec- were extremely segregated. tors, particularly skilled crafts, appear to pre- Projecting these results and assuming that sent lininidable barriers to women's entry. the factors Bielby and Baron identify are Although Bielby and Baron found little causes not correlates of segregation, there evidence of change, a case-by-case analysis is reason to believe that segregation could of those companies dila did move toward decline as the national economy shifts from greater integration led them tt: tentatively a manufacturing to a service base and the identify several factors responsible fur, or at number of workers covered by collective least facilitative of, such change. High on bargaining drops. Countering.these trends, their list are rapid growth in company size however, are trends toward larger firm size and technological changes. To a lesser ex- and greater bureaecratization. Unfortu- tent, the presence of a sizable female work nately, the main effects of each factor can fOre also helpful. Surprisingly, federal he disentangled :mly multivariate analy- intervention in the form of contract com- ses and then only if the individual factors plianee regulations during the early period. an. not }Uglily correlated with one another. appears to hi've had no impact on segrega- Ultimately, itis the complex interpla-of tion. By enabling an empirical assessment these factors that will determine the direc- of the correlate's (if not the causes) of seg- tion of segregation in the 1950s, and itis regation. Bielby and Baron's unique firm- difficult to predict this outcome given our level fbcus contributes immensely to a bet- current level of understanding. Certionk, a ter understanding of the overall downward continuation of the downward trend that trend in segregation that Beller documents. Boller documents cannot be taken for Beller attribute's the decline primarily to granted, given the sconario that Bielby and equal employment opportunity legislation Baron depict. and enfOreement activities, citing the en- Confiminding these structural change's that actmnt of Title' VII and the equal Rights may affect segregation are the haep;es noted

96 COMMENTARY 89 by Rosenfeld in women's family and house- accomplished easily or without resistance hold responsibilities, which have been brought (see, for example, Roos and Reskin). on largely by inflation,unemployment, and Turning to the research implications of divorce. As their paychecks become increas- these papers, Bielby and Baron's paper re- ingly critical to household survival, women minds us of the distinction between re- can even less afford, if indeedthey ever could, search questions and researchable ques= "women's work. lions. On the face of it, such guiding questions Given that naturally occurring trends have as "Why is there job segregation?"and "What opposing implications for segregation and causes it?" are plausible until it is recog- that many previous interventions appear to nized that the level of job segregation may have had little effect, what policy implica- not vary substantially. As a virtual constant, tions can be drawn from these papers? Bielby job segregation defies social science, as all and Baron's contention is persuasive that, our empirical methods determine causeand with such widespread segregation, any 'in- effect through covariation of variables. Thus, tervention is warranted. What, then, should in the absence of variation, we cannot really be the focus of intervention? Beller views "explain" segregation. the sex composition of occupations as es- This predicament may direct us to explore pecially amenable to manipulation, presum- phenomena that do vary to see what they ably through the use of quotas and various -can tell us about sex segregation. For ex- affirmative action strategies. Enhancing the ample, particular occupations and jobs have success of this strategy wouldbe the changes changed sex type, i.e., are no longer mainly in younger workers' job preferences that are female or male. Newly emerging occupa- already taking place. tions such as computer programming, as Bielby and Baron's results suggest that Beller finds, are becoming rapidly identified the solution may not be straightforward with one sex. Bielby and Baron's work also Fundamental forms of workplace organiza- points out that, although the level of sek, tion appear to be in need of alteration if regation does not vary much from firm to segregation is to abate. Some of the crucial firm, specific jobs may he female in one firm factors they identify (e.g., firm size, degree and male in another. These examples lead of job differentiation, and unionization) are to interest in historical or comparative in- outside the scope of government interven- quiries into the determinants of the sex com- tion as typically conceived. If their analysis position of a job or job family. let; is correct, the greatest hope for future in- A new research question dictates changes tegration may lie not in the public policy in methods. The importance of firms in Bielby domain but in the hands of workers them- and Baron's work establishes the importance selves, through either more enlightened of using jobs rather than occupations as the collective bargaining or other mechanisms unit of analysis and the concomitant need of workplace democracy. The complex or- for firm-specific data. This is not merely be- ganization of the contemporary workplace cause the use of job data uncovers more appears to play a major role inmaintaining segregation, but because job-level analysis segregation. This organization can perhaps picks up organizational context and varia- be changed only from within. tions in hiring and allocation that occupa- Women moving into professional and tional-level analysis does not. Firms appear managerial positions may be a catalyst in to exert a strong effect on occupations, and transforming the traditional organizational disregarding their context may obscure more structures that have functioned so long to than it illuminates. For example, it is dif- exclude them. As other papers in this vol- ficult to know what the occupational ume suggest, however,change will not be Rosenfeld measured represent, especially as 1

97 90 PAMELA STONE CAIN

we do not know whetherthey were volun- New York, 1377) and others, and, in their tary or involuntary moves. Moreover,the careful attention to selected firms (outliers), high rate of movement across sex-typed oc- they blend u.se study and survey method- cupations may be misleading. A move to a ologies. The richness of their analysis should sex-atypical occupation is not necessarily a settle the debate between those who ad- , move to an integratedjob. vocate one method over another.'Clearly we Focusing on jobs, not occupations, raises need both, especially in this area, because problems of data availability. Few individ- of the problems of obtaining valid survey ual-level surveys have any information on data on jobs and firms from either workers respondents' jobs or firms. At firms them- or employers. selves, it is often difficult to obtain coop- Finally, how do these studies enlighten eration and access. These problems point to the major debate of whether segregation is 3^ the need for new data collection efforts to the result of employee choice or employer supplement existing surveys. More broadly, discrimination? These papers offer evidence the difficulties Beller encountered in com- for both positions. Within firms, Bielby and piling a series of data on occupational seg- Baron document the existence of powerful regation should alert us to the need for im- mechanisms of control that would support proved national statistical reporting systems an employer-sideexplanation. Among work- that would enable us to monitor this critical ers, Rosenfeld showsconsiderable circula- indicator of sex equity. But prospects for tion across sex-typed occupations. More- better data at the national level are not en- over, she finds someevidence that male couraging, given the current political cli- employees are ma'Aing rational choices in mate. The agency within the.U.S. Depart- their avoidance of lower-paying, female- ment of .Labor, for example, thatcollected dominated occupations as their family re- the data Bielby and Baron used in their anal- sponsibilities increase. The question of ysis has now been all but shut down. causeemployee or employeris un- Bielby and Baron's paper also under- doubtedly too simply framed. The different sZores the need for new analytic strategies. levels of analysis and seemingly contradic- Their survey analysis complements case tory findings of these papers highlight the studies by Rosabeth Moss Kanter (Men and complexity of the etiology 'of job segrega- Women of the Corporation, Basic Books, tion.

4%. a

Occupational Sex Segregation: 1/ Prospects for the 1980s

ANDREA H. BELLER and

. KEE-OK KIM I-IAN

Occupational segregation declined 2 to 3 tegration and 100, compkte segregation. The times as rapidly during the 1970s as during number tells the proportion of women (or the 1960s (Beller, in this volume). While of men) that would have to be redistributed this is encouraging, segregation continues among occupations for the occupational dis- at. a high level overall; as of 1981, 61.7 per- tribution to reach complete equality be- cent of women (or men) would still have to tween the sexes. change occupations for the occupational distribution to become completely integrated ASSUMPTIONS, DATA, AND by sex (Beller, in this volume). What pros- METHODOLOGY pects lie ahead for the remainder of this decade? To attempt to answer. this question, In order to project the index of segrega- we shall make several projections of occu- tion, we need to know the projected.dictri- pational segregation to 1990. These projec- bution of employment across occupations and tions are based on the trends in segregation the projected sex composition within each analyzed and reported in Beller (in this vol- occupation. Occupational employment pro- ume). jections for 1990, constructed by the Bureau The best measure of occupatimal segre- of Labor Statistics (BLS), are published in gation, the segregation index, is computed Volume 2 of The National Industry-Occu- as follows: pation Employment Matrix (Department of Labor, BLS, 1981). These projections are St = 1/2Emit fit (1) based upon the bureau's intermediate labor force projections and assume a 4.5 percent where mit is the percentage of the male labor unemployment rate in 1990. These projec- force employed in occupation i in year t, tions are discussed further in Appendix A. and fie is the percentage of the female labor In constructing projections on the sex force employed in occupation i in year t. composition within occupations, we employ The index may take on a value between a number of techniques and entertain a va- 0 and 100, where 0 represents perfect in- riety of alternative assumptions. The basic

91 99 92 ANDREA H. BELLER AND KEE-OK KIM HAN

assumption underlying most of these pro- pends on the initial proportion of males as jections is that the sex composition of all well as on the growth in the occupation over occupations or of each occupation individ- the period. It is easier for women to enter ually continues to change during the 1980s growing occupations than to enter stable or along the path established during the 1970s. declining ones. Moreover, the effect of the The data on which we base our projections initial proportion on the present proportion come from the Current Population Survey might depend on the growth rate of the oc- (CPS), collected monthly by the Bureau of cupation. This model also has some draw- the Censtis either from the March Annual backs. Because it averages over all occu- Demographic File (ADF) or from the un- pations, it will overestimate change in some published annual averages (AA) computed occupations and underestimate change in by BLS from the monthly data. First, we others. Because it is linear rather than lo- make three projections based upon infor- gistic, the projected values for the propor- mation about the labor force as a whole, tion of males could exceed 1.0 or be nega- employing linear and logistic models. Then tive. we make four projections based upon infor- To eliminate the effects of averaging, in mation for age cohorts; the equations are our two other labor force projections we as- specified according to the linear spline model. sume that each occupation's sex composition All of these projections are based upon 255 is a function of time. 3-digit census occupations. Linear Individual Labor Force Projection (P2) Linear Group Labor Force Projection (P1) First, we specify the percentage of males First, as in Blau and Hendricks (1979), in each occupation as a linear function of we assume that the sex composition ir all time: occupations changes over time according to the same linear function. We estimate an Pt = a + bt + (3) equation in which .the proportion of males where t is 1, 2, ...11 for years = 1971 to 1974, in each occupation in 1981 is a linear func- 1977, and 1981.' But, since p, is a fraction tion of the proportion of males in that oc- between 0 and 1, the linear function might cupation in 1972, of the percentage change not be a good model, particularly near the in employment in that occupation over the extremes. period, and of the interaction between the proportion of males and the percentage change in employment. Using the AA data, the followingfollowing equation is estimated: /equation P1.81 = + 'YAE1.81- 72 Eqs. (3) and (5) require a time series data set. Data 8(M,72 * AE1.81-72) + et, (2) for 1971 to 1974 and 1077 are from the Annual De- mographic Files (ADF) of the monthly CPS, while 1981 where po is the percentage of males in oc- data are from the unpublished BLS annual averages of cupation t in year t = 1972 or 1981, and the monthly CPS. The former are the onl) years for AE1.81- 72 is the percentage change in em- which we have the ADF, while the latter is the most ployment in occupation i between 1972. and recent year of data available. The AA data are somewhat more reliable statistically than are the monthly data. 1981. Eq. (2), which requires data only for the end points, This model has a logical rationale. The is estimated with AA data alone, since we can remain proportion of males in an occupation de- within a single. statistically more reliable data set.

100 p.

OCCUPATIONAL SEX SEGREGATION: PROSPECTS FOR THE 1980s 93

needed to identify work experience cohorts, LogistiC Individual Labor Force Projection we make our projections based upon age (P3) cohorts. We project to 1990 the sex com- The logistic equation constrains the value position of individual occupations for each of p, to lie between 0 and 1: age group. Then we use the BLS projections on cohort size to aggregate over groups and Pr = (4) to obtain the sex composition of each oc- 1 + cupation for the labor force as a whole. Com- where fl is a vector of parameters to be es- bining these with the BLS occupational em- timated, and t is the year as above. Eq. (4) ployment projections, we compute the can be rewritten in the "log odds ratio" form: projected segregation index. The details of 'this approach are described in Appendix A. In( = (3t . (5) The advantage of using these projection pi methods is that we can incorporate specu- This equation, which can be estimated by lations. about what might happen under al- ordinary least squares (OLS), will have het - ternative scenarios for such factors as federal eroscedastic residuals, increasing the stand- efforts on affirmative action. The disadvan- ard errors; however, this is not important tage is that projections for specified small for puiposes of prediction.2 subgroups of a population are likely to be A different approach to projecting the less reliable than are projections for the whole segregation of the work force in 1990 is to population. We make four projections based piece together information about population upon age cohorts: conservative, moderate, subgroups. Beller (in this volume) examined moderately optimistic, and optimistic.3 All occupational segregation by work experi- cohort projections are based. upon trends ence cohort and found that (1) new entrarts between 1971 and 1977 only; the latter is into the labor market are less segregated the most recent year for which we have the than is the rest of the labor force, and (2) ADF data containing the demographic de- between 1971 and 1977, occupational seg- tail to identify age cohorts. regation declined for the entering cohort as it aged and between entering cohorts, Conservative Age Cohort Projection (P5) Changes in occupational segregation can be projected by cohort and aggregated to the The conservative projection assumes that labor force. In order to accomplish this, we no further change occurs for a given cohort need to know the age7sex specific compo- after 1977, partly because equal employ- sition of the civilian labor force in 1990; for- ment opportunity (EEO) efforts have slowed tunately it has been projected by the BLS down. Each cohort maintains the same sex (Department of Labor, BLS, 1979). Since composition within occupations as it ages, we do not have the additional data for 1990 and the youngest cohort (16 to 24 years of

2 This equation should be modified to include an 3 The assumptions about behavior and policy under- equation error, interpreted as a surrogate for omitted lying each of these projections do not generate the variables, in addition to the usual error term. We will particular mathematical models we use. Rather the assume that the variance of the equation en-or equals models postulate change as a linear function of time, 0, as discussed in Medoff (1979). In his empirical results since affirmative action policies and related behavior the estimates of the equation modified to include equa- have changed over time. More complex mathematical tion error were not much different from those that were models might also be consistent with the underlying not. assumptions about change.

101 JI

94 ANDRE H. BELLER AND KEE-OK KIM IA

age) has the same occupational sex compo- position of highly male (more than 85 per- sition in 1990 as in 1977: cent male) and highly female (lessthan 15 percent male) occupations will change atdif- Pi,j+1(1996)= Pi,j (1977), (6) ferent rates than will occupations with sex and compositions in between. We estimate the following equation: pi(1990) = Pi.i (1977), (7) pc1(1977) = a1 + 1317)1,1(1971) where j is age cohorts 1 to 5, defined by the + '00(1971) + 81M1,1(1971) + ui , (8) age groups in the 1990BLS projections, shown in Appendix A. where The assumption of no further change for F1,1 = 0.00 if pi .1 = 0.15 to 1.00, imply no change an individual cohort need not pci if pf,1 < 0.15, for the labor force as a whole as long as = 0.15 younger cohorts are lesssegregated than older and ones are. As thelabor force ages, younger, M1,1 = 0.00 if pi.1 = 0.00 to 0.85, less segregated cohorts replace older, more 0.85 if pi,i > 0.85. segregated ones. Nonetheless, the assump- = pc' tion of no change for all cohortsafter 1977 This equation is estimated for a six-year is quite conservative inlight of the decline period, 1971 to 1977, and thus can be used in segregation through 1981shown by the to predict the value of pi3(1983) using1977 aggregate labor force data (Beller, inthis data for the independent variables. The 1977- volume, Table 2-1). Thus, these conserva- 1983 growth rate in the proportion of males tive assumptions may be viewed asyielding in occupation i for this cohort canthen be a lower-boun`cLestimate onthe projected de- used to predict pi,1(1990). cline in the index of segregation during the 1980s. Optimistic and Moderately Optimistic Age Cohort Projections (P8 and P7) Moderate Age Cohort Projection (P6) The optimistic projection (P8) is con- The moderate projection is constructed structed under the assumption that affirm- under the assumption that the rate ofchange ative action, attitudes, and other factors con- in the sex composition of occupationsfor the tinue to change during the 1980s at the same youngest (entering) cohort willbe.the same rate as during the 1970s.Actually, this is between 1977 and 1990 as it was between quite optimistic given what we already know 1971 and 1977a period of considerable about the Reagan administration's proposed change. We might expect this if youthful .cutbacks in affirmative action and de-em- attitudes and aspirations have changed, but phasis on enforcement. Thus, we consider equal opportunity efforts subside so that, aS this t6 be an upper-bound estimate on how it becomes older, the rest of thelabor force much change could occur under the best of remains as segregated as it was in 1977.This circumstances. For P8, we assume that as projection applies Eq. (6) to oldercohorts each 1977 cohort ages to 1990 its rate of and projects change for the youngest cohort change in percentage of males in each oc- according to a linear spline function esti- cupation is the same as for the similar cohort mated on 1971 and 1977 data. The linear as it aged between 1971and 1977, and that the rate of change between entering cohorts e spline allows different segments of a continuous linear function tohave different slopes in 1977 and 1990 is the same asbetween (Poirier, 1976). It is likely that the sex corn- entering cohorts in 1971 and 1977. For these

102 1

OCCUPATIONAL SEX SEGREGATION: PROSPECTS FOR THE 1980s 95 projections, we apply Eq. (8) to the young- the value of pi,j +1(1990), the optimistic pro- est cohort and estimate the following equa- jection, P8. tion for older cohorts: p1,j4.1(1977) = aj + pipci(1971) PROJECTIONS OF OCCUPATIONAL + yjFki(1971) + BjMi, j(1971) + uj (9) SEGREGATION, 1981 TO 1990 If the mechanisms have been put into place, Segregation indexes computed using BLS then Reagan's policies may succeed only in occupation projections and our linear pro- reducing btl not in eliminating change. We jections on the sex composition of occupa- consider it moderately optimistic (P7) to as- tions for the labor force yield only a modest sume that the rate of changefor each cohort decline in occupational segregation between during the 1980s is one-half the rate during 1981 and 1990. As shown in Table 6-1, col- the 1970s. The value of p,, j.,. 1(1983) provides umn 1, the segregation index, is projected the moderately optimistic projection, P7, and to decline by 1.69 percentage points to 59.97

TABLE 6-1Actual and Projected Segregation Indexes, 1972, 1977, 1981, and 1990 Employment Year Unstandardized Standardized 1972 68.32 67.23 1977 64.15 64.02 1981 61.66 61.66 1990 Labor force Linear-group (P1) 59.97 59.35 Linear-individual (P2) 60.37 59.51 Logistic-individual (P3) 56.06 55.20 Combined-individual (P4) 59.91 59.03 Age cohort Conservative (P5) 62.11 60.89 Moderate (P6) 57.29 56.02 Moderately optimistic (P7) 50.02 49.09 Optimistic (P8) 42.20 41.33 NOTES: P1 was computed based on the following equation for projectingpercentage of males in occupation i:

p,,m, = -.016 + .970 ps,72 .0364 ELM- 72 .042 (PI,72AE1.81 -12) (165) (7561) (1 64) (1.36) R2 = .966, N = 262, t-values in parentheses. P2 and P3 were computed based upon separate -linear or logistic equations for each occupation,respectively. P4 was computed by selecting the equation for each occupation with the highest R2 If either functional form was significant; if neither was significant assuming the 1981 value. P5 assumes no further change after 1977 in the percentage of males in each occupation l'or each age cohort; P6 assumes no change for all but the youngestcohort, which experiences a linear change for all but the youngest cohort, which experiences a linear rate of change; P7 assumes a linear rate of change of percentage of males in each occuaptionfor each age cohort at one-half the rate during the 1970s; P8 assumes a linear rate of change at the same rate as during the 1970s. Employment standardized- indexes are standardized to 1981 employment totals. Projected segregation indexes include only 255 occupations because the BLS employment projections were unavailable for 7 of the occupations 'included in the analysis of trends in Beller (in this volume). SOURCES: Annual Demographic Files of Current Population Survey, 1972 to 1975 and 1978, computer tapes; and Bureau of Labor Statistics, annual averages of monthly Current Population Surveys, 1972, 1977, and 1981, unpublished. tabulations.

10.3 96 ANDREA H. BELLER AND KEE-OK KIM HAN

or by 1.29 percentage points to60.37 according to the linear-group (P1) and linear- cording to the linear-group (P1) and linear- ized index during the 1970s. Thus, accord- individual (P2) projections, respectively. ing to logistic trends, changes in the sex These projected rates of decline in the over- composition of occupations toward a less all index are much slower than during the segregated work force will be larger duri-4 past decade. A logistic rate ofgrowth (P3) the 1980s than during the 1970s; however, in the sex composition of individual occu- projected adverse-changes in the occupa- pations combines with the BLS occupation tional distribution will more than offset these projections to predict a decline in the seg- more favorablechanges in sex composition. regation index from 61,66 in 1981 to 56.06 From these data we conclude that projected in 1990. This projected 5.6 percentage point changes in the sex composition of occupa- decline is around 84 percent of the actual tions during the 1980s based upon changes 6.66 percentage point decline between 1972 between 1971 and 1981 will decrease oc- and 1981. cupational segregation less than in the past To decompose these changes, we stand- in part because of opposing changes in the ardize the projected segregation indexes to occupational distribution toward increased the 1981 occupational distribution.4 Shown segregation. in column 2, the employmentstandardized However, these projections of occupa- segregation indexes lie slightlybelow the tional sex composition may be somewhat off, unstandardized indexes indicating that the because they project the female share of the occupational distribution is projected (by the labor force in 1990 as higher than the BLS BLS) to change slightly toward more seg- projects them (Department of tabor, BLS, regated occupations between 1981 and 1990. 1979, Table 5, p. 7). The BLS projects the This contrasts with the slight change in the female share of the labor force to grow from occupational distribution toward less seg- 41.0 in 1977 to 45.5 in 1990, whereas P1 regated occupations during the 1970s. The projects the female ,rare (aggregatedfrom projected employment standardized segre- occupational shares) to be 49.3 in 1990, P2 gation index based upon logistic trends(P3) projects it to be 48.2, and P3 projects' it to declines by 6.46 percentage poirlts during be 50.3. Since projections for a population the 1980s to 55.20, or by more than the 5.57 ,(the.Iabor force) tend to be more accurate percentage point decline in thestandard- than projections for a specified subset of that or by 1.29 percentage pointsto 60.37 ac- population (occupations), the BLS projections suggest that our own projections somewhat overestimate the female share of tie employment standardized inch% of segregation labor force in 1990.5 is defined as follows: What, then, do these projections tell us? The sex composition of some occupations s: = 1/2E . 4 changed so rapidly during the 1970s that the where rate cannot be sustained duringthe 1980s (MdT) (T_1) (100) on the basis of projectedgrowth in the fe- E(M/T)(T_ I) male labor force. For which subset of occupations will the female rate of entry de- (F,,IT) (T ,) (low cline during the 1980s? Will it decline further ft L (Farn) (T,g -1) F is the number of females in occupation i in year t. 5 This difference may he somewhat mitigated by the M is the number of males in occupation i in year t, fact that the BLS has consistently underestimated the and T equals F + M equals total employment in growth rate of the female labor force (Lloyd and Niemi. occupation i in year t. 1979. p. 311, n. 19; Smith. 1977. p. 23).

104 OCCUPATIONAL SEX SEGREGATION: PROSPECTS FOR THE 1980s 97 in the large, traditionallyTemale occupations pations shows no significant trend in the so that not only will the proportion of the percent female, more than one-fourth, 27.5 female labor force that is in them decline, percent, shows a significant increase. These but also the female proportion of these oc- assumptions yield the combined-individual cupations will decline? If so, occupational projection (P4) of the segregation index of segregation will decline more than is pro- 59.91, only slightly below P2. jecte' Or will it decline in the traditionally male occupations to which access has been MAJOR COMPONENTS OF PROJECTED, increased so recently? If so, occupational CHANGE IN THE INDEX segregation will decline less than is projected. Obviously incentives can be created Many of the trends toward decreasing by public policy for movement in one di- segregation begun in the 1970s (Beller, in rection or the other. If affirmative action this volume) are projected to continue into continues to promote equal opportunity for the 1980s. However, counterbalancing these women in nontraditional occupations, then will be a major new source of increasing traditionally female occupations will be: segregation. According to the P2 linear pro- come relatively less attractiveIf opportu- jections, a number of predominantly male nities decline in nontraditional jobs, then crafts and.operative and laborer occupations female occupations will look relatively, are projected to grow (some rapidly) and to attractive. As our experience during the'1970s account for increasing proportions of the male suggests, federal policy can significantly af- labor force during the 1980s. According to the P1 linear projections, if women take a fect occupationalet segregation (Beller, 1982a, b). larger share of the growth in these jobs as Each of these projection methods has ad- they did in the male occupations that greiv vantages and disadvantages, and each makes rapidly during the 1970s, then these occu- projections that are quite reasonable for many pations may not increase overall segrega- individual occupations. The projected fe- tion. Several traditionally female occupa- male percentages of P1 to P3, as well as the tions are projected to continue to account estimated annual linear trend of P2, are pre- for decreasing proportions of the female la- sented in Appendix B, Table B-1. As these bor force, while some increase in segrega- data show, the logistic growth model, P3, tion will result from the expansion of the does very well in capturing the acceleration predominantly female health services oc- or deceleration in the rate ofchange in per- cupations. centage of females near the tails of the dis- According to P2, occupations projected to tribution; however, it overprojects the fe- continue contributing to declines in the seg- male share ofhe labor force the most. The regation index during the 1980s are ele- linear-individual projection, P2, overpro- mentary school teachers; telephone opera- jects the female share the least. Since the tors; cooks, except private household; child truth probably lies somewhere in between, care workers, privatehousehold; and maids we construct a combined estimate. Foreach and servants,, private household. Declines occupation we choose the individual equa- in the segregation index during the 1980s tion with the highest R2, where at least one are projected from some new sources aswell: is significant at the 10 percent level accord- secondary school teachers; managers and ing to the F-statistic. Where neither the lin- administrators, not elsewhere classified; ear nor the logistic equation issignificant, bookkeepers; waiters; and hairdressers and we assume that no change occurs in percent cosmotologists. Also projected to decrease female between 1981 and 1990. As shown the segregation index during the 1980s are in Table B-1, While the majority of occu- two female occupationsregistered nurses 105 98 ANDREA H. BELLER AND KEE-OK KIM HAN

and bank tellersthat increased it during to women, the P1 projections show that the the 1970s. percentages of females could increase among The occupation of miscellaneous clerical carpenters (to 8.3 percent); auto mechanics workers is projected to continue increasing (to 6.8 percent); heavy-equipment mechan- the segregation index by a large amount. ics (to 9.4 percent); welders and flame-cut- Also projected to increase the index during ters (to 11.2 percent); machine operatives, the 1980s is the rapidly growing secretarial miscellaneous (to 34.1 percent); truck driv- occupation, which decreased it during the ers (tO 9.2 percent); and construction labor- 1970s. Several female health services oc- ers (to 9.1 percent). Increasing the tate at cupations are projected to grow and to be- which women enter those blue-collar oc- come more segregated: health aides,onurs- cupations that are projected to grow during r ing aides, and practical nurses.6 the coming decade should be.'a major focus As mentioned before, the biggest change of any public policy designed to reduce oc- projected for the 1980s that will hinder fur- cupational segregation.' It; ther declines in occupational segregation is substantial growth in a number of highly mare crafts and operative and laborer oc- PROJECTIONS BASED ON AGE COHORTS cupations. The following occupations, with The four segregation indexes based upon the projected percentage of females in pa- projections of occupational sex composition rentheses, will add a large amount to the within age cohorts vary considerably de- 1990 segregation index if their female per- pending upon the assumptions. Based upon centage grows during the 1980s at the same conservative assumptions, P5 projects only linear rate as during the 1970s: carpenters a slight decline in the index of segregation, (3.0 percent); auto mechanics (0.7 percent); from 64.15 in 1977 to 62.11 in 1990, slightly heavy-equipment mechanics (2.8 percent); above the actual 1981 index. (The projected welders and flame-cutters (5.3 percent); ma- indexes for each age cohort, as well as the chine operatives, miscellaneous (27.7 per- indexes for 1971 and 1977, appear in Ap- cent); truck drivers (4.0 percent); and con- pendix B, Table B-2.) Note that these construction laborers (4.5 percent). Of these, servative assumptions yield a projection only women made significant inroads during the slightly above P1, P2, and P4. The con- 1970s only into the occupations of carpen- servative projection appears especially so in ter, heavy-equipment mechanic, and. truck light of the actual 1981 segregation index. driver. In all except machine operatives, the Based upon moderate assumptions that percentage of females is projected to grow further declines occur only for the youngest between 1981 and 1990, but by nearly im- cohort after 1977, P6 projects a decline in perceptible amounts. However, growth in the index of segregation to 57.29 in 1990, these occupations need not hinder declines closest to the logistic projection, P3, of 56.06. in occupational segregation; if these male (P6 projects a female share of the labor force blue-collar occupations respond to gross th of 43.3 percent in 1990.) While the linear in the same way as the male white-collar spline equation predicts a substantial drop occupations did during the 1970s, allocating a higher share of new than of existing jobs Whether this projected growth in blue-collar jobs will he realized is questionable because of the present 6 These occupations were identified on the basis of high unemployment rate ontrasted with the 4.5 per- the linear trends for each individual occupation; they centage rate embedded in the IlLS occupation projec- carry the assumption that the percentage of females tions for 1990. Moreover, Pis Carey (1981. p. 42) points during the 1980s follows the same trend as during the out, job growth in blue-oillar occupations is more sen- 1970s regardless of any changes in total occupational sitive to the underlying assumptions of the projections employment that might occur. than in other major occupational categories.

-406 OCCUPATIONAL SEX SEGREGATION: PROSPECTS FOR THE 19808 99 in segregation for the youngest cohort (see dicted for the 1980s. The truth may well lie Table B-2), even changes of substantial mag- in between. In the next section, we focus nitude restricted to a single cohort have a on changes in thecharacteristics of the sup- limited impact on the overall index. It would ply side of the professional labor market. take many years of a continued influx of less segregated cohorts for the overall occupa- PROJECTIONS TO 1990 OF SEGREGATION tional distribution to show a major decline AMONG PROFESSIONAL OCCUPATIONS in segregation. BASED ON COLLEGE MAJORS The moderately optimistic and the optimistic age cohort projections, P7 and P8, In our final set of projections on segre- depart from the others in projecting a sig- gation among professional occupations we , nificant decline in the index of segregation break with previous methodological pat- during the 1980s. Based upon the assump- terns. So far we have projected trends in tion that the total change in percentage of the sex composition of occupations either for males for each occupation as a cohort ages the whole labor force or for a subgroup of between 1977 and 1990 is the same as for that population. Here we use an aggregate the similar cohort between 1971 and 1977, time series regression framework in which i.e., one-half the rate of change, P7 projects the segregation indexes are variables. Since the index of segregation overall to decline professional occupations generally require a by 11.68 percentage points to 50.02 in 1990. college degree, segregation among profes- P7 projects the female share of the labor sional occupations is hypothesized to be a force at 47 percent, which is close to the function of segregation by college field of BLS projection o:45.5. Projection P8 is very study among recent bachelor's degree re- optimistic and assumes that the rate of change cipients. As long as the sexes face equal op- in percentage of males by occupation for portunity in the job market, to the extent each cohort between 1977 and 1990 is the that a greater similarity arises in the edu- same as between 1971 and 1977for the com- cational preparation of men and women, parable cohort, i.e., double the total change. segregaticl in professional occupations should P8 predicts a rather substantial drop in the decline. We have chosen to specify a rela- segregation index of nearly 20 points to 42.20 tionship where segregation in professional and projects a female share of the labor force occupations in year t is a function of seg- just over 50 percent. regation in college majors in year t3, a lag The age cohort projection difibr from the of three years. While many of last year's labor force projections in their emphasis upon college graduates fill this ybar's job vacan- source of change. They assumethat segre- cies, several previous years' graduates may gation is a characteristic of cohorts of indi- also, especially where postgraduate educa -. viduals rather than of occupations or of the tion or training is involved.8 We thus spvify labor force. Thus, change is based more upon the following equation: the characteristics of the supply side of the = a ÷ b Sr_.3 + et , (10) labor market than, as in the earlier projections, upon the demand side. If, on the one where S° is the index of segregation among hand, segregation were primarily the result of the choices of each sex, then the optimistic projections suggest that occupational 8 We postulate a relationship for a single year, be- segregation could decline significantly dur- cause the number of years for which wehave compa- ing the 1980s. If, on the other hand, seg- rable time series data is limited. If more years of data were available, a more complex distributed-lagmodel regation were primarily the result of em- in which occupational segregation in year t was a func- ployer practices and other demand-side tion of several previous years of segregation among factors, modest further declines are pre- college graduates might be desirable.

107 100 ANDREA II. BELLER AND KEE-OK KIM II AN professional occupations in year t,ST__3 is gation indexes for professional occupations the index of segregation among bachelor's increase (in 1982) and then decline very degree recipients bymajor field of study in slowly thereafter. In fact, the projected seg- year t3, and t is 1974 to 1981. regation index for professional occupations The data on the size and sex composition of 48.36 in 1990 based on the NCES pro- of college majors are taken from the National jections lies only slightly below the. actual Center for Education Statistics' (NCES, 1980) 1981 index-49.61. This value is slightly Projections of Education Statistics to 1988- above one for professional occupations com- 89, and the data for professional occupations puted with the earlier combined-individual are from the published AA data. (Employ- projection (P4), 47.05. We suspect from the ment and Earnings, various issues). The distinct break in the series between 1978 model and data are discussedin more detail and 1979 that the NCES projections under- in Appendix C. estimate the extent of change in the sex com- First, we computed a time series of seg- position of college majors.'° regation indexes for college major among Assuming that.the number of majors by bachelor's degree recipients and for profes- field has been correctly projected by NCES sional occupations, which appear in Table but that the sex composition has not, we 6-2, columns (1) and (2), respectively. With estimate the trend in the sex composition these data we estimate Eq. (10); the esti- for each college major as a linear function mated equation with t-values in parentheses of time and then project it into the future. is:9 We estimate equations of the following form for each major field: =23.88 + 0.727 Sren_3. (23.50) (29,20) (11) PF? = a. + + et , (12) 112=0.99, N = 8. where PF, is the percentage of females among Initially we used the NCES projections bachelor's degree recipients in a major field

.of the number of female and male college in year t,t = 1969 to 1978, and f3 is the graduates by field of study through 1989 estimated trend. (NCES, 1980, pp. 66-71) to compute pro- The estimated trends in percentage of fe- jected segregation indexes for college ma- males by major iield of study are presented jors, as shown ih Table 6-2, column (3). In in Appendix B, Table B-3, along with the contrast to the increasingly rapid decline in actual 1969 and 1978 values and the pro- segregation among college majors. during the jected 1989 values. These estimates 'reveal 1970s (column [1]), the NCES projections two major trends among college students: show a slight increase in segregation in 1979 and only a modest decline thereafter (column [3]). Then, using Eq. (11), we computed the projected segregation indexes for 10 To check this we computed the actual segregation professional occupations, which appear in index among college fields of study of bachelor's degree recipients for 1979-1980 directly from data from Earned column (4). Likewise, the projected' segre- Degrees Conferred 1979-80 (NCES, 1981). (The categories used for 1978-1979 data were not comparable with those published in the NCES projections volume.) The actual index for '1980 was 34.53, below that based 9 We also estimated this equation for 1-, 2-. 4. and upon NCES projections. While these data are not strictly 5-year lags. and found the explanatory power greatest comparable with earlier data, .because the NCES sup- for the 3-year lag. Given the linear trend in the data, plemented the data from Earned Degrees Conferred a 3-year lag is approximately equivalent to a 5-year with information from additional sources (NCES, 1980, moving average. p. 49), they are nevertheless suggestive. .

108 1 r. .1

pCCOPATIONAI; SEX SEGREGATION: PROSPECTS FOR THE.1980s 101 a /te.' TABLE 6-2Actual and Prokd Segregation Indexes for College 'Majors and ,Professional Occupations ° Actual 14ojection I Projection II Projection,,III College Professional College Professiozial College Professional College Professional Majors Oecupationg Majors Occupations Majors Occupations Majors Occupations Year (1) (2) (3)' (4) (5) (6) (7) (8) 1969 46.08 1970 45.45 `1971 44.56 1972 44.17 58.94 1973 43.32 1974 41.99 '56.13 1975 40.42 11 1976 38.77 .55. 1977 36.92; 54.07 1978 35.62 53.32 1979 51.98 35.98 36.91 35.22 1980 34.04 50.98 35.66 36.30 44.04 51.19 1981 49.61 35,54 35.67 3.85 50.33 1982 35.35 50.03 .33.16 50.70 31.6°7 49.47 1983.. 35.33 49.79 34.18 50.26 30.48 48.61 1984 34.89 49.71 33.32 49.80 29.30 47.75 1985 34.46 49.57 3/.47 49.43 28.12 46.89 1986 34.18 49.56 31.70 '48.72 26.93 46.03 1987 33:69 49,23 30.97 48.09 25.75 c 45.17 1988 33.36 48.92 30.55 47F48 24.56 44.31 1989 32.94 48.72 30.43 46.92 4335 43.45 1990 48.36 46.39 42.59 NOTES: Projection I is computed based on NCES projections of degree recipients by major field of study. Projection II is computed based on projecting the previous linear trend (1969 to 1978) in the female proportion of degree recipients in each major field. . Projection III is computed basedon projecting the ,previous linear trend (1969 to .1978) in the segregation index for college majors. SOURCES: Cols. (1) (except 1980), (3), (5), add (7) are computed from NCES (1980, pp. 66-71). Col. (1), 980; NCES (1981, pp. 19-24). Col. (2), 1974 to 1981 is computed from Department of Labor, BLS, Employment and Earnings, 1974 (March), 1975 (June), and 1976 through 1981 (January). Col: (2), 1972 is computed from BLS unpublished annual averages,

(1) a decline in the percentage of females in cent per year). In eaci. of these fields, be- nearly all traditionally female fields of study tween 1969 and 1978, the number of women (public affairs and services, library sciences, !grew to around 25 percent of all majors; letters, and education); and (2) an increase according to our 'projections, they will be- in the representation of women in all other come 40 to 50 percent of these majors by fields. The largest upward trends are in the 1989. The female proportion of psychology traditionally male disciplines of architecture majors also grew by 1.9 percent per year (2.1 percent per year); agriculture and nat- during the 1970s. For such trends to con- ural resources (2.5 percent per year); ac- tinue the same conditions must prevail in counting (2.4 percent per year); business and the future as in thesecent past, even at the management (2.0 percent per year); and higher number of female majors. That is,

computer and information sciences (1.5 per- wotrien must not,. encounter substantial re-

109 102 ANDREA H.BELLER AND KEE-OK KIM HAN sistaoce or increased discrimination as their future. Under the assumptiiin of a linear rate numbers increase." of decline in the index of segregation among Using these projections of sex composi- college majors, estimated in an equation like tion by major, we recomputed the projected (12), we obtain. the projections in Table 6- segregation indexes for college majors and, 2, column (7). From them we project the with Eq. (11), for professionals, and present segregation indexes for professional occu- them in columns (5) and (6) of Table 6-2, pations in column (8). This most liberal pro- respectively. These projections show a rate jection predicts an index of segregation for of decline in the segregation index for col- College majors of 23.38 in 1989 and for lege majors that seems more reasonable, al- professional occupations of 42.59 in 1990.'2 though the level is even higher in the initial This projection lies just above one com- years than based upon the NCES projec- puted from the logistic- individual projection tions. The index 'of segregation among col- (P3), 40.51. Thus, if we were to feed it back lege majors is projected to decline to 30.43 into the overall index, this most optimistic in 1989 and among professional occupations projection for professional occupations would to 46.39 in 1990. Relating this value back to be consistent with that of logistic trends, the projected indexes of segregation for which project an overall decline in segre- professional occupations computed from gation during the 1980s slightly less thin Previous methods, it lies between that of during the 1970s. Greater declines in seg- the combined-individual projection (P4), regation in professional occupations are pre- 47.05, and the projections of the linear-in- dicted by the moderately optimistic projec- dividual (P2), 45.26, and of the 'moderate- tion (P7), 38.'07, and by the optimistic age cohort (P6), 45.43. Since there is strength projection (P8), 32.28. in numbers, we feel confident that this is a It is hardly surprising that the optimistic reasonable moderate estimate: it projects a projections predict greater declines than these decline of around 3.2 percentage points in projections because the optimistic projections the segregation index for professional oc- assume changes in all cohorts, while these cupations between 1981 and 1990, or around take account of changes just among new en- one-third of the .magnitude of decline be- trants. Thus, even with continued declines in tween 1972 and 1981. segregation in the colleges (on the supply side), We can force the segregation index of col- segregation among professional occupations lege majors to decline after 1978, as we be- would still decline less during the 1980s than lieve it did (see note 10), by estimating its during the 1970s and would still be substantial historical trend and projecting it into the in 1990. At present rates of change, complete integration by sex among college fields of study (an index value of 0) would be attained in the 'I According to Becker's (1971) theory of discrimi- year 2009, at which point (in the year 2012) nation, they could encounter more discrimination as the segregation index for professional occu- their numbers increased ifthey had to entermore dis- pations would equal 23.88; then, one-fourt.i criminatoryinstitutions, nolding constant thelevel of of all women (men) would still have to change discrimination. Increases in.the numbers of women in traditionallymale fieldsof study also probably results jobs to eliniinate segregation. This result points from their perception of improved job opportunities in to the fact that it is easier to change the flow related occupations. If such opportunities were to decline, say as a result of lesser efforts in affirmative action, a chain of responses could be set into motion whereby women would not increase their entry into 12 The index this method projects for 1980, 34.04, is traditionally male fields of study and would not in- quite close to the one computed from actual NCES creasingly prepare for traditionally male occupations. data for 1980, 34.53 (see note 10 above).

I 0 I

OCCUPATIONAL SEX SEGREGATION: PROSPECTS FOR THE 1980s 103 of new workers than to change the stock of CONCLUSION incumbent workers. Policies that promote af- r firmative action on the demand side could aid Although significant declines in the oc- in bringing about such changes in occupa- cupational segregation of the sexes occurred tional segregation, even among professionals. during the 1970s, as measured by the index That occupational segregation among pro- of segregation, projections from several fessionals declined more than projected based,. methods predict only a modest further de- on projected declines in segregation among cline in the index during the 1980s. Al- college majors (compare 1980 and 1981 data though women are projected to continue enin Table 6-2, columns [2] and [8]) is con- tering the traci.itionally male professional and sistent with the explanation that, relative to managerial oesupations, they will also con, young men, young women are quitting less' tinue to enter the highly female clerical oc- than previously, due to enhanced oppor- cupations, which are projected to grow by tunities for advancement.° In order for the More than 2 million jobs. Also, women are segregation indexes to decline by more than not projected o enter several of the pre- is due to the influx of less segregated college dominantly male crafts and operative and graduates alone, less quitting must occur laborer occupations that are projected to grow among women already in nontraditional considerably during the 1980s. Projected professional occupations. This interpreta- changes in the fields of study of female col- tion of the data is consistent with findings lege graduates predicNeelines in segrega- from our earlier study (Beller, in this vol- tion among professional occupations con- ume) that show that declines in segregation sistent withmo more than moderate declines were large not only among new entrants but in the overall index of segregation. Only those also among those with a few years of pre- projections that. assume that all cohorts ex- vious work experience. It is quite logical perience declines in segregation during the that while entry positions are important, 1980s at the same rate as during the 1970s much advancement occurs in the subse- predict a large decline in the overall index quent early, years of a career. Thus, these of segregation. If these projections are cor- findings support our earlier findings, which rect, any significant decline in occupational -suggest that the benefits from affirmative segregation during the 1980s cannot be ex- action have not been confined to new en- pected without major policy initiatives. trants but. have been felt among recent entrants as well. 14 APPENDIX A PROJECTIONS DATA AND METHODOLOGY 13 A recent study (Osterman, 1982) shows that affirmative action reduces female quitting. Further, Lloyd BLS Occupational Employment and Niemi (1979, p. 72) show that more and more Projections Women are remaining in the labor force, and thus experienced workers make up a larger proportion of the female labor force. The BLS occupational employment pro- 1 An alternative explanation, suggested by a referee, jections that we use, which are based upon is sex differences in the probability of college graduates the classification system of the 1970 Census moving into professional jobs. Since relative wages of of Population, are not the most recent ones; many professional jobs have fallen, their increased in- however, the newer projections are based tegration could be the result of males moving into other occupations, such as business. While this explanation upon a different data matrix, which does not seems plausible, segregation declined among the man- contain the same occupational categories agerial occupations as well. . (Departrnent of Labor, BLS, 1981, p.1).

111 104 ANDREA H. BELLER AND KEE-OK KIM HAN

Unlike the Census, the new Occupational cohort. Since the occupation projections are Employment Statistics (OES) survey con- based upon these labor force projections, ducted by the BLS is a survey of jobs, not they should be consistent with them. of people. Projections based,on such a matrix avoid the problem of having to specify an unemployment rate,but the primary dis- Methodology of Projections Based on Age advantage is the Jack of consistency with CPS Cohorts data, which are based on a count of persons rather than of jobs. Tht.! first set of these In order to make projections by age co- new projections appears in Carey (1981). hort, we need to know the sizes of age-sex As stated in the text of this paper, the cohorts for the projected year to supplement BLS occupational employment projections information on the projected occupational used assume that a stable long-run unem- distribution' and the projected sex compo- ployment rate close to 4.5 percent will be sition G2 each occupation. Age-sex-specific achieved by the mid-1980s. This seems ex- projections on the size of the civilian labor ceptionally optimistic from the vantage point fprce in 1990, created by the BLS, are pre- of the end of 1982 when, in September, the 'tented in Employment Projections for the economy hit a high unemployment rate of 1980's. They project the size of age cohorts. 10.1 percent. How would an unemployment which we grouped into four categories: 16 rate higher than 4.5 percent affect occupa- to 24, 25 to 34, 35 to 44, and 45 + . Unfor- tional employment projections foi 1990? Ac- tunately these categories cannot be trans- cording to Carey (1981, p. 42), job growth formed into cohorts based on potential labor in blue-collar occupations is more sensitive market experience, which creates more to the underlying assumptions than is job meaningful and presumably more homo- growth in other major occupational cate- geneous groupings for labor force analyses. gories. ,The need for additional blue-collar For example, the age group 16 to 24 contains workers is very much affected by the de- a mixture of high school graduates already mand for manufactured goods as well as by in their chosen field for several years and changes in productivity. Job growth in the individuals who are still students and who white-collar and service categories generally work part tune at jobs unrelated to their is less sensitive to the underlying assump- future occupations. In the analyses by ex- tions than is blue-collar job growth. Most perience cohorts presented in another pa- likely the employment estimates would ov- per, v.z., excluded individuals with 0 years erestimate the size of blue-collar occupa- of potential work experience (defined as tional employment. This consideration should AgeEducation6), hoping to eliminate be kept in mind in examining the projec- students. Since we do not know the sizes of tions of occupational segregation, because age-sex experience cohorts for the projected many blue-collar occupations arehighly male, year, we must base our cohort projections and the larger they are the more they add on age rather than on work experience co- to the index of segregation. horts. Thus, we grouped our Annual De- The BLS 1990 occupational employment mographic File data for 1971 and 1977 into projections -assume that the size, sex and age cohorts. Segregation indexes by age co- age composition of the labor force will change hort for 1971 and 1977 shov.., in Table B-2 as indicated by the intermediatelabor force differ from those by experience cohort (Beller, projections published by the BLS in Em- in this volume, Table 2-2). The youngest age ployment Projections of the 1980's." We use cohort is more segregated than is the young- those projections for our projections by age est experience cohort after exclusion of in-

12 OCCUPATIONAL SEX SEGREGATION: PROSPECTS FOR THE 1980s 105 . dividuals with 0 years of poteintial experi- than do the projections based upon the whole ence. labor force, which direct our attention more To project what happened to a specific to the characteristics of employers, or the cohort as it aged over time, we aged each demand side. Of course, supply changes in group backward to 1977 by subtracting 13 response to perceived changes iv, demand. years. This yielded age groups in 1977 of 16 Women's aspirations change as they per- to 21, 22 to 31, and 32 to 41. We addedthe ceive reductions in the barriers to their en- age group 42 + to complete the labor force. try into traditionally male occupations (Res- We also added the youngest complete co- kin and Hartmann, 1984). hort, 16 to 24 years of age, so that we could Once we create projections of the per- use it for the projections as well. Aging the centage of males in each occupation by age groups backward by 6 years to 1971 yielded group, we must combine them to create pro- age groups 16 to 25, 26 to 35, and 36 to 45. jections of the percentage of males in each We added the group 46+ and the youngest occupation for the labor force as a whole. age cohort, 16 to 24. This construction can To obtain a labor force estimate from the be diagrammed as follows: cohort estimates, we use the BLS projections on the age-sex distribution of the labor force in 1990. We know the projected num- Group 1971 1977 1990 ber of males, females, and total labor force +6 +13 for each age group. We also know from BLS 1 16-2i--016-24-016-24 projections the projected distribution of em- 2 16-25 .. ployment across occupations for the civilian 3 26-35 '16.22-31 25-34 labor force as a whole. We need to combine 4 36-45 32-4135-44 5 46 + + 45+ these two pieces of information to determine the occupational distribution of employment for each age group, and we must First, we must project the percentage of make assumptions to do this. The simplest males in each occupation for each age group assumption is that the occupational distri- in 1990. To do this we choose three alter; bution is the same for each age group as it native assumptions, which allow for (1) no is for the labor force as a whole; this is equiv- change between 1977 and 1990; (2) changes alent to standardizing the employment dis- in the youngest cohort, group 1, only; and tribution of each subgroup to the employ- (3) changes in all cohorts. Changes are pro- ment distribution of the whole. (Thus, looking jected to occur at the same rate between ahead, when we use the projected sex com- 1977 and 1990 as a given group ages as the position within each occupation of each age rate for the similar age group as it aged be- group to compute segregation indexes, we tween 1971 and 1977. The methodology of will see differences in the indexes that are these three projections (labeled conserva- based upon differences in their occupational tive, moderate, and optimistic) is described sex composition alone.) Anotherpossible as- in the text of this paper. sumption is that each age group has the same This method of projecting assumes that occupational distribution of employment in the sex composition of occupations is char- 1990 as it .did in 1977. This, too, is imper- acteristic of the particular age cohort in fect, because the occupational distribution question rather than of the labor force or is projected to change. Consequently we the occupation. That is, it directs our atten- opted for the first approach. tion to the characteristics of individuals, or We computed total employment,_ number the supply side of the labor market, more of males, and number of females in occu-

113 106 ANDREA H. BELLER AND KEE-OK KIM HAN

pation i in age group j in 1990 according to n; and F is the number of females in oc- the following: cupation i for age group.' according to cohort projection n. Tu = x Ti , (At) The- numbers are then aggregated over. age groups to obtain the values for the civilian labor force in 1990: Mi) = Tu x pg, (A2) M7 = E Mu (A4) FU = T - Mg, (A3) where Tu is total employment in occupation FT = Fg, (A5) i for age group j, where j = 1,...,5 as in the diagram above; E, Ise total employ, ment in (A6) occupation i, given by BLS projections; Ti Mn + F." is civilian labor force in age grotip j, given I by BLS projections; pit) is proportion male Using these projections of proportion of males in occupation i for age group j according to by occupation and the BLS occupational cohort projection n, where n = 1,2,3; Ali; employment projections for 1990, Ei, we is the number of males in occupation I for compute four projected segregation indexes age group J according to cohort projection for 1990, P5 to P8.

TABLE, B-1Actual and Projected Proportion Female for Detailed Occupations

Actual Annual .Projected 1972 1981 Raf te 1990 Occ. o Code Occupation Name Change*(P1) (P2) (P3) I. Proportion Female Increasing 1 Accountants .2171 .3850 .0152 .4291 .5239 .5476 3 Programmers, computer .1989 .2941 .0093 .3247 .3822 .4113 10 Engineers, chemical .0000 .0615 .0056 .1043 .0964 .4129 12 Engineers, electrical and electronic .0105 .0378 .0028 .0817 .0594 .2666 13 Engineers, industrial .0235 .1126 .0124 .1560 .2442 .6959 25 Foresters and conservationists .0417 .1053 .0090 .1650 .1879 .7742 31 Lawyers .0396 .1423 .0100 .1910 .2284 .3644 55 Researchers and analysts, operat. and sys. .0991 .2551 .0188 .2741 .4188 .5883 56 Personnel and labor relations .3097 .4977 .0133 .5342 .6028 .6063 64 Pharmacists .1349 .2517 .0088 .3076 .3240 .3563 65 Physicians, medics, and osteopaths .1006 .1376 .0030 .1909 .1620 .1681 85 NEC, health technicians .5821 .5787 .0100 .6052 .6721 .6666 151 Technicians, chemical .1169 .2574 .0155 .3022 .3965. .4874 '152 Draftsmen .0629 .1929 .0120 .2459 .2884 .3763 153 Technicians, elect. engineers .0549 .1124 .0079 .1515 .1882 .3061 171 Operators, rap) .3514 .5738 .0380 .5988 .9790 .8860 192 Public relatiolis .2989 .4545 .0173 .5073 .5721 .5847 202 Officers and manag., bank .1874 .3735 .0165 .4222 .5204 .5530 220 NEC, managers, office .4190 .7024 .0233 .7300 .9356 .8628 222 NEC, pub. admin., office and admin. .2006 .2900 .0057 .3100 .3414 .3496

NOTES: Column (P1) contains the linear-group projection, column (P2) contains the linear-individual projection, and column (P3) contains the logistic-individual projection. NEC = not elsewhere classified. *Coefficient estimate on linear time trend in individual occupation projection, where estimate is significant.

114 OCCUPATIONAL SEX SEGREGATION: PROSPECTS FOR THE 1980s 107

TABLE B-1(Continued)

Actual Annual Projected Rate 1972 1981 1990 '.Occ of Code Occupation Name Change(P1) (P2) (P3) 225 NEC, purchasing agents and buyers .1326 .3038 .0135 :3426 .4148 .4648 231 Retail trade sales manag. and dept. heads .2736 .4036 .0119 .4732 .5299 .5422 233 Sales manag., exc. ret. trade .0292 .1362 .0105 .1944 .2242 .4391 245 NEC, manag. and admin. .1214 .1966 .0080 .2424 .2724 .3049 260 Adv. 'agents and sales work .2273 .4683 .0256 .5063 .7100 .7241 265 Ins. agents, brokers, and underwriters ..1179 .2363 .0128 .2867 .3508 .4173 286 Vendors and carriers, news .2444 .2973 .0211 .3193 .4734 .6537 270 Agents and brokers, real est. .3668 .5000 .0114 .5490 .5916 .5934 281 Sales rep., manuf. Indust. .0700 .2000 .0114 r .2596 .2870 .3553 282 Sales rep., wholesale trade .0460 .1199 .0071 .1681 .1837 .2506 301 Bank tellers .37%5 .9355 .0043 .9501 .9749 .9586 305 Bookkeepers .8794 .9116 .0039 .9327 .9517 .9399 315 Dispatchers and starters, vehicle .1628 .3805 .0249 .4065 .6186 .6883 323 Expediters and production controllers .2308 .4096 .0113 .4561 .5056 .5159 331 Mail carriers, post office .0667 .1548 .0090 .2026 .2444 .3091 333 Messengers and office helpers .1410 .2766 ".0202 .2934 .5129 .7250 343 Computer and peripheral equip. operators .3776 .6407 AMA .6596 .9335 .8555 360 Clerks, payroll, and timekeep. .7174 .8097 .0135 .8367 .9446 .8965 361 Postal clerks .2669 .3802 .0101 .4025 .4824 .4919 375 Clerks, statistical .7090 .8006 .0090 .8364 .8913 .8667 381 Storekeepers and stock clert:s .2290 .3501 .0099 .3958 .4550 .4689 390 Agents, ticket station and express .3178 .4722 .0188 .4941 .6625 .6732 412 Bulldozer operators .0000 .0099 .0013 .1239 .0232 .1623 415 Carpenters .0049 :0180 .0013 .0831 .0303 .0610 422 Compositors and iypesetters .1647 .3506 .0210 .3803 .5491 .6181 430 Electriciai,s .004:1 .0168 .0012. .0720 .0255 .0496 441 NEC, blue-collar supervisors .070/ .1134 .0033 .1606 .1434 .1562 455 Locomotive engineers .00(4) .0213 .0019 :0784 .0334 .1108 470 Air cond., heating, refrigeration .13J00 .0047 .0007 .0628 .0125 .0434 481 Mechanics, heavy equipment .0070 .0179 .0012 .0936 .0285 .0472 482 Mechanics and install. appliance home .0152 .0455 .0033 .1178 .0673 .2108 550 Structural metal workers .0000 .0122 .0011 .0787 .0191 -.0488 552 Teleph. install and repairers .0194 .0982 .0076 .1378 .1684 .3495 554 Teleph. linemen and splicers .0000 .0506 .0060 .0771 .1072 .8681 601 Asbestos and insulat. workers .0000 .0377 .0056 .1000 .1037 .8507 631 "Meat cutters, butchers. exc. manuf. .0348 .0809 .0038 .1307 .1135 .1402 664 Shoemaking, machine operatives .6184 .7260 .0122 .7223 .8539 .8248 666 Furnace tenders and stokers .0123 .0128 .0012 .0507 .0201 .0527 703 Bus drivers .3414 .4732 .0122 .5038 .6043 .6071 705 Delivery and route workers .0247 .0856 .0057 .1420 .1393 .2172 715 Truck drivers .0042 .0286 .0018 .0923 .0401 .0686 740 Animal caretakers, exc. farm .4125 .5699 .0251 :6387 .8100 .7822 750 Carpenters' helpers .0160 .0167 .0020 .1150 .0351 .3307 754 Garbage collectors .0118 .0278 .0011 .1731 .0347 .0385 762 Stock handlers .1687 .2479 .0093 .2911 .3267 .3530 764 Vehicle washers and equip. cleaners .0909 .1576 .0042 .2149 .1920 .2017 785 Not specified laborers .0396 .1122 .0073 .2181 .1778 .2881 903 Janitors and sextons .1051 .1899 .0066 .2407 .2503 .2759 910 Bartenders .2786 .4725 .0192 .5156 .6604 .6688 (Continued)

115 108 ANDREA H. BELLER AND KEE-OK KIM HAN

TABLE B-1(Continued)

Actual Annual Projected Rate 1972 1981 1990 Occ. of Code Occupation lame Change* (PI) (P2) (P3) .4814 .6146 .6230 932 Attendants, recreation and amuse. .3120 .4477 .0155 .2062 .2017 .2406 962 Guards .0461 .1366 .0086

II. Proportion Female Decreasing .4759 .2106 .2364 1,!1 Teach., adult ed. .3768 .4267 - .0219 .8627 .8003 .7922 142 Teach., elem. ed. .8505 .8359 - .0038 .8984 .8480 385 Telephone opvrators .9668 .9302- .0040 .9469 .0000 .0000 502 Millwrights .0000 .0000- .0029 .0594 .5119 .4772 624 Grader: and sorters, manuf. .7727 .6389- .0162 .6869 .0955 .1142 822 Farm laborers, wage workers .1535 .1594- .0095 .1898 .4344 .4318 912 Cooks, exc. private household .6236 .5228 - .0109 .5616 .8626 .8474 915 Waiters .9173 .8967 - .0038 .9211 .5135 .5096 933 NEC, attendants, personal serv. .846 .5773 -.0117 .6077.

III..Proportion Female-No Significant Trend .1080 .0418 .0768 2 Architects .0303 .0440 4 Syst. analysts, computer .1081 .2584 .2993 .3121 .3266 .0124 .0245 6 Engineers, aero- and astrq-. .0000 .0122 .0467 .0652 .0251 .0963 11 Engineers, civil .0065 .0164 .0396 .1713 14 Engineers, mechanical .0000 .0243 .0652 .0544 .0335 .0524 23 NEC engineers .0000 .0290 .8914 32 Librarians .8278 .8580 .8745 .9020 .4525 .4563 44 Scientists, biological .2778 .4035 .4609 .2866 .3452 45 Chemists .1008...2090 .2577 .9678 .9662 75 Nurses, registered .9763 .9680 .9890 .7781 .7675 76 Therapists .5826 .7049 .7271 .7426 .7451 80 Clinical lab. techs. .7063 .7724 .7948 .5765 .5681 83 Radiologic techs. .6618 .6863 .7299 .0839 86 Clergy .0163 .050 .0949 .0766 .4380 90 NEC, religious workers .5745 .4717 .5278 .4281 .2927 .3820 .4014 91 Economists .1176 .2484 .5349 .4369 .4373 93 Psychologists .3800 .4870 .6787 .5984 .5980 100 Social workers .5856 .6397 .6288 .6511 .6487 101 Recreation workers .4457 .5798 .8194 .4974 .4971 140 Col. teach., not specified .2703 .3788 1.0000 .9744 .9693 143 Teach., pre- and kindergart. .9681 .9833 .5289 .4933 .4932 144 Teach., secondary sch. .4964 .5138 .7632 .7232 .7205 145 NEC, teach., exc. col. .7409 ."380 .4794 .5386 .5551 150 Technicians, ag. and biol., exc. health .2927 .4400 .0563 .0188 .0743 161 Surveyors .0141 .0114 .2970 .3083 .3218 162 NEC, techs., eng. and science .1437 .2469 .0769 .0152 .0278 163 Pilots, airplane .0000 .0125 .5648 .5559 .5558 174 Counselors, voca. and education .5000 .5323 .4610 -5434 .5495 180 Athletes and kindred .3077 .4351 .3242 .4180 .4384 183 Designers .1909 .2953 .4908 .4901 184 Editors and reporters .4110 .5050 .5465 .1960 .2062 185 Musicians and composers .3058 .2535 .3232 NOTES: Column (P1) contains the linear-group projection,column (P2) contains the linear-individual projection, and column (P3) contains the logistic-individual projection.NEC = not elsewhere classified. *Coefficient estimate on linear time trend in individual occupationprojection, where estimate is significant.

11 6 OCCUPATIONAL SEX SEGREGATION: PROSPECTS FOR THE 1980s 109

TABLE B-1(Continued)

Actual Annual Projected Rate ' 1972 1981 1990 Occ. of Code Occupation Name Change(P1) (P2) (P3) 190 Painters and sculptors .4264 .5121 .0000 .5424 .5559 .5558 191 Photographers .1558 .2424 .0000 .2868 .1829 .1776 194 NEC, writers, artists, entertainers .3333 .3853 .0000 .4272 .3115 .3171 195 Research, not specified .2791 .3936 .0000 .4107 .4774 .4896 201 Assessors, controllers, treasurers .3448 .5000 .0000 .5311 .6337 .6356 205 Buyer, whole. and retail .3292 .4346 .000: .4735 .5175 .5198 210 Manager,,credit and collect .2535 .3788 .0000 .4042 .4803 .4986 212 Administrators, health .4746 .4954 .0000 .5442 .4941 .4937 215 Inspectors, (O. construct., pub. admin. .0619 .1019 :0000 .1545 .2487 .3676 216 Manag. and superintendent, building admin. .4265..5063 .0000 .5572 .5134 .5134 223 Officials, lodges, soc. and unions .1875 .2845 .0000 .3034 .3114 .3140 224 Mail super. and postmasters .3409 .3636 .0000 :4121 .5071. .5347 230 Manager, rest., bar, cafeteria .3239 .4045 .0000 .4356 .3893 .3893 235 Sch. ac'enin., college .2530 .3504 .0000 .3506 .3923 .4001 240 Sch. admin., sec. and elem. .2624 .3668 .0000 .4010 .4255 .4313 262 Demonstrators .9375 .9619 .0000 .9663 .9726 .9720 264 Hucksters and peddlers .7304 .7964 .0000 .8295 .9441 .9106 271 Sales agent, stock and bond .0990 .1667 .0000 .1915 .2186 .2461 283 Sales clerks, retail trade . .6887 .7130 .0000 .7504 .7298 .7302 284 Sales work, exc. clerks, retail .1302 .1965 .0000 .2502 .2032 7v.2040 285 Sales work, service and construe. .2941 .4304 .0000 .5146 .4648 .4854 303 Billing clerks .8456 .8808 .0000 .9286 .9654 .9362 310 Cashiers .8667 .8637 .0000 .8960 .8715 .8715 312 NEC, supervisors, clerical .5779 .7073 .0000 .7469 .7731 .7631 313 Collectors, bill and account .4833 .6444 .0000 .6740 .7942 .7702 314 Clerks, counter, exc. food .7386 .7642 .0000 .8051 .7804 .7831 320 Enumerators and interviewers .8205 .7544 .0000 .7842 .5988 .4914 321 NEC investigator and estimator .4339 .5444 .0000 .5754 .5879 .5875 325 File clerks .8493 .8371 .0000 .8619 .8068 .8027 326 Insur. adjusters, examiners, and investigators .3519 .5753 .0000 .6189 .7291 .7181 330 Library attend. and assist. .7518 .8255 .0000 .8600 .9261 .9012 332 Mail handlers, exc. post office .4375 .4767 .0000 .5136 .4977 .4978 341 Bookkeep. and billing operators .9130 .8936 .0000 .9329 .8085 .7311 345 Key punch operators .8975 .9383 .0000 .9494 .9866 .9701 355 NEC, office machine operators .6949 .6833 .0000 .7156 .5438 .4820 364 Receptionists .9702 .9742 .0000 .9961 .9792 .9770 370 Secretaries, legal .9908 .9888 .0000 1.0000 .9851 .9733 372 NEC, secretaries .9909 .9914 .0000 1.0000 .9921 .9922 374 Clerks, shipping and receiving .1486 .2272 .0000 .2734 .2667 .2769 376 Stenographers .9040 .8611 .0000 .8790 .8164 .7413 382 Teacher aides .8932 .9303 .0000 .9634 .9680 .9536 391 Typists .9608 .9634 .0000 .9891 .9745 .9728 392 Weighers .2791 .3421 .0000 .3996 .2969 .3035 394 Misc. clerical workers .7513 .8200 .0000 .8716 .8911 .8778 402 Bakers .2895 .4148 .0000 .44i7 .4737 .4762 404 Xolermakers .0000 .0000 .0000 .0840 .0000 .0010 405 kbinders .6250 .5600 .0000 .6064 .5671 .5783 410 and stonemasons .0059 .0000 .0000 .0842 .0000 .0002 413 '''Cabinetmakers .0500 .0267 .0000 .0904 .0279 .0258 (Continued) 117 D

110 ANDREA H. BELLER AND KEE-OK KIM HAN

TABLE B-1(Continued)

Actual Annual Projected Rate 1972 i981 1990 Occ. of Code Occupation Name Change(P1) (P2) (P3) 652 Lathe and mill machine operatives .0488 .0594 .0000 .1331 .0568 .1,583 653 NEC, precision machine operatives .1475 .1754 .0000 .2521 .0835 .0994 656 Punch and stamp press operatives .2739 .3208 .0000 .4124 .3544 .3556 662 Sawyers .0496 .0909 .0000 .1591 .1553 .6861 421 Cement and concrete finishers .'.127 .0000 .0000 .1030 .0000 .0010 424 Crane, derrick, hoist oierators .0467 .0070 .0000 .0941 .0124 .0068 425 Decorators and window dressers .5977 .7222 .:0000 .7627 .8934 .8542 433 Install. and repair elect. power lines .0098 .0086 .0000 .0618 .0044 .0053 436 Excavat., grading, exc. bulldozer .0000 .0062 .0000 .0740 .0113 .0315 452 NEC, inspectors .0305 .0839 .0000 .1364 .1619 .2269 454 Job and die setters, metal .0106 .0532 .0000 .1211 .0661 .0283 461 Machinists .0054 .0388 .0000 .0861 .0769 .2243 471 Aircraft .0000 .0325 .0000 .0916 .0414 .0486 472 Auto body repairers .0000 .0098 .0000 .0675 .0217 .1132 .0066 473 Auto mechanics .0058 .0058 .0000 .0685 .0068 475 Repairers, data process. machine .0222 .0700 .0000 .1298 .0930 .2669 480 Farm implement .0000 .0000 .0000 .0995 .0004 .0011 484 Office machine .0000 0400 .0000 .1086 .0481 .0456 485 Radio and television .0081 .0367 .0000 .1216 .0406 .a501 503 Molders, metal .0962 .1731 .0000 .2276 .1990 .2099 510 Painters, construct. and maintenance .0188 .0575 .0000 .1149 .1114 .2060 522 Plumbers and pipe fitters .0027 .0044 .0000 .0595 .0092 .0124 530 Printing press operator .0563 .1091 .0000 .1595 .1350 .1699 534 Roofers and slaters .0000 .0000 .0000 .0477 0000 .0006 535 Sheetmetal workers and tinsmiths .0139 .0397 .0000 .1069 .0577 .0911 545 Stationary engineers .0105 .0165 .0000 .0639 .0316 .1634 551 Tailors .3226 .4091 .0000 .5666 .5151 .5344 561 Tool and die makers .0058 .0237 .0000 .0878 .0276 .0337 563 Upholsterers .1364 .2222 .0000 .2612 .1425 .1499 .6389 575 NEC, craft and kindred workers .0690 .1746 .0000 .2648 .4046 602 Assemblers .4671 .5231 .0000 .5787 .5462 .5461 604 Bottling and can operatives .3455 .4231 .0000 .4710 .4176 .4183 610 Checkers, examiners, inspec. manuf. .4847 .5356 .0000 .5743 .5239 .5240 611 Clothing-ironers and pressers .7683 .8067 .0000 .8334 .7846 .7831 612 NEC, cutting operatives .2773 .3137 .0000 .3606 .3047 .3049 .0790 615 Drywall install. and lathers .0120 .0,127 .0000 .1034 .0272 621 Filers, sanders, polishers, buffers .2213 .3186 .0000 .3614 .2806 .2813 .7333 622 Furnace tenders, smelters, pourers .0429 .0323 .0000 .1058 .0965 623 Garage work and gas sia. attend. .0458 .0559 .0000 .1048 .0797 .1100 .8241 625 Produce graders and packers, exc. fact. and farm.7143 .7333 .0000 .7699 .8337 .6594 630 NEC, laundry and dry clean. opera. .6970 .6614 .0000 .7044 .6600 633 Meat cutters, butchers-manuf. .3258 .2917 .0000 .3297 .2832 .2832 634 Meat wrappers, retail trade .9000 .8980 .0000 .9357 .8358 .7171 640 NEC, mine operatives .0070 .0192 .0000 .0610 .0336 .1333 .0330 641 Mixing operatives .0202 .0390 .0000 .0998 .0287 642 Oilers and greasers, exc. auto .0435 .0526 .0000 .1189 .0719 .2183 .6352 643 Packers and wrappers, exc. meat and produce.6090 .6339 .0000 .6827 .6354 NOTES: Column (P1) contains the linear-group projection, column (P2) containsthe linear-individual projections and column (P3) contains the logistic-individual projection. NEC = notelsewhere classified. *Coefficient estimate on linear time trend in individual occupation projection,where estimate is significant.

118 'OCCUPATIONAL SEX SEGREGATION: PROSPECTS FOR THE 1980s 111

TABLE B-1(Continued)

Actual Annual Projected Rate 1972 1981 1990 Ch.sc. o Code OccupationNrni. Change*(P1) (P2) (P3) 644 Painters, manufactured articles .1461 .1667 .0000 .2314 .1469 .1458 645 Photogriphic process workers .4691 .5177 .0000 .5679 .5503 .5501 650 Drill press operatives .2267 .2500 .0000 .2936 .2607 .2618 651 Grinding machine operatives .0538 .1119 .0000 .1499 .2038 .269ft 663 Sewers and sdtchers .9583 .9603 .0000 .9908 .9558 .9565 665 Solderers .7674 .7292 .0000 .7299 .6354 .6065 672 Spinners, twisters, and winders .6071 .6735 ,.0000 .7174 .6720 .6727 674 NEC, textile operatives .5282 .5520 .0000 .5959 .5060 .5056 680 Welders and flame-cutters .0361 .0465 .0000 .1125 .0526 .0530 681 NEC, winding operatives .4658 .4546 .0000 .5402 .5856 .5873 690 Machine operatives, misc. specified .2814 .2913 .0000 .3408 .2769 .2773 692 Machine operatives, not specified .2148 .2781 .0000 .3387 .3625 .3640 694 Misc. operatives .3166 .3526 .0000 .3643 .3553 .3553 695 Not specified operatives .3036 .3452 .0000 .4101 .4835 .4974 706 Fork lift and tow motor operatives .0099 .0579 .0000 .1220 .0909 .2624 711 Parking attendants .0303 .1000 .0000 .1922 .1283 .5642 712 RR brake operators and couplers .0000 .0000 .0000 .0525 .0163 .0088 713 RR switch operators .0000 .0000 .0000 .0553 .0175 .0092 714 Taxicab drivers and chauffeurs .0904 .0932 .0000 .1361 .0902 .0936 751 'Construction laborers, exc. carpenters' helpers.0049 .0215 .0000 .0905 .0454 .0875 753 Freight and material handlers .0604 .0966 .0000 .1544 .1106 .1139 755 Gardeners and groundskeepers, exc. farm .0221 .0463 .0000 .1002 .0917 .1719 760 Longshore workers and stevedores .0000 .0227 .0000 .0857 .0299 .0654 761 Timber cutting and logging .0123 .0101 .0000 .0302 .0000 .0015 770 NEC, warehouse laborers .0267 .0595 .0000 .0715 .0723 .0783 821 Farm supervisors .0357 .0645 .0000 .1211 .1937 .2039 901 Cleaners, lodging quarteis- exc. .9786 .9657 .0000 1.0000 .9454 .9174 902 NEC, building interior cleaners .5509 .5534 .0000 .5992 .5830 .5826 911 Waiters' assistants .1367 .1982 .0000 .2802 .2882 .3110 913 Dishwashers .3578 .2892 .0000 .3645 .3540 .3.527 914 Food counter and fountain workers .8208 .8370 .0000 .8752 .8788 .8709 916 NEC, food service workers, exc. private house..7377 .7342 .0000 .77:13 .7387 .7391 921 Dental assistants .9787 .9784 .0000 1.0000 .9761 .9840 922 Health aides, exc. nursing .7973 .8454 .0000 .8912 .9321 .9084 925 Nursing aides, orderlies, attend. .8344 .8665 .0000 .9034 .8682 .8698 926 Practical nurses .9650 .9772 .0000 1.0000 .9652 .9131 942 Child care workers, exc. private .9579 .9544 .0000 .9839 .9084 .8760 944 Hairdressers and cosmetologists .9089 .8936 .0000 .9171 .8647 .8490 950 Housekeepers, exc. private house. .7094 .6992 .0000 .7424 .6274 .6095 952 School monitors .8750 .9722 .0000 .9962 1.0000 .9328 954 Welfare service aides .8235 .8837 .0000 .9173 .9086 .9035 960 Crossing guards and bridge attend. .6327 .6222 .0000 .6625 .5411 .5378 961 Firemen, fire protection .0050 .0095 .0000 .0775 .0057 .0065 964 Police and detectives .0264 .0557 .0000 .1125 .0856 .0987 965 Sheriffs and bailiffs, .0508 .0725 .0000 .1230 .1099 .3588 980 Child care workers, priv. house. .9797 .9774 .0000 .9853 .9690 .9721 984 Private house cleaners and servants .9719 .9519 .0000 .9700 .9395 .9320

119 S

112 ANDREA H. BELLER AND KEE-OK KIM HAN

TABLE 11-2Actual and Projected Segregation Indexes by Age Cohort, 1971 to 1990

1971 1977 Projected 1990 Group Age U ES Age U ES Age (P5) (P6) (P7) (P8)

1 16-24 68.64 67.14 16-24 64.68 61.85 16-24 61.96 37.08 46.60 37.08 2 16-25 68.31 67.33 16-21 65.49 66.04 -- 3 26-35 68.94 68.90 22-31 62.67 61.81 25-34 66.18 66.18 47.82 39.61 4 36-45 70.85 70.77 32-41 66.30 67.4.3 35-44 61.91 61.91 55.02 49.63 5 46+ 68.62 69.16 42+ 66.88 67.61 45+ 68.19 68.19 58.17 48 90 Total - 68.14 - - 64.15 - - 62.11 57.29 50.02 42.20 NOTES: The age groups are constructed so that each group can be followed as it ages over time. Thus, the difference between the 1971 and 1977 intervals is 6 years and between the 1977 and 1990 intervals, 13 years. U = unstandardized, and ES = standardized to the employment of the whole labor force in the giveri year. SOURCES: 1971 and 1977: Current Population Survey, Annual Demographic Files, 1972 and 1978, computer tapes; 1990: Department of Labor, BLS (1979, Table 4, p. 5); Department of Labor, BLS (1981, Table 5, pp. 495- 502).

TABLE B-3Actual and Projected Proportion Female for College Majors, 1969 to 1989

Actual Projected Percent Female Percent 1969 1978 Trend' Female, 1989 Field of Study (1) (2) (3) (4) Social sciences 36.3 41.0 0.005 45.5 Psychology 42.9 58.8 0.019 79.4 Public affairs and services 71.6 49.4 -0.026 10.8 Library sciences 93.6 88.5 -0.003 87.4, Architecture and environmental design 4.3 23.7 0.021 47.0 Fine and applied arts 59.1 62.0 0.004 66.4 Foreign language's 73.1 75.9 0.003 79.7 Communications 42.2 46.9 0.007 51.0 Letters 63.4 57.1 -0.008 46.9 Mathematics and statistics 37.4 41.1 0.005 47.5 Computer and information sciences 13.0 25.7 0.015 40.7 Engineering 0.7 7.4 0.006 11.5 Engineering technologies 0.5 2.8 0.003 6.4 Physical sciences 13.6 21.3 0.009 30.4 Biological sciences 28.0 38.4 0.012 50.6 Agriculture and natural resources 3.8 24.6 0.025 50.1 Health care professions 76.9 80.5 0.004 84.0 Accounting 7 8 29.4 0.024 52.7 Business and management , 9.1 26.4 0.020 45.4 Educatior, 75.8 72.5 -0.004 67.2 Other 53.1 58.4 0.007 66.1 All fields 43.7 47.1 - 52.3b a 'The trend value is the estimated coefficient on year in an OLS regression equation In which the percentage of females is regressed on time. All trend values are significant at the .05 level except communications, which is significant at .10 level. b Computed front the projected sex composition for each major field. SOURCE: NC:ES (1980, pp. 70-71).

120 OCCUPATIONAL SEX SEGREGATION: PROSPECTS FOR THE 1980s 113

is 49.61 in 1981, while that based on un- APPENDIX C PROJECTIONS FOR published detailed data is 50.55. As is com- PROFESSIONAL OCCUPATIONS BASED mon, the aggregation tends to mask some UPON COLLEGE MAJORS segregation, but the effect here is small. It The model used to project segregation is even smaller in 1977, when the segre- among professional occupations based on gation index based on 23 aggregated cate- segregation among college majors departs gories is 54.07 and based on 59 detailed oc- from the other models used in this paper. cupations is 54.35. The index based on In this model the segregation index itself, unpublished aggregated datain 1972 is 58.94, rather than the sex composition of each oc- while that based on detailed data is 59,44. cupation, becomes the data point. We chose Thus, the effect of this aggregation appears this model because attempts to relate the to be to somewhat overstate the decline in sex composition of a professional occupation segrega tic n UL professional occupations dur- to that of a specific college major proved ing the 197, unsuccessful. Thus, we hypothesized, that the overall degree of segregation among college graduates would affect the overall de -'ACKNOWLEDGMENTS gree of segregation in professional occupa- Discussions with Francine Blau, John tions. Boyd, and Kenneth Stolarsky stimulated the The data used for these projections for developmenof ideas. Helpful comments on professional occupations are slightly differ- an earlier draft were provided by Marianne ent from the data used earlier in this paper. Ferber, Barbara Reskin, and several re- To obtain a continuous time series, data were viewers. Computational assistance by John taken from the published BLS AA data (Em- Boyd and Alex Kwok is gratefully acknowl- ployment and Earnings, various issues). The edged. female proportion of occupational employment was first published in 1974. Initially each published category represented ah in- REFERENCES dividual occupation or a combination of oc- Becker, Cary. S. cupations having a minimum employment 1971 The Economics of Discrimination. Chicago: University of Chicago Press. estimate of 150,000; the sex distribution was Beller, Andrea H. included only where the basis of the esti- 1982a"The Impact 'f Equal Opportunity Policy on mate was at least 15,000. By this criterion Sex Differentials in Earnings and Occupa- the sex distribution was not 'published for tions." American Ec9notnic Review, Papers and Proceedings (May): 171-75. many detailed categories. Consequently to 1982b"Occupational Segregation by Sex: Determi- obtain comparable data over time, more ag- nants and Changes." journal of Human Re- gregate categories of professional, occupa- sources 17 (Summer):371-92. tions were used in these analyses. Twenty- Blau, Francine D., and Wallace E. Hendricks 1979 "Occupational Segregation by Sex: Trends and four separate categories were used for 1978 Prospects." Journal of Human Resources 14 to 1981, 23 for 1575 to 1977, and 18 for 1974. (Spring):197-210. 'We also used 'unpublished AA data for 1972 Carey, Max L. 1981 "Occupational Employment Growth Through and used the same 24 categories as in 1978 1990' Monthly labor Review 104(8):42-55. to 1981. Thus, the segregation indexes used Department of Labor, Bureau of Labor Statistics here differ slightly from the other ones re- 1979 Employment Projections for the 1980..s. Bul- ported in this paper, which include 59 dis- letin 2030. Washington, D.C.: U.S. Govern- aggregated categories. For example, the ment Printing Office. June. 1981 'The National Industry-Occupation Employ. segregation index for professional occupa- ment Matrix. 1970. 1978 and Projected 1990:" tions based on the published aggregated data Bulletin 2086, Vol. 2. Washington, D.C.: U.S.

121 114 ANDREA H.BELLER AND KEE-OK KIM HAN

Government Pri. '!ng Office. April. Employ- of Female Quit Rates." Review of Economics ment and Earnings. Various issues. and Statistics 64(4):604-12. 0 Lloyd, Cynthia,andBeth Niemi Poirier, Dale 3. The Econometrics of Structural Change, With 1979 The Economics of Sex Differentials. New York: 1978 Columbia UniVersity Press. Special Emphasis on Spline Functions. Am- sterdam: North-Holland. Medoff, James L. Reskin, Barbara F., and Heidi I. Hartmann 1979 "Layoffs and Alternatives Under Trade Unions in U.S. Manufacturing." AmericanEconomic 1984 Women's Work, Men's Work: Sex Segregation on the Job. Washington, D.C.:National Acad- Review 69 (June):380 -95. emy Press. National Center for Education Statistics 1980 Projections of Education Statistics to 1988-89. Smith, Ralph E. By Martin M. Frankel and Debra E. Gerald. 1977 "The Impact of Macroeconomic Conditions on Employment Opportunities for Women." Pa- April. I per No. 6, prepared for U.S.Congress, Joint 1981 Earned Degrees Conferred 1979-80. Septem- Economic Committee, 94th Cong., 2d sess. ber. Osterman, Paul 1982 "Affirmative Action and Opportunity: A Study. al

122 =110

Part II Explaining Segregation: Theoretical Perspectives and Empirical Evidence

N..

123 Occimational 3egregation and Labor Market Discrimination

FRANCINE D. BLAU

The post-World War II period has wit- rates -)f married women (with husband pre- nessed a rapid, growth in female labor force sent) from 21.6 percent in 1950 to 51.0 per- participation and a steadAarrowing of sex cent in March 1981 (U.S. Dept. of Labor, differences in the'extent of participation .in BLS, 1982) was a major factor in the expan- work outside the home. In 1950, 86.8 per- sion of the female labor force, cent of the (adult) male population partici- These trends appear to have been accom- pated in the labor' orce as compared with panied by an increase in the labor force at- 33.9 percent of the (adult) female population tachment of women. This is suggested by (1f.S. Dept. of Labor, ETA, 1981). By March the marked rise in the labor force partici- 1982, 76.6 percent of males and 52.1 per- pation rates of married women (with hus, cult of females were labor force partici- band present) with preschool-age children pants.' The large increase in participation from 11.9 percent in 1950 to 47.8 percent in March 1981 '(U.S. Dept. of Labor, BLS, November 1981). Further evidence of the I Black women's participation rated have historically increasing stability of women's participation been considerably higher than those of white women, is provided by the decrease in labor force although the differential has declined in recent years. turnover among women that has occurred In 1955 the labor force participation rate of black and over the last 20 years, particularly since the other nonwhite women was 46.1 percent in comparison late 1960s (Lloyd and Niemi, 1979).2 At the with 34.5 percent for whites. By 1980 the participation rates were 53.4 and 51.3 percent for nonwhites and same time, the gap between men's and whites, respectively. The participation rates of black women's educational attainment (average males have fallen more sharply than those of white males over this period. In 1955 the participation rate of both white and nonwhite males was about 85 percent; by 1980 the participation rates of whites and non . 2 Labor force turnover is measured by the ratio of whites were 78.3 and 70.8 percent, respectively (U.S. the labor force experieri6e rate (the percentage of women Dept. of Labor, BLS, 1980, 1981). These differing ra- who were in the labor force at some time during the cial trends in participation rates are an important factor year) to the annual average labor force participation in evaluting the recent gains in black earnings noted rate (the percentage of women who where in the labor below. See Brown (1981). force at any particular point in time or survey week).

117 124 118 FRANCINE D. BLAU years of schooling) has beeneliminated (Lloyd fered for sex differentials employment and Niemi, 1979). patterns and earnings. We focus on the var- These dramatic shifts in the quantity and ious neoclassical'and.institulional models of quality of labor supplied to the market by sex (and race)discrimination and on the al- women do not appear tohave been accom- ternative explanation provided by the hu- panied by a noticeable improvement in the man capital model,'which emphasizes the relative economic status of women workers. voluntary choices of women. In the second In fact, the, median earnings of year-round, section we first evaluate the empirical evi- full-time women workers as a percentage of dence on the degree of sex discrimination men's fell from 63.9 to 60.2 between 1955 in the labor market, and then turn to an and.1981 (Lloyd and Niemi, 1979; U.S. Dept. assessment of the role of employment seg- of Commerce, 1982). Most of the decline regation by sex in producing differential out- had occurred by the early /1960s, and the comes for men and womenworkers. earnings ratio has been roughly stable since then. The occupational distributions of men and women continue to differ significantly. ECONOMIC EXPLANATIONS..OF SEX A small movement toward greatersimilarity DIFFERENTIALS' IN OUTCOMES appears to have occurredbetween 1960 and 1970 based on census data (Blau and Hen- Theories of Discrimination dricks, 1979). Some estimates suggest that the rate of decline in segregation may have While most of the discrimination models accelerated between 1972 and 1981, al- discussed here were developed to explain though the magnitude of segregation re- or at least wereillustrated in terms of mained substantial (Beller, this volume).3 racial differences, we here apply them to Throughout this period over 60 percent of sex differences. Thefoundation for the mod- the female (or male) labor force would have ern neoclassical analysisof labor market dis- had to change jobs to eliminate the over- crimination was laid by Becker (1957). For representation of women in certain occu- simplicity, it is assumed that male labor and pations and their correspondingunderre- female tabor are perfect substitutes. That is, presentation in others (Blau andHendricks, men and women areequally productive and 1979; Beller, this volume). thus deserving of equal wages in the absence The precise role of labor market discrim- of discrimination. Discriminatory tastes may ination in producing these sex differentials exist in employers, coworkers, and(or cus- in occupational distributions and earnings tomers. has been the subject of considerable debate Employers with "tastes for discrimina- among economists. Inthe first section we tion" against women will hire women work- consider the explanations that have been of- ers only at a wage discountthat is sufficiently large to compensate them for the disutility of employing women. Becker also showed that even if employers themselves have no tastes for discrimination against women, 3 Lloyd and Niemi (1979) find no change in the de- profit-maximizing behavior by employers may gree of segregation over this period using censusdata for 1970 and Current Population Survey (CPS) data for result in sex discrimination if employees or 1977. However, using comparable CPS data. Beller (in customers have such discriminatory tastes. this volume) finds that segregation continued to decline 'Male employees with tastes for discrimi- over the 1970s. Full solution ofthis disparity in re- nation against women will work with them sults may await the availability of the 1980 census data. In any case both studies suggest that the magnitude of only at a wage premium that is stifficient to segregation remained quite high. compensate them for the disutility of female

1 2 5 OCCUPATIONAL. SEGREGATION AND LABOR MARKET DISCRIMINATION 119 . coworkers.4 Customers with tastes for dis- them as 'pipefitters. Men may be willing to crimination against women will buy prod- work with women in complementary (co- ucts or services produced or sold by women operative) or subordinate positions but dis- only at a lower price. -Intuitively we would like interacting with women as peers or expect this type of discrimination to be more superiors. Customers may be delighted to important in sales or service occupations where purchase nylons from female clerks but avoid face-to-face contact with the customer/client women car salespersons or attorneys. These occurs. As a consequence of coworker or cus- discriminatory tastes may be held independ- tomer discrimination employers may, under ently of beliefs that women would be less pro- certain circumstances, discount female wages ductive than men in nontraditional pursuits. to compensate for the higher costs (coworker This latter possibility if considered under no- discrimination) or lower revenues (customer tions of statistical discrimination below. discrimination) attendant upon employing While such reasoning 'makes Becker's women. model more compatible with the large mag- A definition of wage discrimination flow- nitude of occupational segregation that we ing from Becker's work has guided much of observe in the labor market, problems re- the empirical analysis of labor market dis- main. First, as Blau and Jusenius (1976) point crimination. Wage discrimination (the mar- out, a high degree of segregation is still un- ket discrimination coefficient) may be de- likely given (1) the wage flexibility generally fined as the difference between the actual assumed in neoclassical models and (2) the ratio of male to female wages and the ratio existence of a large ceteris paribus wage dif- that would exist in the absence of discrim- ferential between men and women. (The ination assuming perfect substitutability, empirical .work considered below provides this would be wage parity (Becker, 1957, p. support for the existence of a substantial pay 126). In empirical work, where the wages gap that is not accounted for by the pro- of hetet ogeneous male and female labor are ductivity-related characteristics of men and compared, this is approximated by the no- women.) For example, let us consider the tion of pay differentials that are not ac- case in which discriminatory tastes reside in counted for by productivity differentials. employers. Employers whose tastes for dis- While the type of discrimination defined crimination are so strong that they exceed by Becker does not necessarily predict that the marketwide discrimination coefficient occupational segregation by sex will occur, will not hire women. Employers who are it may be made compatible with occupa- exactly compensated for the disutility of hir- tional segregation if we postulate that tastes ing women by the market discrimination for discrimination against women vary across coefficient will be indifferent to whether theip occu)ational categories. The issue may be employ men or women and will presumably more one of socially appropriate roles than hire both. Employers with relatively weak of the desire to maintain social distance that tastes for discrimination who are over- Becker emphasized. Employers may have compensated by the marketwide discrimi- no compunctions about hiring women as nation coefficient will hire only women. secretaries but may be reluctant to employ A high degree of occupational segregation by sex due to discrimination will not be observed unless most employers of workers in jobs are in the first category, that is, 4 The notion of employee discrimination is developed unless most employers have such strong tastes by Bergmann and Darity (1981) in terms of productivity reductions due to employee hostility rather than of for discrimination against women in male direct increases in costs due to compensating differ- jobs that they are not sufficiently compen- entials. sated by the large ceteris paribus pay dif-

126 .120 FRANCINE D. BLAU

ferentials that appear to exist between men From this perspective one may question why and women. This seems to be unlikely. a high degree of occupational segregation by Differences in tastes for discrimination sex appears to be associated with large dis- among employers (employees, customers) can criminatory pay differentials. perhaps more plausibly produce sex segre- Even if discrimination is made compatible gation by firm in the Becker model (Arrow, with occupational segregation in the Becker 1973). That is, women would tend to be em- .model, gegregeion does not play a causal ployed by less discriminatory employers Olio rote in generating the., sex pay differential, are overcompensated by the prevailing sex Rather, both occupational and pay differ- pay differential. A problem that arises here is entials are due to tastes for discrimination the stability of this situation in the face of among employers, coworkers, and/or cus- competitive forces (see below). Women may tomers. Eliminating occupational segrega- also, in this model, find better employment tion (were it possible to do so) would not opportunities working with less discrimina- eliminate the pay .differential. Indeed, re- tory employees or selling to (serving) less dis- ducing segregation might require still fur- criminatory customers (clients). - ther discounting of female wages, since it Second, in the Becker model discrimi- would necessitate women's making inroads natory pay differentials are in some sense into areas characterized by stronger discrim- the price paid by the discriminated group inatory tastes on the part of employers, co- for associating with the discriminators. In workers, and/or customers. general, differences between the two groups Bergmann (1974) has developed an anal- in factor endowments make such association ysis in the Becker tradition that gives a more profitable even in the face of discrimination. central role to employment segregation. In However, sufficient opportunities in a seg- Bergman .1's"overcrowding" model,8 dis- regated context can eliminate the need for criminatory employer tastes result in the pay differentials. This may be illustrated by segregation of male and female labor into the case in which 'tastes for discrimination two sets of occupation. While such segre- reside in coworkers. In the case of perfect gation need not result in sex pay differen- substitutes, for example among workers in tials, differentials will occur if job oppor- the same occupational category, complete tunities (demand) in the female sector are sektegregation by firm is expected, since small relative to the supply of female labor. sexualy integrated work forces are more Employers who do hire women will utilize costly (i.e., men must be paid a premium the labor-intensive production techniques to work with women). The necessity for wage that their lower wages make profitable. Thus, differences is obviated by such segregation, however, since men and women do not work together (and thus it is not necessary to com- costs to change history matters, Given historically'ris- pensate male workers for the disutility of ing female participation rates, women, as relatively new working with women). Discriminatory pay entrants, may find males already in place in many sec- differentials will arise in this case only if for tors. However, women have always been heavily con- some reason (e.g., costs of adjustvient due to centrated in a few female-dominated activities, even when they constituted a small proportion of the labor personnel investments in workers [Arrow, force. Assuming discrimination played a role in this 19731 ) complete segregation is not possible.5 segregation, it appears that the notion of the personnel investment tends to require both employer and employee discrimination. Further, once we have personnel investments men and women are no longer in fact 5 Of course, one may wonder why men rather than equally productive, although they may be potentially women are the recipients of these personnel invest-. equally productive (see below). merits. As Arrow (1973) points out where there are 6 See also Edgeworth (1922) and Fawcett (1918).

127 OCCUPATIONAL SEGREGATION AND LABOR MARICET DISCRIMINATION _ 121 in contrast to Becker's (1957) analysis, seg- One obvious solution to the problem raised regation may play acausalrole in producing above is that noncompetitive elements are discriminatory pay differentials.Further, responsible for the persistence of discrimi- discrimination may cause both payandpro- nation. Becker (1957) hypothesized that em- ductivity differentials between potentially ployer discrimination should, on average, equally productive male and female labor be less in competitive industries than in mo- women are less productive than men be- nopolistic ones. In the case of sex discrim- cause, as a result of segregation and crowd- ination the focus has tended to be on im- ing, they have less capital to work with. The perfections in the factor market rother than Bergmann formulation does not overcome in the product market. Madden (1973) has the problem with the Becker model, noted developed Robinson's (1933) monopsony earlier, that an extreme distribution of em- model to explain sex differences in wages. ployer tastes is necessary to generate the Monopsony describes the situation in which high level of segregation we observe. How- labor faces a single buyer. A price- discrim- ever, while Bergmann postulated employer inating monopsonist will pay fem *. labor discrimination as the source of segregation, less if itis less elastically supplied to the the overcrowding concept may be linked to firm than is male labor. Assuming that the any postulated reason for segregation. It is supply curve of labor to the firm is positively thus a persuasive explication of the wage sloped, the elasticity of labor supply to the consequences of segregation, regardless of firm is the percentage increase (decrease) in its cause. labor hours supplied to the firm in response Another question that has been raised to a given percentage increase (decrease) in about the Becker analysis, particularly with the wage offered by the firm. A lower elas- regard to the case of employer discrimina- ticity of labor supply for women thus means tion, is the issue of the survival of discrim- that the quantity of labor supplied by women ination in the long run under perfect com- to the firm isless responsiveto wage changes petition (Arrow, 1973). Assuming that than is the case for men. employer tastes for discrimination against The persuasiveness of this explanation for women vary, the least discriminatory firms aggregate pay differentials by sex is unclear, that hire the highest proportion of (lower- a priori. One issue relAtes to Madden's (1973) priced) female labor will have lower costs argument that female labor is less elastically and thus higher profits. Capital will flow supplied to the firm. On the one hand, as toward these .firms, and, assuming constant Madden (1973, 1976) argues, such factors as returns to scale, only the least discrimina- occupational segregation and the power of tory (lowest-cost) firms will survive. The ap- 'male unions may limit women's alternatives parent persistence of sex (and race) discrim- and thus decrease their wage elasticity of ination in the labor market over time has supply to the firm, all else equal. Supply- given rise to additional analyses of discrim- side factors, such as the tendency for women ination, which we shall consider below. to engage in less job search than men do or However, this criticism of the Becker model to seek jobs that are closer to home, could is a double-edged sword in that it creates also contribute to this result. On the other skepticism among many economists that la- hand, as Blau and Jusenius (1976) argue, the bor market discrimination is indeed respon- aggregate female labor supply curve (to the sible-(in whole or part) for the observed sex market) is more elastic than the male labor differences jn Market outcomes. Perhaps the supply curve-(to the market). This has con- best devOlopcd alternative explanation is the sequences for the elasticity of supply to the human capital model considered in the next firm in that home work provides a viable section. alternative for women at the margin of labor 128 122 FRANCINE D. BLAU force participation.? Furthermore, if men are First, employers may discriminate against more likely than women toacquire firm- women because of real orperceived average specific training, that would also lower their productivity or productivity-related behav- mobility relative to women. A second issue ioral differences between men and women. relates to the degree of monopsony that ac- In this case sex is assumed toprovide in- tually exists in the labor market. Pure mon- formation regarding expected productivity. opsony (one buyer) in alabor market is un- Aigner and Cain (1977) argue that economic doubtedly quite rare. However, Madden discrimination does not exist if the employ-. (1973) argues that there are considerable er'sperception of the average sex difference monopsonistic elements in the labor maris correct since on average each group is ket. This is an empirical issue deserving of paid in proportion to its productivity. How- more attention. Moreover, itis not clear ever, they acknowledge asdisquieting the that the case of few buyers of labor can be result that at each ability level women will analyzed in the same way as can the case of receive lower pay than men,9 Othershave one buyer. called this discrimination in that the "indi- It seems likely that the monopsony ex- vidual is judged on the basis of the average planation is more applicable to specific oc- characteristics of the group . .to.which he cupations and labor markets than tothe ag- or she belongsrather than upon his or her gregate sex pay differentia1.8 Anideal case own characteristics"(Thurow, 1975, p. 172).10 might be the employed female (or male) Implicit in this view seems to be the as- Ph.D. with an employed Ph.D. spouse in a sumption that other personal characteristics one-university town. Nonetheless;-this the- besides sex are readily available and that ory does set up amechanism by which oc- while the sex difference exists, on average cupational segregation may lower women's it would not be present in a ceterisparibus wages relative to men's in this case by comparison. If so, perhaps this mightbe reducing women's options. more appropriatelyconsidered mistaken be- A second approach to explainingthe long- havior on the part of employers. But, as run existence ofdiscrimination is the notion Aigner and Cain (1977) point out,discrim- of statistical discrimination (Phelps, 1972; ination based on employers' mistaken be- Arrow, 1972a, b, 1973; Aignerand Cain, liefs is as unlikely (or even more unlikely) 1977). Statistical dis'crimination provides a to persist over time inthe face of compet- motivation for discrimination that is com- itive forces as is discriminationbased on em- patible with profit-maximizing behavior on ployer tastes. So the question of the per- the part of employers. It sterns from im- sistence of discrimination in thelong run perfect information and may take one of two remains. forms.

9 This assumes that the variances of the measurement Women are more likely than men are to quit their error and of the productivityildicator are the same for jobs to leave the labor force, while men are morelikely males and females. than women to quit to change jobs (Barnes and Jones, I° See also Blau and jusenius (1976) and Piore (1971). Lewin and England (1982) argue that it is the explicit 1974). b Some preliminary evidence in support of the mon- use of ascriptive characteristicslike race or sex in per- opsony view as an explanationfor male-female wage sonnel decisions that constitutes the discriminatory as- differentials across urban areas, based on data for white pect of statistical discrimination, even whenthe em- males and white, never-married females, is presented ployer perceptions are correct. From a normative by Cardwell and Rosenzweig (1980). Note that in re- perspective this is certainly correct, but it is not clear stricting their analysis to never-marriedfemales they that it counters Aigner and Cain's (1977)argument that focus upon women with the lowest value of nonmarket such behavior does not constitute economic discrimi- tune. nation.

129 OCCUPATIO A SEGR CATION AN LABOR MARKET DISCRIMINATION 123

Arrow's (1973, 1976) notion of perceptual empirical studies (see below). As they note, equilibrium sheds some light on this issue. dispersion in risk aversion among employers Arrow's model allows for the endogeneity of should result in the bidding up of women's qualifications workers ,become qualified wages, just as the existence of less discrim- by making some type of investment in thee- inatory firms should erode discriminatory selves where the decision to invest depends pay differentials in the Becker-type taste- on the gain from qualifying. In this case em- for-discrimination model. Further, the ex- ployers' perceptions of sex differences in istence of a large risk discount of women's qualifications may become self-confirming wages should trigger a market for test in- even when there is no intrinsic sex differ- struments or indicators that are equally re- ence in ability or behavior. Multiple equi- liable for them. libria may result. For example, if employers' At present neither the role of occupa- view of female job instability leads them to tional segregation nor the issue of the per- give women less training and to assign them sistence of discrimination in the long run to jobs where the cost of turnover is mini- appears to be satisfactorily understood at the mized, women may respond by exhibiting theoretical leve1.12 Perhaps the most prom- the unstable behavior employers expect. This ising notion advanced here is Arrow's idea in turn confirms employer perceptions. On of perceptual equilibrium, a kind of "vicious the other hand, if employers believe women circle," or feedback theory of discrimina- are stable workers, they will hire women tion, which, as noted earlier, can accom- into positions that are sufficiently rewarding modate and provide some rationale for oc- to inhibit instability (Arrow, 1976)." Here, cupational segregation. 13 The overcrowding pay and productivity differences between concept can be appended to this model to potentially equally productive male and fe- imply a further reduction in wages due to male workers that may persist in the long segregation. A problem, however,' is that run can result from employers' in some sense this model cannot explain the sex segrega- erroneous beliefs. Viewing the matter some- tion that appears to exist among jobs re- what differently, thee employers' ex post quiring similar amounts of skill, stability, "correct" assessment of sex differences in etc. While this may, not be a major coin- average productivity may be seen to result ponent of occupational segregation, itis from their own discriminatory actions. probably a nontrivial component. Moreover, the resulting female sector may Institutional models, such as the internal be subject to a Bergmann-type overcrowd- labor market analysis or the dual labor mar- ling problem, further reducing relative wages ket model (Doeringer and Piore, 1971; Piore, there. 1971)," that give a more explicit role to oc- A second type of statistical discrimination cupations may be helpful here. Blau and may occur even if the two sexes have equal Jusenuis (1976) argue that the major con- average abilities or behavior. Risk-averse tribution of such models is not to suggest employers may discriminate against women if their ability or behavior is less reliably predicted by some indicator(s) than is men's 12 The latter point is emphasized i-. Darity's (1982) (Phelps, 1972; Aigner and Cain, 1977). How- consideration of racial pay differentials.. ever, Aigner and Cain express doubts that 13 A number of other authors have emphasized the risk aversion could cause discriminatory pay importance of feedback effects in analyzing sex pay and 'differentials of the magnitude obtained by occupational differentials. See, e.g., Bergmann (1976), Blau (1977), Ferber and Lowry (1976), Strober (1976), Weiss and Gronau (1981). " See also Cain (1976), Gordon (1972), and Wachter " See also Spence (1973, 1974). (1974). 130 ' 124 FRANCINE D. BLAU

new explanations for sex differentials in ficult one, to determine which view is correct earnings and employment distributions but or what portion of the observed differences is rather to elucidate the linkage between the accounted for by each explanation. two: to suggest why the same set of factors As .developed by Mincer and Polachek that produce earnings differentials is also (1974) and others, the human capital anal- _likely to generate employment segregation. ysis calls attention to the traditional division They argue that under the administered sys: of labor by sex within the family under which tem of the internal labor market, the firm women can expect shorter and more dis- attaches wage rates to occupational cate- continuous involvement in market work than gories rather than to individuals (see also can men. This reduces their long-run payoff Thurow, 1975). Under such circumstances to human capital investments, since they the only way in which the firm can distin- have a shorter work life over which to reap guish between men and women in terms of the returns. Similarly, employers will be re- pay is to assign them to different job cate- luctant to invest in firm-specific training for gories. Within occupational categories, in- women workers. Thus, women may earn less stitutional constraints mandate equal pay for than men both because of their lesser amount equal work, except for 'relatively small dis- of labor market experience and because of tinctions based on seniority and/or merit the lower returns to experience they obtain considerations. Such group treatment of in- (the latter reflecting their smaller invest- dividuals will be most efficient (will result ments per unit of time), Female earnings in the discarding of the least information) if are further reduced by the depreciation of occupational categories are as homogenous their.stock of human capital during the time as possible. Thus, employers are likely to they spend out of the labor force. structure female jobs to fit the perceived av- The human capital model can also provide erage characteristics of women workers. This, a theory of ocr,iupational choice to explain in turn, influences women's behavior and pro- occupational segregation by sex (Polachekr ductivity a la Arrow's perceptual equilibrium. 1976, 1979, 1981; Zellner, 1975; Landes, Bergmann's overcrowding mechanism may 1977). According to this view women will further widen the pay differential between tend to avoid occupations requiring consid- male k; ad female jobs. erable investments in on-the-job training and having high rates of depreciation for time spent out of the labor force. Thus, we would The Human Capital Alternative expect predominantly female jobs to have Theories of discrimination are concerned relatively flat age-earnings profiles. with explaining occupational and pay differ- One problem with this analysis is that the entials between (potentially) equally produc- sexual division of labor within the family is tive men and women. The aggregate figures in turn influenced by the relative market cited earlier in fact compare heterogeneous rewards (wage rates) of husband and wife male and female labor. Some or all of the (Becker, 1965, 1973). This implies that labor observed occupation and pay differences may market discrimination against women could in fact be due to productivity differences be- influence their allocation of time between tween the sexes. The human capital model the home and the market and thus the amount can provide a consistent explanation for oc- and types of their human capital invest- cupation and pay differentials by sex in terms ments. From the standpoint of empirical work of the voluntary choices of women rather than this means that analyses which treat such of market discrimination against them. It then variables as experience and education as ex- becomes an empirical question, albeit a dif- ogenous may underestimate the extent of

131 125 O OCCUPATIONAL SEGREGATION AND LABOR MARKET DISChIMINATION

labor market discrimination (Weiss and Discrimination and Earnings Gronau, 1981). Further, women's lower returns to experience may reflect employers' A crude test of the relative merits of the reluctance to provide opportunities for firm- discrimination and human capital explana- specific training as well as their own vol- tions for sex differences in earnings is pro- untary decisions. In the latter case it would vided by an examination of the time series disc) be important to determine whether the trends in the sex pay differential. We first employers' decisions are in fact justified by briefly consider this time series evidence, ceteris paribus sex differences in quit pro- then move to a detailed discussion of the pensities. more sophisticated cross-sectional analyses of the extent of labor market discrimination. As noted In the introduction, there has ac- EMPIRICAL FINDINGS tually been some deterioration in the rela- As indicated in the preceding section, there tive earnings position of women since the are alternative views as to how labor market mid-1950s. It has frequently and incorrectly discrimination might produce occupational been assumed the,/ increases in the female and pay differences between men and labor force participation rate over time are women. Further, the human capital model indicative of declines in the average level of provides an altt,rnative explanation for sex experience of women workers due to the differences in market outcomes that is con- impact of new entrants (Economic Report of sistent with voluntary decision making by the President, 1974). In fact, as Mincer (1979) women rather than with discrimination explains, the female labor force grows not against them. For the most part empirical only through "widening" (increases in the research has been focused on the question flow of entrants or reentrants) but also through of whether or not labor market discrimina- 'deepening" (decreases in the flow of exi- tion against women (and minorities) exists. ters). The impact on the .average level of Choosing among alternative models of dis- experience of female workers depends on criminuion and understanding the causes the relative magnitude of these two flows. and consequences of employment segrega- In addition, since la or force entry tends to tion have received considerably less atten- be selective of female nonparticipants with tion, particularly from economists. higher levels of previous labor force expe- In this section we first consider in some rience, and labor force exit tends to be se- detail the question of whether or not dis- lective of female participants with lower lev- crimination exists. This is an important in- els of previous labor force experience, labor quiry, since its resolution is necessary in turnover does not dilute average experience order to determine the overall context in levels as much as it might first appear (Blau, which employment segregation by sex takes 1975, 1978). place. There is no point in seeking to de- Lloyd and Niemi (1979) present a variety termine the role of segregation in producing of evidence indicating that the trend has discriminatory pay differentials by sex if in been for women to remain in the labor force fact there is little evidence that such dis- longer and more continuously and that en- criminatory differentials exist. Second, we trants make up a decreasing proportion of explore the existing literature concerning the the female labor force. They conclude that role of occupational segregation in produc- "it appears. . .the work experience differing pay differentials and, more briefly, some ential between the sexes has narrowed in of what has been learned about the causes the past twenty years" (p. 133). Thus, female of such segregation. participation trends do not seem to be re-

132 126 FRANCINE D. BLACI sponsible for the widening pay gap. More- been increasing their earnings position rel- over, they find that "over time,"the gap be- ative to younger men. Between 1970 and tween men's and women's educational 1981 the ratio of the mediamincomes of full- attainment has been eliminated and, among time, year-round women workers to those young cohorts, it appears that women's pro- of men increased from 59 to 60 percent for pensity to enroll in college is roughly similar all women, but rose from 65 to 70 percent to men's" (p. 146). Men and women do con- among 25- to 34-year-olds (U.S. Dept. of tinue to differ sharply in fields of speciali- Commerce, 1980, 1982).. Beller (this vol- gation, although some progress has recently ume) also found that younger cohorts ex- been made in this area as well (Baker, un- perienced greater declines in occupational dated). Further, one may speculate that the segregation than did theorork force as a rising' divorce rate has increased women's whole. Thus far, however, the modest gains incentives to invest in their labor-market- by younger women do not provide over- related. human capital. whelming support for this view. Much will While one would like better data, partic- depend on how this group fares in the next ularly on experience, it seems reasonable to few years. It is also important to point out conclude that human capital factors do not that if employers were willing to reevaluate account for the widening pay gap between their traditional promotion ladders they might men and women. This provides support for find that many older women workers do have the view that discrimination plays a role in the experience necdssary for higher-level producing the differential. It is unlikely that positions. tastes for discrimination against women have During recent years, while women as a increased during this period. However, as group have made little progress in advanc- we have seen, little decline occurred in the ing their earnings positions relative to men, degree of sex segregation in employment black women have advanced relative to white during that time. If increases in the demand women. Black women's median (full-time, for labor in female jobs did not keep pace year-round) incomes have increased from 69 with increases in the, supply of female labor, percent of those of white women in 1964 to relative "crowding" in female jobs may have 90 percent in 1981. Black men have also worsened, exerting a downward pressure on gained relative to white men. Their median female wages relative to male wages. In- full-time, year-round incomes rose from 66 creases in the real wages of women over the percent of those of white men in 1964 to 71 period could have continued to induce in- percent in 1981 (Lloyd and Niemi, 1979; creases in female labor force participation U.S. Dept. of Commerce, 1982). The gains ceteris paribus. in the relative incomes of blacks are partially In addition, it may be argued that women due to substantial increases in their relative face substantial experience and training re- educational attainment but cannot be fully quirements in their efforts to move into high- explained by that factor (Brown, 1982). They level male jobs (Freeman, 1973). This makes may in part be due to the impact of equal it difficult for equal employment opportu- employment opportunity legislation. The nity legislation to open the doors to such more rapid improvement in the relative in- jobs for older women. Further, younger come position of black women (compared women may have had the opportunity to with white women) than of black men (com- incorporate new expectations of greater la- pared with white men) may reflect the large bor force attachment over their life cycles number of entry-level positions in many into their human capital investment deci- typically female jobs, in comparison with siofis to a greater extent than have older the higher experience and training require- women. It is true that younger women have ments in typically male jobs discussed above.

133 OCCUPATIONAL SEGREGATION AND LABOR MARKET DISCRIMINATION 127

We now turn to a review of studies that ferences with respect to some productivity- attempt to measure the extent of labor mar- related characteristics may reflect the indi- ket discrimination explicitly. The general rect effects of discrimination (Blinder, 1973). practice in empirically estimating the mag- For example, as discussed earlier, women may nitudef labor market discrimination against be excluded from high-wage occupations be- a particular grow?, e.g., women, is to as- cause of their sex. Further, if the endogeneity certain the proportion of the sex differential of choice variables is taken into account, we thfit is accounted for by differences in the see that labor market discrimination maydis- productivity-related characteristics of male courage women's human capital accumulation and female workers and to allocate the re- . or alter its type. Measured laborrparket dis- sidual to discrimination, Discrimination may crimination is likely to be underestimated to be measured by the coefficient on a sex the extent that such factors representing other dummy variable in an earnings regression dimensions of discrimination are controlled equation, in which case the impact of the for. other explanatory variables on earnings is A related problem is the interpretation of wnstrained to be the same for each, sex group. sex differences in the coefficientsof earnings More often, this constraint is relaxed by es- regressions. For example, a smaller coeffi- timating separate earnings regressions for cient on labor force experience for women sex or race-sex groups. Inthis case the por- may reflect their decisions to invest inless tion of the pay differentialllue to sex on-the-job training than men do, as pro- ferences in the returns to a given set of char- posed by human capital theorists, or dis- acteristics (i.e., sex differences in coefficients, crimination on the part of employers re- including the constant term) is attributed to sulting in less access to on-the-job training discrimination; the portion of the pay dif- opportunities. I5 Similarly, sex differences in ferential that is due to sex differences in the returns related to marital status may re- endowments of productivity-related char- flect unmeasured differences in labor force acteristics (i.e., sex differences in the means attachment between married men and of the independent variables) is attributed women (Polachek, 1975) or sexdiffeienees, to nondiscriminatory factors (Blinder, 1973). in employer treatment of marital status that There are various problems with this "re- is unrelated to productivity (e.g., the view sidual" approach to measuring labor market that married men deserve higher salaries discrimination. Perhaps the most serious is because they have families to support). the specification problem. On the one hand, These problems of specification and in conventional data sources do not allow for terpretation coefficients reflect a more the measurement of all productivity-related fundamental problem': We would like to characteristics. The absence of actual labor measure the extent of labormarket dis- market experience from the data sets used criminationa demand-side phenomenon. in the early studies of sex discrimination is But wages are influenced by both sup- a prime exampleof an important omitted ply- and demand-side factors. The earnings variable. Type (as opposed to amount) of functions that are typically estimated are es- schooling, which varies greatly between men sentially reduced-form equations, and thus and women, would he another example. If, their coefficients may reflect the influence on average, males are morefavorably en- dowed with the characteristics measured by these omitted variables, the extent of labor I5 Further, one may question the basic premise of market discrimination will be overestimated the human capital model ,that upward-sloping experi- because of imperfect controls for these omit- ence-earnings profiles are indeed due to on-the-job ted factors. On the other hand, group dif- training. See, e.g., Medoff and Abraham (1980, 1981). A

0: 134 128 FRANCINE O. BLAU

of both supply and demand (Butler, 1981; performing the analysis on grouped data Chip lin, 1981). Ideally, one would like to where the grouping criterion is independent specify and estimate 'a structural model with of the measurement errors. But it may not separate supply and demand equations. No be easy to meet this requirement (or to know one has yet attempted to specify and esti- whether or not one has met it), and, as they mate such a model on an economywide ba; acknowledge, the results can be quite sen- sis. sitive to one's choice of a criterion. Another An additional problem with the earnings classical approach to the errors -in- variable regression approach is that, while the theory problem involves the use of instrumental specifies the measurement of discrimination variables. However, the specification of the in terms of pay differences that are not ac- appropriate instruments is difficult gii,en the counted for by productivity differences, the current state of theory and the availability empirical work involves adjustment using of data (Kamalich and Polachek, 1982). proxies like education and experience for An additional problem in using conven- productivity. This giyes rise to an errors-in- tionally estimated earnings regressions to variable problem (Hashimoto and Kochin, measure discrimination most likely pro- 1980; Roberts, 1980). For simplicity, let us duces biases in the opposite direction, that assume that the productivity proxies are is, leads us to underestimate discrimination. measured so that they are positively related Regressions are generally estimated on the to earnings. Then it may be shown that if selected sample of labor force participants. women have lower mean values of the pro- But Gronau (1974) has argued that it is the ductivity proxies than men have (as is gen- wage offers, not the actual wages of males erally the case), the coefficient on a sex and females,. that should be compared. The dummy variable (female= 1) in an earnings distribution'of actual wages represents only regression is biased 'downward, giving an that part of the offer distribution that is ac- exaggerated estimate of the effect of dis- ceptable to job seekers." Thus, according crimination (Hashimoto and Kochin, 1980). to Gronau, mean female wage offers will be Unfortunately, however, there is no ob- overestimated by restriction of the sample 0 vious solution to this problem. Roberts (1980) to labor force participants.18 He provides suggests a procedure that he terms "reverse empirical evidence in support of this con- regression," in which the independent var- tention. Since male participation rates (par- iable (e.g., education) is regressed on the ticularly in the prime work ages) are still dependent variable (e.g., earnings) and a substantially higher than women's, it may sex dummy.16 Such a procedure will prcf- be argued that data on men's wages are con- duce unbiased coefficients only if the de- siderably less affected by selectivity bias.") .pendent variable is measured without error. But earnings as measuredare only dan imperfect, indicator of permanent earnings ca- '7 See also Heckman (1974). pacity, which, one might argue, is the the- Gronau has argued elsewhere (1973) that the value oretically relevant variable (Hashimoto and of time of housewives may be either higher or lower Kochin, 1980). Moreover, measured earn- than the market wage of comparable employed women. If the foimer is true, it is not necessarily the case that ings are only a proxy for the total rewards restriction of the sample to labor force participants ov- for the job, including fringes and the non- erstates women's wage offers. However, the empirical pecuniary benefits of the work (Madden, evidence he presents supports Gronau's (1974) conten- 1982). Hashimoto and Kochin (1980) suggest tion. See also Cogan (1980)., 9 As noted earlier, black male participation rates are lower than those of whites. Thus, the same type of selectivity problem discussed in the text affects race '6 See also Kamalich and Polachek (1982). comparisons among males.

135 0

OCCUPATIONAL SEGREGATION AND LABOR MARKET DISCRIMINATION 129

Thus, observed wage differences between census data, Fuchs (1971)found that sex dif- men and women willunderestimate the ferences in individual characteristics could "true" male-female wage-offer differential. account for only 3 to 15 percent of the dif- On the other hand, over time male partic- ferential. Similarly, Oaxaca (1973a,b),. using ipation rates have been declining while fe- data froM the 1967 Survey of Economic Op- male rates have been increasing (see above). portunity, found that 80 percent of the pay Thus, we may be sweeping out more of the differential between white men and white female offer distribution and less of the male women and 94 percent of thedifferential offer distribution than was formerly done. between black men and black women could This may be a partial explanation for our not be explained by productivity-related in- failure to observe a decline in .the male-fe- dividual characteristics.22 male pay gap over time. As these authors were aware. their lack Bearing these problems in mind, we con- of data onactual labor force experience cre- sider the empirical work on sex discrimi- ated an important omitted-variable prob-- nation. As noted earlier, most empirical work lem.23 The general procedure of estimating in this area has focused on the question of experience as the years elapsed since school whether or not labor market discrimination completion, while fairly accurate for males, exists and has attempted to estimate its mag- is much more questionable fqr females. In nitude. While there are still various unre- addition to their theoretical contributions solved problems in estimating the extent of noted earlier, Mincer and Polachek (1974) labor market discrimination, it is not clear were the first to provide empirical estimates from our consideration of them that they of the impact of labor force experience and result, on net, in overestimates or under- time spent out of the labor force on the earn- estimates of discriminatory pay differentials. ings of women using newly available lon- If the evidence suggests that there is labor gitudinal data. They analyzed retrospective market discrimination against women, then work history data from the National Lon- this will provide some motivation for con- gitudinal Survey (NLS) of mature women sidering the role played by ..segregation aged 30 to 44 in 1967. Mincer and Polachek in employment in producing these discrim- were able to account for 45 percentof the inatory outcomes, `pay gap between white married men and Not surprisingly, the estimate of the sex. women inthat age group on the basis of pay differential that is due todiscrimination human capital variables, including actual la- varies considerably depending on the group bor market. experience and time spent out studied, data set employed, and variables of the labor force. In arriving at this esti- controlled for. We shall focus'our discussion mate, they were aware of the joint deter- on studies usingnational samples of indi- mination of earnings and experience and at- viduals across a variety of occupations.2° For The most part, the earlier studies of male- female pay differentials attributed a sub- 22 Oaxaca's results including controls for occupational stantial portion of the sex tiay differential to and industrial characteristics are considered below. See also Cwartney and Stroup (1973), Sawhill (1973), and discrimination.2' For example, using 1960 Blindef (1973). On the other hand, Darity and Myers (1980) found that while the structural equations for wages for white males and females were significantly 2" For more detailed reviews of the empirical liter- different, they were dot so in the case of black males ature, including occupation-specific studies, seeKohen and females. (1975). Lloyd and Niemi (1979). and Treiman and Hart- 23 For an except!on, see Suter and Miller (1973), who mann (1981). restricted their sample to women from the National 21 The one exception was a study by Sanborn (1964)i Longitudinal Surveys in the 30- to 44-year age group see below. who had worked continuously since school completion.

136 130 FRANCINE D. BLAU

tempted ,to adjust' for simultaneous equations This reasoning implies that the positive 'bias by using a two-stage procedure. While coefficient on experience in an ordinary least Mincer and Polacilek are to be commended squares (OLS) earnings regression is bidsed for their awarene'ss of the simultaneity prob- upward. This is because some of the esti- f. lem, their application of the two-stage pro- mated positive impact of labor market ex- cedure in their exploratory study is far from perience on wages is really due to a positive thorough. For one thing, their specification effect of wages on experience (at higher of the determinants of lifetime labor supply wages, women supply mote,time to the Mar- (the proportion of years the respondent ket), Similarly, the negative:\ Coefficient on worked 6 months or more since school com- home time is biased upward in absolute value. pletion) includes an endogenous variable, This is because some of the estimated neg- number of children. For another, while life- ative impact of home time tm wages is really, - ,- time experience is treated as endogenous, due to a negative effect of wages on home home time and job tenure (seni9rity) are tinie (at higher wages, women spend less entered into the earning; function as ex- time out of the labor market). Correction for ogenous variables (Sandell and Shapiro; 1978; the simultaneous equations bias should thus Mincer and Polachek, 1978). reduce the size of both coefficients in ab- That these problems are serious is sug- solute value. On the contrary, Mincer and. gested by their counterintuitive findings for Polachek (1974) find that, if anything,. rees- the two-stage procedure. In the ingle-pe- timation of the earnings function using two- riod context, economic theory would. lead stage least squares yields "larger positive one to expect a positive' relationshipbe- coefficients for (total) experience and stronger tween labor force participation and own negative coefficients for home'time" (p. S99). wage. An increase' in the market wagavail- Further, in estimating the proportion of able to a woman increases the opportunity the pay gap explained, by the human capital cost of nonmarket activities for her. This variables, the actual mean levels of the var- encourages her 4q substitute marketwork iables are employed. Yet the heart of the for time spent on housework and leisure. endogeneity problem is that wage discrim- flowerer, in a multiperiod,context this pos- ination may have influenced the artaunt of itive substittition effect could be outweighed human capital that women have accumu- by a negative income effect. The negative lated. Thus, itis likely that Mincer, and income effect arises from the fact that the Polachek and analyses modeled on theirs increase in the wage obtained while the overestimate the impact of the human cap- woman is working is like an increase in in- ital variables on the sex pay differential. In come.. At higher-income levelsshe is ex- light of this potential bias it is notable that pected to demand more of all goods from Mincer and Polachek found that over half of which she derives satisfaction,"includingiei- the pay gap between white married men and sure. She may thus consume moreleisure women could not be explained bythe hu- over the life cycle by supplyingless time to man capital variables and was potentially due the market over the life cycle. Note that she to discri- .ination. must work some of the time for the income Another issue that has been raised re- effect to come into play. The evidence sug- garding Mincer and Polachek's (1974) find- gests, however, that for married women's ings is the generalizability of their results participation decision's the substitution ef- for the 30- to 44-year-old age group to the fect dominates the income effect (where the whole female population as is necessary latter is indicated by the response to changes to draw inferences for the aggregate male- in husband's income [Mincer, 19621). Thus, female pay gap. The work of Corcoran (1978, women's labor supply is expected to be pos- 1979) using a full age range from the 1976 itively related to their wages. Panel Study of Income Dynamics (PSID),

137 OCCUPATIONAL SEGREGATION AND BOR MARKET DISCRIMINATION 131 suggests that their findings may not in fact discussed in reference to the Mincer and be generalizable. She found that the wages Polachek (1974) study. This suggests that, of women aged 30 to 44 are much more at least with regard to this consideration, strongly affected by labor force withdrawals she has underestimated the effect of labor than are those of the broader age range (Cor- market discrimination. coran, 1979). Women in this 30-to 44-year An additional issue raised by the human age group are more likely than are women capital model is the interpretation of sex dif- in general to have recently reentered the ferences in returns to experience and mar- labor market after a prolonged period of ital status. For example, Mincer and Pola- nonparticipation. Corcoran suggests that her chek (1974, p. S103) argue that findings are consistent with the notion that women's wages are temporarily depressed The association of lower [female] coefficients with by labor force withdrawals because of mis- lesser work experience is not fortuitious: a smaller fraction of time and energy is devoted to job information about job opportunities and/or advancement (training, learning, getting ahead) about their own value as workers.24 per unit of time by persons whose work attach- In addition to providing work histories, ment is lower. Hence, the 45 percent figure in the PSID data provide measures of labor the explanation of the gap by duration-of-work force attachment (i.e., absenteeism due to experience alone max.12s)dewed as an under- own illness; absenteeism due toillness of statement. others; self-imposed restrictions on work hors and'or job location; voluntary part- To what extent do sex differences in returns time work). For an unrestri_led age group to experience and marriage reflect employer of women, Corcoran (19Th) found that ad- discrimination and to what extent do they justments for schooling, work history, and reflect women's choices? While a definitive labor force attachment accounted for 36 per- answer is not available, the evidence sug- cent of the wage gap between white men gests that discrimination most likely plays a and white women and 27 percent (he wage role. gap between white men and black women. With respect to returns to experience, a In addition to the difference in age group, study by Duncan and Hoffman (1979) is par- some of the difference betweenCorcoran's ticularly interesting. Using direct measures and Mincer and Polachek's findings with re- of on-the-job training from the 1976 PSID, spect to the importance of the human capital they find that men and women receive about variables may reflect a growing work force the same payoff to on-the-job training. How- attachment of women over the period ever, "past years of work experience have a spanned by the 1967 NLS and the 1976 PSID high payoff in training for men, especially (Mincer, 1979). These findings of large white men, but have very little effect on the unexplained wage gaps for white and black chances that women will receive training" women are impressive in light of Corcoran's (p. 601). They see their results as consistent detailed controls for productivity-related with a view that firms have different pro- factors. In addition, since her findings are motion practices on the basis of sex (and based on OLS estimates, she has not ad- race). Thus, the observed lower returns to justed for the simultaneous equations bias experience of women may reflect to some extent employer discrimination in permit- ting women access to on-the-job training opportunities. 24 Using longitudinal panel data from the NLS on the On the other hand, Sandell and Shapiro 30- to 44-yea age group. Mincer and Ofek (1982) find (1980), analyzing data from the NLS, have direct evidence of rapid wage growth upon reentry. Unlike Corcoran (1979), however, they attribute it to found that young white and black women the "repair" of previously eroded human capital. who plan to work at age 35 have expenence-

138 132 FRANCINE D. BLAU wage profiles that "begin at a lower point With respect to worker and employer firm- and have a steeper (initial) slope than those ,specific training investment decisions, it is of their 'no-work-plans' counterparts" (p. 343). job (rather than labor force) stability that is Thus, some of the sex difference in work the issue. Some, evidence in favor of the experience coefficients may reflect women's Arrow notion in this case is provided by the own investment decisions based on their work findings of Blau and Kahn (1981), who used expectations. However, Sandall and Shapiro data on young men and youngwomen from also find that the returns to job tenure (sen- the NLS to analyze sex differences in quit- iority), which they take as an indication of ting. They found that, all else equal (in:: investment in firm-specific training, do not eluding job related characteristics), white and differ significantly between those who plan black women were no more likely to quit to work at age 35 and these who do not. This their jobs than were men of the same race. in turn suggests that part of the differential They also found that a high proportion of access to on-the-job training opportunities the observed sex differential in quitting was by sex implied by Duncan and Hoffman's associated with job rather than personal findings may reflect the inability or the un- characteristics.26 Similar findings were re- willingness of employers to distinguish be- ported by Viscusi (1980) for a larger age range tween the committed and uncommitted group. from the PSID.26 Both Blau and Kahn and As our previous discussion suggests; there is Viscusi found that, all else equal, blacks were some difference of opinion as to whether such less likely to quit their jobs than white work- statistical discrimination would represent eco- ers of the same sex. Blau and Kahn also nomic discrimination or not. However, it is report some support for one of the models likely to result in committed women getting of statistical discrimination presented ear- less training and receiving lower returns to lier in that female quits were found to be job tenure than do committed men. less accurately predicted th..n men's by the In evaluating the implications of Sandell explanatory variables." No basii for statis- and Shapiro's findings for sex differences in tical discrimination was found in the case of earnings, it is instructive to consider Ar- race. row's (1973) notion of perceptual equilib- With respect to the interpretation of sex rium. Given the set of market opportunities she can reasonably expect and her expected value of nonmarket time, each young woman 25 Job characteristics include current wage, long-run determines her work plans for age 35. Since earnings opportunities associated with the job, collec- women with given characteristics in the tive bargaining coverage, and industry and occupation dummy variables. Sandell and Shapiro sample presumably face se Ragan and Smith (1981) find that sex differences similar opportunities, the differences in their in industry turnover rates explain a substantial portion work plans probably reflect differences in of the sex difference in earnings among individuals. their expected value of nonmarket time However, as they acknowledge, since their data refer (perhaps due to differences in tastes). How- to the industry's history and not the individual's, their possibility findings are consistent with the possibility that women ever, this does not preclude the are restricted to low wage/high turnover jobs (e.g., that,if confronted with a different set of those requiring little specific training). Osterman (1979) market opportunities, substantially more of found no sex differences in absenteeism, all else equal, the women would be committed to market for a sample of professional workers. work. It is even possible that, given the male 27 While Osterman (1979) reports no basis for statis- set of job opportunities (with similar returns tical discrimination on this ground, Kahn (1981) shows that Osterman does not employ the correct indicator to experience and job tenure), they would of predictability. When the correct indicator is used. exhibit the same degree of labor force com- Kahn finds that women's absenteeism is less accurately mitment as that of "males. predicted than men's.

139 OCCUPATIONAL SEGREGATION AND LABOR MARKEDISCRIMINATION 133 differences in the returns to marital status, in treatment, if it exists, would have tobe Hill (1979) uses explicit data on experience, classed as discriminatory. human capital investment, and labor force The evidence reviewed here strongly sug- attachment from the 1976 PSID to explore gests that labor market discriminationdoes this issue further. She finds that, allelse indeed play a role in producing the observed equal, marriage has strongvositive wage ef- male-female pay differential. While it is dif- fects for (white and black) men, whilethe ficult to pinpoint the exact portion of the sex effects of marriage on (white and black) pay gap due todiscrimination, the findings women's wages are not significantly differ- suggest that over half of thedifferential can- ent from zero. Malkiel andMalkiel (1973) not be explained by sexdifferences in pro- and Osterman (1979) report similar findings ductivity-related factors. Some .considera- for the wage effects of marriage fromstudies tions (e.g., omitted variables) suggestthat of sex differences in the wages of profes- this may be an overestimate of the magni- sional workers, also including good controls tude of discrimination. On the other hand, for work experience and attachment. A sup- other considerations (e.g., the impact of la- ply-side explanation for these findings can- bor market discrimination on the incentives not be entirely ruled out: giventhe tradi- to acquire experience,training, etc.) sug- tional division of labor within thehome, gest that it may be an underestimate.Thus, married men may be more highly motivated we are still left withfairly strong evidence or hardworkingthan single men with similar of the importance of discrimination. measured characteristics, while married women may beless highly, motivated or Occupational Segregation and Earnings hardworking than their single counterparts. But, as Osterman points out, howdo we Having determined the overall labor mar- then explain the absence of a negative effect ket context in which employment segrega- of marriage on the earnings of women? tion by sex takes place, we are nowable to Moreover, Hill examined the effect of mar- turn to a consideration of theempirical evi- ital status and number of children on wages dence regarding the role played by such seg- and a as more explicitcontrols for experience, hu- regation in producing pay differences man capital investments,and labor force at- consideration of some recent evidence on tachment were added to the wage regres- the causes of such segregation. According to sions. She finds that the wage effect ofmarital t le discrimination modelsconsidered earlier, status among all race/sex groups remainsquite segregation may work to lower women's earn- stable and concludes that marital statusdoes ings because of a lesser provisionof on-the- not serve as a proxy for theseproductivity- job training and fewer incentives for worker related factors. If marital status is not serv- stability in female jobs (Arrow's perceptual ing as a proxy for these obviousand impor- equilibrium model, institutional models) and/ tant factors, it seems doubtfulthat it is serv- or because ofovercrowding, While these no- ing as a proxy for moresubtle traits like tions suggest that it would be instructive to motivation. However, Hill doesfind that look at the relationship between earnings number of children is to some extent proxy particularly the discriminatory pay gap and for these factors. Both Ostermanand Hill segregation, there are three majorproblems feel that employers may believe that mar- in arriving at an empirical estimate. ried men deserve higher salariesbecause of Ther first is a data problem. Most data their greater financial responsibilities. Given sources, even detailed censusdata, tend to traditional views of men's and women's eco- aggregate some male and femalejobs into nomic roles, they do not feelthe same is apparently integrated categories (Hartmann true of married women.Such a difference and Reskin, 1982). Further, insofar as men

140 134 FRANCINE D. BLAU and women are segregated by firm within pation on pay by contrasting the wages of occupational categories, aggregation across women in male and female jobs, ceteris par- firms will result in an underestimate of the ibus, may result in an underestimate. On extent to which men and women are in seg- the other hand, the internal labor market regated work settings (Blau, r977). The im- (institutional) model suggests that when pact of both of these factors is revealed in a women are able to obtain employment in recent study by Bielby and Baron (this vol- male jobs within a firm, they should be paid ume) of 393 California establishments. Us- at about the same rate as men. This implies ing the employers' own job classifications, that pay comparisons of women in male and they found that 51 percent of the firms were female jobs can provide good estimates of completely sex-segregated with respect to the impact of occupation on wages. How- job classification: no men and women shared ever, since data are generally aggregated the same job title. An additional 8 percent over firms, women may earn less than men of the firms were single-sex establishments. do in male jobs if they work for low-paying The mean index of segregation of the re- firms. Further, women may be segregated maining 41 percent of firms was 84.1. It has by job level within the same occupational been found that the portion of the sex pay category. gap associated with occupational differences Judging the impact of occupation by com- is larger, the finer, the occupational cate- paring the wages of men in male and femalkr:, gories employed (Treiman and Hartmann, jobs does not appear to be satisfactory either. 1981). Thus, it seems reasonable to con A problem here is that men are not believed elude that aggregation problems result in an to be discriminated against in or excluded underestimate of the impact of employment from male jobs, so the question arises as to segregation on the sex pay gap. why they are employed in the female sector. The second problem, is more conceptual. If it is because they have. very strong non- The logical way to determine the impact of pecuniary (nonmonetary) tastes for female occupational characteristics or categories on jobs, then their wages may reflect a differ- pay, all else equal, is to control for sex or ential due to overcrowding. The pay differ- to look within sex groups. Yet it seems pos- ential between men in male and female jobs sible that the existence of overcrowding in may also reflect overcrowding if bad luck, female jobs may lower .the 7ges of women poor information, or job rationing in male in male jobs. Even when vrmen work in jobs (e.g., due to unions) are responsible for male jobs their opportunity \set may differ the employment of men in female jobs. On from that of their male coworkers: the lower- the other hand, men may work in female paying alternatives they face in the female jobs because they have found a niche there sector may reduce their supply price to the that pays comparably with what they could firm. This is a potential wage spillover effect earn in male jobs, e.g., due to employment of overcrowding. Further, women may face in a high-paying firm or at a high level in discrimination in the male sector that lowers the occupation hierarchy. in that case a their relative wages there. Indeed, in Berg- comparison of men's wages in male and fe- mann's original formulation of the over- male jobs will not, reflect an overcrowding crowding model it is the exclusion of women differential. from male jobs due to discrimination that These two considerations suggest that causes the overcrowding of the female sec- empirical estimates may understate the con- tor.It makes intuitive sense that women tribution of employment segregation to the remain concentrated in female jobs because sex pay gap. A third point works in the op- they have little to gain by obtaining male posite direction. Workers in higher-paying jobs. Thus, measuring the impact of occu- jobs (or firms) may have unobserved char-

141 OCCUPATIONAL SEGREGATION AND LABOR MARKETDISCRIMINATION 135

'a. acteristics that are associated with higher of the sex pay gap on the basis of controls for productivity: what appear to be occupation such factors as detailed occupation, age, and effects on wages may actually be due to education.29 The greater magnitude of the ex- unobserved productivity differences among plained differential in the Sanborn study than workers. in the studies of Fuchs (1971) and Oaxaca The union-impact literature suggests some (1973a,b) cited earlier, in which occupational fruitful alternative approaches for examining controls were not included, is an indication the impact of crowding, since many analo- of the importance of occupational category in gous problems arise in investigatingthat is- determining earnings. Further, using data from sue. For-example, papersby Kahn (1980) the 1974 PSID sand the 1967 NLS, England and Flanagan (1976) suggest that one might (1981, 1982) found that, all else equal, the look at the effect on the wages of men and percentage of females in the occupation is sig- women in male andfemale jobs of changes nificantly negatively related to female earn- over time or of differences acrosslabor mar- ings.° kets, in the degree of crowding. This type Table 7-1 illustrates the impact of occupa- of approach could provide an empirical es- tional category within an individual firm a timate of the impact of crowding that takes large fiduciary institution.3' In equation (1), spillover effects into account. The selectivity only controls for sex and race are entered into problem of comparing the wages of women *he regression, thus providing an estimate of (or of men) in male and female jobs might the gross or unadjusted sex-race' differentials. be overcome by using a technique devel- Equation (2) includes controls for productiv- oped by Lee (1978) in his investigation of ity-related individual characteristics but no the union-nonunion differential. controls for occupational characteristics. This While existing studies may deal inade- quately with the problems raised here, it is still instructive to review the findings in this Including adjustments for even more narrowly de- area. Using data from the1980 census, Trei- fined occupational. categories from the BLS, as well as man and Hartmann (1981)found that 35 to. sex differences in turnover,absenteeism, and work ex- 39 percent of the earnings difference between perience, Sanborn (1964) was able to explain 71 percent associated with sex dif- of the pay gap. The problem with this portion of his men and women was analysis is that he most probably engaged in double ferences in the distribution of their employ- counting. He assumed that the sex differences in age ment among 479 detailedcategories.° Oc- and education that prevailed within the census cate- cupational differences appear to be a significant gories also prevailed within the more detailed BLS factor in explaining the.sex pay gap, even when categories. Further, he adjusted within occupations for estimates ofaggregatesex differences in turnover, ab- other productivity-related factors are con- senteeism, and work experience. Leaving aside the trolled for. For example, Oaxaca (1973a,b) issue of endogeneity, labor quality differences between found that the inclusion of controls for major men and women are likely to beconsiderably smaller occupation and industry and for class ofworker within occupations than in the aggregate, since it is increased the portion of the sex pay gap ex- these traits thit sort people into occupations. Indeed, in the presence of discrimination women may be more plained from 20 to 37 percent in the case of qualified than men in specific occupations. See Ham- whites and from 6 to 39 percent in the case ilton (1973) for some evidence consistent with the latter of blacks. In this case job characteristics ac- possibility. counted for some 20 to 35 percent of the dis- 3° See also Roos (1981), Ferber and Lowry (1978). criminatory ray gap. Using 1950 census data, Cabral et al. (1981), Stevenson (1975), and Jusenius (1977). Sanborn (1964) was able to explain 43 percent 3 This analysis was part of the statistical evidence developed by Janice Madden and me in an employment discrimination case. Madden (1982) alsoprovides 214 See also Chiswick et al. (1974). a discussion and analysis of thesedata.

142 1

136 FRANCINE D. BLAU

TABLE 7-1Regiession Analysis .of the Salaries of Active Employees in 1978 (standard errors) Dependent Variable: Natural Log of Annual Salary Independent Variables (1) (2) (3) Personal characteristics Female - 0.5659° - 0.3236° -0.07854 (0.0183) (0.0151) (0.0083) Black - 0.3573° - 0.1985° -0.0948° (0.0243) (0.0185) (0.0092) Other minority - 0.2091° -0.0943° -0.0459° (0.0359) (0.0267) (0.0130) Education (highest grade completed) 0.09410' 0.0176° (0.0035) (0.0021) Age (in years) 0.0512° 0.0268° (0.0042) (0.0022) Age (squared) -0.00059° -0.00030° (0.00005) (0.00003) Firm experience (in years) 0.0258' 0.0088° (0.0029) (0.0014) Firm experience (squared) - 0.00027° -0.09018° (0.00010; (0.00005) Job characteristics Hay points - - 0.0007° - - (0.00001) Firm officer - - 0.3628° - - (0.0132) 1 Other exempt - - 0.2332° - - (0.0116) Constant term 9,7748 7.1928 8.2253 R square 0.4156 0.6873 0.9294 F statistic 476.0479° 550.1990° 2146.5542' Number of employees 2012 2012 1806 ° Significant at the 1 percent level on a two-tailed test.

gives the total effect of discrimination (oper- another managerial or professional occupation ating both through unequal pay for equal work (exempt from the Fair Labor Standards A,A). and Unequal access to higher-paying jobs). A relatively small proportion of the discrim- Discrimination is estimated to account for 57 inatory sex differential (controlling for race) - percent of the gross pay differential in the case 24 percent ( -0.0785/ 0.3236) - is due to of women (controlling for race) and 45 to 56 pay differences within similar occupational percent of the gross pay differential it 1e case catcgories. The remainder, 76 percent, is due of blacks and other minorities (controlling for to sex differences in distribution among oc- sex). Controlling for occupational character- cupational categories within the firm. Occu7 istics in equation (3) gives us an estimate of pational differences explain somewhat less than pay differences betweenequally qualified male half of the discriminatory pay gap in the case and female (black and white) workers in sim- of blacks and other minorities (controlling for ilar job categories. In this case the job char- sex). acteristics include hay points an employer There are some problems with these spec- evaluation of the value of the job to the firm - ifiaations. For one thing age rather than actual and two dummy variables indicating whether labor market experience is used as an explan- the individual is an officer of the firm or is in atory variable due to data availability, (The

1 4 3 OCCUPATIONAL SEGREGATION AND LABOR MARKET DISCRIMINATION 137 firm experience variable does, however, inantly male occupations differ in the way measure actual firm experience.) However, predicted by the human capital model. She the studies reviewed earlier strongly suggest finds that the earnings of women in pre- that the discriminatory differential would per- dominantly female occupations do not show sist even if we were able to control for actual lower rates of either depreciation or appre- labor market experience. For another, job ciation than do the earnings of women in grade (hay points) rather than occupational occupations employing more males.33 Fur- category is employed, making this similar to thar, she finds women who have discontin- the type of investigation one would undertake uous work histories are no more likely to be in a study of the issue of comparable worth. in predominantly female occupations than But there are some advantages to the use of are women who have been employed more hay points as an overall measure of job level continuously. Indeed, since she finds that that cuts across male and female jobs. It over- women earn less in female jobs at all levels comes one of the practical problems with ef- of experience, she concludes that "the evi- forts to ascertain the size of intraoccupational dence does not support the contention of sex pa; differentials: paucity of data on one human capital theorists that women maxi- sex group or another within a job category mize lifetime earnings by choosing female duethe very sex segregation 'by occupation occupations" (England, 1981, p. 18). that we seek to study.32 These results support If we provisionally conclude that the hu- the notion that studies of the impact of oc- man capital analysis of occupational choice cupation undertaken at the level of the firm discussed above does not explain oy6upa- and utilizing job categories more closely ap- tional segregation, at least at the aggregate proximating the job titles used by the em- level, what does? Many potential candidates ployer will reveal a greater impact of job cat- remain, ranging from premarket discrimi- egory on wages than aggregate analyses. nation (e.g., by families through the socialization ,process, or by schools through the actions of teachers, guidance counselors, or The Causes of Occupational Segregation admissions committees) to the exclusionary As discussed earlier, the human capital practices of employers (due to their own model provides an explanation for occupa- tastes, statistical discriminations, and/or the tional segregation by sex in terms of wom- tastes of employees or customers). Clearly, en's optimizing behavior, given the tradi- considerable additional work needs to be tional division of labor by sex within the done to narrow the field and/or to attach family. Polaehek (1'979, 1981) provides some relative weights to these competing expla- support for this view when jobs are cate- nations. gorized according to variants of the census major occupational groups. The problem with Pay Differentials Within Occupations his approach is that these major occupational categories combine predominantly The relatively flat earnings profiles of male and predominantly female jobs (Eng- women in female jobs are consistent with land, 1982). England (1981, 1982) explicitly examines whether women's earnings patterns in predominantly female and predom- See a..so Beller (1982). She finds that when one examines detailed (three-digit) census occupations, the evidence for the human capital model Is mixed in that the expected signs on the labor supply variables are 32 Employer job evaluation schemes may, however, not always obtained. King (.1977) finds little evidence understate the relative value of predominantly female of flatter age earnings profiles for women in female as occupations (Treiman and Hartmann, 1981). compared with male professions. 144 138 FRANCINE D. BLAU the notion advanced in the institutional view manly sought by and attracted to the higher- and implied by Arrow's notion of perceptual wage establishments, whilefemale workers equilibrium that employers would structure for the most part find employment in the female jobs to fit the average perceived lower-paying firms, which, regardless of their characteristics of women workers. How- preferences, are less able to compete for ever, the finding that women inmale jobs male labor. also have relatively flat earnings .profiles might Blau tests this model using unpublished at first appear inconsistent with the notion 1970 wage data from the Bureau of Labor that the interlial labor market mandates equal Statistics on extremely narrow, white-collar pay for equal work by sex.However, it should occupational categories (e.g., accounting be noted that the census categories are ag- clerk, class A) in three northeastern cities. gregated across job levels and firms. Thus, She argues that within such narrow .cate- firms may pay women and men in the same gories male and female labor is likely to be job category at roughly the same rate, but fairly homogeneous. 35 Blau finds that within promote women more slowly.34 Further, it .occupations men and women are segregated may be that women and men inthe same by establishments to an extent in excess of census job category aresegregated by firm. what would be expected on the basis of Blau (1977) examines the extent of employ- chance. Within firms, occupational pay dif- ment segregation by sex withinoccupational ferences are found to be relatively small, categories and its relationship to intraoc- and sex pay differentials within occupations cupational pay differentials within the con- are primarily due todifferences in pay rates text of an institutional model. Herfindings among (rather than within)firms.36 Further, suggest that pay differentials between men men tend to earnless when they work with and women in the same occupational, cate- women, which is counter to what wewould gory may reflect hiringdiscrimination by expect on the basis of the Beckermodel if firms. discriminatory tastes were located inem- Blau postulates that institutional and mar- ployees. ket forces determine a wage hierarchy of Blau finds evidence of a wage hierarchy firms within the local labor market that is of firms that is consistent across occupations consistent across occupational categories. She and sex groups. Controlling for occupational argues that, while employer tastesfor dis- mix, the representation of women in the crimination against women are fairly wide- firm is found to be consistent across occu- spread, the ability to exercise them is con- pations and inversely related to the wage strained by the firm's position in the wage standing of the firm, Note that these find- hierarchy. That position is determined by ings also conflict with the Becker model. In a variety of factors and cannoteasily be al- the case of employee preferences it is not tered to accommodate employer prefer- expected that men will earn more when they ences regarding the sex compositionof spe- work with relatively fewer women. In the cific occupational categories. Thus, in each case of employer preferences it is not ex- occupational category male workers are pri- pected that the firms that hire relatively the

34 Note that such a sex difference by job level would (1981), and Malkiel and Malkiel (1973). not support the human capital view in that itwould 33 See footnote 29. not be economically rational for women to opt totake 3° For other studies reporting differences in the dis- the lower-paid training positions, but not to reap the tribution of men and women by firm that are associated gains of moving up the job ladder. For findings sug- with pay differentials, see Buckley (1971), McNulty gesting that women have lower promotionprobabili- (1967), Bridges and Berk (1974), Talbert and Bose (1977), ties, see Duncan and Hoffman (1979), Cabral etal. Allison (1976), and Dussault and Rose-Liike (1980). 145 OCCUPATIONAL SEGREGATION A D LABO MA KET D SCRIMINATION 139

fewest women (presumably the most dis- cupational differences in producing male-fe- criminatory firms) will pay women the high- male pay differentials. One area of particular est wage rates.37 concern is the issue of the impact of crowding in female jobs on the wages of women in male jobs. Finally, the question of the indirect ef- CONCLUSIONS fects of discrimination on the qualifications of Various explanations have been offered for women (and minorities) is another area upon the pay and occupational differences between which future research could fi-uitfully be fo- male and female workers. Some emphasize cused. labor market discrimination, while others, most notably the human capital model, focus on ACKNOWLEDGMENTS the voluntary choices of women. A review of the empirical literature strongly suggests that, Portions of this chapter draw on my work all else equal (including fairly refined meas- entitled "Discrimination Against Women: ures of work experience and labor force at- Theory and Evidence," in William A. Dar- tachment), women do earn less than men. ity, Jr., ed., Labor Economics: Modern Views This suggests that labor market discrimination (Boston: Kluwer-Nijhoff, 1984). I am grate- does indeed play a role in producing the lower ful to the publisher for granting permission earnings of women. However, we lack a widely for the reproduction of some of that material accepted economic theory of the role of oc- here. cupational segregation in producing this dif- I am grateful for the comments and sug- ferential and of the persistence of sex discrim- tostions of Barbara Reskin, William Darity, ination in the labor market over time in the Jr., Katherine Abraham, Andrea Beller, face of competitive forces. Perhaps it is time Barbara Bergmann, Charles ,own, Paula now to devote less of our empirical efforts to England, Marianne Ferber, Claudia Gol- ascertaining the existence of discrimination and din, Joan Huber, Robert Hutchens, Law- more toward determining which model ofdis- rence Kahn, Michael Reich, Harvey Rosen, crimination is most consistent w h the data and two anonymous referees. I am indebted and the mechanisms by which these discrim- to Susan Schwochan for excellent research inatory outcomes are produced. On the basis assistance. of the existing evidence it appears that sex segregation in, employment is an important REFERENCES mechanism for producing sex differences in earnings and that the occupational differences Aigner, D. J., and G. C. Cain 1977 "Statistical Theories' of Discrimination in La- between men and women do not seem to be bor Markets," Industrial and Labor Relations consistent with optimizing behavior on the tgaiew, 30, 2 (Jan.) 175-187. part of women. However, considerably more Allison, E. K. work is needed to understand the causes of 1976 "Sex Linked Earnings Differentials in the Beauty Industry," Journal of Human Resources, 11. 3 O sex differences in occupationaldistributions (Summer) 383-390. fully and to determine the role of such oc- Arrow, K. 1972a"Models of Job Discrimination," in A. H. pascal, Ed., Racial Discrimination In Economic Life (Lexington. Mass: D. C Heath and Co.) ' Note that the comparisons made in the text be- 83-102. tween the results expected on the basis of the Becker 1972b'Some Mathematical Models of Race in the and institutional models rest fairly heavily on the as- Labor Market," in A. H. Pascal, Ed., Racial sumption that labor is fairly homogeneous within these Discrimination in Economic Life (Lexington, detailed occupations. Otherwise, variations in labor Mass: D. C. Heath and Co.) 187-204. quality might account for these interflrm differences in 1973 "The Theory of Discrimination.** in 0. Ashen- pay rates. fetter and A. Rees, Eds., Discrimination in La-

1 4 6 140 FRANCINE D . BL41.I

bor Markets (Princeton, New Jersey: Princeton, in the Labor Market: An Appraisal," Signs: University Press) 3-33. Journal of Women in Culture and Society, 1, 3, Part 2 (Spring), 181.199. 1976 "Economic DiMensions of Occupational Seg- regation: Comment I," Signs, 1, 3, Part 2 Blau, F. D., and L. M. Kahn (Spring), 233-237. 1981 "Race and Sex Differences in Quits by Young Workers," Industrial and Labor Relations Re- Baker, C. O. Undated Earned Degrees Conferred: An Examination view, 34, 4 (July) 563-577. of Recent Trends, National Center for Edu- Blinder, A. S. cation Statistics. 1973 "Wage Discrimination: Reduced Form and Structural Estimates," Journal of Human Re- Barnes, W. F., and E. Jones sources, 8, 4 (Fall) 436-455. 1974 "Differences in Male and Female Quitting," Journal of Human Resources, 9, 4 (Fall), 439- Bridges, W. P., and R. A. Berk 451. 1974 "Determinants of White-Collar Income: An Evaluation of Equal Pay for Equal Work," So- Becker, G. cial Science Research, 3, 3 (Sept.), 211-234. 1957 The Economics of Discrimination (Chicago: University of Chicago Press). Brown, C. 1981 "Black/White Earnings Ratios Since the Civil 1965 "A Theory of the Allocation of Time,: Eco- nomic Journal, 75, 229 (Sept.), 493-517, Rights Act of 1964: The Importance of Labor Market Drop-outs,"NBER Working Paper #617 1973 "A Theory of Marriage," Journal of Political Economy, 81, 4 (July/Aug.), 813-846. (Jan.). 1982 "The Federal Attack on Labor Market Dis- Beller, A. H. crimination: The Mouse That Roared?" Re- 1982 "Occupational Segregation by Sex: Determi- search in Labor Economics, 5, 33-68. nants and Changes," Journal of Human Re- sources, 17, 3 (Summer), 371492. Buckley, J. E. 1971 "Pay Differences Between' Men and Women Bergmann, B. R. in the Same Job," Monthly Labor Review, 94, 1974 "Occupational Segregation, Wages and Profits 11 (Nov.), 36-39. When Employers Discriminate by Race or Sex," Eastern Economic Journal, 1 (Apr.-July) 103- Butler, R. J. 1982 "A Note on Estimating Wage Discrimination 110. in the Labor Market,"Journal of Hunian Re- 1976 "Reducing the Pervasiveness of Discrimina- sources) 17, 4 (Fall), 606-621. tion," in E. Ginzberg, Ed., Jobs for Americans (Englewood Cliffs, N.J.: Prentice Hall) 120- Cabral, R., M. A. Ferber, and C. A. Green 1981 "Men and Women in Fiduciary Institutions: A 141. Study of Sex Differences in Career Develop- Bergmann, B. R., and W. Darity, Jr. ment," Review of Economics and Statistics, 63, 1981 "Social Relations in the Workplace and Em- 4 (Nov.) 573-580. ployer Discrimination," Proceedings of the Thirty-Third A »nue/ Meeting of the Industrial Cain, G. Relations Research Association,. University of 1976 "The Challenge of Segmented Labor Market Theories to Orthodox Theory: A Survey," Jour- Wisconsin, Madison, 155-162. nal of Economic Literature, 14, 4 (Dec.), 1215- Blau,F. D. 1257. 1975 "Longitudinal Patterns of Female Labor Force Participation," in H. S. Parries et al., Dual Cardwell, L. A., and M. R. Rosenzweig Careers: A Longitudinal Analysis of the Labor 1980 "Economic Mobility, Monopsonistic Discrim- Market Experience of Women, V. 4 (Colum- ination and Sex Differences in Wages," South- bus, Ohio: Center for Human Resource Re ern Economic Journal, 46, 4, (Apr.), 1102-1117. search, Ohio State University, Dec.), 27-55. Chiplin, B. Equal Pay in the Office (Lexington, Mass: 1981 "An Alternative Approach to the Measurement 1977 of Sex Discrimination: An Illustration Ftom D. C. reath and Co.). University Entrance," Economic Journal, 91, 1978 "The Impact of the Unemployment Rate on 394 (Dec.), 988-997. Labor Force Entries and Exits," in Women's Changing Roles at Home and on the Job. A -.Chiswick, B. R., J. A. O'Neill, J. Fackler, and S. W. special report of the National Commission for Polachek Manpower Policy, Report No. 26 (Sept.), 263- 1974 "The Effect of Occupation on Race and Sex Differences in Hourly Earnings," Business and 286. 'Economic Statistics Section Proceedings of the Blau, F. D., and W. E. Hendricks American Statistical Society, 219-228. 1979 "Occupational Segregation by Sex: Trends and Prospects," Journal of Human Resources, 14, Cog. n, J. "Married Women's Labor Supply: A Compar- 2 (Spring), 197-210. 1980 ison of Alternative Estimation Procedures," in Blau, F. D.. and C. L. Jusenius J. P. Smith, Ed., Female Labor Supply: The- 1976 "Economists' Approaches to Sex Segregation

147 OCCUPATIONAL SEGREGATION AND LABOR MARKET DISCRIMINATION 141

ory and Estimation (Princeton, N .J. : Princeton Freeman, R. B. University Press) 90-118. 1973 "Changes in the Labor Market for Black Amer- Corcoran, M. icans, 1948-1972," Brookings Papers On Eco- 1978 "The Structure of Female Wages," The Amer- nomic Activity, 67-131. ican Economic Review, papers and proceed- Fuchs, V. R. ings, 68, 2 (May), 165-170. 1971 "Difference', in Hourly Earnings Between Men and %*omen," Monthly Labor Review, 94, 5 1979 "Work Experience, Labor Force Withdrawals and Women's Earnings: Empirical Results Us- (May),419-15. ing-the 1976 Panel Study of Income Dynam- Gordon, D. M. ics," in C. B. Lloyd, E. Andrews, and C. L. 1972 Theories of Poverty and Underemployment Gilroy, Eds., Women in the X.abor Market (New (Lexington, Mass: D. C. Heath and Co.). York: Columb:a University Press). Gronau, R. Darity, W. A., Jr. 1973 "The Intrafamily Allocation of Time: The Value 1982 "The Human Capital Approach to Black-White of Housewives' Time," American Economic . Earnings Inequality: Some Unsettled Ques- Review, 63, 4 (Sept.), 634-651. tions," Journal of Human Resources, 17,1 1974 "Wage Comparisons A Selectivity Bias," (Winter), 72-93. Journal of Political Economy 82, 6 (Nov./Dec.), Darity, W. A., Jr., and S. Myers, 111.9-1143. 1980 "Changes in Black-White Income Inequality, Gwartney, J. D., and R. Stroup 1968-1978: A Decade of Progress," Review of 1973 "Measurement of Employment Discrimination Black Political Economy, (Summer), 354-379. According to Sex," Southern Economic Jour- Doeringer, P B., and M. J. Piore nal, 39, 4 (Apr.), 575-587. 1971 Internal Labor Markets and Manpower Anal- Hamilton, M. T. , ysis (Lexington, Mass: D. C. Heath and Co.). 1973 "Sex and Income inequality Among the Em- Duncan. G. J., and S. Hoffman ployed," Annals of the American Academy of 1979 "On-the-fob Training and Earnings Differ- Political and Social Sciences, 409 (Sept.), 42- ences by Race and Sex," Review of Economics 52. and Statistics, 61, 4 (Nov.), 594-603. Hartmann, H. I., and B. Reskin Dussault, G., and R. Rose -Lithe .1982 "Job Segregation: Trends and Prospects," pa- 1980 "La Discrimination 1 l'Egard des femmes et Is per presented at a Conference on Occupational Segregation sur les Marches du Travail: Le Cas Segregation, Ford Foundation, New York City, des Employes de Bureau h Montreal," in La - June. Discrimination d l'Egard des femmes sur le Hashimoto, M., and L. Kochin Marche du Travail, Collection Tires a Part, 1980 "A Bias in the statistical Estimation of the Ef- No. 25 (Quebec: Universals Laval) 475-504. fects of Discrimination," Economic inquiry, 18 Economic Report of the President (July), 478-486. 1974 (Washington, D.C.: U.S. Govt. Printing Of- Hecknian, J. J. fice, Feb.). 1974 "Shadow Prices, Market Wages and Labor Edgeworth, F. Y. Supply," Econometrlca, 42, 4 (July), 679-694. 1922 "Equal Pay to Men and Women for Equal Hill, M. S. Work," Economic Journal, 32 (Sept.) 431-457. 1979 "The Wage Effects of Marital Status and Chil- England, P. dren," Journal of Human Resources, 14, 4 (Fall), 1981 "Wage Appreciation and Depreciation: A Test 579-594. of Neoclassical Economic Explanations of Oc- Jusenius, C. L. cupational Sex Segregation," mimeo (Dec.). 1977 "The Influence of Work Experience, Skill Re- 1982 "The Failure of Human Capital Theory to Ex- quirement, and Occupational Segregation on plain Occupational Sex Segregation," Journal Women's Earnings," Journal of Economics and of Human Resources, 17, 3 (Summer), 358- Business, 29, 2 (Winter), 10'Z -115. 370. Kahn, L. M. Fawcett, M. G. 1980 "Union Spillover Effects on Unorganized La- 1918 ."Equal Pay for Equal Work," Economic Jour- bor Markets," Journal of Human Resources, nal, 28 (March), 1-6. 15, 1 (Winter), 87-98 Ferber, M. A., and H. M. Lowry 1981 "Sex Discrimination in Professional Employ- 1976 "The Sex Differential in Earnings: A Reap- ment: A Case Study Comment," Industrial praisal," Industrial and Labor Relations Re- and Labor Relations Review, 34, 2 (Jan.), 273- view, 29, 3 (Apr.), 377-387. 275. Flanagan, R. J. Kamalich, R. F.,, and S. W. Polachek 1976 ''Wage Interdependence in Unionized Labor 1982 "Discrimination: Fact or Fiction? An Exami- Markets," Brookings Papers on Economic Ac- nation Using an Alternative Approach," South- tivity, No. 3. ern Economic Journal, 49, 2 (Cct.).

148 142 FRANCINE D. BLAU

King, A. C. 1979 "Wage Differentials: Comment," in C. B. Lloyd, 1977 "Is Occupational Segregation the Cause of the E. S. Andrews, and C. L. Gilroy, Eds., Women Flatter-Experience-Earnings Profiles of in the Labor Market (New York: Columbia Women?" Journal of Human Rfsourcea, 12, 4 University Press) 278-285. (Fall), 541.549. Mincer, J., and H. Ofek Kohen, A. I., with S. C. Breinich, P.,Shields 1982 "Interrupted Work Careers: Depreciation and 1975 "Women and the Economy: A Bibliography Restoration of Human Capital," Journal of Hu- and a Review of the Literature on Sex Differ- man Resources, 17, 1 (Winter), 3-24. entiation in the Labor Market," Center for Hu- Mincer, J., and S. Polachek man Resource Research, College of Adminis- 1974 "Family Investments in Human Capital: Earn- trative Science, Ohio State University. ...- ings of Women, " Journal of Political Economy, Landes, E. M. 82; 2 (Part 2) (March/Apr.), S76-S108. "Sex Differences in Wages and Employment: 1977 1978 "An Exchange: Theory of Human Capital and A Test of the Specific Capital Hypothesis," the Earnings of Women: Worsien's Earnings Economic Inquiry, 15, 4 (Oct.), 523-538. Reexamined," Journal of Human Resources, Lee, L. 13, 1 (Winter), 118-134. 1978 "Unionism and Wage Rates: A Simultaneous Oaxaca, R. Equations Model With Qualitative and Lim- 1973a"Male-Female Wage Differentials in Urban ited Dependent Variables," International Eco- Labor Markets," International Economic Re- nomic Review, 19 (June), 415-433. vieW, 14, 3 (Oct.), 693-709. Lewin P., and P. England 1973b"Sex Discrimination in Wages," in 0. Ashen- 1982 "ReconceptualIzing Statistical Discrimination," felter and A.''Reei, Eds., Discrimination in paper presented at the annual meetingsof the Labor Markets (Princeton, N. J.: Princeton Southwest Social Science Association, San An- University Press) 124-151. tonio, Tex., March. Osterman, P. Lloyd, C. B., and B. T Niemi 1979 "Sex Discrimination in Professional Employ- 1979 The Economics of Sex Differentials, (New York: ment: A Case Study," Industrial and Labor Columbia University Press). Relations Review, 32, 4 (July), 451-464. Madden, J. F. Phelps, E. S. The Economics of Sex Discrimination (Lexing- 1973 1972 "The Statistical Theory of Racism and Sexism," ton, Mass: D. C. Heath and Co.). 1 American Economic Review, 62, 4 (Sept), 659- 1976 "Economic Dimensions of Occupational Seg- 681. regation: r :mment III," Signs, 1, 3, Part 2 Fiore, M. J. (Spring), 245-250. 1971 "The Dual Labor Market: Theory and Impli- 1982 "The Measurement of Employment Discrim- cations," in D. M. Cordon, Ed., Problems in ination: Reduced Forms, Reverse Regression, Political Economy: An Urban Perspective (Lex- Comparable Worth and the Definition of La- ington,. Mass: D. C. Heath and Co.) 90-94. bor Markets," Proceedings of the Social Sta- Polachek, S. W. tistics Section at the Annual Meeting of the 1975 "Potential Biases in Measuring Male-Female American Statistical Association, Cincinnati, Discrimination,"Journal of Human Resources, Ohio, Aug. 10, 2 (Spring), 205-229. Malkiel, B. G., and J. A. Malkiel 1976 "Occupational Segregation: An Alternative 1973 "Male-Female Pay Differentials in Professional Hypothesis," Journal of Contemporary 'Busi- Employment," American Economic Review, 63, ne as , 5, 1 (Winter), 1-12. 4 (Sept.), 693-705. 1979 "Occupational Segregation Among Women: McNulty, D. J. Theory, Evidence, and A Prognosis," in C. B. 1967 "Differences in Pay Between Men and Women Lloyd, E. S. and C. L. Cihoy, Eds., Women Workers," Monthly Labor Review, 90 (Dec.), in the 7..abor Market (New York: Columbia 40-43. University Press) 137-157. Medoff, J. L., and K. G. Abraham 1981 "Occupational Self-Selection: A Human Cap- 1980 "Experience, Performance, and Earnings," ital Approach to Sex Differences in Occupa- Quarterly Journal of Economics, 95, (Dec.), tional Structure," Review of Economics and 703-736. Statistics, 63,11 (Fab.), 60-69. 1981 "Are Those Paid More Really More Produc- Ragan, J. F., Jr.. and S. P. Smith tive? The Case of Experience:Journal of HL- 1981 "The Impact of Differences in Turnover Rates man:sources, 16, 2 (Spring) 186-216. on Male/Female Pay Differentials," Journal of Mincer, J. Human Resources, 16, 3 (Summer), 343-365. 1962 "Labor Force Participation of Married Women," Roberts,,II. V. in N B ER, Aspects of Labor Economics (Prince- 1980 Statistical Biases in the Measurement of Em- ton: Princeton University Press) 63-97. ployment Discrimination," in E. R. Livernash,

149 .2

OCCUPATIONAL SEGREGATION AND LABOR MARKET DISCRIMINATION 143

Ed., Comparable Worth: Issues and Aherne- Talbert, J., and C. E. Bose

. tives (Washington, D.C.: Equal Employment 1977 "Wage Attainment Processes: The Retail Clerk Advisory Council) 173-196. Case," American Journal of Sociology, 83, 2 2 Robinson, J. (Sept.), 403-424. 1933 The Economics of Imperfect Competition (Lon- Thurow, L. C. don: Macmillan and Co., Ltd.). 1975.Generating Inequality (New York: Basic Books, Roos, P. A. Inc.). 1981 "Sexual Stratification in the Workplace: Male- Treiman, D. J., and H. I. Hartmann, Eds. Female Differences in Economic Returns to 1981 Women, Work, and Wages: Equal Pay for Lobs Occupation," Social Science Research, 10, 3 of Equal Value (Washington, D.C.: National

(Sept.), 195-224. Academy Press); . Sanborn, H. U.S. Department of Codnerce, tureau of the Census 1964 "Pay Differences Between Men and Women," 1980 A Statiktiad Portrait of Women in the 'United Industrial and Labor Relations Review, 17, 4 States: 1978, Current Population Reports, (July), 534-550. Special Studies, Series P-23, No. 100 (Feb.). Sandell, S. H., and D. Shapiro 1982 MoneY Income and Poverty Status of Families 1978 "An Exchange: Theory of Human Capital and and Persons in the United Ststes: 1981 (Ad- the Earnings of Women: A Reexamination of vance Report), Curientfopulation Reports Se- the (Evidence," Journal of Human Resources, ries P-60, No. 134 (Washington, D.C.: U.S. 13, 1 (Winter), 103-117. Government Printing Office, July). 1980 "Work Expectations, Human Capital Accu- U.S. Department of Labor, Bureau of.Labor Statistics mulation, and the Wages of Young Women," 1980 "Perspectives on Working Women," June. Journal of Human Resources, 15, 3 (Summer), 1981 News Release 81-522 (Nov. 15), 335-353. 1982 Employment Earnings (Apr.). Sawhill, I. V. U.S. Department of Labor, Employment and Training 1973 "The Economics of Discrimination Against Administration Women: Some New Findings,"Journal of Hu- 1981 Employment and Training Report of the Pres- man Resources, 8, 3 (Summer), 383-396. ident. Spence, A. M. Viscusi, W. K. 197 "Job Market Signaling," Quarterly Journal of 1980 "Sex Differences in Worker Quitting," Review Economics, 87, 3 (Aug.), 355-374. of Economics and Statistics, 62, 3 (Aug.) 382S- 1974 Market Signaling (Cambridge, Mass: Harvard 398. University Press). Wachter, M. Stevenson, M. H. 1974 "Primary and Secondary Labor Markets: A Cri- 1975 "Relative Wages and Sex Segregation by Oc- tique of the Dual Approach," Brookings Papers cupation," C. LloycL.Ed. Sex, Discrimination, on Economic Activity, no. 3, 637-694. and the Division of Labor (New York: Colum- Weiss, Y., 'and R. Gronau bia University Press). - 1981 "Expected Interruptions in Labour Force Ear- Strober, M. H. ticipation and Sex-Related Differences . in 1976 "Toward Dimorphics: A Summary Statement Earnings Growth," Review of Economic Stud- to the Conference on Occupational Segrega- ies, 58 (Oct.), 607-619. tion," Signs, 1, 3, Part 2 (Spring), 293-302. Zellner, H. Suter, L., and H. Miller 1975 "The Determinants of Occupational Segrega- 1973 "Income Differences Between Men and Ca- tion," in C. B. Lloyd, ed., Sex, Discrimination reer Women," American Journal of Sociology, and the Division of Labor (New York: Colum- 78, 4 (Jan.), 962-974. bia University Press) 10-145.

150 Toward.aGeneral Theory of Occupational Six Segregation: The Case of Public SchoolTeaching

.MYRA H. STROBER 11,.../

Occupational sex segregation has several ing and librarianship. Also, men spend less interrelated dimensions. First, there is seg- time on housework and child care thando regation between paid occupationsarid those women, and men engage infewer different that are unpaidthat is, the percentages of household tasks (Walker and Woods, 1976; women and men inpaid employment are Robinson, 1977; Stafford and Duncan, 1977). . unequal. In 1980 the civilian labor fOrce Although this paper sometimes touches ticipatiOn rate for women over age 20 was on issues of women's participationin the 51 percent; foi men it was 7,9 percent paid labor force and occupational stratifi- (Monthly Labor Review, 1981,A4:60). Sec- cation, its focus is on the second type of ond, there is segregation across occupations occupational segregation: segregation across within paid employment:labor market seg- occupations within paid employment. A the- regation. The inOex ofdissimilarityindicates ory of occupational segregationby gender that in 1977 about 64 percent of American within the labor.market must deal with three men (or women) wouldhave had to change central questions: (1) How does an occu- their occupations in order to achieve equal- pation become primarily male or female? ity in the gender distribution across occu- (2) Once an occupation is gender typed, what pations (see Lloyd and Niemi, 1979;Gross, forces help keep it that way? (3) How do 1968; Blau and Hendricks, 1979). In few occupations change their gender designa- occupations are women represented in action? cordance with their representation in the labor force as a whole. Third, within any TOWARD A GENERAL THEORY OF single occupation," women and men are not OCCUPATIONAL SEX SEGREGATION distributed equally across the occupational hierarchythat is, there is occupational Extant sociological and economiciheol :es stratification. Women are clustered at the of occupational segregation by gender in the lower levelsi men at the upper levels. And labor market stem from remarkably diver- this is often true even in occupations that gent world views and locate the causesof are overwhelminglyfemale, such as teach- segregation in a variety of different actors 6 144 151 TOWARD A GENERAL THEORY OF OCCUPATIONAL SEX SEGREGATION 145 with diverse motivations.' Status attainment tions, and wor en's Wagei in those occu- theory in sociology and human capital the- pations are thereby depressed. ory in economics pinpoint women's own be- Although the world view of the dud labor havior as the primary cause of their segre- market or internal labor market theories is gation into occupations with low status and much less oriented toward individual choice low pay. Women's own values, behaviors, and market processes than is neoclassical aspirations, attitudes, sex-role expectatidns economics, these theories also locate the (status attainment theory), educational cre- source of occupational segregation in em- dentials, and interrupted work histories (hu- ployer behavior. Employers create seg- man capital theory) are seen as the causes ments in the labor market, either to take of their occupational designations and low advantage of profit opportunities (the view pay rates. of the non-Marxist dual labor market theo- The view that women's own behavior is rists) or to prevent the development of worker central is clearly articulated by Matthaei solidarity (the view of the Marxists among

(1982:194), who argues that job segregation the dual labor market theorists). . exists because ". . . womenwished to work Feminists have viewed all of these .theo- in jobs done by women." Employment in ries as inadequate, largely because the the.: women's work preserved women's sense of odes have paid insufficient attention to the their femininity. Kessler-Harris's book (1982) centrality of gender relations in the society on the history of wage-earning women in at large. I have argued that, although the the United States, while less exclusively profit motive may explain employers' de- supply-side oriented, also stresses the role sires to augment tha division of labor, it does of women's choices in producing occupa- not explain why that division turns into one tional segregations. blisedon gender (Strober and Best, 1979).2 Economic theories of discrimination and Hartmann (1976:138) has argued that to ex- statistical discrimination, on the other hand, plain job segregation by gender one must locate the source of inequality of earnings examine patriarchy as well as capitalism. and occupational distribution in employers Hartmann defines patriarchy as "a set of so- and their "taste" for discrimination (discrim- cial relations which has a material base and ination theory) or their wish to minimize the in which there are hierarchical relations be- risk associated with employing women (sta- tween men, and solidarity among them, tistical discrimination theory). Discrimina- which enable them to control women. Pa- tion theory, however, recognizes that the triarchy is thus the system of male oppres- tastes of workers and customers may be im- sion of women." In Hartmann's view, male portant factors contributing to the formation capitalists and male workers oppress women. of employers' tastes. The "overcrowding" And Hartmann as well as Milkman (1980) explanation for occupational segregation point to the role of male worker organiza- builds on the theory of discrimination and tions (i.e., trade unions) in ir'..ting and' points out that, as a result of employers' operationalizing their tastes for exclusion, women are crowded into certain occupa- 2 Blau and Jusenius (1976) noted that "sex is an obvious basis for differentiation, due to employers' distaste for hiring women in male occupations and/or real or perceived quality differences between male and fe- Sociological theories and those of Marxist econo- male labor" (pp. 192-193). But the unanswered ques- mists and feminists are reviewed in Sokoloff (1980). tion remains: Why do employers have a distaste for mmoinic theories are reviewed in Blau and jusenius women, and why do they perceive them as being less 0976). Cain (1976). and Amsden (1980). qualified, or why are women "less qualified"?

152 146 MYRA H. STROBER maintaining occupational segregation by der designation of an occupation: its origin, gender.3 its maintenance, and its change, if any. Yet In addition to these more formal theories each of the theories and many of the expla- of occupational segregation by gender, there nations contain threads'of truth. What I have are numerous explanations that seek to ex- done is selected the strongest threads from plain the gender designation of a job in terms each and woven them into a new theory. of the job's characteristics. Oppenheimer Orthodox adherents to various schools of (1970) has hypothesized that women fill those thought may be uncomfortable in finding jobs that require relatively high levels of aspects of their theories woven into a new prejob, as opposed to on-the-job, training, fabric, but I have purposely borrowed such as nursing and teaching. There is also sights when their observations contributed a rich folklore maintaining that women's jobs to the overall explanation, without too much are those requiring dexterity or those that concern about the insights' ideological par- women traditionally performed in the home. entage. I call the theory "general" because None of these explanations provides a strong it may be used to explain the origins, main- basis for a theory of gender segregation be- tenance, and changes in the gender assign- cause for each hypothesis it is so easy to ment of jobs in general, i.e., for all occu- point to counterexamples. Why are certain pations. occupations that require a great deal of pre- My theory has two central tenets. It injob training, such as law and medicine, "re- corporates the concept of patriarchy, al- served," in the United States, for men? If though I define patriarchy to make it ap- women are so dexterous, why are there so plicable in ,a non-Marxist framework. My few female brain surgeons? If jobs that used theory maintains that decisions concerning to be, or are, performed in the home be- the gender assignment of jobs are made by come women's jobs in the market, why are men. In particular, I argue that, within the most chefs, bakers, and food servers men? constraints laid down by race and class, it is Moreover, the difficulty with ,putting forth male workers who decide which occupations a theory of gender segregation based on the they will inhabit. Male employers set wages inherent characteristics of a job is that the and working conditions but, except when analyst then finds it impossible to explain the job explicitly or implicitly requires fe- shifts in gender assignments or differences male claracteristics, male employers allow in the assignment of jobs in different coun- male workers to decide which jobs will be tries. If the major reasons for a particular theirs. The remaining jobs are offered to job assignment are the job's inherent charwomen; if sufficient numbers of women do acteristics, how can the gender assignment not wish to work in the job, the employer change while the inherent characteristics re- recruits immigrants. Sometimes new jobs main the same? Or, how can an occupation appear to be "designed" for women, that is, be assigned to one gender in one industrial- it appears that men are not given "first dibs" ized countryltut to the other gender else- on these jobs. In such cases, employers de- where? sign the jobs for women and do not offer Neither the formal theories nor the ad hoc them to men first because they believe that explanations offered thus far can answer the most men would deem the jobs undesirable three major questions concerning the gen- relative to existing ones. The second key aspect of my theory is that, in deciding which new jobs to claim for themselves and which jobs to leave for Rubery (1978). while not mainly interested in sex segregation, has also noted the importance of unions women, male workers (again within the im- in perpetuating dual labor markets. portant constraints laid down by race and

153 TOWARD A GENERAL THEORY OF OCCUPATIONAL SEXSEGREGATION 147 class) attempt to maximize their economic strategy that guards against the develop- gain. They compare the wages, hours, and ment of worker solidarity. working conditions of the new job with those I have argued elsewhere (Strober, of existing jobs.4 If the new job is superior, 1976:295) that ideology, i.e., "social, legal, they claim it and move in; if not, the job cultural and economic conventions," includ- becomes women's work. Thus, whether a, ing "subtle pressures from family, employ- particular job in a particular locale is initially ees, customers and 'the community'," have male or female is a function of when the job enforced certain hiring taboos preventing came onto the market. employers from attempting to maximize Suppose a technological or organizational profits. We need to move further and briefly change occurs and a new job develops. New define the concept of patriarchy, for it is plant and equipment are put into place and patriarchy that is the source of these con- workers are to be hired. In the short run,, ventions and pressures. Based on Hart- in accordance with neoclassical assumpd mann's work, I define patriarchy as a set of tions, technology and capital are fixed. The personal, social, and economic relationships employer estimates the product or service that enable met. to have power over women demand and determines the need for labor. and the services they provide. This is a pre- Based on the existing wage structure of the liminary definition, and I am aware that it firm or industry and the wagee for similar needs specificity and refinement.° jobs in the local labor market, het assigns a One reason for discussing the concept of wage rate to the job and proceeds to ad, patriarchy here is to demonstrate that male vertise for workers .5 If "qualified" men show employers are not simply profit maximizers. up, they are hired. Qualifications, of course, They are simultaneously pursuing profit may well be based on racial and class char- maximization and the maintenance of male acteristics as well as on objective criteria. privilege; that is, there is a tension between Let's take up the case(s) where an insuf- patriarchy and profit maximization. This ficient number of men apply for the job. But tension is often latent; indeed, employers first let's ask why male capitalists and/or may not even be conscious of it. But the fact managers give male workers firstdibs on is that employers permit male workers to jobs. Why should employers use male work- choose their jobs because employers want ers for a particular job when they could hire to maintain patriarchy. They recognize that female workers at a lower wage rate? Neo- if male privilege is threatened in the work- classical theorists such as Arrow (1973) have ing class or among professionals or lower- responded to this question by citing various level managers (by allowing women to have market rigidities that prevent profit max- jobs that men want), upper-level managers, imization. Marxist dual labor market theo- entrepreneurs, and capitalists would. soon rists such as Reich et al. (1973) have implicitly argued that capitalists are willing to sacrifice short-term profits for a long-run e It may be, as Whitehead (1979) has suggested, that patriarchy is not the best term to use to describe women's subordinationin gender relations. Witit4ead argues that the term patriarchy connotes the power of illau (1977) points out that within an occupation, a husband over his wive(s), children, and property and ten tend to work in high-wage firms and women in is only one specific form of male dominance. She has

. v-wage firms. She argues that the" high-wage em- also suggested that the term implies an unchanging, ployers are thus able to hire preferred workersmen. historically constant form of subordination. I use the S In thispaper I refer to an employer as "he" because term heee mainly because it has been used in earlier believe that this designation reflects reality and, for work and do not mean for it to refer only to the rela- purposes of explicating this theory, it is important to tional aspects of gender, nor do I assume it to be his- consider the gender of employers. torically constant. 154 148 MYRA H. STROBER find their, own male privilege's under siege. interfere with male bondinga However, some Male employers believe that they benefit women might be hired, with theexplicit or from keeping women subordinate to and de- implicit understanding of both employer and pendent on men in all classes of society, so employees that although hired they are none- that women need to be married to men and theless not full members of the "group." A so that, because oftheir dependence, women sense of marginality might beconveyed through will continue to provide children as well as lower pay, ineligibility for promotion, or in- domestic services for men. Male employers eligibility for, membership in the relevant (as well as male workers) recognize that pa- union.9 triarchal relations must be maintained at the Suppose, however, that few men apply workplace if they are to remain unthrea- for the new jobs at the existing wage rate. tened in the household. Thus, male man- If the employer believes that an adequate agers are willing to sacrifice somepotential supply of native male labor can be attracted profits by allowing men to choose the jobs from other industries or other parts of the they wish. Male managers are willing to trade country, he may raise the wage rate slightly off some profit opportunities to maintain the to attract men. The employer will be more system of patriarchy. likely to do this if he believes that women Let us move to the question 'ofhow jobs may not be interested inholding the job, are allocated bygender when an insufficient e.g., because holding such ajob might vi- number of men apply for positions in a new olate existing norms and/or if he believes occupation. The employer has several op- that native men would perform the job sig- tions: (1) raise the wage rate and try to at- nificantly better than either women or for- tract (more) men; (2) hire womenwho apply eign workers. On the other hand, if financial and/or encourage (more) women to apply, at constraints prevent him from raising wages, the existing wage rate; (3) recruit male or and he thinks women can perform thejob female immigrants and hire them at the ex- adequately, he might try to hire women. If isting wage rate.? having women perform the job in question The option exercised will depend first on violates existing societal norms about which how many men have already been hired. If, jobs are acceptable for women of a particular for example, a "significant" number ofthe race or class, so that few womenapply, the positions have been filled by men,employ- employer might engage in a campaign to ers may prefer thefirst option: raising wages alter those norms. If women's employment and attracting more men, for employers are in the new occupation violates normsonly Anlikely to hire women to fill the remaining moderately, the employer's efforts to change jobs in an occupation that men have claimed, the norms would probably be successful apd even if men haven'tfilled all the job vacancies. To do so would erode theprinciples of patriarchy and finale privilege. Itmight alsO 8 The issue of taboos against the sexes working to.. violate societal taboos about the two sexes gether needs further investigation. For a discussion of working together on certain jobs and/or might male bonding, see Bradford et al. (1975). oMeyers (1980) has detailed the exclusion of women from the teachers' union in France during the 1890s, Of course, to options 2 and 3 could be added options when women were a minority of teachers. However, specifying that the wage rate be lowered when re- once women became the majorityof teachers, the men cruiting women or immigrants. Orie could arguethat remaining in the profession saw the need for gender during certain periods of our history the decision to cooperation and permitted women to Join the union. recruit blacks was made on grounds similar to thede- Other examples of women's exclusion from unions are cision to recruit immigrants, given in Hartmann (1976) and Milkman (1980).

155 TOWARD A ENERAL THEORY OF OCCUPATIQNAl SEXSEGREGATION 149 women would hired. Norms clearly can we will find a wage differential, with men be changed, but,xcept during wartime or earning higher wages than women. After all, times of social revolution, they change fairly one of the reasons why the mendeclined' slowly. Thus, if according to existing norms the job in questioh was probably its low rel- a job is consideredflagrantlyjunsuitable for ative wage. Thus, within race and class cat- women it is likely that in theshort run, egories, the fact that men have the first choice ordinary campaigns to alter these norms of occupations leads not only to occupational would not bring forth an adequate supply of sex segregation but also to agender wage women workers. In such a situationthe em- differential. Patriarchy combined with men's ployer would begin to 'recruit workers from desire to maximize their economic gain leads abroad. '° to higher relative wages for men. This discussion can be summarized by As time passes and an occupation be- noting the conditions under which foreign comes solidly female, employers maylower men, or perhaps men from a nativeracial the wage rate relative to others or fail to minority, would be recruited to fill a new increase it as fast as others, thereby further occupation: (1) enough native majority men increasing the gender wage differential. As did not apply to fill an occupation; (2) financial noted earlier, when a wage rate is 'first set constraints prevented the employer from for a new occupationwhen employers think raising the wage and/or the employer did that men will enter the occupationthe wage not believe that native majority menwould rate is 'set in accordance with the firm's or perform the job any better then women or industry's internal wage structure and in ac- foreign or minority men; and (3) employment cordance with wages for similar jobs in the of women in the occupation flagrantly vio- local labor market. However, we may hy- lated existing norms. (A discussion of con- pothesize that, once an occupation becomes ditions under which foreign or minority a female occupation,' employers willoften women might be recruited goesbeyond the lower its wage rate. First, the original wage scope of this initial sketch of thetheory.) rate, which was set in comparison to existing Next, comes the question of the devel- male wages, will be too high in comparison opment of the wage differential between to the wages for other female jobs in the women's and men's jobs. I have argued thus firm and in the local labor market." Second, far that, if men do not apply for a new job employers will see no reason to pay women at an existing wage rate and if employersdo male rates: women do not "need" the money not decide to raise the wage rate in aneffort as much as men, only men require a"family to attract men from other areas in the nation wage''women need only enough income or from other jobs, employerswill offer the to support themselvesand women, be- job to women at the same rate at which it cause they are often geographically immo- was offered to men. Supposethen that a bile and/or excluded from other higher-pay- sufficient number of women apply and that ing occupations have a less wage-elastic all of the new job slots are filled at the posted supply curve than men and therefore can wage rate, At this point, if we comparethe be retained at a lower wage. wage rate earned by women inthis job to In the theoryhave proposed, women's the wage rate earned by the men who considered taking the job but decided not to,

" For a discussion of how firms' internal wage structures differ by gender and how job evaluation tech- 1° Alternatively, the employer might turn to a native niques cement these differences, see Treiman and racial minority. Hartmann (1981).

156 150 MYRA H. STROBER job choices play a rather insignificant role. and are.rejecting male occupations. When By and large, women move into only those these nv\gative 'sanctions disappear, e.g., jobs that men leave for them. Turning to during wktimewomen more readily, and the issues of stability and change in the gen- often enthikiastically, enter the higher-paid der assignment of occupations, we find that male occupations. women's opportunities for choice are again If an occupation is occupied by women, overshadowed by men's actions. I do not the barrier to integration is largely male be- want to argue that women make nochoices havior. Women put up no resistance to men at all. They do, and it is interesting to in- entering "their" occupations, for they know vestigate their constrained choices and the that if more men enter both the prestige circumstances under which some women are and earnings of the occupation are likely to able to contribute to a modest breakdown rise. But men are reluctant to enter female of occupational segregation. However, like occupations, primarily because of their low the origin of occupational segregation, the wages but also because they fearridicule by stability and change in that segregation is other men and aspersions on their mascu- determined overwhelmingly by men's choices linity if they do. and men's behavior. In recent times, of If both genders initially take part in an course, equal employmentopportunity leg- occupation, call it X, eventually one of the islation and affirmative action orders have two will come to dominate. Which gender served to increase women's choices and de- achieves primacy in X depends on the at- crease the scope ofmen's exclusionary be- tractiveness of alternative occupations for havior. Th ?following discussion, however, men. If, as time goes on, X isdeskilled (i.e., examines the dynamics of occupational seg- requires lower skills) and wages fall, or if regation in the absence of legislationand new occupations are created that menfind executive and court orders. more attractive economically, m snwill move Once a job has been inhabited by one out of X. On the other hand, if newand gender or another, it becomes "typed" as existing occupations come to be seen as less male or female, and strong forces act to attractive economically than X, men will maintain its gender assignment. If men oc- move into X. It is alsopossible that women cupy an occupation, theymight actively and will move to other occupations, in which collectively seek to keep women out, fearing case X may become an occupationfor which that if women enter they will lower the earn- foreign labor is recruited. One observes that, ings of the job by accepting lower wage rates in the musical chairs of occupationalshifts, or, if they are paid the same as men,di- there is a clear hierarchy of players: men minish patriarchal hegemony. For although get first choice of job opportunities. One is men act as individual maximizersin choos- also impressed with the interdependence of ing their occupations, once theybegin to occupational gender assignments. Whether work in an occupation and identify with it, a particular occupation remainsgender typed they act collectively to maintain its gender or changes its gender assignmentdepends designation. It is true, of course, that women not only on its wages and workingconditions will rarely choose to enter a male-typed oc- but also on those in alternative occupations. cupation, fearinga diminution of their per- It seems to be that as occupations shift ceived femininity and thus a reduction of from one gender designation to another there their prospects for marriage, which until re- are important "tipping" points.Once an oc- cently was their primary avenue to eco- cupation becomes significantly m'Ille(or fe- nomic gain. Women behave in this way pri- male), it quickly tips and fairly soon there- marily because they fear negative sanctions after becomes overwhelmingly male (or from men not because they have free choice female). just what percentage constitutes this

.157 TOWARD A GENERAL THEORY OF OCCUPATIONAL SEX SEGREGATION 151

tipping point probably varies by occupation location. For example,' in 1870, when women and historical period. constituted 60 percent of all teachers na- The reasons for the existence of a tipping tionwide, women made up less than half the point are evident from our discussion of the teacher population in 26 states. In Wash- forces that maintajn the stability of an oc- ington, D. C., womenmade up 92 percent cupation's gender designation. Once it is clear of the teachers in 1870, but in neighboring that an occupation is significantly male, men Virginia women filled only 35 percent of the actively prevent women from entering and teaching jobs. women become reluctant toapply. By the Why did women come to constitute an same token, Once it is clear that an occu- increasingly larger percentage of teachers in pation is significantly female,it will be the period from the mid-nineteenth century shunned by men. In other words, the ex- to the end of World War I? In the context pectation that occupations will not be mixed of my theory outlined earlier, the question but will be either.male or female helps bring should be rephrased to read: why did men about the fulfillment of that expectation. choose to leave the teaching profession during that period? We can begin by noting that the latter half of the nineteenth century HOW TEACHING BECAME A FEMALE witnessed a substantial increase in the de- OCCUPATION mand for teachers as a result of population To what extent can my theory be applied growth, increased commitment to universal to the case of public school teaching in the education, and a desired decrease in the late nineteenth and early twentieth centu- typical number of students in each class. For ries? StAtistics for the nation as a whole on teaching to have remained a male profes- the percentage of women teachers are un- sion, the percentage of all male workers en- available for the years prior to 1870. In Mas- gaged in teaching would have had. to be in- sachusetts, which was at the vanguard of the creased. movement of women into teaching, women Nonetheless, although demand for teach- made up 56 percent of the public school ers was increasing, school boards were not personnel in 1834 and 78 percent in 1860 willing to raise wages to attract male teach- (Vinovskis and Bernard, 1978). In Ohio, a ers. In fact, teachers, especially in rural areas, more typical state with respect to women in were paid very low wages, often on a par teaching, women made up 39 percent of the with those of common laborers. This may teachers in 1840 and 46 percent in 1850 have resulted from a disinclination to set (Woody, 1929). For the United States as,a high tax rates and/or an ideological deval- whole, about 60 percent of all teachers were uation of education and teachers' roles. On women in 1870. By 1900 thefigure was up the one hand, the educational requirements to 70 percent, and by 1920 it had reached (literacy and a working knowledge of the a peak of 86.percent (U.S.Office of Edu- three Rs) demanded of teachers at that time cation's Biennial Survey of Education for were modest by modern standards. Onthe 1870, 1900, 1920).1' other hand, school boards required a native- These figures conceal considerable vari- born middle-class appearance and behavior ability by geographic region and rural/urban and good moral character. Nevertheless, it was virtually impossible to support afamily with middle-class standards on a, teacher's salary. Most men who taught school did So 12 The two major sources of statistics on teachers by as a stepping stone to another career or on gender are the annual and biennial reports of the U.S. a part-time basis while pursuing other work. Commissioner of Education and the decennial census reports. Most women who taught school were young

158 152 MYRA S ROBER

and single with few financial responsibilities raised and the schoolterm lengthened, the for others. Men often taught daring thewin- average teaching salary remained inade- ter term, when the older boyswere in at- quate to support a family. Men also disliked tendance, while women were more likely to losing their former classroomautonomy. And teach during thesummer term. at the same time, and perhaps mostimpor- Miring the latter half of the nineteenth tantly, attractive job opportunitieswere de- century, although starting somewhat earlier veloping for men in business andin other in Massachusetts, teaching began to undergo professions. a revolution in the organization of schools As men left teaching, school boardsturned and schooling. The revolution beganin ur- more and more toward women. Fora va- ban areas and spread to the countryside. riety of reasons, womenwere ready to move Schools became more formalized in three into teaching. First, manyyoung women important ways. First, as schoolsgrew in possessed the required education.By the size, classes became graded, i.e., children middle of the nineteenthcentury, women were taught in groups divided by age. Sec- and men had virtually thesame literacy rates, ond, once schools were large and graded, and girls were almostas likely to be attend- they were bureaucratized. A curriculumwas ing school as were boys.'3 Second,young developed for each grade, anda large num- girls were moving increasinglyinto the paid ber of school management functions were work force. As the production ofmany goods required. Third, as a result ofa knowledge and services movedout of the home and explosion, the growth of the middle class, began to be supplied through the market, and the increasing complexity of work, the the domestic services ofyoung women were high school evolved. Moreover,states beless frequently needed by theirparents. At gan to regulate education, lengthening school first, young women did pieceworkin their terms and formally credentialling teach- own homes. They also worked as domestics ers. Teachers were often required to at- in other people's homes. Finally, whenNew tend summer institutes to maintain theircre- England mill owners soughtyoung women dentials. to work in their factories,young women These changes tended to make teaching moved into these positions.But most other less attractive tomen. When teaching was jobs were closed towomen. Thus, although a relatively casual occupation that could be men were moving out of teaching because engaged in for fairly-short periods oftime, it was attractive to men ina variety of circumstances. A farmer could easily combine teaching in the winter with caring forhis 13 Lockridge (1974), using the abilityto sign one's farm during the rest of theyear. A potential name on one's will as a measure of literacy, found that by 1850, except in the South, bothmen and women minister, politician, shopkeeper,or lawyer over the age of 20 were almost universally literate. could teach for a short period of time to gain The causes of this silent revolutionin girls' school visibility within the community,However, attendance is an interesting topic in itsown right, for once standards were raised for teacher cer- in colonial times girls were generally excluded from tification and school termswere lengthened district schools. Vinovskis and Bernard(1978) noted and combined into that in 1850 in New England about 80percent of white a continuous year, men males and about 75 percent of white females began to drop out of teaching (Morain, 1974). ages 5 to 19 attended school. Attendance rates for 1850for ',oth In urban areas, where teachingwas first for- sexes and the ratio of female /male school attendance malized, and later in ruralareas, most men descends, however, as wemove through the Middle found that the opportunitycast of teaching Atlantic, North Central, South Central,and South At- was simply too great. Even though annual lantic regions, respectively. In the South Atlantic,about 41 percent of white males and about 35percent of white salaries were higheronce standards were females ages 5 to 19 attended school.

159 /

TOWARD A GENERAL THEORY OF OCCUPATIONAL SEX SEGREGATION 153 the opportunity cost of remaining was too, women were generally the teachers. A myth high, single young women with middle-class grew that women had more difficulty than backgrounds found teaching an attractive al, men in "handling" the older boys.Thus, it ternative to other paid work or to remaining may be that in rural schools men received at home and assisting with domestic chores. a pay premium for their supposed discipli- Young women might well have moved into nary abilities. It may also be that on the teaching without any assistance from idea, whole men had more experience in teaching logical campaigns. But perhaps to ensure and received a return on that experience. women's interest in teaching or to make their In urban markets, however, the men who entry more palatable to their parents, future remained in teaching did not perform the husbands, and pupils, a major ideological same jobs as their female counterparts. The crusade was waged in favor of women's entry labor market for teachers in urban areas was into teaching. Advocates of women as teach- highly stratified, ancl men had the higher- ers, such as Catharine Beecher, Mary Lyon, paying, more prestigious jobs: principals, Zilpah Grant, Horace Mann, and Henry vice-principals, and high school teachers. Barnard, argued that not only were women That management jobs were reserved for the ideal teachers of young children (be- men even in an occupation that was over- cause of their patience and nurturantqual- whelmingly female is an important obser- ities) but also that teaching was ideal prep- vation. It may be that school boards, which aration for motherhood. They also proclaimed perceived women as impermanent mem- the virtues of women's willingness to teach bers of the work force, believed they could at lower wages than those required by men. decrease their 'management training costs Indeed the arguments in favor of women by training only men for managerial posi- teaching were so compelling that one won- tions. However, the fact is that even when ders haw it was that any men remained in women did maintain their attachment to the teaching. But a few did remain and at higher work force they were rarely trained or hired wages than those paid to women(Strober for management positions, thus suggesting and Best, 1979). that other considerations beyond training Why didn't teaching become a completely costs were operative in school boards' de- female occupation? And why were the men cisions. No doubt a desire for patriarchical who remained in teaching paid higher wages hegemony at the local level was a factor in than women teachers? It is useful to answer school boards' decision to hire male man- these questions by looking separately at ru- agers for schools and especially for the su- ral and urban labor markets. perintendency, a post that brought its in- In rural areas, one-room schoolhouses often cumbent into frequent contact with local male persisted even after school terms were leaders in business and politics. lengthened and credentialling was formal- What quantitative evidence exists to sup- ized at the state level. Women and men port this theory? As any cliometrician has tended to do the same job, and the gender by now ascertained, it is not possible to pro- wage differential tended to besmall, al- vide a definitive empirical testing of the the- though invariably in men's favor. As already ory put forth here. The required data on noted, prior to the lengthening of the school alternative wage rates simply do not exist. term, there had generally been two separate There is, however, econometric evidence short terms, one in the winter and one in consistent with this theory. the summer. Men tended to teach in the In regressions designed to explain the winter term and had the older boys in their cross-sectional variance in the percentage of class. During the summer term, when men women teachers in a sample of counties for and older boys were engaged in farming, 1850 and 1880, Strober and Lanford (1981)

1.60 151 MYRA 11. STROBER found that the length of the school year and ure' of 36 percent. In elementaryschools in the number bf teachers per schoolboth 1978, men constituted about 17 percent of measures of the relativeformalization of all teachers, an 'increase of about 11 per- schoolwere positive and significant de- centage points from the 1945-1946 figure of terminants of the percentage of female 6 percent. Tlie detailed reasons for this teachers. At the same time, the female/male change remain to be explored (for a brief salary differential was found to be lower the discussion, see Tyack and .Strober, 1981). greater the percentage of women in teach- HoWever, my theory suggests that men in- ing (Strober and Lanford, 1981). Inpooled creasingly saw teaching as economically at- cross-sectional regressions for the period tractiver(increased unionization has perhaps 1870-1970, Lanford (as cited in Gordon, 1980) helped in this regard) and that, in accord- found support for the hypothesis that greater ance with the principle that menshould have formalization of the educational system in- first choice in job opportunities, tho'se re- creased the proportion of female teachers. sponsible for teacher hiring were happy to In a case study of school personnel in San readmit them to the profession. Francisco in 1879, Strober and Best (1979) described their results as follows (p. 234): CONCLUSION Holding education and experience constant, sex In this paper I have sought to outline a played a significant role in determining the po- new, general theory of occupational sex seg- sition and type of school of employment. We also regation. The theory suggests that occupa- o- concluded that education and experience were tional sex segregation as well as the female/ less important than position and type of school male wage differential results from two main explaining salary variation by sex and that, jor principles. First, although Male employ- holding constant education, experience and po- within sition, a greater percentage of the F/M salary' ers set wages and working conditions, ratio stemmed from sex differentials in pay across the constraints set down by race and class, types of schools than from sex differentialswithin male employers allow male workers to de- types of schools. cide which occupations they will inhabit. Second, in deciding which jobs to claim for Margo and Rotella (1981) found for the themselves and which to leave for women, Houston school system in the 1892-1923 pe- male workers, again within the constraints riod that "although some of the prevalence laid down by race and class, attempt to max- of males in administrative posts and some imize their economic gain by comparing the of the size of the female/male salary differ- economic package presented by any partic- ential is explained by differences in expe- ular occupation with the economic packages rience and education, maleness itself was a offered by other occupations. Thus, occu- valued attribute in school personnel" (p. 20). pations become male or female not because L; the post-World War II period, teach- of their inherent characteristics but because ing has maintained its gender designation, of the interaction of patriarchy and male but the percentage of men in the profession workers' utility maximization. has increased markedly. Moreover, women I have also used this general theory as a in teaching are no longerprimarily young framework for explaining the changes in the and unmarried. In 1978 about one-third of gender composition of the teaching profes- all teachers, elementary and secondary com- sion in the late nineteenth and early twen- bined, were men. In high schools in 1978, tieth century. Stated most simply, teaching men constituted slightly morethan h" of became afemale occupation largely because all teachers (54 percent), an increase of about men moved out of it. Asschools and school- 18 percentage points from the 1945-1946 fig- ing became more formalized, teaching be-

16j 'TOWARD A'GENERAL THEORY OF OCCUPATIONAL SEX SEGREGATION 155 came less attractive to men while atthe same Arrow, Kenneth 1973 The theory of discrimination. Pp. 3-33 in Orley time more lucrative job opportunities were Ashenfelter and Albert Rees, eds., Discrimi- developing for men in business and in other nation in Labor Markets. Princeton, N.J.: professions. Although the quantitative evi- Princeton University Press. dence summarized in the paper is consistent Blau, Francine D. 1977 Equal Pay in the Office. Lexington, Mass.: D. with the theory, it is difficult to obtain his- C. Heath. torical data that would provide a definitive Blau, Francine D., and Wallace E. Hendricks test .of the theory, 1979 Occupational segregation by sex: Trends and As noted earlier, recent changes in equal prospects. journal of Human Resources 14:197- employment opportunity legislation and af- 210. Blau, Francine D., and Carol jusenius firmative action orders have complicated the 1976 Economists' approaches to sex segregation in dynamics of occupational sex segregation. the labor market: An appraisal. Pr. 181-199 in Martha Blaxall and Barbara B. Reagan, eds., Clearly, future theoretical work will need Women and the Workplacei The Implications to look carefully at these interventi(ns. Based of Occupational Segregation. Chicago: Uni- on what we have learnedhere, however, we versity of Chicago Press. predict that, in order to be successful in Bradford, David L., Alice G. Sargent, and Melinda S. Sprague changing the gender assignment of jobs, any 1975 Executive man and woman: The issue of sex- intervention strategy (such as equal em- uality. P. 52 in Francine E. Gordon and Myra ployment opportunity efforts) would have to H. Strober, eds.,' Bringing Women Into Man- do more than merely attack hiring and pro- agement. New York: McGraw-Hill. Cain, Glen G. motion rules. It would have to concern itself 1976 The challenge of segmented labor market the- with gender relations in society as a whole, ories to orthodox theory: A survey. journal of because patriarchal ideology and supply and Economic Literature 14:1215-1257. demand factors in the labor market are inex- Gordon, Audri 1980 A Dynamic Analysis of the Sexual Composition tricably interwoven. It would appear that, of Public School Teaching: 1870 to 1970. Un- unless there is widespread agreement on published Ph.D. dissertation, Stanford Uni- the virtues of breaking down patriarchal re- versity. lations, male employers and male workers Gross, Edward 1968 Plus ca change .. .? The sexual structure of will find ample opportunities for frustrating occupations over time. Social Problems 16 :198- the goals of governmental interventions in 208. the job market. Hartmann, 'Heidi I. 1976 Capitalism, patriarchy, and Job segregation by sex. Pp. 137-169 in Martha Blaxalllind Barbara B. Reagan, eds., Women and the Workplace: ACKNOWLEDGMENTS The Implications of Occupational Segregation. Chicago: University of Chicago Press. The following people provided helpful Kessler-Harris, Alice comments on the first draft of this paper: 1982 Out to Work: A History of Wage-Earning Women In the United States. New York: Oxford Alice Amsden, Francine Blau, Martin Car- University Press. noy, Regina Cortina, HeidiHartfnann, Henry Lloyd, Cynthia B., and Beth T. Nierni Levin, Karen Oppenheim Mason, Aline 1979 The Economics of Sex Differentials. New York: Quester, Judith Steihm, Joan Talbert, Deb- Columbia University Press. Lcdaidge, Kenneth orah Thresher, and David Tyack. 1974 Literacy in Colonial New England: An Enquiry Into the Social Context of Literacy in the Early Modern West. New York: Norton. REFERENCES Margo, Robert A., and Elyce J. Rotella 1981 Sex Differences in the Market for School Per- Amsden, Alice H. sonnel: Houston, Texas, 18924923. Paper pre- 1980 introduction. Pp. 11.38 in Alice H Amsden. sented at the annual meeting of the Social Sci- ed., The Economics of Women and Work. New ence History Association. Nashville, Tenn., York: St. Martin's Press. October.

162 156 MYAA H H. STROBER

Matthaei, Julie A. Stroberj Myra H. 1982 An Economic History of Women in America: 1976 Toward dimorphics: A summary statement to Women's Work, The Sexual Division of Labor the conference on occupational segregation. P. and the Development of Capitalism. New York: 295 in Martha Blaxall and Barbara B. Reagan, Schocken Books. eds., Women and the Workplace: The Impli- Meyers, Peter V. cations of Occupatioral Segregation. Chicago:. University of Chicago Press. 1980 From conflict to cooperation: Men and women teachers in the Belle Epoque. Pp. 493-505 in Strober, Myra H., and Laura Best Donald N. Baker and Patrick J. Harrigan, eds., 1979 The female/male salary differential in public The Making of Frenchmen. Waterloo, Ontario: schools: Some lessons from San Francisco, 1879. Historical Reflections Press. Economic Inquiry 17:217-236. Milkman, Ruth Strober, Myra H., and Audri Gordon Lanford .1980 Organizing the sexual division of labor: His- 1981 The Percentage of Women in Public School torical perspectives on "women's work" and Teaching: A Cross - Section Analysis, 1850-1880. the American labor movement. Socialist Re- Paper presented at the annual meeting of the view 49:95-150. Social Science History Association, Nashville, Morain, Thomas Tenn., October. 1977 The Entry ofWomen.Into the School Teacher Strober, Myra H., and David Tyack Profession in' Nineteenth Century Iowa. Un- 1980 Why do women teach and men manage? A published paper, Iowa State University. 'report on research on schools, Signs: Journal Oppenheimer, Valerie K. of Women in Culture and Society 5:494-503. 1970 The Female Labor Force in the United States: treiman, Donald I., and Heidi I. Hartmann, edi. Demographic and Economic Factors Govern - 1981 Women, Work, and Wages: Equal Pay for Jobs ing its Growth and Changing Composition. of Equal Value. Washington, D.C.: National Population Monograph Series, No. 5. Berke- Academy Press. ley, Calif.: Institute of International Studies. Tyack, David B., and Myra H. Strober Reich, Michael, David M. Gordon, and Richard C. 1981 Jobs and gender: A history of the structuring Edwards of educational employment by sex. In Patricia 1973 A theory of labor market segmentation. Amer- Schmuck and W. W. 'Charters, eds., Educa- ican Economic Review 63:359-365. tional Policy and Management: Sex Differen- tials. New York: Academic Press. Robinson, John P. 1977 Hoy Americans Use Time. New York: Praeger. Vinovskls, Maris, and Richard M. Bernard 1978 Beyond Catherine Beecher: Female education Rubery, Jill in the antebellum period. Signs: Journal of 1978 Structured labour markets, worker organization and low pay. Cambridge Journal of Eco- Women in Culture and Society 3:856-869. nomics 2:17-36. Walker, Katherine E., and Margaret E. Woods 1976 Time Use: A Measure of Household Production Sokoloff, Natalie J. of Family Goods and Services. Washington, 1980 Between Money and Love: The Dialectics of Women's Home and Ma, ket Work. New York: D.C.: Center for the Family of the American Home Economics Association. Praeger. . Stafford, Frank, and Greg Duncan Whitehead, Ann 1979 Some preliminary notes on the subordination 1977 The Use of Time and Te.hnology by House- holds in the United States. Working paper, of women. IDS Bulletin l0:10-13. Department of Economics, University of Woody, Thomas Michigan. 1929 A History of Women's Education in the United States. New York: Science Press.

163 c- ,t) Commentary:Strober's.Theoty of Occupational Sex Segregation

KAREN OPPENHEIM MASON

As Francine Blau's review of the litera- to refining our ideas about the causes of oc- ture in Chapter 7 makes clear, neoclassical cupational segregation. In*What follows, then, economists have had relatively little to tell I begin by criticizing Strober's theory. I then us about the causes of sex segregation inthe discuss sociological and economic ideas that, American economy. .Economists have paid together with Strober's theory, help explain far more attention to the consequences of the job segregation of the sexes in the Amer- occupational sex segregation than to it. causes, ican economy. and,' as often as not, have attributed its -ex- Strober's theoryilays the ultimate blame istence to unspecified "tastes" or amorphous for occupational sex segregation on the pa- institutional factors outside the economy. triarchal system in which men enjoy wom- Even when the nature of these tastes or en's sexual, child-rear0g, and domestic institutional factors has been made -con- services in thehousehold.The immediate crete, the explanations offered have often blame for segregation, however, is laid on failed to withstand the test of logic or em- employers, most of whom are men. Ein- pirical analysis. ployers are said to strive toward two goals In light of this, Myra Strober's attempted (goals that Strober recognizes as potentially -leap beyond the neoclassical in Chapter 8 is contradictory): (I) profit maximization and to be applauded. By explicitly recognizing (2). enforcing the economic dependency of that men exploit women and derive various women on men. The latter is of interest to advantages from doing so, Strober has sought male employers because it provides the ma- to explain the existence and persistenceof terial base for the patriarchal system, i.e.> sex segregation in an economythat in many it forces women to become dependent wives other ways may operate according to neo- and mothers (employers are said to worry classical market principles. Although Strob- about maintaining women's dependency on er's attempt is not entirely success,al, it is men in social classes other than their own nonetheless valuable. Because her theory is because threats to patriarchy in the working straightforward and provocative, it provides class may lead to threats to patriarchy within an important stimulus todiscussion and hence the managerial or capitalist class). In hiring

157 164 1.58 KAREN OPPENHEIM MASON

'workers for their establishments, employers when new and better-paying jobs opened #re thus said to concernthemselves not only up, leaving the oldline of work to be filled With minimizing their wage bill (the usual by women: school teaching and clerical work neoclassical assumption) but also with min- (see Oppenheimer, 1970:77-79; Davies, 1982: imizing the risk that women of a given race 56-58). and class will earn more than the menof Nevertheless, while containing sound that race and class and consequentlywill no elements, Strober's theory has several prob- longer be dependent on them and forced to lems. Although it attempts. to go beyond marry them. standard neoclassical assumptions, it retains How do employers meet thedouble goals enough of these assumptions to encounter of profit maximization and of ensuring wom- one of' the most seriousproblems that neo- en's dependency on men? They do so,Strobes classical explanations for sex discrimination argues, by offering all newjobs first to men, tend to face, namely, the seemingly coun- at a wage determined byconditions in the terfactual prediction that sex segregation will local labor market. These jobs are offered gradually disappear. Moreover, whether to women only if men refuse totake them. Strober's theory can explain the extremely Strober in turn argues that .men's willing- high levels of occupational sex segregation ness to take particularjobs depends entirely observed in our economy and preserved On economicconsiderations: Unlike their through decades of industrial change is un- bosses, male workers are strictly profit max- clear. And for the historical period for which imizers. Thus, if they can earn more money it seems intended, Strober's theory is in sev- elsewhere, male workers will turn down the eral key respects surprisingly implausible. jobs that a particular employer offers them. Finally, whether an entirely new theory of The result is that the poorer-paying jobs are occupational sex segregation is needed is open left for women, the better-payingjobs hav- to debate. Lei me elaborate oneach of these ing been snatched up by men.Strober's the- points. ory this suggestswhy there is a wage gap between the sexes as well as occupational THE PROBLEM OF THE DEMISE OF SEX sex segregation. SEGREGATION There are. several points in Strober'sthe- ory with which I agreeand for which there As Blau (Chapter 7 in this volume)and is fairly good empiricalevidence. Women others (e.g., Stevenson, 1978) have noted, in this society are without question econom- most neoclassical theories ofoccupational ically disadvantaged compared with men. segregation imply that it will graduallydis- and this situatior hardly an accident of appear becausenondiscriminatory employ- history or nature.!) ,.re are obvious ideo- ers or firms will be. 4 aneconomic advantage logical (Williams and Best, 1982), legal (Ka- over other firms andwill consequently ex- implied dis- no..vitt.1969). and informal mechanisms perience greater growth. This (Bernard. 1971:88-102) that maintainthe appearance of segregation isproblematic, system producing thisdi...advantage and that however, because occupational stir-ration continue to do so even in the faceof major has in fact remained firmly entrens.ned in protest movements. Mostemployers today our economy for over a century(Oppen- are men and as men canhe suspected of heimer, 1970:64-77: Gross, 1968; Williams, having a stake' in the' system of male privi- 1979; Matthaei, 1982:187-232). lege IGoode. 1982). There also are atleast Strober's theory- suffers the same prob- two fairly clear historicalexansles of oc- lem. In an economy in which employers typ- cupational aband(1111nellt and SUCCeSSi011 ically of new jobs to men first,withhold- IllstallCe %. Inch ownleft one line ohs (irk ing them from women until such time as

165 COMMENTARY 159 men have refused to take them, anemployer egories, according to "qualifications." Eni- who offers jobs to women first is likely to ployers and the market then function to cre- minimize his wage bill much more success- ate a one-to-one mapping between the two fully than will discriminatory employers, be- queues, with the best-qualified male worker cause women's wages willhave been driven getting the best-paying job and the best- downward by other employers' discrimina- qualified female worker gettinga lower-pay- tory practices. Over time, then, any em- ing job than any male worker. ployer whose desire for profits outweighs his Although this model implies the existence desire for maintaining patriarchy is likely to of job segregation between the sexes (or at undercut competing firms, thereby experi- least does so if we assume a tight connection encing greater growth. The long-run impli- between jobs and wage rates), the magni- cation is that such firms will succeed and tude of the segregation implied will depend discriminatory firms will fail, meaning that on where the divide between the sexesfalls occupational sex segregation should gradu- in the job queue. If the divide happens to ally disappear. coincide with the divide between two different jobs, there will be perfect segregation of the sexes. if it does not, however, there THE PRObLEM OF EXTREME will be some degree of job integration, the SEGREGATION precise degree depending on the number of There is considerable evidence to suggest positions in the job that straddles the divide that the occupational segregation of the sexes between men and women in the worker in our economy is extreme. Studies of de- queue (if there is a very large numberof tailed occupationswhich .aggregate jobs positions, a substantial port' n of the work and hence tend to underestimate the true force may end up emplo ed in the inte- extent of job segregationshow that at least grated job).' two-thirds of male or female workers would To be sure, the notion that an entire labor have to switch occupational categories for market can -be ordered into a single job and the sexes to achieve identical occupational a single Worker queue is unrealistic. Never- distributions (e.g., Williams, 1979). More- theless, even if we imagine labor markets over, studies conducted at theindustry or consisting of multiple job and worker queues, firm level (e.g., Blau, 1977; Bielby and Baron, a mapping process of the kinddescribed here Chapter 3 in this volume) suggest an even seems likely to result in somedegree of job higher level of segregation. To be plausible, integration, possibly far more integration than then, a theory of job segregation must read- is in fact observed in our economy. Only if ily explain not only the existence of some there are separate male and female job and sex segregation in our economybut also the worker queues, as some scholars believe existence of extreme segregation. A closer look at Strober's theory raises doubts on this score. l Throughout this commentary, Idistinguish between a jobwhich is a particular bundle of tasks as- The basic tenets of Stroher's theory can signed to individual workers in a given industry or be interpreted in terms of a system of job firmand a position, which represents one slot in a and worker queues and their mapping onto given job. There are certain jobs that involve only one each other within the labor market. In the position (e.g., President of the United States), but most simplest case, there is one job queue and jobs involve multiple positions. Hence, it is quite possible in the queue-matching system that Strober's the- one worker queue. Inthe job queue, jobs ory implies for the &vide between men and women are ordered according to wagelevel, while within the worker queue to fall in the middle of a in the worker queue, workers are ordered multipositioned job, thereby producing sexual integra- according to gender, and within gender cat- tion of that particular job.

166 160 KAREN OPPENHEIM MASON there are (e.g., Edwards, 1975), is the high tal taboos about the two sexes working to- level of segregation observed in our econ- gether on certain jobs." Although this is an omy likely to occur. Thus,whether the basic intriguing suggestion, just how common 'such tenets of Strober's theory necessarilyimply taboos are and whether they can explain the the existence of high levels of job segrega- existence of high levels of segregation in the tion is unclear. workplace is unclear.3 Strober herself seems Perhaps in recognition of this possibility, to think that the enforcement of suchtaboos Strober supplements her theory's basic tenis probably not the sole explanation for why ets with a set of ad hoc argumentsabout the employers prefer to hire men for jobs that. actions both employers and male workers are already heavily male; shedescribes the are.t likely to take once ajob has become taboos as "possible" and pertaining, only to substantially but not entirely occupied by "certain jobs." Moreover, whether job segmen. Strober argues thatemployers in this regation and physical separation are nec: situation can pursue two courses. They can essarily coterminous is unclear. Indeed, it either fill the remaining positions with women is probably just as common for workers in workers, but pay them less than the male a given job to work at somedistance from workers in the job in order to keep the women each other (e.g., parts inspectors who work in an economically inferior position, orthey in different locales within a plant) or for male can raise the job's wagelevel in order to try and female workers in distinct jobs to work to attract men into the remainingpositions.2 in close physical proximity to each other (e.g., While the first of these strategies will clearly a female secretaryand her male boss) as it maintain women's economic inferiority to is for workers in the same job to work side men, it will not result insexual segregation by side while those in different jobs work of the job. Thus, for her theory to imply a apart from each other. Thus,although re- high level of job segregation, Strober must search into taboos against mixed-sex work argue that employers will preferthe second groups would be useful (as Strobernotes), strategy to the first. the ability of these taboos to explain the high Strober offers three' reasons why employ- level of segregation observed in our econ- ers might want to raise wagesin order to omy seems doubtful. make heavily male jobs entirely male. The Strober's final reason for suggesting that first reason, which Strober states very cur- employers will fill an already largely male sorily, is that hiring women (even at an injob with men is little more compelling than ferior wage) "would erode the principles of the first two reasons. It is that hiring women patriarchy and male. privilege." It is unclear "would. . .interfere with male bonding." why the hiring of women at lower wages I am uncertain what "male bonding" is sup- would erode patriarchy, since in Strober's posed to refer to (it smacks of reductionist theory the inferior earnings of women are apparently sufficient to maintain patriarchal relations. This first reason, then, is uncon- 3 The reason for the existence of these taboos is also vincing. not entirely clear. Presumably, a fear of illicit sexual Strober's second reason for suggesting that relations is what motivates any desire to ensure that employers will prefer to hire men is that men and women do not work side by side. Adesire on hiring women "might violate possible socie- men's part to maintain social distance from their status inferiors (women), however, might he just as important. In this case, however, the functionali.e.,job separation of the sexes would presumably be more 2 There is actually a third option that I am ignoring important than would their physical separation; and here, only because it does not change Stroher's theory male workers would he as concerned with maintaining materiallythat is, that the employer's attempt to hire patriarchy as with maximizing their economic gain (con- foreign or immigrant labor. trary to Strober's main theoretical tenets).

167 161 COMMENTARY of oc- arguments about men'spsychological "needs" about preserving the sexual "purity" are.) And or "masculineinstincts," the existence of cupations than the male workers which is questionable). It is hard tobelieve what will male workers gainby preserving that employers operating in acompetitive "patriarchal hegeniony" (which isundefined monopolizing all and largely unregulated economy(the sit- but which I assume means workers uation that Strober seems tobe referring to) the positions in a given job)? Male would deliberately raise their wagebill sim- may, of course,fear that their boss's desire his desire to ply because hiring women might"interfere to earn profits will overcome lead- with male bonding. "4 Thus, in thefinal anal- bolster the patriarchal system, thereby employers to fill sub- ing him to replace maleworkers with less ysis, the willingness of this possi- stantially male jobs with more expensivemale expensive female workers. But problem workers rather than turning tocheaper fe- bility raises a more fundamental how employ- male workers remains unclear. with Strober's theory, namely, between the goals of If it is implausible that employersare the ers resolve the tension that high levels of job profit maximization and maintainingpatriar- ones likely to ensure to segregation exist, then inStrober's theory chy. In the end, then, Strober's attempts who do so (the only ensure that hertheory implies the existence it must be male workers raise as other possible creators of segregation are of a high level of job segregation they answer. women, but inStrober's theory women are many questions as assumed to be powerless in thelabor market). Strober's description ofhow male THE HISTORICAL CONTEXT workers reinforce or increase segregationis Once an occupation hasbeen Although Strober does not spell outthe as follows: her theory is sup- "typed" as male (a process that seems tobe precise historical period primarily inevitable, though for reasonsStrober does posed to cover, it seems to pertain "will actively and col- to conditions in themid to late nineteenth not make clear), men prior to lectively seek to keep women out,fearing and early twentieth centuries, i.e., intervention into .the that if women enter, theywill lower the any significant state lower wage free market economy for the purposeof reg- earnings of the job by accepting this rates or, if they are paidthe same as men, ulating employment conditions.5 In context, Strober's theory seemsimplausible diminish patriarchal hegemony." because the econ- It is unfortunate thatStrober does not in three respects. First, explicate these ideas further, since a num- unanswered. Why, ber of questions remain the example, workers fear wom- 8 This is certainly the period to which for example, must male of school teaching pertains and is alsothe period during en's incursion into "their" occupationsif em- wtich much of the sex typing found intoday's labor ployers are no less interested inensuring rif.;rket was established (Oppenheimer,1970;64-120; women's economic inferioritythan the male Snyder and Hudis, 1976). Moreover,Strober's con- employment workers are? (Indeed, Stroberinitially im- cluding remarks about the irnpact of equal concerned opportunity legislation suggestthatstate intervention plies that employers are more in the free marketplace maychange the terms of her theory (although only if that interventiontakes particular forms). Finally, even if Strober intendsher theory to he atemporal, it is important torecognize that the It plausibly might do so weremale workers to press the economy has a history and that through organized protests sex segregation of for the exclusion of women change over (something that Strober later sug- the processes influencing the economy can or political action have cho.;-n to interpret t' :n this section of her argu- time. For all of these reasons, I gests they may indeed intended apparently concerned only Strober's theory as though it was primarily ment, however, Strober 1. early twentieth century with employers' own intent sts. notwith the necessity to describe late nineteenth and conditions, even if Strober herself did nothave this of giving in to certain political pressuresfrom male explicitly in mind when creating the theory. workers. 168 162 KAREN OPPENHEIM MASON omy to which she refers was more highly nation would have helped minimize some competitive than is the monopoly economy employers' wage bills, there is every reason of the twentieth century, the assumption to think employers would have preferred that employers would be willing or able to this approach to the "first dibs to men" ap- forgo profits in the interests of maintaining proach that Strober outlines. Only with the domestic patriarchy seems especially open creation of state-enforced regulations re- to question. One can envision firms rela- quiring equal pay for equal work does the tively insulated from free market pressures approach Strober describes become more making this choice, but it is much harder to plausible. envision small, struggling firms in the un- A final point that is implausible in the regulated marketplace doing so.6 nineteenth-century context is the idea that Also relatively difficult to accept in the male employers would try to maintain wom- nineteenth-century context is the notion that en's economic inferiority in social classes other employers wishing to ensure that women than their own. Strober's argument here is earn less than men would bother with the akin to a domino theory of political change. complex procedure that Strober describes, If working-class women are allowed to be- i.e., offering jobs to men first, waiting to see come economically independent tom the if all positions are filled by men, etc. In the men of their class and are consequently freed nineteenth century, there seems to have been from the need to marry and serve them, littlefeeling that employed women de- then upper-class women might be inspired served the same wages or job opportunities to follow their working-class sisters down as men (Smuts, 1959:110-142; Kessler-Har- the road to independence. Thus, even though ris, 1981:54-70). Hence, a much easier strat- most male employers would never seek to egy for the employer wishing to pay women have their sons marry working-class women, less would have been to do just that: pay they are said to be motivated to maintain women less than men, regardless of the job. gender inequality in the working class out To be sure, in some situations this strategy of concern for their own position vis-à-vis might have required inventing separate job upper-class women. titles for women and men, so that the dis- While there may be some validity to this parity in wages between the sexes could be idea in historical periods when feminist con- Masked or socially justified. But employers sciousness transcends class barriers, the idea seem perfectly capable of inventing separate seems implausible for the late nineteenth job titles or pay grades when they want to century, when most upper-class women ap- (e.g., Newman, 1976). Thus, even if we are parently had little sense of identification with willing to believe that nineteenth-century their working-class sisters or at best had a employers were interested in ensuring that highly patronizing identification that made women were paid less than men, whether clear their own social superiority. Certainly, they would have used the system Strober upper-class women in this period were will- describes is questionable. Indeed, because ing to exploit the working-class women they a straightforward system of wage discrimi- hired as domestic servants, often treating them with little consideration or with the sense that the servant girl had much in com- It is interesting that the particular occupation Strober mon with the lady of the house (Katzman, studies is found in the public sector and hence does 1978:158-173). Moreover, although nine- not involve the same competitive forces that affect pri- teenth- century working-class women did not vate sector employment. The assumption that employ- earn as much money as did working-class ers are willing (and able) to forgo profits in order to help maintain patriarchy may be inure realistic in the men, in some parts of the country these public sector than in the private one. women worked in large numbers (Mason et

1 C 9 COMMENTARY 164 al.,1978). Yet despite this,upper-class chy could be maintained in other ways, why women in these areas remained firmly de- would employers cut into profits in order to yoted to the "cult of domesticity" and, by offer jobs to men first?) implication, to the patriarchal system. Although women's inferior earnings may Working lass women's relative independ- have contributed to the patriarchal system ence thus seems to have posed little threat in nineteenth-century society, there were to upper-class women's commitment to re- other institutional factors that also did so. maining ladies and hence economically de- Among these were the laws and judicial pendent on their husbands.' This makes the precedents that made women the legal and idea that upper-class men would have cut political (as well as economic) inferiors of into profits in order to ensure the economic men (Kanowitz, 1969:35-93) and a host of inferiority of working-class women seem im- norms,cusioms, and culturally transmitted plausible. beliefs that taught women to orient themselves exclusively toward careers as wives and mothers and to otherwise behave in a FURTHER PROBLEMATIC ASSUMPTIONS manner consistent with their label as the Strober's theory contains two other as- "weaker" sex (Kessler-Harris, 1982:49-53). sumptions that are questionable. The first These forces seem to have been no less im- is that the wage gap was the only or by far portant than the dismal wages and dead-end the most important prop for the patriarchal jobs available to women in convincing them system in the late nineteenth and early that marriage and motherhood were the most twentieth centuries. Although Strober no- satisfactory careers (Matthaei, 1982:101-140; where states this assumption explicitly, it is Kessler-Harris, 1982:20-72, passim). Given implicit in her argument that, in order to this, whether upper-class men would have maintain patriarchy, employers acted to enacted against their economic self-interests sure the existence of a wage gap. (If patriar- in order to produce a wage gap between working-class women and men seems questionable. The other problematic assumption im- ' This is not surprising, since it was working-class 'plicit in Strober's theory is that, in the con- women whc, tended to accept middle-class norms about text of the labor market, men were more a woman's place, rather than the reverse. Most evidence suggests that when women did go to work in the interested in maintaining the domestic late nineteenth century, they justified doing so in terms vision of labor between the sexes than any of their future domestic roles or current family needs other aspect of the patriarchal system, in- (Smuts, 1959; Matthaei, 1982; Kessler-Harris, 1981, cluding the general male prerogative to con- 1982). In other Words, the working-class girl who worked trol women and receive deference from them. in a factory or went into service did not see herself as becoming independent from a potential husband (though True, even when acting as employers or as she did sometimes see herself gaining partial inde- workers, men may have been interested in pendence from her family of origin). Rather, work for ensuring that women, as a class, were kept most working-class girls was an interlude between in an economically inferior position and hence childhood and marriage, during which they helped sup- forced to participate in a patriarchal family port their families, increased their prospects for a "good" marriage by being able to afford a good wardrobe, or system. But it seems equally likely that, when learned skills that were argued to be helpful to their acting as employers or workers, men were future roles as housewives and mothers. It is, there- concerned about keeping women in their fore, not surprising that upper-class women failed to place on the job, i.e., ensuring that no woman become feminists in the face of high labor force par- would outrank or could give orders to a (na- ticipation rates among their working-class sisters; they already enjoyed the amenities to which these working- tive white) man. class women aspired. To the extent this was true, it makes little

170 164 KAREN OPPENHEIM MASON sense to think that employers offered all new of sex segregation and will then attempt an jobs to men first. To be sure, if a job in- integration of these three ideas. volved authority over other workers or, given The first idea or set of ideas is the most the temporary nature of most women's work amorphous, but it is also the most impor- (Matthaei, 1952:195-196), required conti- tant. It is that an "ideology of gender" or nuity of employment, then employers no normative/cultural system lies behind and doubt offered it to men. However, if a job guides both men's and women's behavior was routine, easily learned, and neither re- and in so doing tends to separate the roles quired long-term employment nor involved and activities of the Sexes, including their control over other (male) workers, then em- occupational roles (Reskin and Hartmann, ployers had no reason not to offer it first to 1984:Ch. 2; di Leonardo, 1982). Basic pre- women (Matthaei, 1982:196). Indeed, there cepts in the American ideology of gender are historical cases in the textile and clothing include (1) the assumption that the sexes are industries, as well as in clerical and sales inherently different from each other in char- jobs, in which this appears to be exactly acter, temperament, and capacity; (2) the what happened (Kessler-Harris, 1982:142- specific perception that women are naturally 179). This makes a theory that rests on the suited to be mothers, soothers, supporters, assumption that all new jobs are offered first and/or pets, while men are suited to be ad- to men untenable. venturers, leaders, fighters, and doers (Wil- liams and Best, 1982); (3) the evaluation that the feminine is of lower prestige (less im- IS A NEW THEORY OF OCCUPATIONAL portant) than the masculine and that for men SEGREGATION BY SEX NEEDED? to engage in feminine activities or pursuits Strober begins her paper by arguing that is consequently highly stigmatizing (more so existing theories of sex segregation in the than for women to engage in masculine ac- workplace are inadequate. None of these tivities or pursuits, although that, too, is theories, she states, is "capable of answering stigmatizing); and (4) the assumption that the three major questions concerning the men have the right to control women and gender designation of an occupation: its or- women have the obligation to acquiesce in igin, its maintenance, and its change, if any." this control (Collins, 1971). Athough Strober is narrowly correctno Because gender is usually the first social single theory in existence at the time her identity learned by children, precepts such paper was written could adequately explain as these are strongly and often uncon the origins and persistence of sex segrega- sciously cherished by both sexes. To the ex- tion in our economyI am not convinced tent that this is true, such precepts are likely that a new theory of occupational segrega- to shape virtually every decision made by tion is needed. The old theories, although women and men that affects the jobs into incomplete and not always systematic, which they areTuited. This means that, nonetheless offer considerable insight into once a set of °ca. ,Itions has been "typed" the segregation of the workplace, especially or labeled as appropriate for one sex only, when considered together. Indeed, in my these labels are likely to persist over time judgment, these "old" ideas more persua- (unless revolutionary forces disturb them). sively suggest why sex segregation is both The fact that most. dividuals in society extreme and enduring than does Strober's are socialized to an ideology of gender that theory. To argue this claim, I will first re- emphasizes the oppositeness of the sexes view three of the most important ideas in may also help explain why occupations are the sociological, economic, and historical lit- sex-typed or labeled in the first place. If eratures about the origins or maintenance one's social respectability and sense of self-

1 71 COMMENTARY 165

respect depend on behaving in a manner which they are inherently unsuited. The no- appropriate to one's gender (and inappro- tion of statistical discrimination thus sug- priate to the other gender), then working at gests how the ideology of genderand the jobs that members of the opposite sex work reality of most women's and men's lives at may be degrading. Matthaei (1982:194) is likely to influence employers' behavior has argued that this was a critical influence and thereby *tribute to the sexual seg- on job segregation in the nineteenth cen- regation of the work force. While it is clear tury: that statistical discrimination cannot explain all forms of job segregation (e.g., that which Employment in a clearly masculine job, doing occurs among unskilled workers doing equally a job done by other men, fortified a man's sense of manhood; competition with these men to do heavy or light tasks), this concept nonethe- the job well, or unification with them against the less points to an important proce:s likely to "big man," the employer, actively expressed and contribute to the segregation of male and

measured his manhood....[On the other hand,] female workers. doing a job that women also performed expressed The final set of ideas relevant to under- a man's similarity with the opposite sex, showed standing the sex segregation of the economy him to be womanly and feminine. derives from Edna Bonacich's (1972) theory Likewise, women wished to work in jobs done of the split labor market. The basic tenet of by women. A woman's femininity was already this theory, which was originally created Ito threatened by her presence in the labor force, explain the existence of ethnic antagonism, the masculine sphere....If a woman was forced is that there often are three significant classes to seek wages outside of the home she would seek jobs which were clearly woman's work. in conflict within capitalist labor markets, not just the two that Marx identified: (1) the In other words, individual men and women, capitalists or employers, who are concerned in seeking work, may have created or con- with maximizing profits (a point on which tributed to their own segregation by avoid- Marxist and neoclassical economists seem to ing employment in sexually integrated jobs. agree); (2) the high-priced "established" Other ways in which the 14 logy of gender workers who, through political organization may have influenced the ox,cupational seg- and struggle, have managed to wrest some regation of the sexes in the nineteenth and degree of economic 'security from the cap- early to mid twentieth centuries are noted. italist class and who are interested in main- below. taining or improving this economic security; The second idea that can help explain oc- and (3) the low-priced workers, i.e., socially cupational sex segregation in our society is identifiable groups who, for a variety of rea- the concept of statistical discrimination. Be- sons, are unable or unwilling to demand as cause this concept is reviewed elsewhere in high a wage as the established workers earn this volume (see Chapter 7), I will not at- and whose primary concern is simply find- tempt a full summary here. Suffice it to note ing a job, rather than achieving a particular that when employers pay for the training of level of economic security. workers, the perception that women are more In Bonacich's theory, there are three basic likely than men to leave the labor force in dynamics in the split labor market: (1) the order to marry or rear children may explain capitalists try to minimize their wage bills why employers are reluctant to hire women and consequently try to replace high-priced for these positions. Employers may also be labor with low-priced labor; (2) the low-priced reluctant to hire women for supervisory po- workers try to find jobs; and (3) the high- sitions if they believe (as they are likely to) priced workers struggle to protect them- that this vijates the natural order between selves from the incursion of the low-priced the sexes or places women in positions fo' workers. Bonacich argues that established Men and women seek jobs in which only members of their own sex work or that ars'otherwise vetted as acceptable for their sex it Mid-nineteenth century Occupational "gatekeepers" ideology of "separate offer skilled jobs, jobs Men and women spheres" and the public involving on-the-job N.- work in different versus domestic sphere authority, commitment to jobs/occupations division of labor the firm, long-term employment between the sexes only to men

Employers offer unskilled or "preskilled" Women enter the labor force jobs to women as temporary, supplemental in order to workers and hence work for lower wage.bill lower wages than men are willing to accept Male workers try to ensure that "cheap" female workers won't lower their wages by restricting women's access to desirable jobs 4)

FIGURE 9-1Schematic. integration of ideas about the etiology of job segregation between the sexes. 173 COMMENTARY 167 workers use two strategies in attempting to ideology of women's and men's separate protect themselves from the low-priced spheres was already well established (Kes- workers: exclusion and the creation of caste sler-Harris, 1982:20-72). For married adults systems. In other words, established work- the division of labor between the sexes ers can either try to exclude low-priced la- matched this ideology, with women devot- bor from the labor market entirely (this is ing themselves for most of their married lives sometimes done via restrictive immigration to domestic work and men to work in the laws) or they can try to restrict low-priced paid labor force. That this ideology and di- labor to a narrow range of poorly paid oc- vision of labor was already in place seems cupations in which they have little interest. to have had three consequences. The first In the nineteenth century, women were was that whet\ young women and men sought low-priced workers compared with native work they tended to look for jobs that were white males. Women typically worked tem- known as women's or men's work, or, if the porarily, before they were married, and jobs were very new, that gaVe every indi- usually worked as supplemental earners (in cation of becoming women's or men's work many occupations, including school teach- (e.g., because the employer advertised for ing, the wages they earned were often in- members of one sex only, as was typical in sufficient to live on; Matthaei, 1982:187-232). the nineteenth century).. In other words, the For these reasons, women were 'usually will- first path through which the division of labor ing to work for much lower wages than were between the sexes and the, ideology of sep- men; they were also less frequently involved arate spheres influenced the sexual segre- in labor actions. (though there were notable gation of the work force was by influencing instances in which women formed unions or young men's and women's own bccupational participated in strikes; e.g., Daw ley, 1976). choices (Matthaei, 1982:194). While the "cheapness" of female labor did Second, as the concept of statistical dis- not always threaten male workers, there are crimination suggests, the ideology of sepa- well-documented cases in which it did and rate spheres and the division of labor be- in which organized groups of male workers tween the sexes also influenced the actions responded by attempting to ensure that their of various occupational "gatekeepers" own jobs could not be taken over by women employers; schoolteachers, employment (e.g., by pressing for the passage of protec- agencies, and the householders who hired tive labor legislation; see Hartmann, 1976; domestic servants. One such influence may Matthaei, 1982:217). One effect of this was have been the one Strober emphasizes, to segregate women into certain poorly paid namely, a tendency to favor men in the hir- occupations that native white males were ing process in order to maintain the primacy uninterested in. While the actions of male of wage work as part of the masculine sphere. labor unions in response to the perceived However, occupational gatekeepers' preju- threat of "cheap" female (and immigrant) dices probably had other effects on the oc- labor cannot alone explain the occupational cupational segregation of the sexes as well. segregation of the sexes, it seems to have Most important was a tendency to offer jobs been one force that helped create and main- to one sex only according to that sex's sup- tain this segregation. posedly unique talents and traits or accord- Figure 9-1 depicts what I think is the most ing to the structural position they were to historically accurate and sociologically rea- occupy within the workplace. For example. sonable integration of these three sets of in the period after the Civil War, women ideas. In the mid-nineteenth centurythe vere preferred to men as domestic servants, period when women as well as men began partly because they could be hired for less entering wage work in large nurnbers the money but also because' they usually had

174 168 KAREN OPPENHEIM MASON prior experience. That they were normally The cheapness of female labor may also supervised by a woman also meant that re- have contributed to the segregation of the lationships between mistress and servant sexes through another route. This was by were much less strained than if the servant encouraging employers to hire only women was a male (Matthaei, 1982:198) for jobs for which skills and on-the-job au- Likewise, men were preferred for indus- thority were minimal and the costs of labor trial jobs that involved on-the-job authority turnover were also low. In other words, em- or that required a commitment to the firm ployers sought female workers not only be- or the establishment (Matthaei, 1982:196). cause they perceived them to be inherently N. only was it highly inappropriate for a suited to particular kinds of work but also woman to supervise others outside her own because women workers were inexpensive.8 home, but also women typically worked only While it is impossible to gauge the impact sporadically and usually quit altogether when this had on occupational segregation com- they married. Consequently, they could not pared with the impact of male and female he expected to take on a responsible position workers' own occupational choices and the that required considerable on-the-job train- choices of employers dictated by their per- ing or at least employers were unwilling to ceptions of women's versus men's traits or offer theth such jobs (most women may have patterns of employment, it seems clear that been uninterested in them, too; see Mat- it contributed to the segregation of the sexes thaei, 1982:194). In various ways, then, the in the nineteenth and early to mid twentieth perceptions of the sexes that occupational centuries. gatekeepers developed during their social- In summary, several distinct processes ization to the ideology of separate spheres stemming from the acceptance of a partic- or by simply observing how most women ular definition of masculine and feminine ,and men in fact led their livesled them roles and temperaments in American society to hire or direct members of each sex to the appear to underlie the segregation of the jobs that seemed appropriate for them. sexes within the workplace. If Matthaei The third influence of the sexual division (1982), Kessler-Harris (1982), and other his- of labor and ideology of separate spheres on torians are to be believed, women and men the job segregation of the sexes was to themselves helped create the segregation of cheapen the price of female labor compared the workplace by seeking jobs in which only with male labor. The fact that women were their own sex worked. The tendency to willing to work for less money than were choose a job labeled-appropriate for one's most native white men can be seen as a own sex was exacerbated by the actions of direct outgrowth of the gender division of employers and other occupational gatekeep- labor during this period of history. Because ers who, in keeping with the same precepts women viewed wage work as a temporary of masculine and feminine behaviorsand condition designed to help their families (or the rea; differences between women's and to pay.for extras such as new clothes), they were willing to work for lower wages than were men, for whom work was acentral, 8 Marxist theorists also argue that employers hired lifelong commitment (Matthaei, 1982:193- men and women for different jobs as part of a strategy 197). As Bonacich's theory and the historical of labor market segmentation designed to keep the record both suggest, this led organized groups working class weak by creating divisions within it (e.g., of male workers to work for women's re- Edwards, 1975). As Strober notes, while employers may indeed have used this strategy, why they chose moval from certain jobs, something that no gender as one of the bases on which to segment the doubt contributed to the overall segregation labor market is not readily explained by their desires of the sexes in the workplace. to weaken the power of the working class.

175 COMMENTARY 169 men's motives for working and patterns of sion of this paper and Madelon Weberand employmentfrequently sought workers Mary Scott for their typing assistance. of only one sex for particular jobs. Finally, organized groups of male workers further REFERENCES AND BIBLIOGRAPHY reinforced the segregation of the sexes by Bernard, Jessie acting to ensure that women could not enter 1971 Women and the Public Interest. Chicago: Al- their occupations and thereby lower their dine-Atherton. wages. Because individualworkers, em- Blau, Francine D. 1977 Equal Pay in the Office. Lexington, Mass.: ployers, and organized groups of male work- Lexington Books. ers all acted in ways thatproduced a sepa- Bonacich, Edna ration of the sexes within the workplace,the 1972 "A theory of ethnic antagonism: The split labor extreme degree to which the American market." American Sociological Review 37:547- should not 559. economy is sexually segregated Collins, Randall be surprising. Nor should the incredibly slow 1971 "A conflict theory of sexual stratification." So- speed at which sex segregation has changed cial Problems 19:3-21. over the past century(Williams, 1979). Davies, Margery W. 1982 Woman s Place Is at the Typewriter: Office The main implication of the views I have Work and Office Workers, 1870-1930. Phila- presented for the future of occupational seg- delphia: Temple University Press. regation between the sexes is very similar Dawley, Alan ends her 1976 Class and Community: The Industrial Revo- to the point with which Strober lution in Lynn. Cambridge, Mass.: Harvard paper. Occupational segregationis unlikely University Press. to disappear or even lessenappreciably un- de Leonardo, Micaela less major revisions occur in our ideology of 1982 "Occupational segregation and cultural analysis." Washington, D.C.: Paper presented at gender and the division of labor between the Workshop on job Segregation by Sex, the.sexes. To be sure, some changes in gen- Committee on Women's Employment and Re- der ideology and in the male-female division lated Social Issues, National Research Council, of labor have occurred during the past four May 25, Edwards, Richard C. 'decades (e.g., Mason et al., 1976; Waite, 1975 "The social relations of production in the firm 1981). And there are starting to be some and labor market structure." Pp. 3-26 in Rich- noticeable changes in occupational segre- ard C. Edwards, Michael Reich, and David M. Gordon, eds., Labor Market Segmentation. gation as well (see Chapter 2 in thisvolume), Lexingtod, Mass.: D. C. Heath. perhaps as a result of the ideological shifts. Goode, William J. However, unless we give up our idea that 1982 "Why men resist." Pp. 131-150 in Barrie Thorne with Marilyn Yalom, eds., Rethinking the men and women areinalienable opposites, Family: Some Feminist Questions. New York: more dissimilar than alike, we areunlikely Longman. to see the disappearanceof occupational Gross, Edward segregation between the sexes. Ultimately, 1968 -Plus ca change ...? The sexual structure of occupations over time." Social Problems 16:198- job segregation is just a part of the generally 208. separate (and unequal) lives that womenand Hartmann, Heidi men in our societylead, and, unless the 1976 "Capitalism, patriarchy, and job segregation overall separateness is ended, the separate- by sex." Pp. 137-169 in Martha Blaxall and Barbara Reagan, eds., Women and the Work- ness within the occupationsI system is un- place: The Implications of Occupational Seg- likely to end, either. regation. Chicago: University of Chicago Press. Kanowitz, Leo 1969 Women and the Law: The Unfinished Revo- ACKNOWLEDGMENTS lution. Albuquerque: University of New Mex- ico Press. I thank Barbara Reskin and Louise Tilly Katzman, David M. for their helpful comments on an earlier ver- 1978 Seven Days a Week; Women and Domestic

176 170 KAREN OPPENHEIM MASON

Service in Industrializing America. Urbana: Demographic and Economic Factors Govern- University of Illinois Press. ing Its Growth and Changing Composition. Kessler-Harris, Alice Berkeley: University of California, Institute of 1981 Women Have Always Worked: A Historical International Studies, Population Monograph Overview. Old Westbury, N.Y.: The Feminist Series, No. 5. Press. Reskin, Barbara, and Heidi I. Hartmann, eds. Men's Work, Women's Work: Sex Segregation 1982 Out to Work: A History of 'Wage-Earning 1984 Women in the United States. New York: Ox- on the job. Committee on Women's Employ- ford University Press. ment and Related Social Issues. Washington, D.C.: National Academy Press. Mason, Karen' Oppenheim, John L. Czajka, and Sara Arber Smuts, Robert W. Women and Work in America. New York: Co- 1976 "Change in U.S. women's sex-role attitudes, 1959 1964-1974." American Sociological Review 41 lumbia University Press. (August): 573-596. Snyder, David, and Paula M. Hudis Mason, Karel. Oppenheim, Maris A. Vinovslds, nd 1976 "Occupational income and the effects of mi- Tamara K. Hareven nority competition and segregation: A reana- 1978 "Women's work and the life course irk Essex lysis and some new evidence." American So- County, Massachusetts, 1880." Pp. 187-216 in ciological Review 41(April):209-234. Tamara K. Hareven, ed., Transitions: The Stevenson, Mary Huff Family and the Life Course in Historical Per- 1978 "Wage differences between men and women: spective. New York: Academic Press. Economic theories." Pp. 89-107 in Ann H. Matthaei, Julie A. Stromberg and Shirley Harkess, eds., Women Working: Theories and Facts in Perspective. 1982 An Economic History of Women in America: Women's Work, the Sexual Division of Labor, Palo Alto, Calif.: Mayfield Publishing Co. and the Development of Capitalism. New York: Waite, Linda J. Schocken Books. 1981 "U.S. women at work." Population Bulletin Newman, Winn 36(2). 1976 "The policy issues. Presentation III." Pp. 265- 'Williams, Gregory 272 in Martha Blaxall and Barbara Reagan, eds 1979 "The changing U.S. labor force and occupa- Women and the Workplace: The Implications tional differentiation by sex." Demography of Occupational Segregation. Chicago: Uni- 16(Febsitia'ry):73-87., versity of Chicago Press. Williams, John E., and Deborah L. Best Oppenheimer, Valerie Kincade 1982 Measuring Sex Stereotypes: A Thirty-Nation 1970 The Female Labor Force in the United States: Study. Beverly Hills, Calif.: Sage Publications. .

1 '7 e

Work Experience, 1 0Job Segregation, and Wages MARY CORCORAN, GREG J. DUNCAN, and MICHAEL PONZA

Women are a vital part of today's labor tune, full year. Women and men also hive force, and work is clearly an important part very , different. occupations. Treirnanand of their lives. Women constituted more than Hartmann (1981) show that 70 percent of two-fifths of the labor force in 1978, and the men and 54 percent of the women in almost 60 percent of all women aged 18 to the labor force are concentrated in occu- 64 were employed in 1978. Almost all women pations dominated by their own sex. Unlike work at some point in their lives, and their men, women areheavily concentrated in a earnings are often necessary to ensure ad- few job categoriessecretarial work, sales, equate family support. In 1978 nearly two- teaching, nursing, and various service oc- . thirds of the wo'men working were either cupations. presently unmarried or married to men The most prominent economic explana- earning less than $10,000 per year (in 1977).' tion linking labor supply patterns and wages There is considerable evidence that men's is human capital theory.2 Human capital it- and women's work participation patterns self is 'defined as worker skills or qualifica- differwith men working continuously after tions acquired through schooling or on-the- completing school and women moving in job training. An individual worker's stock of and out of the labor force to accommodate human capital can be increased by the proc- family and child-rearing duties. Women earn ess of investment. Investmentshave an op- considerably less than men do. Since 1930 portunity cost (in terms of forgone earnings the median salary of full-time, full-year as well as of the direct costsof training) and women workers has beenabout 60 percent a return (in the formof higher subsequent of the median salary of men who work full earnings). Human capital theory particu-

This paper was supported by a grant from the Alfred P. Sloan Foundation. Human capital theory is quite similar to the func- These figures are taken from U.S. Department of, tionalist theory of Davis and Moore. It has been argued Labor, Women's Bureau, 10 Facts on Women Workers, that this theory underlies much of the empirical work Washington, D.C., August 1979. in social stratification (see Horan. 1978).

171 g

9 172 MARY CORCORAN GREG .DUNCAN AND MICHAEL PONZA lady emphasizes investment in formal WORK HISTORY, WAGES, AND JOB schooling (Becker, 1975) and in on-the-job SEGREGATION: THEORETICAL MODELS tra'ning (Mincer, 1974). Workers are presumed to choose freely among jobs with dif- Work Experience and Earnings ferent amounts of training, and wages are presumed to reward pastnvestments in ed- In the human capital model, investments ucation and training in a similar we./ for all in on-the-job training are considered tc be workers. critical determinants of wages (see Becker, In recent years human capital theory has 1975; Ben-Porath, 1967; Mincer, 1974; and been expanded to deal with the structure of Rosen, 1972). On-the oh training has a cost, female wages. Some of its proponents have since time spent in training is assumed to argued that the sex division of labor within be time diverted from production, and pro- the home generates sex differences in pat- duction presumably determines earnings. terns of investment in work-related human On-the-job training also has a return in the caliital and that this in turn generates sex form of higher later earnings. The following difkrences in Wages and the sex segregation function describes this hypothetical rela- of occupations (Mincer and Polachek, 1974; tionship: Polachek, 1976, 1979, 1981; Mincer and Ofek, 1-i 19h2). These arguments have focused par- E, = + E rC, = Y, + C1, (1) ticularly on sex differences in patterns of I = 0 labor force participation. This paper investigates human capital where E, is earnings capacity in year t, E, theory's predictions about the relationships is earnings that would be received in the between patterns of work, wages, and job absence of any postschool training, C, is the segregation. The paper is in four major parts. dollar cost of investments in human capital The first summarizes human capital thet:.- in the year, Y, is earnings in the tth year, retical models. In the next section we use C, is dollar cost of investments in the t" 13 years of data from the Panel Study of year, and r is rate of return to investments Income Dynamics (PSID) to describe and in human capital. compare men's and women's patternsof work If we assume that total benefits of an in- participation. These comparisons focus on vestment increase as the payoff period in- aspects of lifetime work experience that are creases and that the marginal costs of in- hypothesized to he important for women vestments are upwardly sloping in a single duration of work and nonwork periods, the time period, it can be shown that a profile extent to which labor market experience in- of investment ratios (C "Ed that are large at volves part-time work, and the sex typing first and then decline over time maximizes of experience (i.e., the extent to which past the present value of expected lifetime earn- experience was in female-dorninated occu- ings (see Ben-Porath, 1967). That is, the pations). The third section reviews past re- proportion of one's earnings capacity in- search en the extent to whicl. different as- vested in on-the-job training will be high in pects of work experience influence wages the early years and then will decline rapidly. and sex typing of a person's current job, and The human 'capital model assumes that it also includes the results from our own workers freely choose among a variety of analyses of these issues based on 13 rears jobseach with a different combination of of PSID data. Finally, we discuss the thetraining and productive work. It generally oretical and policy conclusions drawn from views training and productive work as mu- our reviews of past researchand from our tually exclusive activities, and, thus, ac- owr:scarch. cording to the model, employers will pay

179 WORK EXPERIENCE JOB SEGREGATION AND WAGES 173

less for jobs with training than for similar Mincer and Ofek (1982) have since re- jobs that do not provide training. An im- vised this initial model to account for "res- plication of the model is that wages grow toration" or "repair" of depreciated human with experience because workers are ac- capital. They argue that the "reconstruction quiring additional skills as they increase their of (previously eroded) occupational skills is experience and so their wages grow not just more efficient than the construction of new because of seniority. human capital." That is, it costs less to repair human capital than to build it. This restoration phenomenon leads Mincer and Ofek Labor Force Withdrawals and Wages to distinguish the short-run and long-run Mincer and Polachek (1974) extend the consequences of nonparticipation. In the human capital model to account for the pos- short run (say, the first year following an in- sible depreciation of human capital that may terruption), one would expect sharply lower result from the discontinuity of women's work wages than those received just pr'or to thc. experience. They argue that during periods interruption, followed by a period of rapid of labor force withdrawal for child-bearing wage growth during which human capital is and child-rearing, prolonged nonparticipa- restored. Thus, the long-run effects on wages tion in the paid labor market can cause the of nonwork time may be considerably smaller skills acquired at school and work to become than the short-run effects, less valuable. Since the empirical work of Mincer and The following function adjusts the basic Ofek and our own replication of it show that human capital wage model to account for wage "rebound" following an interruption is depreciation or obsolescence effects: an important phenomenon, it is useful to e- consider alternative explanations of it. Cor- Es = E, + E (rCi 8A), (2) coran (1979) and Corcoran and Duncan (1979) suggest that time out may lead to a tem- where E,,Et, Ea, r, Ci are defined as in Eq. porary reduction in wages because of tem- (1), 8, is the depreciation rate of human capi- porary mismatches between worker skills and tal in year i, and Ei is earnings capacity in jobs. Women workers lack complete infor- year i. mation about job opportunities when they The total benefits of investments in on- do return to the labor force, and it takes the-job training increase with the length of time for them to discover jobs that are best the payoff period but decline with the length matched to their skills. Employers also have of periods of nonparticipation thai follow in- imperfect information about the productiv- vestments. This suggests that optimal in- ity of their new empleyees, and the learning vestipQnt patterns will differ depending on process for them is time-consuming.3 One the continuity of market activities. Contin- uously employed workers should concentrate investments early in their careers. Morgensen (1978), Jovanovic (1979), and Prescott Workers who interrupt their work careers and Visscher (1980), for example, explain that earnings rise with experience with a firm because firms learn wil! defer investments in on-the-job training about worker productivities in various Jobs (instead of until they reenter the labor marketcter workers' acquiring skills through experiende). This completing these activities so as to minimize learning process results in the more senior workers the loss from depreciation. Since such work- being matched more accurately to jobs commensurate ers have a shorter payoff period, their over- with their skills than less experienced workers. Better job matches allow the senior workers to exhibit higher all volume of investment should be lower productivity tri average, and if the market rewards than that of workers who remain contic a- productivity, these diffeiences may account for their ously in the labor force. :iigher average earnings.

180 174 MARY CORCORAN, GREG1. DUNCAN, AND MICHAELPONZA common mechanism forthis sorting process 1976) do not make unambiguous predictions is to hire new workers in at low wagesbut about the effect of part-time work on human then to promote them rapidly as they suc- capital investment and wages, but there are cessfully complete their probationary pe- reasons for believing thatless training is ac- riods. In neither of these cases are the work- quired in part-time work than in full-time ers "restoring"depriviated skillsin the work. First, because part-time work means Mincer -Ofek sense. Wage increases accom- fewer hours in the labor market than full- pany improvedinformation of employees time work does, women who expect towork about their job opportunities or improved part time in the future have ashorter ex- information of employers about the produc- pected work life and hence less incentive to tivity of their employees. invest in on-the-job training. In this case Note that Mincer and Ofek have altered both the overall volume of investment and the original Mincer and Polachek model the rate of investment would be lower for substantially. Since depreciated human cap- those who plan to work part time than for ital can be restored, it no longer follows that those who plan to work full time. If current intermittent workers will necessarilydefer part-time -work patterns are associated with investments in on-the-job traininguntil all thelikelihood of future part-time work, then interruptions are over. This decisionwill de- current part-time workers will bemaking pend on the relative sizes of thedeprecia- fewer investments. Second, if employers tion and restoration effects.{Similarly, the suspect that part-time workers are morelikely relative sizes of these two effects will also to leave than arefull-time workers, they determine the long-run wage costs of labor might restrict training opportunities in part- force withdrawals. If these long-run costs time work. Employers would be mostlikely are small, thendepreciation may account for to restrict opportunities forfirm-specific little of the wage gap between menand training. Finally, just as it is arguedthat women. skills depreciate during Periods of nonwork, skills could depreciate more (or appreciate less) during part-time work than during full- Part-Time Work Experience and Wages time work, since part-time workinvolves Women are considerably more likelythan fewer hours of work (i.e., more hours of men are to work in part-timejobs, a fact that nonwork). The depreciation from nonuse may lead toconsiderable differences in the would be greater if the nature of part-time amount of on-the-job training women ac- work precluded workers from maintaining quire and, therefore, in theirrelative wage their market skids.If formal training is growth. The most general human capital scheduled when part-time workers are not theories (Heckman, 1976; Blinder and Weiss, at work, then there will be less wagegrowth resulting from the acquisition of new skills for them. Two sources of data with crudedirect A qualification suggested to us byJacob Mincer is measures of on-the-job trainingdo show a necessary here: For the intermittentworker, each in- positive relationship between workhours and terruption carries with it a positiveprobability of not training. Duncan and Hoffman (1979) found returning to the labor market. Thus,the exmcted payoff period is diminished by more than the interruption with the 1976 wave of the PSID that adult each time it looms. So although wage leisdue to de- workers aged 18 to 64 who worked less than preciation can be made up. the intermittentworker's SO hours per week reported training periods decision about whether and when to invest willdepend attached to their jobs that were only about jointly on the relative sizes of the depreciationand half as long as those of workers workingbe.. restoration effects and on the probabilityof returning tween 40 and 50 hours. Staffordand Duncan to work.

181 1 WORK EXPERIENCE, JOB SEGRECATIOkAND WAGES 175

(1979) found qualitatively similar, although withdrawals from the labor force, then it is less statistically significant, differences in rational for women who expect to taketime training by labor supply forworkers in the out from the labor force towork in fields 1975-1976 Time Use Study. where there is less chance of atrophyi.e., in fields with low depreciation ratesbut also with low return? to experience.Thus, such Labor Force Withdrawals and Job women will experienceless atrophy than will Segregation; Human Capital Explanations women who expect morecontinuous work The 1974 Economic Reportof thePresi- participation. By selecting jobs that are easy dent speculated that sex differences in pat- to leave and reenter, women canthus more terns of work participation maybe the cause easily combine the dual demands or career of the sex segregation of jobs. Thisline of and family. Polachek's model can explain sex reasoning has been extensivelydeveloped segregation only if typically "female"jobs by Zellner (1975) and Polachek (1976,1979, are those wherethere is the least atrophy. 1981).8 Since Polachek's explanationsub- Note that if depreciated skills can be re- sumes Zellner'smodel, we will concentrate stored (as Mincer and Ofek argue), this on his model inthe following discussion. weakens the force of Polachek's arguments. Polachek (1981) defines atrophy as theloss Webster's clefinesatrophy as "a wasting of earnings potential that occurswhen skills away or progressivedecline." Thus, the cas- are not continuouslyused. He shows that if ual reader might assume that Polachek's the cost of labor force withdrawals(the atro- atrophy rate is equivalent to Mincer and phy rate) varies across occupations,and if Polachek's depreciation rate. But atrophy, lifetime labor force participation differs across as defined byPolachek, picks up two things individuals, then a worker will choose "that depreciation (i.e., reduction in work skills occupation which imposes thesmallest pen- due to nonuse) and forgoneappretation (i.e., alty, given his desired lifetiine participa- the loss in expected earnings growth due to tion." This model treats the "lifetime as a missing a year of work).' unit." Thus, this model implicitly assumes Depreciation and the gruwth of earnings that workers tend to work in the same sort with experience are quite different proc- of occupation throughout theirlives or at esses. Depreciationimplies that the level of least over long periods of time. Lifetimework work skills is lower following an interruption participation is assumed 'to beexogenously determined, and atrophy rates areassumed to vary cross occupations. We are grateful to Siv Gustafson for first pointin4 This model provides a human capital ex- out this distinction to us in a personalconversation .n planation for the sex segregation of thelabor 1978. England (1981, 1982) is the first authorwho cleay market. If work skills do atrophyduring makes this distinction in a published paper. England first referred to the loss in c:-,..ected earningsgrowth as "forgone appreciation," andprovides an excellent discussion of Polachek's models. 5 England (1981, 1982) provides an extensivediscus- Zellner's explanation of the sex segregation of oc- England sion ei .nese models. This section isinformed by her cupations rests solely on forgone appreciation. (1981) points out: "Zellner assumes that occupations work, initial salaries 6 Polachek notes that "this assumption canbe relaxed can be divided into those that offer.high initial by posing the problem within a dynamiccontrol frame- and flat earnings profiles and into those wilklow work" but goes on to say that "evenwithin such a salaries and steep earnings profiles.Woinekthecause likely framework the same conclusions hold for occupations of their shorter expected work lives. will be more with chosen at a given stage of the life cycle"(Polachek, to maximize lifetime earnings in the occupations 1981, p. 64). Note that this relaxationstill) implies-oc- high initial salaries and flat wage growthi.e., in cupational immobility over a life-cycle stage. male' occupations." 182 176 MARY CORCORAN:GREG 1. DUNCAN. AND MICHAEL PONZA than it was just prior to that interruption. If mittent work participation, and part-time this earnings loss is long lasting, it is obvious work. Most job segregation models, human why women who expect prolonged labor force capital or otherwise, implicitly assume con- withdrawals should enter fields with low de- siderable immobility between "female" and preciation rates. Polachek's model also im- "male" jobs. As a first step toward testing plies that women who expect prolonged these models, we use 13 years of data from withdrawals should enter fields with high the PSID to assess the accuracy of these initial salaries but fairly flat earnings growth 'descriptions of women's work behavior and rates. This second decision only makes sense occupational immobility. if we assume that, all else equal, jobs with high earnings growth pay less initially than Patterns of Labor Supply, 1967 to 1979 jobs without such earnings growth. (Eng- land [1981] makes thii point quite clearly.) Every year PSID respondents report their To summarize, Polachek provides an in- own and their spouse's work hours. Cor- genious human capital explanation for job coran et al. (in press) examined labor supply segregation. If Polachek's general model is for adult men and women, aged 23 to 47 in correct, then women who anticipate pro- the first year of the pan !, who lived in their longed nonwork time should work in fields own households.8 As t1/2 i.ected, there were with low atrophy rates and should experi- dramatic differences between the sexes in ence less depreciation and less wage growth the frequency and regularity of work and in than do otherwise similar workers: Pola- the extent of part-time or part-year work. chek's model also hasiveral implications Differences between the races were much for the nature of typicaily female and male less dramatic within the groups of men and occupations. First, since women's choice of women. Black women acquired more ex- a "female" or "male" occupation reflects life- perience than white women did, while black time participation plans, we would expect men acquired less of it than did white men. that the sex typing of women's occupations Between 70 and 80 percent of the two change little over a prolonged period of time. groups of women were absent from the labor In additio I, we should find that deprecia- force for at least 1 of the 13 years.° The tion and /car earnings growth will be lower in "female" occupations and that women who expect discontinuous careers will choose In terms of the PSID sample, this group consists "female" rather than "male" occupations be- of all individuals who were household heads or wives cause discontinuity is penalized less. Be- in each pf the 13 year- s. c'iminated from this analysis cause of this choice, women with discontin- are children and the small group of other relatives of uous work careers will be concentrated in the household head (e.g., brother or sister). The lower age restriction was imposed to avoid sample selection "female" occupations. Occupational immo- problems associated with the decision to leave the pa- bility and low wage growth are also pre- rental home and form one's own household. The upper dicted by any job segregation model that age restriction eliminated from the sample individuals presumes that women are locked into a set who would i.ave reached the early retirement age of of female-dominated jobs that do not pro- 62 by the end of the panel period. The results are reported in Corcoran et al. (in press). vide productivity-enhancing experience, We used 250 hours during a calendar year to define whether an individual was in or out of the labor force during that year, and we use ' 500 hours to separate WOMEN'S WORK AND OCCUPATIONAL part- from full-time workers. This procedure has its HISTORIES disadvantages. An individual with a 40-hour-per-week job who drops out of the labor force altogether for six The human capital models summarized months out of a calendar year will be classified as a above predict that women's low wages result part-time worker during that year without having a from a low overall volume of work, inter- spell of nonwork. In one sense this individual was a 183 WORK EXPERIENCE OB SEGREGATION AND WAGES 177 comparable fractions for white and black men job is only .39 for women aged 30 to 44 years were about 10 and 16 percent,respectively. in 1967. This suggests there may b.consid- Even when they did work, women were erable mobility between "male" and "fe- much less likely to work full time. Less than male" job sectors. one-tenth of these adult women worked full- We further tested this assumption of in- time during the entire 13-year period. tersectoral immobility by calculating pat- The total volume of work experience ac- terns of occupational segregation over the quired by men was much higher than for years 1975 to 1979. Our measureof occu- women. On average, menworked in almost pational segregation is based on 2.-digit oc- twice as many of the years as women did, cupation and 2-digit industry categories. in nearly twice as many weeks as women Thus, it has the advantage of accounting for did, and for nearly three times as many hours both occupational and industrial segmen- as women did duringthis period. tation by sex. L° We define a female-domi- The part-time nature of the work of women nated job as an industry-occupation group was highlightedwhen we examined hours with more than 50 percent women workers. worked per week and weeks worked per Industry-occupation groups with less than year during-the workspells. The average 50 percent female workers are designated work week of men exceeded 40 hours, male-dominated jobs. amounting to 46 hours for white menand To investigate the dynamics of job seg- 43 hours for black men. In contrast, white regation over the period from 1975 to 1979, and black women averaged 36 hours per week we selected a sample of womenaged 23 to during work spells. Similar differences 57 in 1975 who worked in the first and last showed up in the number of weeks worked years of that period andwho may or may per year, with men averaging47 weeks and not have wor!ted during the three years in women averaging 42weeks. As a result, the between." About 70 percent of white women total number of hours averaged by men dur- workers held female-dominated jobs in 1975. ing their work spells w almost twice as If job segregation was completely rigq.d, then high as for women during their spells. we would expect to observe that so mefrac- tion spending all of their working years in jobs dominated by women. Table 10-1 shows Patterns of Occupational Segregation, that this is clearly not the case. Only half of 1975 to 1979 the white women spent all of their working Both Polachek's (1981) model and seg- years in the five yearsbetween 1975 and mented labor market models assume little 1979 in jobs dominated by women. Less than mobility between "male" and "female" oc- one-sixth of these white women spent all of cupations over a prolonged period of time. their working years in jobs dominated by Yet England (1982) reports that the corre- men, leaving more thanone-third who lation between percent female in detailed switched job types at least once. Switches census occupationcoding of first job and between female- and male-dominated jobs percent female in detailedcoding of 1967 for black women were almost as common--

full-time worker and in another sense this indiVidua; experienced a spell of nonwork during that year. Our 1° Our procedure for determining whether a given measure considered part-timeworkers to be those either job was "female - dominated" or "male-dominated" is working a limited number of hours per week or those detailed in Corcoran et al. (in press). working during only part of the year. Our measure of " As before, the sample consists of household heads nonwork spells required that such spells be long enough -and wives in this age range and thus excludes a small to take an individual away from work forvirtually an number of adults who are related to the head of the entire year. household in some other way. 134 178 MARY CORCORAN, GREG .EDUNCAN, AND MICHAEL PONZA

TABLE 10-1Dynamics of Occupational Segregation Subgroup Women Who Worked at Least 250 Hours in 1975 and 1979 Men Occupational Change White Black All White Black All Fraction spending all working years in .51 .61 .52 .07 .18 .08 female-dominated jobs' (871) (538) (1,409) (1,563) (606) (2,169)

Fraction spending all working years in .15 .09 .14 .79 .67 .78 male-dominated jobs (871) (538) (1,409) (1,563) (606) (2,169)

Fraction switching at least once in .34 .31 .34 .15 .18 .16 either direction (871) (538) (1,409) (1,563) (606) (2,169)

Fraction of those in female-dominated Jobs initially who switched to male- .31 .25 .30 .60 .37 .56 dominated Jobs (647) (423) (1,070) (207) (106) (313)

Fraction of those in male-dominated jobs initially who switched to female- .44 .55 .45 .43 .13 .09 dominated jobs (224) (115j (339) (1,300) (456) (1, 756) NOTES: Table reads: 51 percent of the 871 white women who worked at least 250 hours in 1975 and 1979 spent all of their working time in female-dominated jobs. The number of observations is given in parentheses below each estimate. a A job is designated as female dominated if the percentage of women comprising it is greaterthan or equal to 50. Otherwise it is designated as male dominated. SOURCE: Panel Study of Income Dynamics.

31 percer of the black women were coded In addition, our classification procedure for as switchii.z from one type tothe other. identifying female- and male-dominated oc- These figures on the extent of switching cupations is a crude one and undoubtedly between job types mix together both kinds ,misses some true switches that would be of changes. A more relevant statistic on the caught with a more refined set of occupa- issue of whether women who take female- tional and industrial codes. On the other dominated jobs remain in them is the frac- hand, errors in the coding of occupation and tion of women who began in female-domi- industry may create the appearance of a nated jobs and switched out of them. That switch when in fact there was none.13 It is fraction is 31 percent for white women and impos4ible to say whether the net effect of 25 percent for black women. these considerations is to increase or to de- These figures on 'switches between job crease the estimated extent of switching be- types deserve careful scrutiny. On the one hand, they are likely to understate the true amount of movement between job typesbe- some did not work in some of the middle three years, cause the time span overwhich such changes giving them fewer thLa five years in which a switch ci;r be observed. can be observed islimited to only fiYears. I2 13 Appendix Table B. I in Corcoran et al. (in press) sheds some light on this by showing minparable mobility figures for the case when male-dominated jobs 12 Women who worked continuously in t.4 same sec- are defined as less than 40 percent female, and female- tor between 1975 and 1979 but switched in 1980 or dominated jobs are defined as greater than 60 percent had switched in 1974 are classified here as persistent female. Less mobility is found with this more restric- residents in one sector. Also, while most of the women tive definition, but the extent of mobility is still sub- in this sample worked in every one of the five years, stantial.

185 WORK EXPERIENCE013 SEGREGATION AND WAGES 179

tween job sectors. However, it is almost cer- fected by labor force withdrawals than are tainly true that the extent of switching is wages of women in a broader age range. substantial, a fact that is inconsistent with A.rc:.,ent paper by Mincer and Ofek (1982) Polachek's assumption of occupational im- suggests that some of these inconsistencies mobility and with any other labor market in past research arise because cross-sec- model based on rigid segmentation by sex. tional analyses tend to confound the short- This result also suggests that analysts should run and long-run effects of nonparticipation. be wary of using a woman's current occu- Mincer and Ofek use eight years of National pation as a measure of her past occupational Longitudinal Survey (NLS) data to explore history. how time out of the labor force affected wage growth for a sample of women aged 30 to 44 in 1967 who were married sometime dur- WORK HISTORY AND WAGES: ing 1967 to 1974. They found a large short- EMPIRICAL EVIDENCE run loss in wages immediately following an interruption (ranging from 3.6 to 8.9 percent per year out of work), followed by a The Depreciation Effect period of rapid wage growth. Estimates of Empirical evidence on whether deprecia- long-run wage losses were moderae (0.4 to tion lowers wages consists mostly of cross- 1.1 percent per year out of work). A repli- sectional data where earnings of different cation and extension of this analysis was con- individuals with different work histories are ducted by Corcoran et al. (in press) using compared, after statistical adjustment to make 13 years of information from the PSID. Short- the individuals as similar as possible. This run depreciation effects were estimated to evidence has produced contradictory results range from 2.5 to 4.7 percent per year de- about the size of the depreciation effect pending on the exact form of the model and i.e.,about the extent to which wages de- on the definition of the sample when the dine with time out of the labor force once age range was identical to the one used by one controls experience and tenure. Mincer Mincer and Ofek (30 to 44), and the effects and Polachek (1974) reported that 1967 wages were estimated to be similar in magnitude dropped by 1.2 percent per year out of work when the age range was extended to be- for white married women aged 30 to 44 with tween 23 and 47 years. Long-run deprecia- children. Sandell and Shapiro (1978) repli- tion was estimated to be between 1.0 and cated the Mincer-Polachek analysis after 1.5 percent per year in the replication, with correcting for coding errors in women's re- son' of these coefficients not statistically ports of employment behavior. They re- significant at conventional levels. ported that wages declined only 0.4 percent This recent work reconciles the disparate per year of nonparticipation and that this estimates of depreciation. The past analyses effect was insignificant. Corcoran (1979) rep- of wage effects of withdrawals based on cross- licated the Mincer-Polachek analysis for a sectional data are likely to pick up both short- national sample of wives aged 30 to 44 with run and long-run effects. Married working children, taken from the 1975 PSID and ob- women aged 30 to 44 who have interrupted tained similar results to those of Mincer and work are likely to have recently returned to Polachek. But Corcoran (1979) and Cor- the labor force,- and so short-sun effects may coran and Duncan (1979) also reported that have a large weight in analyses run on this the decline in wages was much smaller (0.6 group (e.g., that of Mincer and Polachek. percent per year out of work) for working 1974). Analyses run on women. in a broader women in a broader age range (18 to 64 age range (e.g., that of Corcoran, 1979; and years). These results suggest that wages of that of Corcoran and Duncan, 1979) are likely married women aged 30 to 44 are more af- to put more weight on the long-run effect 180 MARY CORCORAN, GREG] . DUNCAN, AND MICHAEL PONZA and so will provide lower estimates of de- intermittent workers will concentrate in fe- preciation. male jobs. The Mincer and Ofek (1982) estimates of restoration came from the estimation of a The Restoration Effect cross-sectional wage equation. Corcoran et In their original paper, Mincer and Po- al. (in press) were able to use more complete lachek (1974) suggested that the optimal information about the amount and sex typ- timing of investment in on-the-job training ing of work experience before and after work would differ depending on the continuity of interruptions and also estimated a wage- market activities. In particular, workers who change equationa specification with sev- interrupt their work careers for nonmarket eral statistical properties that make it pre- activities will defer investments in on-the- ferred to a wage-level equation. Since this job training until they reenter the labor mar- work has only recently been completed and ket after completing these activities so as to addresses many of the important issues con- minimize the loss from depreciation. Min- sidered in this paper, we summarize our cer and Ofek (1982) haveconsiderably re- analysis here. Readers interested in the devised this hypothesis by arguing that "details are referred to Corcoran et al. (in press). preciated" or "eroded" human capital can We selected the adult women in the PSID be cheaply and rapidly "restored" soon after sample and used the 13 years of work history labor market entry. They show that postin- obtained for them. We developed our wage terruption wages grow at rougi.ly 2.5 per- equation by identifying the first (F) and last cent per year of experience, on average,and (b) wage observation for all women who that growth rates in the first year following worked ileast 2 of the 13 years." A cross- an interruption rangefoam 5.8 to 6.4 per- sectional wage equation at time F would be cent per year depending on the exact spec- of the form: ification of their model. This growth rapidly erases estimated short-termlosses from the InWir = aoF. airSa. "217e0iF depreciation associated with short spells out Ct3phoiF ITFZIF OiF, (3) of the labor force and is much larger than the wage growth of comparable continuous where WiF is the wage rate of theith indi- workers. They further demonstrate that vidual in time F, aoF is a constant in time growth in tenure accounts for less than half F, S4F is the education level of the Ith indi- of this wage growth and interpret this to vidual in time F, eow is years of work ex- mean that the remainder isdue to growth perience for the Oh individual in time F, how (repair) of general training. is years of nonwork for the ith individual in The nature and causes of wage rebound time F, 'Ca. is a vector of observed produc- following work interruptions are crucial ele- tivity-related characteristics for the ith in- ments in understanding the wage conse- dividual in time F, ZiF is a vector of unob- quences and job choice offemale labor sup- servable individual-specific productivity- ply patterns. If depreciation is, quickly related characteristics, workplace charac- repaired, it longer follows that intermit- teristics, and labor market differences for tent workers will defer investments in on- the ith individual in time F, and Ow is the the-job training until all interruptions are completed. Thus, the observation that the wages of women grow more slowly inthe 14 Throughout this section we use 250 hours as the years following the completionof schooling cutoff point to distinguish individuals in the labor force because of the reduPed incentives to invest from those out of the labor force. The sensitivity of in human capital may no longer hold. It also these results to change in this definition are detailed weakens the plausibility of the reasoning that in Corcoran et al. (in press).

'187 WORK EXPERIENCE, JOB SEGREGATION, AND WAGES 181 stochastic disturbance term for the Ph in- Although the dependent variable in Eq. (6) dividual in time F. is wage change rather than wage level, the PSID information on the work history be- parameters on the experience variables (a4 tween F and L can be used to distinguish: a7) correspond to the parameters in the eL, years of work experienceaccumulated cross-sectional Eq. (4). The key advantage between F and L since the last completed to the change formulation is that estimates interruption; e*, years of work experience of these nnrqmeters are free from the sta- accumulated between F and L prior to the tistical pruilems caused by retrospective re- most recent completed interruption; hL, years ports and by unchanging, unmeasured var- of nonwork between F and L during most iables correlated with the included recent completed interruption; and h*, years (measured) explanatory variables. An addi- of nonwork accumulated between F and L tional advantage is that many more women prior to most recent completed interrup- meet therequirementsof working at least tion. 2 of the 13 years than work in a single year, The cross-sectional wage relationship at and, therefore, selection bias problems are time L (allowing the parameters to chazge) much lest; severe in estimating change Eq. is given by (6) than in estimating a cross-sectional equation.35 inWs, = aoL a1LSIL + Table 10-2, columns 1 and 2, shows es- a2Leou, etaLhow timates of wage-change Eq. (6), which is the a4e*i + aseu + longitudinal analogue to the Mincer-Ofek ada*1 + ct7hL, + cross-sectional equation. The work segment following the most recent interruption (el) PLXic (4) was entered quadratically toallow for a more The short-run depreciation and rebound rapid growth at first. Results show the es- effects are given by parameters a, and a5, timated rate of wage growth immediately respectively. Subtracting Eq. (3) from Eq. following the last interruption to be a little (4), suppressing the subscript i, denoting over 5 percent per year forwhite women changes from F to L as "A,- and adding and and 8 percent for black women, with the subtracting otiLSF, P.LXF, and ITLZF results rate of growth declining to zero afterabout in the following general equation for wage 10 years for both groups, which is close to change: the maximum observed value for the eL variable in the sample. Depreciation during the AinW = Aao + aiLAS + most recent interruption is estimated tobe- ACE SF + 4101.2e0E tween 4 and 5 percent per year, so theinitial Aa3hoF a4e* + aseL + wage rebound following an interruption more ei6h* + ce7hL + + than makes up for the wages lost during a year out of thelabor force. Ap.XE + 1TLAZ AITZb. + AO . (5) Effects of Prospective Interruptions on If one assumes that the cross-sectional ef- Wage Growth fects of the explanatory variables are invar- iant between F and L and, further, thatthe Since the profitability of investments is unmeasured characteristics remain constant affected by the length of time over which for the same individual, then the wage change equation simplifies to: 13 See Corcoran et al. (in press) for more detailed A/nW = otiLL, 4 ct,e* + a5e/, discussions of the development of the wage-change + et,h* + a7hi, + ALAX + AO (6) equation and of selection Lim adjustments.

18 182 MARY CORCORAN, GREG I. DUNCAN. AND MICHAEL PONM

TABLE 10-2Basic Wage Growth Regression Independetit Variable White Black White Black h*: years out of labor force prior to most recent .016 -.016 .013 -.015 interruption (.029) (.030) (.029) (.030) hi.: years out of labor force during most recent -.038" -046** - .035* -.070" interruption (.013) (.016) (.015) (.020) e*: years in labor force prior to most recent reentry .012t .030* .012 .623* (.007) (.008) (.009) (.011) et: years in labor force during most recent spell .052' .080 .051t .077 (.023) (.026) (.027) (.032) -.0027t -.0041* - .0025 -,0045 (.0016) (.0019) (.0015) (.0022) NT79: Did not work in 1979 - - - .017 -.124 - - (.099) (.103) NT79*hi. - -.030 .059t - - (.029) (.032) NT79*eL - - .043 I (.056) ( - ,ii:1 NT79*eL2 _ .0009 _ - 'q .'ti; ) (.0090) R2 (adjusted) .021 .057 .024 .035 Number of observations 837 521 837 521 NOTE: Standard errors are in parentheses. Significant at .05 level. Significant at .01 level. t Significant at .10 level SOURCE: Panel Study of Income Dynamics.

benefit's are received, the human capital ing at the end of the 13-year period, w we can model predicts that otherwise identical test directly whether such workers' jobs pro- workers will invest less if they anticipate vided them lower wage growth and lower labor market withdrawals than if they do depreciation. In contrast to the self-re- not. This assumption is also implicit in Po- ported intentions of respondents used in the lachek's argument that workers who antic- articles listed above, this procedure tests for ipate time out Will choose occupations with the effects of actual labor force behavior in low atrophy rates. Sandell and Shapiro (1980) period t + 1 on wage profiles in period t. test this prop,sition by estimating whether We did this by creating a dummy variable NIS women who expected to be out of the (NT79) for whether did not work in 1979, labor force at age 35 had flatter experience- interacting this dummy with hL, eL, and earnings profiles before then than did women el, and adding these four variables to the who expected to be working PI age 35. Al- basic wage-change Eq. (6). The results of though most of their key parameter :.2.sti- this aelysis are reported in Table 10-3, col- mats are in the expected direction, nore umns 3 and 4. In general, white wctnenwho are significant at the 5 percent levelfor white did not work in 1979 had the same wage or black women. increment for additional years of experience This hypothesis is an important one fiir as women who did work ini979 and similar the human capital model; it deserves testing wage loss with time out as didotherwise in ti'e context of the wage-change models similar white women who were working in developed here. Since we know whi2h 1979. But one result for black women does women in the PSID sample were notwork- conform with human capital prediceons. 0

189 r

WORK EXPERIENCE OB SEGO CATION. AND WAGES 183

TABLE 10-3Effects of Part-Time Work and Female-Dominated Work on Wage Growth Independent Variable White Black White Black White Black

h*: years out of labor force . .023 - .005 .022 -.026 .021 -.018 prior to most recent (.029) (.030) (.029) (.029) (.029) (.029)

interruption . hi.: years out of labor force -.0341". -.037* - .035"' - .049'0' - .036"' - .048* during most recent (.013) (.015) (.013) (.015) (.013) (.015) interruption Years of full-time e* .016* .024* .007 .034." .005 .015 (.008) (.008) (.007) (.008) '(.008) (.001) Years of part-time e* .001 .018 -.002 .023* -.003 .008 (.011) (.011) (.011) (.011) (.012) (.011) Years of full-time el, .029"" .030** .062s' .037 (.008) (.010) (.021) (.024)

(Years of full-time eL)- squared -.0036* - .008 (.0018) (.0021)

Years of part-time eL -.001 .017 - .021 .150* (.010) (.013) (.028) (.029)

(Years 9f part-tune e1)- squared .0016 - .014" (.0025) (.003)

Years of full-time ei, in male- .026* .008 dominated jobs (.012) (.016)

Years of part-time eL in male------.009 .024 dominated jobs (.019) (.028)

Years of full-time el, in female- .020* .021* rinminated jobs - - (.009) (.011)

Years of part-time el. )n female- - .004 ,6-.003 dominated jobs (.012) (.014)

Number of cases 837 521 837 521 837 521 112 (adjusted) .025 .049 .028 .096 .023 .044 * Significant at .05 level. Significant at .01 level. SOURCE: Panel Study of Income Dynamics.

Black women who were not working in 1979 that of otherwise similar black women who exhibited no wage loss during prior labor were working in 1979. force withdrawals.16 Their past wage return to experience, however, did not differ from Part-Time Work Experience and In:ermittency Corcoran (1979) and Corcoran and Dun- The depreciation estimate for black women who can (1979) included a retrospective measure did not work in 1979 .s the sum of the coefficient on ILL term (.070) and the coefficient on the NT79*hL of experience in a dross-sectional equation term (4-.059). for a national sample of women. They report

190 184 MARY CORCORAN GREG .DUNCAN AND MICHAEL PONZA that part-time work experience was signifi- tained in each of the annual interviews and cantly less valuable than full-time work ex- do not rely on retrospective reports by either perience. women workers or their husbands. Jones and Long (1979) used data from the The results of the estimation of the aug- NLS to estimate two cross-sectional wage mented wage-growth equation are shown in equations that included interactions be- Table 10-2, columns 1 to 4. Full-time work tween work experience segments and does indeed appear to be associated with whether the work was for part of a week. significant wage growth, while part-time work Although the signs of the coefficients they does not. When the two measures of eL are estimated were consistent with the hypoth- entered linearly, the wage growth associ- esis that part-time work leads to slower wage ated with full-time experienc, is positive and growth, only two of the 12 coefficients were significant for both white anti black women, statistically significant at conventional lev- while the wage growth associated with years els. Their measures of the part-time nature of part-time work in the most recent spell of the work segments were very rough, how- of employment was insignificant for both ever, and may have biased some of the coe groups of workers. With years of experience ficient estimates. prior to the most recent spell of nonwork Corcoran et al. (in press) also investigated (e*), the full-time work variables have larger the effects of part- and full-time work on the coefficients than do the part-time variables, wage growth of women workers withwith some although these differences were not signif- simple modifications to their basic wage- icant at conventional levels. A narabolic change equation. In Eq. (6), full- and part- specification for the el, measure gives ex- time years in the e* and eL segments were pected results for whit:. womenthere is a not distinguished. Since the volume of work parabolic rebound for full-time but not part- hours was ascertained for each of the 13 yearstotime work. For black women, there is par- under investigation, that information can be abolic wage rebound for part-time work used to classify years of experience that in- a puzzling result. volved part-time work (less than 1,500 hours) and full-time work (1,500 hour., or more). Sex Differences in Work History and the Four variables were formed with this infor- Sex-Based Wage Gap mation: (1) the number of years of e* that were part-time (e*-part), (2) the numberof Two sets of analysts have extensively ex- years of e* that were full-time (e*-full),and amined the relationship between work his- (3) the number of years of eL that were part- tory and the sex-based wage gap on a na- time (e1- part), and (4) the number of years ;onally representative sample of women. of el, that were full-time (eL full). This de- Mincer and Folachek (1974) estimated that composition of e* and eL yields the following sex differences in experience and time not equation: working accounted for about 45 percent of the wage gap between employed married AinW aiLAS men and women aged 30 to 44 years in 1966. ash* + a,hL + About half of this difference was due to the a8e*-part + depreciation effect." Corcoran and Duncan a9e*-full + a ioeL-Part 17 Mincer and Polachek used data from two different a sources to make this comparison. This led to some 1.LLAX + AO. (7) inconsistencies between male and female variables. The sample of men was taken from the 1966 Survey of Eco- As with the snore basic measures of e* and nomic Opportunity (SEO). This survey does not meas- all four of these new variables arc ob- ure either work experience or tenure directly. Men

191 WORK EXPERIENCE OR SEGREGATION AND WAGES 185

(1979), using a broader age range and a more that there is a negative correlation between extensive list of work history measures, found the effect of years out of the labor force on that sex differences in work history ac- the probability of working in an occupation counted for between one-third and two-fifths and the atrophy 'rate in that occupation. of the wage gap between working white men Even if we ignore the issue of mobility and working women aged 18 to 64 in 1975. across occupations, Polachek's evidence dc ^ This occurred largely because women had not unambiguously support his hypothesis. acquired less tenure and were more likely Take Polachek's first findingthat home time to have worked part time. The depreciation affects the probabilities of currently being effect and intermittency did 'little to explain in a particular occupation. This finding can the wage gap between women and white only explain sex segregation of jobs ifwomen =men. with extensive home time were more likely to work in female-dominated occupations than were otherwise similar women without ex- Work 11i8tory and Occupational tensive home time. Polachek used his es- Segregation: Empirical Evidence timates of these effents to obtain a projected Polachek'i (1981) argument implies that population-wide occupational distribution women choose "female" jobs because such for women 30 to 44 if they had worked con- jobs penalize discontinuous labor force par- tinuously since school completion. He re- ticipation less than "male" jobs do. This ex- ported that the proportion of women profes- planation presumes that there is immobility sionals and managers (currently malebetween occupations; that depreciation and/ dominated fields) would increase and that or wage growth are lower in "female" jobs the proportion of women in household and than in "male" jobs and that women who service work (currently female-dominated expect discontinuous careers are concentra- fields) would decrease. On the other hand, ted in "female" jobs. We have already demon- his figures indicated an increase in the pro- strated that there is substantial mobility be- portion of women employed in clerical work tween "male" and "female" job sectors over (a female-dominated field), a decrease in a five-year perioda result that contradicts women employed in crafts (a male-domi- a basic assumption of Polachek's explanation. nated field), and a decrease in the propor- The empirical evidence presented by Po- tion of women in sales (an integrated field).18 lachek for this argument is indirect. Pola- chek (1981) has shown that the probability of currently working in a given occupation (defined by 1-digit census categories) is af- 18 Probably the best way to evaluate the quantitative fected by years out of the labor force and importance of this evidence is to estimate theent to which occupational sex segregation would be re- that the size of this effect differ: by occu- duced if women's home time were zeroi.e., if men pation. He has also demonstrated that the and women had the same work participation patterns. relationship between wage growth and years We estimated this by applying Duncan and Duncan's out of the labor force (home time) differs (1955) segregation index to Polachek's sample. This in- across occupations (i.e., occupations have dex measures ".he minimum proportion of one group that would have to be shifted for its occupational dis- different atrophy rates). He has further shown tribution to be equal to that of the other." The segregation index for Polachek's sample is .50. Then we calculated this segregation index on the projected occupational distribution calculated b'y Polachek under were assumed to have no interruptions. These two re- the assumption that, women worked continuously. If sults are not inconsistent. In both cases, sex differences his theory is of -red, occupational sex segregation should in work history explain a large but not major part of be considerably reduced. Under the assumption that the sex -hosed wage gap. Sample differences likely ac- women do not withdraw from the labor force, the seg- count for differences in. the importance of depreciation. regation index is .48a reduction of only 2 percent,

192 if -

f..

AND MICHAEL PONZA 186 MARY CORCORAN, GREG.' . DUNCAN,

England (1982) investigated theeffect of "inatid by short-run effects, since he restricts only ex- home time on sex typing of current occu- analysis to a five-year period and withdrawals during that pation more directly. She reportsthat the amines the effects of of most recent occupation 'period. If lost skills were rapidlyrestored sex composition and our re- and ,the sex composition offirst occupation (as Mincer's and Ofek's [1982] sults suggest), then Polachek's estimateswill. wereuncorrelated with the proportionof considerably exaggerate the lifetime costs of total time employed sinceschool completion Polachek's atro- for white women aged 36 to 50 years in1973.19 time out. Finally, note that This result does not suggest a stronglink phy estimates pick up bothdepreciation and between labor force discontinuityand the the forgone appreciation effectsof fewer years of current or first job. of experience. sex typing that occu- Now we .turn to Polachek'ssecond find- Even if Polachek's evidence effects of pations differ in atrophy rates werecorrect, ingthat the detrimental wage if home time vary acrossoccupations. He cal- this would only explain sex segregation skills and culated this by regressing thedifference be- there were less depreciation of tween 1972 and 1967 wages onhome time,lower returns to experience infemale-dom- affect wages. inated occupations. England (1981,1982) and other variables\ expected to the NIA Unlike most economic studiesof wage dif- tested this assumption using both sample of. ferentials based on the humancapital model, sample of mature women and a Polachek examines dollar changes in wages women in a wider age rangefrom the PSID. If all oc- England's analyses have the twoadvantages rather than percentage changes. looks directly at cupations had identical percentagewage de- of Polachek's analysis. She home time and sex creases per year outof the labor force, i'o- the relationship between she examines de- lachek would likely obtaindifferences in typing of current job, and dollar wage -change, with highlypaid oc- preciation and returns to experience sepa- natural log- cupations showing greaterdecline. Indeed,. rately. England regressed the dollar wage changes arithm Of wages on experience,education, in Polachek's analysis, in'current oc- for professionals, crafts- time out, and percent female are most negative level), and people, and managersthethree highest - cupation (coded at 3-digit census tested.for signicantfi interactionsbetween paid occupations. and between In order to study wagechange between experience and percent female female. She reports 1967 and 1972, Polachek mustrestrict anal- time out and percent that neither the depreciation rate nor re- ysis to women whoreported a wage in 1967 sample is cho- turns to experience wereaffected by per- and in 1972. Thus, Polachek's occupation. This is basis of work behavior. Thiscould cent female in current sen on the Polachek's ex- possibly lead to selection biasproblems when fairly strong evidence against estimating effects of workbehavior on wage planation. Both Polachek's and England'sempirical change. 2° presume A further problem isthat Polachek's es- tests. of the Polachek argument considerable immobility between occupa- timate of effects ofhome time will be dom-

women from the sample.The wage-growth analysis of '9 England (19).31) reports that Wolfeand Rosenfeld Corcoran et al. eliminates onlyabout one-fifth of the (197K) present some evidence that suggests aweak link sample, sir :e it requires only that womenwork at least between home time and sex compositionof occupation. When we rep- problem for 2 of the 13 years under investigation:* 2() This sample selection bias is a general who worked time only licated our wage-change model for women analysts of women's wages. At any point in results were inconsistent with about half of all adult women are inthe labor force. in 1975 and 1979. the results from the larger and lessrestricted sample. Restricting the sample to womenwho worked in two specific years as Polachek doeseliminates even more

193 WORK EXPERIENCE, JOB SEGREGATION AND WAGES 187

tions over time. Whether their use of cur- this with the information on work hours to Fq rent occupation as a measure of occupational create four new variables: history is appropriate depends upon the va- 1, to number of years of eL that were lidity of this presumption. As we have dem- full-time in male-dominated jobs (erfull-nul), onstrated, there, is extensive mobility be- 2., the numbers of years of eL that were tween "male" and "female" job categories. Part-time in male-dominated jobs (eL -part- This in itself is inconsistent with Polachek's md), model. But it also suggests that the use of 3. the number of years of eL that were current occupation as a proxy for occupa- full-time in female-dominated jobs (eL-full- tional history is inappropriate and may pro- fd), and vide misleading information about whether 4. the numbers of years of eL that were job choice is conditioned by expectation's part-time in female-dominated jobs (ecpart- about future work or whether experience garnered in "female" jobs results in 'lower' fd). wage growth and less depreciation than does This decomposition °fez, yields the following experience garnered in "male" jobs. equation: We used. the longitudinal nature of the A/nW =a1LAS + aioe*-part + PSID to develop more direct tests of the following two predictions of the human cap - ai ie*-full ital model: a lel:full-1nd + otiorpart -md + 1. Wage growth and depreciation are lower for work experience gathered in "fe- aoL-full-fd + male" jobs than in "male" jobs. arreepart-fd. (5) 2. Women workers with extensive time As with the basic measures of eL, these four out and frequent interruptions are more likely new variables are obtained in each of the to have concentrated their work experience annual interviews and do not rely on ret- in "female" jots. rospective reports by either women workers We find virtually no support for the predic- or their husbands. These variables also pro- tions. vide more complete measures of the extent To test the first proposition, we modified to which work experience was acquired in our basic wage-growth equation toinclude "female" jots than do measures of occupa- the sex typing of experience. In Eq. (7) we tion that are taken at a single point in time. did not distinguish whether years in the eL The results of estimating Eq. (8), shown segment involved work in male-dominated in Table 10-3, columns 5 and 6, provide little or female-dominated jobs. Since both in- support for the argument that wage growth dustry and occupation were reported for 9 is much higher for male-dominated work than of the 13 years under investigation, we could for female-dominated work, especially for classify years in eL that involved work in white women. A much more important fac- "female" and "male" jobs.21 We combined tor was whether the work performed in a

21 Industry is coded into 2-digit categories for 1971 inated" Job. (See Corcoran et al,, in press, for a more to 1979. Occupation is coded into 1-digit census cat- complete description of this procedure.) Since we only egories for the years 1971 to 1974 and into 2-digit cat- have measures of occupation and industry for the last egories for all the years thereafter. For each occupa- 9 years of the study, we do not break experience in e* tion-industry subgroup, we calculated a measure of (which tended to occur early on in 1967 to 1979) into percent female. If there were more than 50 percent "male" and "female" components. women in that subgroup, we called it a"female-dam- a

194 I

188 MA Y CORCORAN,GREG 1. DUNCAN AND MICHAEL PONZA

particular kind of job was part time or and sex typing of concurrent occupation. time. For white women a yar offull-time, None of these three measures of labor force male-dominated work was associated with a discontinuity over 1967 to 1979 had a sig- I 2.6 percent increase in hourly wages,while nificant, positive relation to the sex typing a year of full-time,female-dominated work of work experience over that period. 1.9 percent increase was associated with a SUMMARY AND CONCLUSIONS in hourly wages. The differencesbetween these two coefficients were not significant. The wage change models. of Mincer and Ofek (1982) and those from our own work _For black women a year offemale-dominated, full-time work was associated with a yield similar results. Women who drop out 2.1 percent increase in wages.This com- of the labor force have lower real wages when pares to a 0.8pendent increase for a year of they return to work than they had when they full-time, male -dog inatedwork. Again, the left work. However, the period following differenCe wasnotlisignificant. Part-time work the return is characterized by rapid wage experience, whe her in "male" or"female" growth, and the net loss in wages from drop- jobs, had no sig iificant effects on wages for ping out is small. This result reconcilesthe either blacks dr/ whites.22 apparently contradictory findings from cross- We examined whether the sex typingof sectional studies about the size of the de- women's work' experience affected the rate preciation effect, because cross-sectional of depreciation during labor force withi..aw- analyses pick up both short-run and long- als by interacting the two labor forcewith- run depreciationeffects. Short-run effects drawal measures,(h* and /It) with a measure are likely given moreweight in an analysis of the average percent female in each wom- of women aged 30 to 44 (a group likely to an's occupation-industry combination over have recently completed labor force with- the 13-year period. These interaction terms drawals) than in analyses run on women in were always insignificantwhen added to the a broader age range. wage-change Eq. (8). This suggests that de- How does this empirical evidence affect preciation does not ,differ for"male" and our understandingof the process of female 11, "female" jobs. wage determination? Theobserved wage loss The 13 years of PSID data allow adirect and rebound pattern is certainly consistent test of the second propositionthatwork- with the Mincer-Ofek story of humancapital' ers who expectdiscontinuous labor force ca- depreciation and restoration. This pattern is reers will concentrate in"female" jobs. If consistent with other stories aswellthe this hypothesis is correct, then the sex typ- job mismatch argument and the probation- ing of work experience over the years1967 ary period argument.We do not have the to 1979 ought to bepositively related to time necessary data todisentangle these argu- out of the labor force in1967 to 1979, in- ments. Regardless of the reason,the rapid termittency of work participation in1967 to rebound of wage losses after labor force 1979 (measured by number oflabor force withdrawals means that the wage losses as- withdrawals), and whether working in 1979. sociated with these withdrawals cannot ex- Results of this exercise (see Corcoran etal., plain much of the male/female wage gap.23 in press) confirm England'sfinding of no relationship between discontinuity of work 23 The fact that women work fewer years and more part-time years than men work does explain asubstan- tial (one-third to two-fins) part of the wage gapbe- 22 The .0243 coefficient estimated for yearsof part- tween men and women workers in abroad age range time, male-dominated work for black women appears (Corcoran and Duncan, 1979). Note, however, that the to be out of line with the otherresults. This estimate bulk of the sex-based wage gap differences is still unex- is based on a small number ofobservations, as reflected plained by male/female work history differences. in its large standard error.

195 WORK____XSZERIE CE IUB GE'S 189

Women are often urged to choose part - continuous work careers were no more likely time work rather than to stop work alto- to have worked at "female" jobs than were gether to keep their "hands in." Our results women with more continuouswork expe- provide little evidence that the wage con- rience. sequences of these twoalternatives differ. These results also have implications for Part-time, work experience is not rewarded models of job segregation other than the for womenparticularly for white women. human capital model. Many models of job .And the long-run wage penalties due to la- segregation either implicitly or explicitly as- bor, force withdrawals are small. The deci- sume that there is rigid segmentationbe- sion whether to work full time or pith time tween "male" and "female" job sectors and is considerably more important than is the that there are fewer promotion and/or train- choice between part-time work or no work. ing opportunities in the "female'* job sector These part-time results, like the wage loss than in the "male" job sector. The analyses and rebound results, are consistent with reviewed in this paper suggest these as- several quite different labor market scena- sumptions are likely wrong. rios -=a human capital model of lower training during part-time work, an imperfect information model, or an institutional model. REFERENCES AND BIBLIOGRAPHY Again, we do. not have the necessary data Appelbaum, Eileen to disentangle the competingexplanations, 1981 Back to Works Determinants of Women's Suc- since they each involve a different interpre- cessful Re-entry. Boston, Mass.: Auburn House tation of the same employer behavior. Publishing Co. Arrow, K.. We also investigated the -human capital 1972 "Models of Job Discrimination." In Racial Dis- models that explain job segregation as the crimination in Economic Life, edited by A. H. result of women's discontinuous work his- Pascal, pp. 83-102. Lexington, Mass: D.C. tory patterns. Such models emphasize two Heath. 1976 "Economic Dimensions of Occupational Seg- costs of discontinuous ,work participation: regation: Comment I." Signs,1, 3, Part 2 depreciation and forgone wage growth. The (Spring):233-237. 1 rapid restoration of wage losses Ii the period Beck, E.M.. P.M. Horan, and C. M. Tolbert 1978 "Stratification in a Dual Economy: A Sectoral immediately following labor force withdraw- Model of Earnings Determination." American als suggests that the first cost might be quite Sociological Review 43 (October):704-720. small. For these human capital explanations 1980a"Industrial 'Segmentation and Labor Market to hold, three things must occur:(1) there Discrimination." Social Problems 28 (Decem- ber):113-130. should be considerable immobility between 1980b "Reply to Hauser, 'Social Stratification in In- "male" and "female" job sectors, (2) wage dustrial Society: Further Evidence for a Struc- growth and depreciation should be lowfor tural Alternative:" American Sociological Re- work in "female" jobs than for work in "r..tie" view 45 (August):712-718. Becker, G. S. jobs, and (3) women with discontinuous work 1957 The Economics of Discrimination. Chicago: careers will be morelikely to choose "fe- University of Chicago Press. male" jobs than will women with continuous 1975 Human Capital: A Theoretical and Empirical work careers. Analysis, With Special Reference to Educa- tion. Second edition. NeW York: Natio.nal Bu- We find little evidence for any of these reau of Economic Research. propositions. First, there is considerable Beller, A. H. mobility between "male" and "female" job In "Occupational Segregation by Sex: Determi- typcs. We did not find that either wage pressnants and Changes." Journal of Human Re- sources. growth or depreciation varied significantly,., with the sex typing of work experience. These Ben-Porath. Y. 1967 "The Production of Human Capital and the results are consistent with England's cross- Life-Cycle of Earnings." Journal of Political 'sectional work. Finally, women with dis- Economy 75 (August):352-365. 196 190 MARY CORCORAN, GREG]. DUNCAN, AND MICHAEL PONZA

Bergmann, B.R. cupational Sex segregation.". Mimeo, Decem- 1974 "Occupational Segregation, Wages, and Profits ber. When Employers Discriminate by Race or Sex." 1982 "The Failure of Human Capital Theory to Ex- Eastern Economic journal 1 (AprilJuly):103- plain Occupational Sex Segregation." Journal 110. of Human Resources (Spring). Blau, F.D. cronau, R. 1977 Equal Pay in the Office. Lexington, Mass: b,C. 1974 "Wage ComparisonsA Selectivity Bias." Heath. Journal of Political Economy 82, 6 (November/

Blau, FAD., and C.L. Jusenius . December):1119-1143. 1976 "Economists' Approaches to Sex Segregation Gwartney-Gibbs, P. in the Labor Market: An Appraisal." Sig is: 1981 Married Women's Work Experience: Intermit- Journal of Women in Culture and Society 1, tency and Sex-Typed Occupations. Ph.D. dis- 3, Part 2 (Spring):181-199. sertation, University of Michigan. Blinder, A., and Y. Weiss Hanushek, E.A., and J.M. Quigley 1976 "Human Capital and Labor Epply: A Synthe- 1976, "Life -Cycle Earnings Capacity and the OJT In- sis." Journal of Political Economy 84(3):449- vestment Model." Discussion Paper 7904. 472. Rochester, N.Y.: Public Policy Program, Uni- Cain, G. versity of Rochester. 1976 "The Challenge of Segmented Labor Market Hauser, R.M. Theories to Orthodox Theory: A Survey."Jour- 1980 "COmment on Beck et al., ASR, October 1978, nal of Economic Literature 14, 4 (Deem-, 'Stratification in a Dual Economy.'"American ber): 1251-1257. Sociological Review i5 (August):702-711. Corcoran, M. Heckman, J.J. 1978 "The Structure of Female Wages." The Amer- 1974 "Shadow Prices, Market Wages and Labor

ican Eco,nomic Review (papers and proceed- ,Supply." Econometiica 42, 4 (July):679-694.' ings) 68, 2 (May):165-170. 1976 "A Life Cycle Model of Earnings, Learning, 1979 "Work Experience, Labor Force Withdrawals, and Consumption." Journal of Political Econ- and Women's Earnings: Empirical Results Us- omy 84(4):S11-S44, ing the 1976 Panel Study of Income Dynam- Hill, M.S. ics." In Women in the Labor Market, edited 1979 "The Wage Effects of Marital Status add Chil- by C. B. Lloyd, E. Andrews, and C.L. Gilroy. dren." Journal of Human Resources 14, 4 New York: Columbia University Press. (Fall):579-594. Corcoran, M., and G: J. Duncan Horan, P.M. 1979 "Work History,-Labor Force Attachment, and 1978 "Is Status Attainment Research Atheoretical?" Earnings Differences Between the Races and American Sociological Review 43(August):334- Sexes." Journal of Human Resources 14, 1 341. (Win ter):3-20. Jones, E.B., and J.E. Long Corcoran, M., G.J. Duncan, and M. Napa 1979 "Part-Week Work and Human Capital Invest- In "Work Experience, Job Segregation, and ment by Married Women."Journal of Human presiWages." In Five Thousand American Fami- Resources 14(Fall):563-578. liesPatterns of Economic Progress, edited Jovanovic, B. by G.J. Duncan and J.M. Morgan. Ann Arbor: 1979 "Job Matching and the Theory of Turnover." Institute for Social Research. Journal of Political Economy 87, 5, Part I (Oc- Doeringer, P. B., and M. J. More tober):972 -989. 1971 Internal Labor Markets and Manpower Anal- ysis. Lexington, Mass: D. C. Heath. Jusenius, C. L. 1977 "The Influence of Work Experience, Skill Re- Duncan, G.J., and S. Hoffman quirement, and Occupational Segregation on 1979 "On-the-Job Training and Earnings Differ- Women's Earnings." journal of Economics and ences by Race and Sex." Review of Economics Business 29, 2 (Winter):1137-115. and Statistics 61, 4 (November):594-603. Kahn, L.M. Duncan, 0.D., and B. Duncan 1980 "Wage Growth and Endogenous Experience." 1955 "A Methodological Analysis of Segregation In- Industrial Relations 19, 1 (Winter):50-63. dexes." American Sociological Review 20 (April):210-217. Landes, E.M. 1977 "Sex Differences in Wages and Employment: Edwards, R. A Test of the Specific Capital Hypothesis." 1979 Contested Terrain; The Transformation of the Economic Inquiry 15, 4 (October):523-538. Workplace in Ameriia. New York: Basic Books. Lazear, F. England, P. "Age, Experience, and Wage Growth." Amer- "Wage Appreciation and Depreciation: A Test 1976 1981 ican Economic Review (September):548-558. of Neoclassical Economic Explanationc of Oc-

197 WORK EXPERIENCE. 108 SEGREGATION. AND WAGES 191

Lloyd, C.B., and B.T. Niemi 1981 "Occupational Self-Selection: A Human Cap- 1979 .The Economics of Sex Differentials. New York: ital Approach to Sex Differences in Occupa- Columbia University Press. tional Structure." Review of Economics and Medoff, J., and K. Abraham Statistics 63, 1 (February):60-69. 1981 'Are Those Paid More Really More Produc- Prescott,.,and M. Viischer tive? The Case of Experience." Journal of Hu- 1980 "Organizational Capital." jouinal of Political man Resources 16 (Spring). Economy, 88, 3 (June)446-461. Mincer, J. Rosen, S. 1974 Schooling Experience and Earnings. New York: 1972 "Learning and Experience in the Labor Mar- National Bureau of Economic Research. ket." journal of Human Resources VII (Sum- Mincer, J., and H. Ofek mer):326-342. 1982 "Interrupted Work Careers." Journal of Hu- Sandell, S.H., and D. Shapiro man Resources 16 (Wintet). 1978 "An Exchange: Theory of Human Capital and Mincer, J., and S. Polachek the Earnings of Women: A Reexamination of the Evidence." Journal of Human Resources 1974 "Family Investments in Human Capital: Earn- ings of Women." Journal of Political Economy 13, 1 (Winter):103-117. 82, 2, Part 2 (March/April):S76-S108. 1980 "Work Expectations, Human eapital Accu- 1978 "An Exchinge: Theory of Human Capital and mulation, and the Wages of Young Women." the Earnings of Women: Women's Earnings Journal of Human. Resources 15, 3.(Summer):- Reexamined." Journal of Human Resources 13, 335-353. 1 (Winter):118-134. Stafford, F.P., and G.J. Duncan Morgensen, D. 1980 "The Use of Time and Technology by House- holds in the United States.:' In Research in 1978 "Specific Capital and Labor Turnover." Bell journal of Economics, 2 (Autumn):572-586 Labor Economics, Vol. 3, edited by Ronald Ehrenberg. Greenwich, Conn.: JAI Press. Oppenheimer, V.K. 1970.The Female Labor Force in the United States: Treiman, D.j., and H.I. Hartmann, eds. Demographic and Economic Factors Govern- 1981 Women, Work, and Wages: Equal Pay for jobs ing Its Growth and Changing Composition. of Equal Value. Washington, D.C.: National Westport, Conn.: Greenwood Press. Academy Press. Piore, M.J. , Weiss, Y., and R. Gronau "Expected Interruptions in Labour Force Par- 1971."The Dual Labor Market: Theory and Impli- 1981 cations." In Problems in Political Economy: An ticipation and Sex-Related Differences in Urban Perspective, edited by D.M. Gordon, Earnings Growth." Review of Economic Stud- pp. 90-94. Lexington, Mass.: D.C. Heath. ies 58 (October):607-619. Polachek, S.W. Wolfe, W., and it Rosenfeld 1978 "Sex Structure of jobs and Job Mobility." So- 1976 "Occupational Segregation: An Alternative Hypothesis." Journal of Contemporary Busi- cial Forces 56:823-844. ness 5, 1:1-12. Zellner, H. 1975 "The Determinants of Occupational Segrega- 1979 "Occupational Segregation- Among Women: Theory, Evidence, and a Prognosis." Women `tion." In Sex, Discrimination, and the Division in the Labor Market, edited by C.B. Lloyd, of Labor, edited by C.B. Lloyd. New York: E.S. Andrews, and C.L. Gilroy. New York: Columbia University Press, pp. 125-145. Columbia University Press, pp. 137-157.

198 O

11 Sex Typing in Occupational Socialization tMARGARET MOONEYMARINI and

MARY C. BRINTON .

1.;

The existence of sex segregation in the labor force participation of women over the labor market is well documented (Gross, last several decades (Oppenheimer, 1970; 1968; U.S. President's Council of Economic U.S. Department of Labor, 1971), the amount Advisors, 1973; Blau, 1977; Williams, 1976, of sex segregation in the labor market has 1979; U.S. Commission oh Civil Rights, 1978; decreased little (England, 1981a). As re- Blau and Hendricks, 1979). Women tend to cently as 1976, more than two-thirds of one R. be concentrated in a relatively small num- sex would have had to change occupations ber of "female" occupations, whereas men to make the occupational distributions of the are employed in awider variety of "male" two sexes equal (U.S. Commission on Civil occupations. More than 40 percent of female Rights, 1978). workers are employed in the 10 occupations Three major typJs of explanations for sex employing the largest numberof women, segregation in the labor market have been whereas less than 20 percent of male work- advanced: (1)` explanations focusing on em- ers are employed inthe 10 occupations employer demands, (2) explanations focusing ploying the largest number of men (U.S. on legal andinstitutional barriers within the Department of Labor, .1975). Women are workplace, and (3) explanations focusing on overrepresented in clerical, sales, and serv- worker characteristics. The first two locate ice jobs; in a few professional and technical the source of sex segregation within the b. jobs (e.g., elementary and secondary school workplace. It has been hypothesized, for ex- teacher, registered nurse; librarian, social ample, that exclusionary behavior by em- worker, medical and dental technician); and ployers results in the overcrowding of women in such jobs as machine operative,where in a limited set of occupations and that th!s they assemble or inspect goods, operate overcrowding reduces the wages of women sewing and other machines, andwork as in those occupations relative to the wages packers and wrappers. Men are overrepre- of the nonrestricted group of men (Berg- sented in managerial, crafts, labor, and firm mann, 1971, 1974). 'Ithas also been hy- jobs and in most professional and technical pothesized that the structure of the labor jobs. Despite a substantial increase in the market, which includes occupations filled

1gh SEX MING Ii OCCUPATIONAL SOCIALIZATION 193

from external sources through the recruit- entry, focusing primarily on socialization ment of new workers and occupations filled practices in the family and school but also from internal sources through the promo- considering messages conveyed by the mass tion of in-house workers, creates institu, media and employment experiences prior to 4ional barriers in the process of job assign- leaving school. In the final section, wellis- ment and promotion that disadvantage cuss the tble that socialization can be inter- tvmen (Doeringer and Piore, 1971; Blau preted to play in producing sei'segregation and Jusenius, 1976). Sex segregation in the in the labor market. labor market has been argued to occur at least in part as a result of "statistical dis- THEORIES OF OCCUPATIONAL CHOICE crimination," whereby individuals are judged AND SEX-ROLE SOCIALIZATION on the basiF, of the perceived average characteristics of the group to which they belong This section provides an overview of the (Thurow, 1975:170-81). Since, on the av- theoretical bases on which sex differences erage, women are viewed as differing from in occupational orientation and job-relevant men in their ability to perform certain types traits have been assumed to arise. We begin of jobs and in their attachment to the labor by outlining general theories of occupational --- market. sex is used as a basis for "statistical choice that have emerged in various disci- discrimination" in the allocation of individ- plines. Since the prediction of sex differ- uals to jobs. ences in outcomes using these theories re- In contrast to explanations of sex segre- quires prior knowledge that the two sexes,, gation that focus on the actions of employers differ on various inputs, we discuss theories and the structure of the labor market, a third of sex-role socialization. These latter theo- set of explanations focuses on the charac- ries, advanced primarily by psychologists, teristics of workers. These explanations at- constitute the basis on which sex differences tribute sex segregation to sex differences in can be predicted by genera) theories of oc- individuals, including, occupational prefer- cupational choice,: ences, skills, and other personal attributes. Women and men are hypothesized to be Theories of Occupational Choice employed in different occupations because they choose different occupations and be- General theories of occupational choice cause they are differentially qualified for abound. Developmental theories such as various types odobs. those of Ginsberg et al. (1951) and Super This paper examines the explanations for (1953, 1957) describe the process of occu- sex segregation that focus on the character- pational selection in terms of general con- istics of workers entering the labor market. cepts of human development. Based on the The4irst section outlines general theories of principles of developmental psychology, oc- occupational choice and points to the need cupational choices are viewed as developing to consider sex-role socialization as an input gradually over time in a series of stages. to these theories. The second section pre- Personality-based theories, such as Hol- sents evidence on the existence of sex dif- land's (1959, 1966, 1973) typology theory, ferences prior to labor market entry in sev- describe career orientations and prefer- eral areas relevant tdoccupational attainment, ences in terms of personality types. Still other including occupational preferences, knowl- psychological theories involve specific ap- edge, values, skills, and dispositional traits. plications of general behavior theory. In In the third section, we examine the so- Krurnboltz's Social Learning Theory of Ca- cialization practices that appear to produce reer Selection (Krumboltz et al., 1976; these sex differences prior to labor market Mitchell et al., 1975), occupational decisions

200 ,194 MARGARET MOONEYMARINI AND MARY C. BRINTON are viewed as an outcome of alifelong series the assumption that individuals seek to max- of learned responses. Other applications of imize expected lifetime earnings, econo- general behavior theory focus more on in- mists. have used human capital theory to ar- formation processing. The decision theories gue that sex differences in expectedlifetime of Vroom (1964) and Kaldor and Zytowski labor force participation produce sex differ- . (1969), for example, are concerned with the ences in occupational choice..Specifically,i process ofdecision making based on the ex- Polachek (1976, 1979, 1981) has argued that pected, consequences of alternative deci- sex segregation in the labor market arises sions. The logic-flow theories of Hilton (1962) because women's expectations of intermit- and Herchenson and Roth (1966) deal with tency in employment cause them to choose the steps individuals go through.in arriving occupations in which the amount of depre- at decisions. Sociological work on occupa- ciation in earnings during periods of absence tional choice, which has arisen out of the from the labor force is low. Zellner (1975), study of social stratification, focuses pri- on the other hand, has argued that sex seg- marily on the status dimensions of occupa- regation arises because women's expectations (e.g., Blau and Duncan, 1967; Sewell tions of intermittency in employment cause et al., 1969, 1970). Work economist's gen- them to choose occupations with High start- erally involves specific applications of gen- ing wages but low wage appreciation. In eral theories of utility maximization, partic- either case, it is implied that women tend ularly the theory of htiman capital, according to enter occupations that require few skills to which occupational selection implies and provide lit `le opportunity for increases varying amounts of investment in human in productivity through experience. capital and affects returns'm the investment Critics of these neoclassical economic ex- (Becker, 1964). planations of sex segregation have pointed In and of themselves, these general the- to a number of theoretical problems. One ories do not explain why males and females is that both male and female occupations select different occupations. Unless the two require differing 'amounts and types of skill. sexes differ on theindependent variables Women a,ncl men are employed in occupa- used as inputs to these theories, sex differ- tions of each skill fype, and within each type' ences in occupationalchoice are not pre- some occupations are more oftenentered dicted.For example, unless the develop- by women than by men. Women's lower mental experiences of the sexes differ, expected lifetime labor force participation developmental and social learning theories explains only the greater tendency of women of occupational choice do not predict sex to be in jobs requiring low skill, not the differences in occupational selection. Simi- concentration of women in a small number larly, unless the adult role expectations of of female occupations within each skill type the sexes differ, psychological and economic (Blau and jusenius, 1976). Within the hu- theories of decision making do not predict man capital framework, the pattern of sex sex differences in occupationalselection. In segregation existing in the labor market can short, regardless of which general theory is be accounted for only by an extreme distri- used, the prediction of sex differences in bution of women's "tastes." Another prob- outcomes requires the input ofadditional lem is that the causal direction of the rela- information that the sexes differ on variables tionship between occupational outcomes and predicting occupational choice. labor force attachment is ambiguous. Al- Attempts to use general theories to under- though it may be that those who anticipate stand why males,and females select different being out of the labor force for a substantial occupations have actually been quite lim- amount of time initially select low-wage oc- ited. The most extensive applications have cupations, it may also be that those who been those of human Capital theory. Under spend a lot of time out of the labor force

20,1 '.

SEX TYPING IN OCCUPAZOGIALIZATLON 195

wind up in low-wage occupations as a result ries, and (4) identification theories. After ex- (Welch, 1979).- amining the sex-role socialization process, Recently, direct tests of the assumptions we consider the content of what is trans- underlying human capital explanations have mitted via that process. That is, we examine presented some disconfirming evidence. the gender-linked behavior patterns that are England (1982) shows that predominantly learned and discuss the division of labor be- female occupations do not penalize inter- tween the sexes that constitutes the basis mittency leis than male occupations and that for many sex differences in behaVior, eat women expecting fairly constant employ- tudes, and personality. ment are norlOre likely to choose male occupations than women planning intermit- PSocial Learning Theories Two basic tent employment. England (1981b) further learning processes, operant conditioning and shows that women have higher lifetime observational learning, are at the heart of earnings if they are employed in predomi- social learning theories. These theories are nantly male occupations, a finding that does based primarily on a mechanistic model not support the contention that women max- (Reese and Overton, 1970).. Sex-typed be- imize lifetime earnings by choosing female havior is seen as resulting from the fact that occupations. Given the lack of empirical reinforcement contingencies depend on the support for human capital explanations of sex of the responder. That is, girlsana boys occupational 'segregation by sex, other ex- are reinforced or punished for differentkinds . planations must be sought. It is possible that of behavior,. and male and female models other general theories of occupational choice display different kinds of behavior. One ma- may be more successful than the human cap- jor tenet of social learning theory is that sex- ital approach in accountingjor sex differ- typed behavior need not be consistent across ences in occupational outcomes, butthese situations but depends on the social context theories have not yet been applied to the in which it occurs. The bases of sex typing study of sex differences. are viewed as arising in the social environ- Since all general theories of occupational ment, not the organism, so that relatively choice require the existence of sex differ- rapid changes can occur if learning condi- env; on predictor variables in order to gen- tions are altered. Sex-role learning is as- erate predictions of sex differences in oc- sumed to take place continuously, although cupational choice, we now turn to a discussion the majority occurs during early childhood. of theories of sex-role socialization. These Cognitive social learning theories use ad- theories provide a basis for understanding ditional constructs to describe the internal the developmental process by which most mental processes that mediate learning, but sex differences inbehavior emerge. cognitions play a Secondary role, and sex typing is conceptualized primarily as a set of behavioral responses. An extensive dis- Theories of Sex-Role Socialization cussion of social learning theory is contained Theories of sex-role socialization explain in Mischel (1970). the process by which individuals learn the behavior that a culture defines as appropri- Cognitive Developmental TheoriesCog- ate for their sex. The theories differ pri- nitive developmental theories derive from marily in the mechanism by which sex-typed Piaget's theoretical framework for under- behavior is hypothesized to be learned. Be- standing child development. Unlike social low we describe the major theories of sex- learning theories, they are based primarily role socialization, including (1) social learn- on an organismic model (Reeseand Over- ton, 1970). Cognitive processes are viewed ing theories, (2) cognitive developmentalI theories, (3) information processing theo- as ongoing processes ofchange. It is as- 202

0 196 MARGARET MOONEY MARINI AND MARY C. BRINTON

sumed that children play an active role in phasized as the source of schemas or the their own development, motivated by a de- means of changing them. The cultural em- sire for competence and mastery over their phasis on gender rather than physical sex world. The child's concepts about masculin- differences is what is seen as making gender ity,femininity, and sex appropriateness, salient. rather than the child's sex:typed behavior, are at the core of sex typing. Such concepts Identification TheoriesFreudian psy- or schema constitute organizing rubrics for choanalytical theory is the basis for all iden- the selection of information from the envi- tification theories of sex-role learning. In ronment and for active processing of that classical Freudian theory, masculinity and input. femininity are acquired through a proceis Developmental changes in sex typing are of identification resulting from castration fear. assumed to go hand in hand with general on the part of the male child and castration developmental changes in Cognitive pro- anxiety on the part of the female child. Al- cesses: To the extent that these changes are though more recent theories of identification inherent in the organism, changes in sex do not placeasmuch emphasis on sexual mo- typing are governed by maturational, inter- tivation, identification with the same-sex par- nal variables in interaction with the social ent continues to be viewed as an important environment. Thus, these theories propose basis for the development of permanent. and organismic as well as environmental influ- global sex differences in personality. Patterns ences on sex typing, and most therefore sug- of behavior are assumed to be integrated, so gest some limits to the degree and rapidity that a child who is feminine in one situation with which sex typing can be changed (Hus- is feminine in artothei. 'In recent years, clas- ,,, ton, in press). Among the most prominent sical theories of identification have fallen into cognitive developmental theories are those &favor, and theorists now emphasize paren- proposed by Kolberg (1966), Block (1973), tal Identification less, viewing parents as one Pleck (1975), and Rebecca et al. (1976). of many socializing influences (Huston, in press). However, there is little empirical evi- Information Processing TheoriesTheo- dence to support either the existence of iden- ries of information processing schema are a tification or the contention that it accounts for hybrid set of theories based on information sex-role learning (Parsons, 1978). processing constructs (Huston, in press). Some reformulations of psychoanalytic They emphasize schemas as cognitivettruc- theory have been undertaken by feminists. tures that guide and organize an individual's These focus on envy of women's Childbear- perception. The schemas are anticipatory ing capacity and caretaking role as the rea- mechanisms that cause an individual to search son for devaluation of the mother and of for certain information and to be ready to women in general (Horney, 1932; Klein, process it. Information inconsistent with the 1957; Lerner, 1974, 1978; Chodorow, 1978). schema may be ignored or transformed. Because the mother as primary caregiver Models of sex typing based on information is perceived as all powerful, men are hy- processing have been proposed recently by pothesized to develop envy, fear, and anger Bem (1981) and Martin and Halverson (1981). in a struggle to free themselves from her. In these nodels sex stereotypes serve as According to Chodorow (1978), they gen- schemas for organizing and structuring so- erally come to see themselves as more dis- cial information. Although schema theories tinct from others as a result. It is suggested are similar to cognitive developmental the- that this basis for sex-role differentiation could ories in focusing on cognitive processes that he altered if the caretaking of young chil- are active and constructive, they differ in dren were shared by males and females. that developmental processes are not em- Again, empirical evidence is lacking to

.,8 203 197 SEX TYPING INOCCUPATIONAL SOCIALIZATION home, a port these reformulations asthe basis for sex garding employment outside the of typing. choice usually ..based on the decision Next we consider .the sex-typed, content whether to add a new role to the traditional of what is transmitted via sex-rolesociali- homemaker role rather than whether to sub- zation. 'Although the division oflabor be- stitute a new'role for the old one(Poloma tween the sexes forms thebasis for many and Garland, 1971; Bahr, 1974; Vanek, 1974; sex differences inbehavior, we present evi- Walker and Woods; 1976; Robinson, '1977; dence to indicate that it alone does not ac- Berk and Berk, 1979). Because ofthe con- count for all gender-based behavior patterns flict between fulfillment of familial role re- transmitted throughlocialization, including sponsibilities and work outside the home, the segregation of women and then within women's investment in family roles nega- the workplace. tively affects their labor force participation and employment in high-status occupations Sex-Rote DifferentiationChildren learn (Rossi, 1965; Sweet, 1973; Waite, 1976; the behavior that is appropriate for their sex Smith-Lovin and Tickamyer, 1978; Marini, via the process of sex-role socialization.Al- 1980). though this learning may occur in a variety Differences in the occupational orienta- of ways, the content of what islearned .de- tions and skills of the two sexesan be ex- pends on the association of gender with par- pected to arise as a consequence of the sex ticular types of behavior in the culture in difference in consistency between familial which a child is raised. A gender-baseddi- and occupational roles. Womenare more vision of labor exists to some extent inall likely to view their work outside thehome societies and forms the basis for many of the.. as a job than as alifetime career and to sex differences inbehavior, attitudes, and choose jobs that permit better coordination personality that are transmitted via sociali- of their responsibilities in the home with zation. In industrialized societiessuch as the their emisloyment (Rossi, 1965; Perucci, United States, the sexual divisiOn oflabor 1970). Because women are less likely to ex- between the market and the home has im- pect to work throughouttheir adult lives and portant implications for theoccupational ori- to be the primary wage earners(Turner, entation and preparation of the sexesprior 1964), their occupational interests focufless to entry into the jobmarket.. For the most than men's on the monetary and statusdi- part, men are expected to supportthe family mensions of jobs and' tend toparallel their financially, and women take the major re- family functions, often involving an orien- sponsibility for home management, child tation toward helping others(Witty and care, and catering tothe emotional needs of Lehman, 1930; Singer and Stefllre, 1954; the family. This division of labor results in .O'Hara, 1962; Lueptow, 1980; ,Herzog, essentialconsistency between men's familial 1982).' The sexual division of laborbetween and occupational roles but producesconflict the market and the home and its effect on between the familial role of women andtheir the sex"difference in consistency between participation in the labormarket. Fulfill- ment of familial roleresponsibilities com- O petes with work outside thehome for the I It should be noted that the lower wages paid to limited supply of a woman's time, energy, women and typically-a,ssociated withwomen's jobs are and emotional commitment. a cause as well as a result ofwomen's orientation toward of employment. Women may not seek to satisfy material Even the entry of increasing numbers ambitions through their own occupations because the women into thelabor force has not changed incomes they can expect to receive are solow; however, this fundamental division of labor.Changes because most Ikomen do Mot rely on their own occu- at in conceptions of the femalerole have re- pations for full material support, they are not aslikely sulted primarily in the need forchoice re- to expect or demand higher wages. 204 g -747"4" -^.` - ,

198" MARGARET MOONEY MARINI AND MARY BRINTON

, -- I familial, and occupational roles may also be samples of college-educated women, in con- seen as promoting the development of dif- trast, has'indicated that married women and ferent personality characteristics and abili-. women with a relatively, large number of ties in the two sexes. Males are socialized children are less likely to beqmployed in to be assertive, authoritative, and compe- male occupations (Almquist and Angrist, tent in occupational skills, whereas females 1970; Klemmack and Edwards, 1973; Bielby, are socialized to-be nurturant, deferent, and 1978a; Brito and Jusenius, 1978; Dayinont competent in domestic skills (Oetzel, 1966i and Tsai, 1981). These findings sugpst that Maccoby and Jacklin, 1974; Block, 1976; a relationship between women's investment Tavris and Offir, 1977; Frieze et al., 1978). in family roles and the sex type of their oc- The extent to Which'women's familial role cupations exists only at the upper end of the responsibilities ac-aunt for sex segregation education distribution, Such a, relationship in the labor market remain an open ques- is likely to'arise because a relationship be, tion. It can be argiied that women's invest- hveen the status and sex type of occupations' pent in faniily roles affects the probability exists at the upper end of the education dis- of their employment in female-typed jobs tribution, where male occupations tend to for several reasons. First, -women's invest- be of higher status than female occupations, ment in family roles may affect the status of Women's fulfillment of traditional family re- the occupations they hold..Ansequently, sponsibilities interferes with employment in there may be a relationship between the high-status male occupations, which place statics and sex type of occupations, with high- heavier demands on Aheir incumbents and status occupations more often being tradi- are, therefore, le§s easy to combine with tionally Male. Second, female jobs may have traditional family. responsibilities. characteristics, such as greater flexibility of The sexual division of labor between the working hours, that make them easier to home and the job market may, therefore, combine with family responsibilities. Third, be seen as forming the basis for many sex women who invest relatively more in family differences in behavior that are transmitted roles may have traditional attitudes that cause via socialization, including sex differences in them to select female occupations More often job-relevant skills and dispositional traits. than male occupations. However, this fundamental diyision of labor Research bearing on the relationship be- cannot account for all sex differences trans- tween women's investment in family roles mitted via socialization. Some sex differ- and the sex .type of the jobs they hold sug- ences, including the tendency for males and gests that the relationship differs depending females to be employed in different occu- on whether a woman has a college educe - pations; have other origins. Regardless of its don. As indicated earlier, England (198''). origins; gender-linked behavior is transmit- trend that women expecting constant e.n- ted via, sex-role socialization; Thus, because plment (as measured by familial ro).! sta- the ocbupational world is sex -segregated, tus) were no more likely t9 choose male oc- children learn to view some occupations as. cupations than women planning intermittent appropriate for their sex and others as inemployment. England's analysis was based appropriate (Looft, 1971a,b; Schlossberg and V on a sample covering the full range of var- Goodman, 1972; Siegel, 1973; Shepard and iation in education. Analyzing a sample re- Hess, 1975; Heilman, 1979). stricted to women who did not go to college, lofferth (1980a) also found that marital sta- Biological Components of Sex Typing tus and children had no effect on the sex type of jobs held by women three, five, and Many theorists have proposed that sex ten years after high school. Research on differences in behavior are at least partially

V 205 'SETYPING IN OCCUPATIONAL SOCIALIZATION .199

O due to genetic, biochemical, and anatomical to suggest that biology may play a small role differences between the sexes. It is gener- in the determination of most sex differences, ally agreed thaean either/or position con- and since it seems likely that the extent to cerning- the effects of biology and sociali- which one sex is better suited to perform zation is too simplistic and that the important sex-typed jol4 has been greatly eitagger- question focuses on the relative role of these ated, the role of biology in the determina- two types of influences in determining sex- tion of occupational segregation by sex is typed behavior. Although the role played indeed likely to be small. by biology is unknown, evidence from three types of studies suggests that socialization Summary rather than biology is the source of most sex differences in behavior, particularly those Theories' used to predict occupational that are likely to have a bearing on occu- choice in various disciplines do not predict pational orientation and performance. First, sex differences in occupationalchoice unless studies of hermaphrodites, whose gendeOs information that the sexes differ on variables biologically ambiguous, indicate that the used to make the prediction is available. gender according to which a child is reared Theories of sex-role socialization advanced is more important for the developmentof within psychology constitute the primary gender identity than genes or gonads (Money basis on which sex differences in occupa- and Ehrhardt, 1972), Second, cross-cultural tional orientation and job-relevant skills are studies of sex-typed behavior indicate that viewed as arising over the early stages of the many personality traits, activities,and oc- life course. These theories describe the cupatkans that are labeled feminine in one process by which gender-linked behavior is society are labeled masculine in another learned. Biology also plays a role in the de- (Mead, 1935; McClelland, 1976; Tavris and termination of some sex differences in Ile- ,havior, but the fat that biologically based Offir, 1977). Third, studies of sex differences O in infancy, when the effects of culture are sex differences may have littlebearing on minimal, rarely find sex differences in be- occupational performance suggests that the havior (Maccoby and Jacklin,1974). Al- effect of biology on occupational choice is though it is difficult to document sex differ- small. ences in infants formethodological reasons (Block, 1976)and some biologically based SEX DIFFERENCES IN OCCUPATIONAL sex differences do not emergeuntil later ORIENTATION PRIOR TO LABOR FORCE agesthe fact that sex differences are rarely ENTRY found in infants does not support the view that sex differences are biologically deter- As a result of sex-role socialization, sex mined (MaCcoby and Jacklin, 1974; Frieze differences in occupational orientation and et al., 1978). preparation arise prior to entry into the la- Not only is the role of biology in the de- bor market. This section examines the der termination of sex differences in various types gree to which occupationalaspirations and of abilities and dispositional traits an open expectations prior to labor market entry are question, but the extent to which jobsthat sex typed and confiders theprobable rela- are thought to requiresex-related traits ac- tionship between this sex typing and sub- tually do require those traits is unknown. sequent sex segregation in the labor market. Consequently, the extent to which biology We will also examine sex differences in may affect sex segregation inthe labor mar- knowledge of the occupational world and in ket via its effects on the 'characteristics of occupational values held prior to labor mar- workers is unknown. Since there is evidence ket entry, Finally, we consider evidence

206 200 MARGARET MOONEY MARINI AND MARY C. MINTON bearing on the existence of sex differences rower than the range of choices made by in abilities, 'such as physical strength and males (Rodman 'et al., 1974; Marini and verbal and quantitative skills, and in dis- Greenberger, 1978). Further analysis of the positional traits, such as aggressiveness, so- 1979 NLS data indicated that 47.5 percent ciability, and self-confidence. It has been of young women aspired to the 10occupa- argued that all of these sex differences are tions most often aspired to by women but determinants of sex segregation in the labor that only 39.5 percent of young men aspired market. More specifically, it has been ar- to the 10 occupations most often aspired to gued that'women and men occupy different by men... positions in the workplace because they Previous research has shown that theoc- choose different occupations and are differ- cupational aspirations of males are alsomore entially qualified for various types of jobs. highly sei typed than those of females (Mar- ini and Greenberger, 1978). This finding is Occupational Aspirations and confirmed by analysis of the 1979 NLS data. Expectations We divided occupations into three sex-type categories on the basis of the percentage of Research on occupational aspirations and female incumbents in the occupation. Oc- expectations held prior to labor market en- cupations with leis than 30 percent women try provides strong evidence that sex dif- were defined as male occupations; occupa- ferences in occupational choice exist. Young tions with 30 to 59 percent women were women are more likely to choose typically defined as sex-neutral occupations, andoc- "female" occupations; whereas young 'men cupations with 60 percent or more women are more likely to choose typically "male" were defined as female occupations. Based 'occupations (Stephenson, 1957; Sewell and on this categorizatiOn, 86.3 percent of males Ornstein, 1964; Doiwan and Adelson, 1966; aspired to male occupations, but only 4.1 Werts, 1966; Astin and Panos, 1969; Marini percent aspired to female occupations. In and Greenberger, 1978; Harren et al., 1979; contrast, 52.8 percent of females aspired to Herzog, 1982). Toexamine the degree of female occupations, and '34.5 percentas- sex segregation in aspirations for the full range pired to male occupations. Similar percentof the Census Bureau's detailed occupa- ages of each sex (9.6 percent of males arid tional categories, an index of segregation was 12.7 percent of females) aspired to sex-neu- calculated using data from the National Lon- tral occupations. These sex differences in gitudinal Survey of Young Americans (NLS). the 'distribution of aspirations by sex type These data were collected in 1979 from a indicate that females are considerably more nationally representative sample of youth ages likely than males to make cross-sexoccu- 14 to 22 and, are described in detail else- pational choices. where. For a measure. o Occupational aspirations for age 35, the index of segregation Discrepancy Between Aspirations and Ex- was 61.0, indicating that 61 percent of one pectationsBy examining both occupa- sex would have to change occupational as- tional aspirations and expectations,some pirations to make- the aspirations distribu- studies have attempted to sort out wishful tions of The two sexes equal. The degree of aspirations from more realistic emiectatiotis, segregation in aspirations also was examined or plans (Burlin, 1976; Marini and Green- by age, but only a small change was ob- berger, 1978; Lueptow, 1981). Expectations served over the age range studied. are more likely to reflect perceptions of con- Not only are the occupational choices of straints such as limited opportunities, the youth highly differentiated by sex, but the sex type of the job, and personal qualifica- range of choices made by females is nar- tions. The discrepancy between aspirations

207 SEX TYPING IN OCCUPATIONAL SOCIALIZATION 201

and expectations therefore provides some tions in the female-dominated,category than indication of the-degree to which Individuals the boys were to shift their aspirations from perceive that constraints may prevent re- the female-dominated category to expecta- alization of their aspirations. Studies of both tions for male-dominated jobs. The girls, aspirations and expectations uniforMly in- therefore,. seemed to perceive the male- dicate that there is greater sex typing of oc- dominated jobs they aspired to as less ac- cupational expectations than of occupational cessible than the boys perceived the female- aspirations. dominated jobs they aspired to. These find- The most detailed study comparing the ings suggest that the sex composition of an sex typing of occupational aspirations and occupation influences the degree to which expectations was carried out by Marini and girls, but not boys, expect to realize their Greenberger (1978), based on data collected occupational aspirations. from a representative sample of eleventh- Further support for the hypothesis that grade students in Pennsylvania in 1968. In the sex type of an occupaional aspiration this study there was virtually no difference influences the degree to which girls feel it between the mean percentage of women can be realizec*s available in a survey by employedinoccupations aspired to (17 per- Burlin (1976) of adolescent girls in a Syra- cent) and expected (18 percent) by boys. cuse high school. She found that more than However, the mean percentage of women one-half of those with discrepant occupa- employed in occupations expected by girls tional aspirations and expectations attrib- (75 percent) was significantly greater than uted the discrepancy to the fact that the or CI the mean percentage of women employed occupation aspiked to was an "inapprop-'ate in the occupations girls aspired to (66 per- occupation for a female." Data from a na- cent). These findings indicate that the girls tional sample of high school students col- expected to enter occupations that, on the lected in 1980 as part of the Monitoring the average, employed a higher proportion of Future project also indicate that the ,irls women than those they aspired to. surveyed more often perceived their sex as Of respondents who aspired to occupa- a barrier to fulfilling their occupational as- tions in which fewer than 50 percent of the pirations (Bachman et al., 1980). When asked incumbents were women (i.e., male-domi- to what extentthey thought their sex would nated occupations),. asmaller percentage of prevent them from getting the kind of work the girls (52 percent) than the boys (94 per- they wNeLlaig.to have, 87.9 percent of the cent) actually expected to enter an occu- males but only 66.4 percent of the females pation of this type. Of respondents aspiring responded "not at all." Experimental re- to occupations in which 50 percent or more search by Heilman (1979) proyides further of the incumbents were women (i.e., fe- evidence that the sexual composition of an male-dominated occupations), the percent- occupation influences the degree to which age of the boys (78 percent) expecting to it is considered a viable career choice. enter a female-dwinated occupation was almost as high as the percentage of the girls Relationship of Occupational Aspirations to (85 percent). In addition, only about 3 per- SubsequentOccupationalBehavior The cent of the girls who aspired to female-dom- degree of correspondence between occu- inated occupations, in comparison with 22 pational aspirations held prior to labor mar- percent of the boys, expected that they would ket entry and subsequent occupational at- instead enter male-dominated occupations. tainments is indicative of the degree to which ,These findings indicate that the girtsAyere individuals realize their occupational aspi- more likely to shift their aspiratiOi from rations. The occupational aspirations of high male-dominated occupations to kxpecta- school students definitely play a role in the

208 L. o.

202 MARGARET MOONEY14 ARIN1 AND MARY 5 BRINTON

determination of adult occupational attain- are difficult to draw, sincestudies done at ment, but the relationship betweenthe sex: different intervals are notcomparable in the type of occupational aspirationsand the sex inclusiveness of the occupational categories type of occupational attainmentshas not been used or in the type of sample studied. estimated. This relationship is of ante est The most readily interpretable estimates because it would indicate the extent to w ch of the relationship between occupational as- the sex segregation of occupational choices 'ratio asuicjasequent occupational at- prior to labor market entry can accountfor tainents are available for the status of oc- the sex segregation in employment that is.. cupations, as measured by the Duncan ... subsequently experienced by a cohort. To Socioeconomic Index (SEI). Analyzing data the extent that sex segregation in occupa- from an 18-year follow-up study of Wiscon- tional choices exists prior to labor market sin high school seniors, Sewell et al. (1980) entry, sex segregation in occupational out- found correlations of .461 for females and comes cannot be attributed tothe direct ex- .543 for males between the status level of perience of sex discrimination in the labor the occupation aspired to in high school and market. However, as we will discuss, dis- the status of the first job held. Somewhat criminatory practices and structural barriers lower correlations of .342 for females and within the labor market may generate a pat- .491 for males were found between the sta- tern of sex segregation that ismaintained tus level of the occupation aspired to in high over time viasocialization. school and the status of the occupation held Most research on the relationshipbe- 18 years later: tween occupational goals and attainments In the absence of information on the re- has focused on the uver.d1 degree of con- lationship between the sex typing of occu- gruence betweenoccupational aspirations pational aspirations and the sex typing of and attainments, where congruence itde- occupational attainments for a sample of in- fined as aspiring to and attaining an occu- dividuals studied while in high school and pation in the same occupational category. again some years later, it is of interest to The findings of such studies depend in part compare measures of sex segregationin oc- on the inclusivenessof the occupational cat- cupational aspirations for a national sample egories used; the more inclusive the occu- of youth with measures of sex segregation . pational categories, the greater thedegree in occupational attainments for the adult of congruence will appear to be.Variability population. Such a comparison permits a among studies in the typeof sample and the crude assessment of the extent to which sex age at whichrespondents were initially segregation in occupational goals approxi- studied also clouds the picture. Estimates mates sex segregation inemployment. of the degree of congruence betweenhigh Measures of sex segregation in respondents' school aspirations and subsequent occupa- occupational aspirations for age 35 were cal- tional attainments range from about 50 per- culated using the 1979 NLS and were then cent (Schmidt and Rothney, 1955) to 80 per- compared to measures of sex segregation in cent (Porter, 1954) in studiesdone 6 months actual employment based on data from the after graduation from high school, toabout U.S. census. 50 percent in a study of women done 5 years As indicated above, the index of sex seg- after high school (Astin and Myint,1971), regation in occupational aspirations deter- to between 15 percent (Kohoutand Roth- mined from the 1979 NLS over the full range ney, 1964) and 25 percent(Kuvlesky and of the Census Bureau's detailed occupa- Beeler, 1967) in studies of men done 10 years tional categories was 61.0. This figure can after high school. Conclusions aboutchanges be compared to a figure of 66.1, measuring in congruence with time afterhigh school the degree of sex segregation in the labor

209 _

SEX TYPING IN OCCUPATIONAL SOCIALIZATION 203

N., TABLE 11-1Percentage Distribution ofOccupationalAspirationp and Attainments by Sex Aspirations° Attainments" Percentage of Female Incumbents Females Males Females Males <30 34.6 86.5 12.5 78.0 30.59 12.6 9.4 17.6 14.7 a60 52.8 4.1 69.9 7.3 Total 100.0 100.0 100.0 100.0 N 4,036 5,073 27,497,081 46,028,117 Data from National Longitudinal Survey of Young Americans (NLS), 1979, is Data from 1970 Census of Population, Vol. 2, pt. 7(a), Table 1. market in 1976 for the same set of Census fined in greater detail in Table 11-1, which Bureau categories (U.S. Commission on Civil presents data on aspirations from the NLS Bights, 1978). The comparison indicates that and data on attainments from the 1970 cen- the aspirations of youth are almost as highly sus. Occupations are divided into three sex- sex segregated as the occupations held by type categories on the basis of the percent- those currently employed. age of female incumbents in the occupation. It was observed earlier that the occupa- It can be seen that the percentage of young tional aspirations of males are more highly women aspiring to typically male occupa- sex typed than those of females. Analysis of tions (34.6 percent) is considerably greater the 1979NLS data indicates that the mean than the percentage, of women actually em- percentage of males actually employed in ployed in those occupations in 1970 (12.5 occupations aspired to by young men is 87.9 percent). The percentage of young men as- percent, whereas the mean percentage of .piring to typically male occupations (86.5 females employed in occupations aspired to percent) is also greater than the percentage by young women is 56.5 percent. These fig- of men employed in those occupations in ures can be compared with ones calculated 1970 (78.0). Again, however, the difference using 1970 census data, which describe the between aspirations and attainments is shown degree to which jobs actually held by women to be greater for women than men. It is and men are sex typed.2 The mean per- difficult to interpret these differences. They centage of males employed in occupations may indicate that females aspiring to typi- held mostly by men in 1970 was 82.3; the cally male occupations are, in fact, less likely mean percentage of females employed in than males aspiring to those occupations to occupations held mostly by women was 70.3. realize their occupational goals. On the other The jobs actually held by both men and hand, they may reflect an increased tend- women therefore have, on the average, a ency on the part of younger women (the somewhat higher percentage of female in- NLS sample) to seek entry into occupations cumbents than the jobs aspired to, but the that are currently male-dominated. difference between aspirations and attain- To conclude, our comparison of the sex ments is greater for women than men. typiNg of occupational aspirations and at- The distribution of occupational aspira- tainments indicates that the degree of sex tions and attainments by sex type is exam- segregation in aspirations is only slightly lower than the degree of sex segregation in employment. This overall similarity between 2 These figures were calculated from data report- the sex typing of occupational aspirations ed in the 1970 Census of Population, Vol. 2. pt. 7(a), and attainments indicates that influences prior Table 1. to labor market entry play an important role

210 .

204 MARGARET MOONEY MARINI AND MARY C. BRINTON in the determination of occupational out- increasing age in any precise way have been comes for individuals.However, young peo- few. Studies of the extent to which children ple of both sexes are more likely to aspire look at work in terms of sex-appropriate cat- to typically male occupationsthan adults of egories indicate little change in occupational the same sex are to be employed in those sex typing with age overthe elerrientary occupations, and this difference is greater school years (Hartley and Klein, 1959; for females than males. To understand more Schlossberg and Goodman, 1972). Two re- fully the relationship between the sex, type cent studies, however,, suggest thatfemales of occupational aspirations and attainments, may become slightly moreliberal during the research on samples of individuals studied elementary years about the jobs they feel over time is needed, It isparticularly im- should be open to both males and females portant that such research beundertaken (Cummings and Taebel, 1980; Umstot, 1980). now, since social changeis likely to be pro- Such an increase in liberality would be con- ducing differences between cohorts. sistent with other studies reporting somewhat less sex stereotyping of occupations in Development Over the Early Stages of the adoleicence and early adulthood than at Life CourseStudies of the occupational as- younger ages (Harmon, 1971;Shephard and pirations of preadolescents (preschooland Hess, 1975). There is also evidence that a elementary school children) indicate that sex sex difference emergeswith age in the de- differentiation in ocCupational goals appears gree to which. occupations are sex stereo- at an early age (Looft, 1971a,b;Siegel, 1973; typed,. !with females becoming more likely Harris, 1974; Hewitt, 1975; Papalia and, than males to view an occupation as appro- Tennent, 1975; Umstot, 1980). Children tend priate for either sex (Shepard and Hess, 1975; to look at activities, includingwork, in terms Nieva and Gutek, 1981:12). of sex-appropriate categories, viewing par- Our own analysis of sex typing in the oc- ticular activities as appropriate only for males cupational aspirations of youth between the or only for females(Hartley and Klein, 1959; ages of 14 and 22, based onthe 1979 NLR, Schlossberg and Goodman, 1972; Tibbetts, found a small decline in the sex typing of 1975; Tavris and Offir, 1977:186; Cummings occupational aspirations dyer this age range. and Taebel, 1980; Umstot, 1980).Girls' oc- The index of segregation, indicating the per- cupational preferences are heavily concen- centage of one sex that would have tochange trated in the occupations of teacher and nurse, occupational aspirations to make the aspi- which often account for 50,to 75 percent of ration distributions of the two sexesequal, their occupational choices (Clark, 1967; Looft, was 67.6 for 14-and 15-year-olds and 61.5 1971a,b;. Siegel, 1973; Hewitt, ..975). The for 20- to 22-year-olds. That only a small range of optionsconsidered by girls is typ- amount of change occurs over this age range ically narrow, whereas boys' choices aredis- is confirmed by the findings of twoearlier persed among more occupations(Looft, studies. Based on analysis of the National 1971a; Siegel, 1973; Hewitt, 1975;Papalia Longitudinal Surveys of the Labor Market and Tennent, 1975). Studies of thediscrep- Experiences of Young Men and Young ancy between aspirationsand expectations Women initiated in 1966 and 1968, respec- further indicate that girls are morelikely tively, Hofferth (1980a) found little change than boys to expect to enter occupationsthat in the sex typing of occupations aspired to are sex typed to the same or ahigher degree for age 35 from grade 9 through the first than their aspirations (Looft, 1971a:366; Pa- three years after high school. Similarly, in palia and Tennent, 1975). an analysis based onthe 1973-1974 assess- Attempts to assess developmentalchanges ment of career and occupationaldevelop- in the sex typing of occupationalchoices with ment conducted by the NationalAssessment

211 SEX TYPING IN OCCUPATIONAL SOCIALIZATION 205 of Educational Progress, Gottfredson (1978) employment of married women. Within this presents data showing little difference in the climate of general attitude change, changes sex typing of occupational aspirationsbe- appear to be occurring in the sex segregation tween 13-year-old and 17-year-old students. of occupational aspirations and plans among Although there may be a small overall high School students. Studies examining such decline in the sex typing of occupational as- changes consistently indicate a decline in pirations among youth, the occupational sex segregation, although the precise amount choices of college students have been found of declineis difficult to assess (Garrison, 1979; to become more sex typed over thecollege Lueptow, 1981; Herzog, 1982). years (Davis, 1965; Astin and Panos,1969; Studying the occupational goals of Wis- Hind and Wirth, 1969). Of entering fresh- consin high school seniors in 1964 at .d p75, men planning careers in male occupations, Lueptow (1981) found that the percentage women were more likelythan men to switch of girls planning to enter predominantly fe- to some. other occupational choice during male occupations dropped from 79.7 per- their undergraduate years. Women were also cent in 1964 to 49.8 percent in 1975. This less likely .to be recruited into a male oc- trend was offset somewhat by choices of a cupation from some other field. In contrast, number of new sex-typed occupations. 'in men were less likely to remain in orbe re- 1975. The overall drop in the proportion of cruited into a female field. This pattern of females expecting to enter predominantly change was paralleled by a siinfiar pattern female occupations was, therefore, only 16.2 of change in undergraduate majors. Men were percent. Changes in the demographic com- more likely than women to remain in orshift position of the schools studied between 1964 to business, engineering, the physical sci- and 1975 and a low response rate in 1975, ences, and mathematics,whereas women howeer, raise some question about the ac- were more likely to remain in orshift to curacy of these estimates. majors in the arts, humanities, and educa- Garrison (1979) examined changes in the tion (Astin and Panos, 1969; Zinberg, 1974; sex segregation of occupational expectations Ernest, 1976). among Virginia high school seniorsbetween It can be concluded, therefore, that sex 1970 and 1976. Using a 7-category measure differences in occupational aspirations ap- of expectations, he found that the index of pear at preschool ages and aremaintained segregation comparing the occupational dis- into adulthood. They, may decrease slightly tributions of the two sexes dropped from during late adolescence and early adulthood 43.6 in 1970 to 38.2 in 1976. Herzog (1982) but have beenitiound to increase during the examined changes in the sex segregation of college years. To accurately document the occupational plans between 1976 and 1980 development of sex differences in occupa- for national samples of high school seniors tional aspirations over the life course, how- surveyed annually. Based on a 15-Category ever, will require longitudinalstudies of co- measure of occupational expectations for age horts over time, which have not yet been 30, she found that the index of segregation carried out. between male and female choices declined from 49.8 in 1976 to 36.3 in 1980:All three Recent Historical TrendsIn the wake of of these studies indicate declining sex seg- the women's liberation movement, attitudes regation in the occupational goals of high about the appropriate roles of women and school seniors, although estimates of the men have been changing (Mason etal., 1976; precise amount of decline in each study de- Spitze and Huber, 1980; Thornton and pend heavily on the way in which occupa- Freedman, 1979). Part of this change in- tional goals are categorized. More detailed oc- volves a more favorable attitude toward the cupational classifications undoubtedly would

212 '206 MARGARET MOONEY MARINE AND MARY C. BRINTON J indicate higher degrees of sex segregation and an occupation. In othewords, boyslearn could affectthe amount of change 'ob- early to direct their attention to the status served over time. and monetary rewards of jobs, whereas girls pay more attention to altruistic concerns and personal fulfillment.. Occupational .Knowledge Three experimental studies that examine Given the narrower range of female oc- the effect of providing occupational infor- cupational aspirations and the sex typing of mation to children (Thompson and Parker, both male and female choices, it is of inter- 1971; Barclay, 1974; Harris, 1974) indicate est to consider whether the two sexesdiffer that knowledge alone plays a limited role in in their knowledge of occupations. As in- determining the occupational choices of dicated above, an understanding of what the young males and females. These studies in- adult world views as tale and female jobs dicate that, unless the presentation of jobs is acquired by children early in life, and the and career information includes examples of ability to identify occupations and describe women in nontraditional roles or encourages them increases rapidly as children enter ad- discussion of sex-role stereotyping, the pro- olescence (DeFleur, 1963; Nelson, 1963). vision of information does little to heighten Research indicates that neither boys n students' awareness of sex typing or to girls are significantly superior in their abili broaden their occupational aspirations to to name and describe occupations (Nelson, include jobs atypically held by their sex. 1963; O'Bryant et al., 1980), although sex Because there appears to be a relationship differences do appear in children's assess- between the job-rilevant information chil- ments of different job dimensions. Boys seem dren process and the values they hold regard- to be more aware of the status and monetary ing occupations, we will now examine the rewards of jobs than are girls (DeFleur, 1963; evidence on sex differences in occupational O'Bryant et al., 1980) and to assimilate this values. information early. Girls do not become as aware of these dimensionsuntil adolescence Occupational Values (O'Bryant et al., 1980).3 In rating the importance of service provided to the com- Sex differences in the values placed on munity by an occupation, each sex gives various dimensions of jobs have been doc- higher ratings to occupations dominated by umented across age groups in studies dating members of their own sex. Topther, these back to the 1930s. Witty and Lehman (1930) studies indicate that although preadolescent reported that, across ages ranging from 8 to males and females have comparable super- 18 years, boys showed a consistently greater ficial knowledge of adult job rolesthat is, tendency than girls to aspire to jobs they they are able to identify roles and describe judged to have high monetary returns. The their duties - =their sensitivity to the re- public respect believed to be associated with wards associated with these roles (respect jobs also played a larger role in the choices and status, money, a feeling of providing of boys than girls. Both of these differences community service) may be conditioned by increased with age, indicating that girls either the values they learn to consider in choosing increasingly looked to marriage as a means of obtaining financial support or became increasingly aware that many financially prof- 3 O'Bryant et al. (1980) used the responses of college itable and highly respected occupations were students as a standard by which to measure the accu- not open' to them. More recently, O'Hara racy of preadolescent responses. It is notknown how well these college students' perceptions would corre- (1962) found financial rewards to be a stronger spond to those of a sample of older adults. motivating force for elementary school boys

213 SEX TYPING IN OCCUPATIONAL SOCIALIZATION 207 than for girls. Singer and Steffire (1954) also dicate similar differences in male and female found significant differences in the impor- job values (Peng et al., 1981). These differ- tance male and female highschool seniors ences. also were found in twoother recent attached to job dimensions, with females at- studies based on smaller, nonnationally rep- taching more weight to the amount of in- resentative 'samples (Brenner and Tornkie- terest and opportunity tohelp people pro- wicz, 1979; Tittle, 1981). Thus, despite some vided by jobs, and males attaching more evidence of declining sex differences in the value to jobs they saw as offering high mon- occupational plans of adolescents in the 1970s, etary rewards, an opportunity towork more which we discussed earlier, sex differences or less on one's own,and the chance to be in occupational values persist. the boss. Since the two sexes differ upon entry into In the last few years, research has con- the labor force not, only in the attitudes, tinued to document these sex differences in knowledge, and values they hold about oc- occupational values.4 Studies attempting to cupations but also in the skills and personal- assess recent changes inthe job dimensions social attributes that affect access to occu- to which males and femalesattach signifi- pations, we will consider evidence on the cance have foundsurprisingly little conver- existence of sex differences in abilities and gence between the. sexesduring the 1970s dispositional traits. (Lueptow, 1980; Herzog, 1982). Lueptow of (1980) compared the occupational values Abilities and Dispasitional.Traits graduating seniors in 1961 and 1975 and found that at both time points males placed sig- Evidence pertaining to sex differences in nificantly greater value on status, money, many abilities anddispositional traits- has freedom from supervision, and leadership been reviewed recently by a number of psy- than did females. Females valued working chologists (Oetzel, 1966; Maccoby and Jack- with people, helping others, using their lin, 1974; Block, 1976; Tavris and Offir, 1977; abilities, and being creative more than males Frieze et al., 1978). Maccoby and Jacklin's did. By 1975 there was some indicationof (1974) extensive review of approximately increased male interest in working with peo- 1,600 studies published, for the most part, ple, but increased female interest inthe between 1966 and 1973 has formed the basis stereotypically masculine-valued dimen- for most later reviews, which expand upon sions of money, status, freedomfrom suit and, in some cases, reinterpret itsfindings pervision, and leadership was notevident. (Block, 1976). In general, research to date Herzog's (1982) analysis of data from the permits few definitive conclusions tobe Monitoring the Future projectreplicated drawn about the existence of sex differ- these differences for successive cohortsof ences, much less their origins(Block, 1976). high school seniors between 1975and 1980. There is some evidence to indicate the ex- In addition, data collected fromsophomores istence of sex differences favoring males in and seniors in a nationally representative quantitative and spatial abilities and sexdif- sample of U.S. high schools as part ofthe ferences favoring females in verbal abilities High School and Beyond survey in 1980 in- (Terman and Tyler, 1954; Tyler, 1965; Oetzel, 1966; Dwyer, 1973; Maccoby and Jacklin, 1974; Block, 1976). However, although these differences appear rather consistently across 4 A study of black inner-city high school students, studies, they are small. It has been esti- however, found no significant differences betweenmales mated that in the large-sample studies re- (Brief and females in terms of valued job dimensions viewed by Maccoby and Jacklin (1974), sex and Aldag, 1975). indicating that sex differences may nut be uniform across racial groups. accounted for only 1 to 2 percent qf the

14 208 MARGARET MOONEY MARI I AND MARY C. BRINTON.

variance in reading performance and 4 per- In conclusion, there is evidence to indi- cent of the variance in mathematics per- cate the existence of some sex differenCes formance (Plomin and Fock, 1981). in personality traits and abilities, but most Sex differences in dispositional traits also appear to be small and, as discussed earlier, have been documented, although the evi- result primarily from sex-role socialization. dence pertaining to these traits is weaker. That is, most appear to be learned and :an Among the most well documented sex dif- be viewed as a product of differences in, the ferences is the tendency for males to be more expectations society holds for the two sexes aggressive than females (Maccoby and Jack- rather than as-a cause of those differences. lin, 1974; Block, 1976). There is also some It has been argued that sex differences in evidence to indicate that males are more personality traits and abilities form the basis doniinant, possessing a stronger, more po- for sex differences in occupational choice and, tent self-concept; are more curious and ex- ultimately, for the allocation of males and plorative; more impulsive; and more active females to different positions in the labor (Block, 1976). Other evidence suggests that market. Although it is possible that sex dif- females may be more fearful and timid, more ferences, particularly in physical character- susceptible to anxiety, less confident in task istics, are the basis for some occupational performance, and more likely to seek help sorting by sex, the relevance of most stere- and reassurance (Block, 1976). Females also otypically ascribed sex differences in per- appear to maintain closer proximity to friends sonality and ability, including physical dif- than do males and to be influenced more by ferences, to occupational performance the social desirability of behavior (Block, remains unknown. That is, it is unknown to 1976). No sex differences have been found what extent jobs that are traditionally held in nurturance and maternal behaviors, such by one sex and that are thought to require as helping and sharing, in general self- sex-related traits actually do. It seems likely esteem, in achievement orientation, and that the extent to which one sex is better in degree of auditory orientation (Maccoby suited to perform sex-typed jobs has been and Jacklin, 1974; Block, 1976). greatly exaggerated. Because sex, differ- Research on sex differences in physical ences in personality traits and abilities are strength and ability has been reviewed less smaller than they are stereotypically as- systematically but indicates substantial dif- cribed to be and are of questionable rele- ferences favoring males in upper body vanceN:, to the performance of most jobs, their strength and smaller differences favoring role in the determination of sex segregation males in leg strength (Wood, 1980). Fe- in the labor market is likely to be minimal. males show somewhat greater tolerance for Ilcat than do males and have more body fat, Summary which gives them an advantage in some activities requiring endurance (Wood, 1980). Marked differences exist between males Although it has not been replicated re- and females in occupational orientation. From cently, early research indicated that boys very young ages, females aspire to and ex- possess greater speed and coordination of pect to enter typically female occupations, gross body movement (Maccoby and Jack- whereas males aspire to and expect to enter lin, 1974). Girls are generally believed to typically male occupations. These sex dif- have better manual dexterity than boys, but ferences in occupational choice have an im- sex differences in dexterity depend on the portant bearing on subsequent sex segre- task observed. Girls have been found to have gation in the labor market, since the degree somewhat better finger dexterity, but they of sex segregation in occupational aspira- do not have better overall manual dexterity tions prior to labor market entry closely ap- (Maccoby and Jacklin, 1974). proximates the degree of sex segregation in 215 4

SEX TYPING IN OCCUPATIONAL SOCIALIZATION 209

the labor market. The sexes also differ in the family. Children not only interact with the importance they attach to various di- their parents earlier and more frequently mensions of jobs, with 'females valuing the than with other adults but they also have intrinsic interest of a job and the opportu- strong emotional ties to their parents that nity it affords to work With and help others reinforce the effects of parental actions. more than males do, and males valuing the Studies of modeling in laboratory settings extrinsic rewards of a job, including status, suggest that parents are effective models for money, and power, more than females do. their children because they are the most Although sex differences in abilities and dis- nurturant and powerful people with whom positional traits exist prior to entry into the a child interacts (Bandura and Huston, 1961; labor market, theAre unlikely to play an Bandura et al., 1963; Mische.l and Grusec, important role in the determination of oc- 1966). In recent years, researchers have cupational segregation. Most differences are sought primarily to address two questions small and appear toiresult from, rather than about parental effects on sex typing. First, 'cause, sex differences in occupational ori- do parents treat boys and girls differently? entation. The-relevance of sex differences Second, do parents serve as role models for in abilities and dispositional traits to occu- the adoption of nontraditional role behav- pational performance is also open to ques- ior? Evidence spertaining to these questions tion, since it is unknown to what extent jobs is reviewed below. that are traditionally held by one sex and that are thought to require sex-related traits actually require those traits. Differential Treatment of Boys and Girls by ParentsBased on a review of research focusing primarily on the sex typing of per- DETERMINANTS OF SEX DIFFERENCES sonal and social behaviors, Maccoby and IN OCCUPATIONAL ORIENTATION Jacklin (1974) concluded that there were As discussed above, sex differences in oc- surprisingly few differences in parents' cupational orientation and joblelevant traits treatment of boys and girls. No differences can be seen as arising largely from the proc- were found in the total amount of interaction ess of sex-role socialization, which begins at between parents and infants or in the amount a child's birth. Messages about what is viewed and kind of verbal interaction. Based on ob- as appropriate for the two sexes are con- servational studies, there was little evidence veyed in myriad ways and constitute the that children of the two sexes received dif- basis on which the two sexes learn to have ferent amounts of parental warmth or re- different expectations about the roles they inforcement for dependency. Similarly, there will fill and learn to behave differently. In was no evidence that parents responded dif- this section, we examine some of the major ferently to aggression in boys and girls. sources of sex-role learning, focusing on fac- However, boys were given more gross motors that are likely to affect the development tor stimulation and were encouraged in of sex differences in occupational orientation physical activity more than girls were. Boys and preparation. Specifically, our review also received more praise and more punish- covers family influences, school influences, ment', particularly physical punishment. messages transmitted by the mass media, Likewise, buys received more pressure than and early employment experiences. girls not to engage in six- inappropriate behavior. The fact that more differences in parental treatment of boys and,girls were not Family Influences documented may be attributable to limita- The earliest and most pervasive influ- tions of the data base reviewed by Maccoby ences on sex-role socialization arise within and Jacklin and to the procedures they used

216 210 MARGARET MOONEY MARINI AND MARY C.!MINTON

. In drawing conclusions fromthe evidence parents indicate that parentssocialize (Block, 1976). achievement differently for boys and girls. de- A more recent review of researchby Hus- Parents have higher expectations and ton (in press), focusing notonly on the sex mand more independence of boys, whereas typing of personal and socialbehaviors but they provide help more readily to girls and also on activities and interests,provides more are more likely tofocus on interpersonal evidence of differential treatment of the two aspects of the teaching situation(Huston, in sexes. Hustonreviewed experimental stud- press). ies in whirl] the child'sgender label was These sex differences in iscialization, manipulated as well as observational studies Which are rooted in parents' general sex-role of parents with their own children, Shefound conceptions, can be expected to havelong- evidence that bog) and girls were encour- term effects on the occupationalbehavior of aged in different types of play activitiesfrom the two sexes. Although the linksbetween infancy upward. Adults made sex-typed toy &rents' socialiadion of personal and social choices for children, offering dolls exclu- behavior and occupational outcomes are in- sively to children, they believed to be girli. direct, childbearing practices that affect the Gross motor activity wasencouraged more quisition of job - relevant skills and thede- in boys than in, girls.Adults played more velopment of occupational expectations bear actively with male infants andresponded on occupationalchoice. As we will discuss more positively tophysical activity in boys below, parents who display less traditional than in girls. In contrast,interpersonal play sex-role behavior, and who are therefore activity and dependent, affectionatebehav- likely to hold less traditional sex-role atti- ior were more often encouraged ingirls than tpdes, produce children whose occupational lines. in boys. Fathers emphasizedsex-typed play behavioris less differentiated along sex activity more than mothersdid and inter- acted with boys more than with girls.Like Parental Role Models The fact that moth- Maccoby and Jacklin (1974), Hustoa (inpress)\ ers and fathers tend to be employed indif- found little evidence that aggression pro- ferent jobs outside the home, to perform vokes different parental responses forboys different tasks within the home, and to have and girls. A considerable body ofevidence, different interests and personal and social however, indicates that boys are given more characteristics provides information to chil- opportunities to play away fromhqme and, dren about what is expected of womenand therefore, more freedom from adult super- men. A consistentfinding of previous revision than are girls. search is that children's sex-roleattitudes In addition to treating the sexesdiffer- are less traditionallystereotyped if the mother ently in terms a play activities, it hasbeen is employed outside the home thanif she is well documented that parentshave-higher not (Huston, in press).Maternal employ- expectations for the adult achievementof ment affects the sex-role attitudesof both their sons than of their daughters(Alexander sexes, but it affects the sex typingof personal and Eckland, 1974; Maccoby andJacklin, and social attributes, interests, and activities 1974; Hauser et al., 1976;Hoffman, 1977; almost exclusive)for girls. Employed Marini, 1978a,b). Recent evidenceindicates mothers have been ound to be particularly that parents also value mathematicsachieve- attractive role modefor their daughters, ment more in sons than indaughters and as evidenced bythect that daughters of estimate the mathematics competenceof sons employed mothers more often want to be to be higher than thatof daughters (Fen- like their mothers and' qay they usetheir nema and Sherman,1977; Fox et al., 1979). mothers as models (HofOnan, 1974; Miller, Observations of the teaching behaviors of 1975). However, the effecs of maternal em- SEX TYPING IN 0' CCElPATIONAL SOCIALIZATION 211 ployment may also arise because maternal view research on a variety of socialization employment is associated with other sex- influences arising within the school, includ; role activities, personality characteristics, and ing the availability of same-sex role models, child-rearing practices of parents. sex stereotyping in textbooks and educa- Although there is considerable evidence tional material, the role of counselors in to indicate that maternal employment fos- channeling students into careers, tracking ters career salience among daughters, par- and vocational education, and training in ticularly if the mOthei has a positive attitude mathematics and science. We also consider toward her employment (Beardslee and 0'. evidence on the success of governmental in- Dowd, 1962; Siegel and Curtis, 1963; Dou- tervention through legislation, such as Title van and Adelson, 1966; Hartley, 1966; White, IX, the Women's Educational Equity Act, 1967; Vogel et al., 1970; Almquist and Angr- and the 1976 Vocational Education Amend- ist, 1971; Baruch, 1972; Altman and Gross- inents. man, 1977; Hoffman and Nye, 1975; Bielby, a 1978b; Kaufman and Richardson, 1982:22- Availability of Same-Sex Role ModelsAl- 27), there is less evidence that it affects the though the impTirtance of role models of one's entry of daughters into traditionally male own sex isa' theme that implicitly runs occupations. Studies of small, restricted through much of the research on, sources of samples of college students suggest that ma- occupational socialization, empirical studies ternal employment increases the probability dealing with the processes involved in role of entry into nontraditional fields by daugh- modeling, particularly in the case of ado- ters (Almquist and Angrist, 1970; Almquist, lescents and young adults, are few. Douvan 1974; Tangri,1972; Xlemmack and Ed- (1976) points to the prbminent role of an wards, 1973). However, studies of larger older-woman model in biographies of women samples that are more representative of the successful in such fields as politics and ac- U.S. college population find little or no pos- ademics and suggests the advantages of itive effect of maternal employment on the womens' colleges in providing a broad array entry of daughters into traditionally male of female-role models. In exa lining deter`- occupations (Bielby, 1978a; Brito and Ju- minants of college females' occupational as- senius, 1978). It appearS likely that the type pirations, Brito and Jusenius (1978) found of employment engaged in by the mother that attendance at either a predominantly rather than her employment per se is the female or predominantly male college was factor that influences entry into traditionally associated with atypical aspirations. More male occupations by daughters. Based on a direct evidence that same-sex role models study of high school students who did not may be salient to the educational and career go to college, Hofferth (1980a) found that choices of college students and' to their ca- for whites the sex type of the mother's oc- reer success is available. Basow and Howe cupation had a direct effect on the sex type (1979) indicate that college seniors reported of the daughter's occupation. This relation- their career choices to be significantly more ship may pertain over the full range of ed- affected by same -sex role models than by ucational attainment, but it has not been opposite-sex models. Tidball (1973) found investigated. that the number of women on the faculty at a college was a good predictor of the number of career women the school would produce. School Influences Fox (1974) found that the distribution of col- The role of schools in promoting or in- lege males and females across fields. of spe- hibiting sex segregation in occupational goals cialization closely mirrored the distribution is a multifaceted one. In this section we re- of same-sex faculty, and the degree to which

218 ".

212 MARGARET'M6ONEY MARINI AND MARY c.'BRINTOly:

female students were voncentrited lit cer- Two groups of researchers,. publishing itain areas reflected the concentrationof fe- comprehensive studies of sex. stereotyping male facultyAlthough priorsocialization in elementary schooltextbooks in the mid- conditioning males and females to major in 1970s, found pervasive evidence of sex bias. sex-appropriate fields iivas..no.doubt afactor Weitzman and Rizzo's (1974) study of the influencing this concentration, the resem- illustrations in a sample of the most widely blance of students' choices to the patternof used textbooks 'revealed that males ap- faculty specialization was striking. Finally, peared in 69 percent of textbook illustra- career success hasbeen linked to .exposure tions, whereas females appeared inonly 31 to a same-sex mentor. In astudy of psy'- percent. These figures became moreskewed chology D.'s, Goldstein (1979) found that the higher the grade level. The lopsided having had a same-sex adviser was signifi- representation of the sexes was most appar- cantly related to academic productivity ent in the science field andleast evident in 0 (measured by number of:publications) 4 years social studies, a more feminine field. Whereas later. Although this result links career suc- men wereportrayed overall in more than cess to exposure to a same-se* mentor, at- 150 occupational roles, women appeared as tribution of a causal relationship is difficult housewives or, when working, in a narrowly because it is impossible to tell how the same circumscribed set of roles such ,as nurse, students Would have performed underthe teacher, librarian, and sales clerk. Women tutelage of an opposite-sex adviser.Taken on Words and Images(1975b), a New Jer- as a whole, thesestudies indicate that same- sey-based group of researchers and con- sex role modelinginfluences occupational sultants, also assessed sex stereotyping in decision making, but more research is needed elementary school textbooks. They exam- to further document the existenceand na- ined 134 readers (from 14 different publish- ture of its effects. ers) in use, in 'three suburban New Jersey school districts. Their findings on the por- Sexterentyping in Textbooks and Edu- trayal of occupational stereotypes were as cational MaterialsSex bias in educational follows: (1) women were portrayed in stories materials was recognized as a serious issue and in illustrations in a total of 26 occupa. in the late 1960s, and'several studies con- tions, whereas men appeared in 147differ- ducted in the early 1971)s documentedthe' ent jobs; (2) aside from therole of doctor, existence of sexstereotypidg, particularly in women wereuniversally portrayed in ,ster- mathematics and science textbooks (Milnar, eotypically feminine roles, such as cook, 1973; Rogers, 1975). With increasingpublic housekeeper, librarian, school nurse, teacher, recognition of the problem;and the passage anic,.! telephone operator; and (3) in terms of of certain legislation, such as Title IX in1972,5 biographies, 'which can be viewed its im- prohibiting the distribution of federal funds portant because they showadults in iignif- 27 to. schools that did notcomply with sex-equity leant roles outside the lime, there were practices, biased representationsof the sexes stories about 17 different women,compared in educational materials wereexpected to with 119 stories.about 88 men. The message decrease. However, textbooks published conveyed by these depictions of adult males throughout he 1970s continued to portray and females is that females have a'narrow the sexes in stereotypical roles. range of jobs to choosefrom and generally make less of a contribution to public life than do males. Other investigators have examined sex 5 Legal judgment regarding the First Amendment stereotyping in textbooks used in particular exclu- and freedom of speech has since resulted in the subject areas. Stern (1976), in an examina- sion of curriculum materials from coverageby Title IX.

219 S. .

SEX TYPING..IN OCCUPATIONAL SOCIALIZATION 213

tion of beginning and intermediate foreign Unfortunately, little is known about the language textbooks published between. 1970 effects of sex-stereotyped educational ma- and 1974, .found that both dialogues and terials on children's attitudes and, ulti- photographs depicted a biased range of oc- mately, on their occupational aspirations. cupational roles for males anefemales and Researchers have been guided by the as- .3 focused much more heavily on males' oc- sumption that reading materials do exert a cupational aspirations than on those of fe- pervasive influence on young readers' views males. Sex stereotyping in mathematics and motivation, but the extent and perma- * textbooks is particularly well documented, nence' of such effects have generally not been Kepner and Koehn (1977) evaluated 24,first-, investigated. Two exceptions are reports by fourth-, and seventh-grade mathematics texts Kimmel (1970) and Nilsen (1977) of exper- put out by 8 major.publishers between 1971 imental reading programs and short-term and 1975 to determine. their representation controlled experiments. Nilsen reports a di- of the sexes. They found that elementary rect correlation between the degree to which mathematics texts tended to depict males in children classify activities as belonging in a greater variety of occupationsthan female and female domains and length of ex- males, both in illustrations and in written posure to the Alpha One reading program, problems, and that sex stereotyping of oc- identified as presenting highly sex-stereo- cupations was pfevalent in these texts. " typed images to children. Kimmel reports These researchers also examined three on several experimental studies that appear widely advertised mathematics textbooks to have an effect on children's stereotypical published between 1975 and 1977. In these attitudes (in this case, of minorities) and contexts females were shown in a wider range cludes cautiously that books may play a sig- of occupations than had been the case in nificant role in conditioning children's atti- previously published textbooks; in fact, tudes, although the duration of the effect of women were shown as moving into typically specific books may not be long. male occupations, such as 'doctor,truck driver, and 'political candidate. However, Counseling and Career GuidanCe There males were not depicted in traditionally fe- is a large body of literature in the fields of male occupations. Another study of sex bias educational and counseling psychology deal- in six widely used elementary mathematics ing with issues related to sex bias in career textbook series published betweetx,1970 and counseling. Although we can deal with this 1975 indicated that the two series published literature in only a cursory fashion here, we most recently showed more occupational role consider findings with implications for the reversals for males and females than those career choices of women. The topics we touch publiihed earlier in the decade (Steele, 1977). on include counselor bias in assessment and Sex-role stereotypes are prevalent not only counseling concerning women's career pref- in textbooks but also in children's picture erences, sex bias in occupationalreference books. Studies by Weitzman et al. ,I972) materials used for counseling purposes, and and Nilsen (1977) have demonstrated the the importance of the counselor in effecting predominance of male over female charac- changes in females' occupational aspirations and outcomes. ters in books that were winners of the pres- - tigious Caldecott Medal, awarded yearly to Studies of counselors'attitudes and a book oriented towardpreschool children. knowledge about women's careers suggest Nilsen documented an actual decline in the that counselors contribute to the sex ster- percentage of female characters during the eotyping of occupations. A nufnber of stud- 1950s and 1960s, from 46 percent in 1951- ies have documented the existence of tra- 1955 to 26 percent in 1966-1970. ditional attitudes on the part of counselors

220 214 MARGARET MOONEY MARINI AND MARY C. BRINTON

'regarding appropriate roles for women. In included whether women need more em- a much-cited experimentalstudy, Thomas ployment alternatives, whether most women and Stewart (1971) found that counselors, can perform the roles of workeLandMime- 'regardless of their sex, perceived conform- maker simultaneously in a satisfactory fash- ing career goals (such as home'economist) ion, and whether women are discriminated as being significantly more appropriatefor against in employment practices. women than deviate careergoals (such as Several researchers have examined career engineer). This perception was correlated information materials published by private with counselors' level of experience; more publishers and the federal government for experienced counselorsverceived no statis- their portrayal of occupational roles for males tically significant difference in the appro- and females (Heshusius-Gilsdorf and Gils- priateness of the goals. Female clients who dorf, 1975; Lauver et al., 1975). Thesestud- purportedly held deviate goals were also ,ies indicate that women tend to be both judged to have a significantly greater need underrepresented and portrayed in tradi- for counseling than those with conforming tionally female occupations in illustrations goals. and accompanying job descriptions. Maki; More recent studies have continued to on the other hand, tend to beportrayed in indicate that school counselors have a re- traditionally male occupations. Two popular stricted view of the occupations appropriate and widely used career orientation textbook for women (Medvene and, Collins, 1976) and series put out by private publishers por- suggest different occupational choices for in- trayed top-management, professional, and tellectuAy gifted males and females (Don- technical positions as being filled almost ex- ahue and Costar, 1977). In a study focusing.,clusively by men, and they portrayed ex- on counselors' perceptions of thevariables tremely skewed sex ratios for traditionally important in college students' career choices, sex-stereotyped jobs such as clerical work- counselors perceived women to be'more in- ers; steward(esse)s; nurses; and construc- , fluenced by such considerations as success tion, large machine, and repair ,workers avoidance, home-career conflict, and atti- (HeshusiuS-Gilsdorf and Gilsdorf, 1975). A tudes of the opposite sex than women restudy of the 1974'4975 Occupational Out- ported themselves to be (Karpicke, 1980). look Handbook, published by the U.S. Bu- Male counselors generally seemed to dis- reau of Labor Statistics, alsofound evidence play somewhat more bias than female cowl- of sex bias in career portrayal (Lauver et al., selors, and at least one study suggests that 1975), although a more recent study of the school counselors may be more biased than 1976-1977 edition reported substantial re- other types of counselors, such as psycho- duction in the amount of sex bias (Farmer therapists (Medvene and Collins, 1976). and Backer, 1977). In a study of the accuracy of counselors' In general, research on counselors' atti-: knowledge of labor force issues relating to tudes and knowledge, and on the career' in- women, Bingham and House (1973)found formation materials they use, indicates that that a sample of secondary school counselors bias exists in the perceived appropriateness demonstrated correct knowledge on only 12 of a variety of occupational aspirations for of 25 icems. On 7 of the items Nast fre- women, in the roles women are assumed to quently missed, significantly more female fill in the labor market, and in the reasons than male counselors responded correctly. perceived to lie behind career choices. Vir- Although the authors identified all items as tually every article in this area concludes factual, those missed moreCfrequently by with a recommendation that school coun- males than females seemed to be especially selors be required to receive exposure to open to influence by attitudes. Examples statistics on female labor force participation SEX TYPING IN OCCUPATIONAL SOCIALIZATION 215

and information on employer biases in var- in the case of assignment to college and ious fields. The provision of accurate occu-'noncollege preparatory tracks, counselors' pational information and training to alert judgments of students' personalities often school counselors to the possibility that their play a large role (Cicourel and Kitsuse, views may be sex biased is undoubtedly im- 1963), a finding that provides a disturbing portant. However; it should be recognized complement to our discussion of counselor that little is known about the degree to which. bias related to the sex of the student. counselors actually influence students. Finally, students frequently are unaware of Understandably, the counseling literature the implications of the tracks into which tends to attribute a strong role to counselors they have been guided or assigned, and as possible agents of social change (Vetter, frustration about plans for postsecondary 1973;.Verheyden-Hilliard, 1977). But stud- education may result (Rosenbaum, 1980). ies that attempt to assess the impact of coun- Analyzing data from the National Longitu- seling on students are few. In one careful dinal Survey of the High School Class of study of the effects of counseling on New 1972, Rosenbaum (1980) showed that: for York high school studehts, Rehberg and both males and females, the actual track had Hotchkiss (1972) found that nonschool influ- a greater influence oncpllege attendance ences sopli as socioeconomic status,intel- than did students' perceptions of the track to which they belonged. Misperceptions 411 ligence, parental encouragement, and pre- ,viously held educational expectations aside, therefore, track placement head an together exerted a much greater influence objective impact on educational outcomes. on students' educational expectationsthan Given the irhportance of tracking as a their exposure to counseling. In another study mechanism that sorts students into-groups of a small sample of high school students, on the, basis of presumed or'stated abilities Tittle (1981) found that only 27 percent of and preferences, the distribution of males those surveyed reported having talked to a and females in different vocational prepara- counselor about work possibilities. tion tracks has potentially important consequences. Studies of the distribution of high Tracking and Vocational EducationEm- school students across educatigpal curricula pirical studies of tracking have typically in the late 1960s and early 1970s indicated focused on the determinants and implica- that, when business and office programs tions of student placement in college and were included in the definition of vocational noncollege preparatory tracks rather than education, 20 to 44 percent of senior girls on specific vocational courses. However,the were enrolled in vocational eduCation pro- findings of these studies have some impor-. grams (Grasso,1980; Harnischfeger and tant implications that are generalizable to all Wiley, 1980; Hofferth, 1980b). Relatively types of tracking that segregate males and few girls were enrolled in the vocational females and thereby affect their occupational education track per se, since it tends to orientations and outcomes. First, a review provide training for entry into blue-collar of the tracking literature by Rosenbaum occupations. However, girls who did enroll (1980) underscores the degree to which in the vocational education track did not students are labeled within the school and differ in ability from those in other noncol- 4 the community once they embark on a lege preparatory tracks. In contrast, boys particular track, regardless of whether track who enrolled in the vocational education placement is a result of their own free choice track tended to be lower in ability than those or counselor assignment(Cicourel and inother noncollege preparatorytracks Kitsuse,1963; Schafer and Olexa, 1971; (Harnischfeger and Wiley, 1980). The overall Heyns, 1974; Rosenbaum, 1978). Second, distribution of females across vocationally 2 " ......

41,

216 MARGARET MOONEY MARI AND MARY C. BRI TQN_

oriented programs differs maikedly from Young Women, collected in1466 and 1968 that of males, and there has been no sub- and in 'three subsequent follow-ups, indi-" stantial change in the sex typing of such cated that those enrolled in a vocational Programs between 1972 and 1978 (American education track were less likely to be Institutes for Research, 1980). Health,home employed in female-typical jobs 10 years - economics,and business and office programs after high school than those enrolled in the are enrolled inprimarily by females (over 75 cgeneral, commercial, and college prepara- percent of those enrolled in 1972 and 1978 tory tracks, which tend to prepare-students were females), andtechnical, agricultural, for white - collar jobi. and trade and industrial programs are Both the studies by Grasso (1980) and enrolled in primarily by males (over 75 Hofferth (1980b), howeier, indicated that percent of those enrolled in 1972and 1978 although females who had been enrolled in were males). Only one program,retail sales, vocational programs were less likely to be isenrolledinby approximately equal employed in female-typed jobs, the jobs numbers of females and males (45 percent they held wereleiseconomically desirable of those enrolled in 1972 and 42 percentof than the jobs hel by females who had been those enrolled in 1978 wtie females). enrolled in business and-office, general, and Since the purpose of vocationally oriented college preparatory programs. Grasso (1980) programs is to preparestudents for particular found that those enrolled in bUsiness and types of occupations, it can beexpested that office programs had higher hourly wages the sex typing of suchpihretfis affects (and yearly salaries) 4 years after high school subsequent occupational segregation by sex. than their peers enrolled in the general A study by Grasso (1980) of females whodid track. (Female vocational education enroll- not go. to college, based ondata from the ment was too low to permit examinationof National Longitudinal Survey of the Labor its effect on earnings.) Hofferth (1980b) Market Experiences of Young Women found that those enrolled in a vocational collected in 1968 and 1972, indicated that education track received lower wages 10 girls in all curricula had highly sex-typed years after high school thanthose enrolled occupational aspirations but that those in in other programs. Black females enrolled business and office programs were the most in business and office programs maintained traditional, with 69 percent aspiring to jobs\a wage advantage 10 years after high school , that were 80 to 100 percent female. In compared with those. enrolled in other accordance with their aspirations, female prdgrams, although white females did not. business and office program students were For females attending high school in the more likely thanothers to hold female - 1960s, therefore, enrollment in a vocational typical jobs 4 years later, with 65 percent. education track that prepares students for holding jobs thatwere80 to 100 percent typically male blue-collar jobs did not result female. Females who had been enrolled in in employment in high-earning occupations. vocational programs, wherevocationalre- It is possible, however, that' increased fe- fers to programs other than white-collar male enrollment in vocational education and clerical programs, were less likely tohold greater awareness of discriminatory prac- female-typical jobs than those enrolled in tices in male-typed blue-collar occupations business and office, general, and college could increase the number of women preparatory programs.Another study of entering the more desirable blue-collar students who did not go to college by occupations and thereby produce earnings Hofferth (1980b), ,based on data from the gains for women. National Longitudinal Surveys of the Labor Planned interventions to decrease sex Market Experiences of Young Men and segregation in vocationally oriented pro- t.

el SEX TYPING IN OCCUPATIONAL SOCIALIZATION 217

grams have shown mixed results. Evenson cialized to view them as such, females' are and O'Neil (1978) reported on several less likely than males to perceive training projects in the 1970s that succeeded in in science and mathematics as useful and to markedly increasing female enrollment in be confident of their ability in these sub- trade and industrial courses. Components jects. that seemed to be related to program success For the elementary school years, when included attention to the training of teachers the two sexes receive comparable training and counselors in ways of eliminating sex in mathematics and science, studies rarely bias, provision of information and individual report sex differences in aptitude ore counseling to students, and the training of achievement in these areas (Fennema, 1974a; students to offer epeer guidance. Programs Maccoby and Jacklin, 1974; Fox et al., 1979).

focusing on change in ,both teachers.and At the secondary and postsecondary school students offered the advantage of an ap- levels, sex differences in performance on proach that was well integrated. In contrast, standardized tests are evident (Maccoby and Waite and Hudis (1980) reported on the Jacklin, 1974), but studies have frequently AMI11.14 more limited success shown by a number of failed to control for differential exposure to programs. Although itis not possible to courses (Fox et al., 1979). Thus, the sex dif- undertake a detailed comparison here, it ferences observed may be attributable to sex appears that interventions concentrating on differences in motivation to study mathe- heightening student awareness of sex ster- matics and the physical sciences rather than eotyping do not result in altered occupational to sex differences in ability. Studies at the choices as much as do programs that also elementary and secondary school levels in- train school staff in issues of sex bias. dicate that males do not report greater liking 41 for mathematics than do females, nor do males Training in Mathematics and Science An show a greater preference for mathematics important difference in the formal education relative to other subjects (Stright, 1960; of the two sexes occurs in the area of training Aiken, 1970, 1976; Callahan, 1971; Ernest, in mathematics and science. 'From high school 1976). Males do, however, show a greater onward, males take more advanced math- preference for science at early irade levels ematics and science courses than females do (Ernest, 1976). At the postsecofidary school (Ernest, 1976; Fennema and Sherman, 1977; level sex differences in attitudes toward both Sherman and Fennema, 1977; Fox et al., mathematics and science are evident (Dre- 1979). This sex difference in technical train- ger and Aiken, 1957; Aiken and Dreger, 1961; ing has important implications for sex seg- Aiken, 1970; Ernest? 1976). regation in the labor market since the poorer It would'appear that, at least at the sec- mathematics and science training of females ondary school level, sex differences in en- prevents them from entering many tradi- rollment' in mathematics 'courses arise not tionally male occupations. As discussed be- from different levels of interest in the. sub- low, recent evidence suggests that sex dif- ject but primarily from the two sexes' per- ferences in mathematics and science training ceptions of mathematics as being differen- arise not so much from sex differences in tially useful to them. Several studies indicate the ability to master these subjects or, at that perception of the future usefulness of least in the case of mathematics, from sex mathematics is an important determinant of differences-in the appeal of the subject as course enrollment in high school (Sherman from the labeling of these fields as male do- and Fennema, 1977; see Fox et al., 1979). mains. Since the physical sciences and Sex differences in the 'expressed usefulne.. mathematics have traditionally been consid- of mathematics have been reported to occur ered male subjects, and children are so- as early as the seventh grade (Hilton and

224 . . . .

218 MARGARET MOONEY MARINI AND MARY C.BRINTON

Berglund, 1974; Fennema and Sherman, Fox et al., 1979) and that they are particu- 1977; Sherman and Fennemp, 1977).There larly likely to interact more with malesthan is also evidenCe that girls are unawareof females in mathematics and science classes al., many of the uses ofmathematics in careers (Levy, 1972; Good et al., 1973; Fox et other than strictly scientific ones (Fennema 1979). A 'bad experience with a teacher is and Sherman, 1977). In addition, sexdif- often the source of a very negative attitude ferences in enrollment in mathematicsand toward mathematics (Ernest, 1976; Poffen- science courses may also existbecause the berger and Norton, 1959), although having two sexes are not equallyconfident of their a good teacheralso is often cited as a positive abilities in these areas. Such sex differences factor in the intellectual development ofgirls in self-confidence are welldocumented (Fox (Anderson, 1963; Ernest, 1976; Fox et al., et al., 1979; Parelius, 1981,1982). There is 1979). When teachers recruit girls formath- some evidence thatthese differences do not ematics programs and havehigh expecta- exist in elementary school (Ernest,1976) but tions for their pelf. ,rmance,they can have that they develop with age (Fennema,1974b; decidedly positive effects (Fox et al., 1979). Ernest, 1976; Fox et al., 1979).There is also Mathematics and science teachers also gen- evidence that they exist regardless ofob- erally provide better role models for boys jective levels of performance.That is, even than for girls from secondary schoolonward when girls get good grades inmathematics since most teachers of thesesubjects are and exhibit higher achievenientthan boys, male (Ernest, 1976). Although, asdiscussed girls perceive themselves tobe less com- above, it is unclear how much influence petent (Fennema, 1974b; Fox etal., 1979). counselors have on students; there is con- just as a variety of socialization agents are siderable evidence to indicate that counselors have been a source of discouragement .responsible for channeling the two sexes into different occupations, a variety of sociali- rather than encouragement to girls wanting zation agents teach childrenthat mathe- to take advanced mathematicsand science matics and science aretraditionally male do- courses (Fox etal., 1979). Such sex ster, be mains. As noted earlier,there is eviuence eotyping by counselors, however, may that parents value mathematicsachieve- declining (Engelhard et al., 1976). interested in , ment more for their sonsthan for their Peer support for females aaugnters and that they estimatethe math- mathematics and science also has been lack- hold a ematics competence of their sons tobe higher ing (Fox et at., 1979). Adolescents than that of their daughters (Fennemaand more negative view ofmathematically gifted Sherman, 197'x;. Fox et al., 1979). There is girls than boys, and high schoolstudents, also evidence that after the sixthgrade both particularly males, view mathematics as a male domain (Ernest, 1976; Fennemaand a sexes get morehelp with mathematics homework from fathers than frommothers Sherman, 1977; Sherman and Fennema, (Ernest, 1976). Learning experiencesof this 1977; Fox et al., 1979). Because the support type within the home canbe expected to of same-sex peers has been found tohave a affect the self-conceptions of maleand fe- positive effect on the mathematicsachieve- critical male children in mathematicsand their en- ment of girls, it is possible that a rollment in such courses (Fox et al.,1979). mass (i.e., a particular sexratio) may be School teachers and counselors also con- needed to provide peer.support for girls and atmosphere in vey messages togirls. that ,may affect their .to maintain an androgynous self-confidence and interest in mathematics mathematics and science classes (Fox et al., and science. There is evidence thatteachers 1979). have different expectations for the two sexes The evidence available on sex differences training sug- in mathematics and science(Ernest, 1976; in mathematics and science

22 ' ""il'91ww?'al.'P'57"Pkv"1"."11411""97427:-.5.1--`- rat: 1,7

SEX TYPING IN OCCUPATIONAL SOCIALIZATION 219 gests that several types of interventions to in the fact that it prohibits discriminatory increase female training in these areas may policies and practices in the treatment of be helpful. Since sex differences in course workers in educational settings receiving enrollment have been found to arise_at least federal funds but also in the fact that it is fn part from sex differences in the perceived the first legislation to specifically protect usefulness of mathematics and science to one's students from sex discrimination. The do- future, earlier education and counseling mains of its coverage of students were orig- programs that make females aware of career inally interpreted broadly to include all ac- opportunities, and the courses necessary to tivities affecting students within educational prepare foi them, may have a significant institutions and agencies receiving federal effect. Previous interventions suggest that funds. It therefore covered admissions.pol- role models are an important component of icies, access and treatment in curricular and such programs (Fox et al., 1979). Increasing extracurricular programs (intluding courses the amount of mathematics and science re- of study, career and course counseling, and quired as part of the basic school curriculum extracurricular activities), and access to stu- also may have a positive -effect, since most dent financial awards. The Women's Edu- sex differences in performance inmathe- cational Equity Act authorized funding at all matics and science and in attitudes toward educational levels for model educational these subjects emerge after it becomes pos- programs of national, statewide, or general sible for students to elect courses. Requiring significance to eliminate sex stereotyping and more mathematics and science training at promote educational equity fOr females. It higher grade levels would reduce the sex thus provided administrative backup for sex- garrin mathematics and science training and equity legislation prohibiting discrimina- might improve the attitudes of girls toward tion. these subjects (Fox et al., 1979:322). Other As a result of a 1984 Supreme Court de- evidence that women planning to major in cision in Grove City v. Bell, Title IX has nontraditional fields show greater attrition been reinterpreted to pertain only to those from these fields during the college years activities within educational institutions that (Astin and Panos, 1969; Zinberg, 1974; Er- directly receive federal funds. This inter- nest, 1976) suggests that provision of psy- pretation, if allowed to stand, would limit 9 chological support by peer advisers (e.g., the jurisdiction of Title IX primarily to ac- other women in nontraditional majors) might cess to student financial aid. Even before slow the rate of attrition ,by women. the reinterpretation of Title IX, a report by the Project on Equal Education Rights (1978), Legislation and Governmental Interven- part of the Legal Defense and Education tion Related to Education Of the legisla- Fund of the National Organization for tion designed to reduce sex discriminaton Women, pointed to some of the difficulties in employment and training, three pieces and red tape in the implementation and en- are particulakly relevant to theoccupational forcement of Title IX. A 1981 report by the socialization of women prior to labor force National Advisory Council on Women's Ed- entry: Title IX of the 1972 Educational. ucational Programs (which was established Amendntents, the Women's Educational by the 1974 Women's Educational Equity Equity Act of 1974 and 1978, and the 1976 Act), however, suggests that there has been Vocational Education Amendments. much progress toward the goals of Title IX, Title IX was passed by Congress in 1972 although the report offers little in the way as part of the Educational Amendments,and of hard evidence on the impact of Title IX. final regulations for its implementation fol- The percentages of degrees earned by lowed in 1975. Its significance lies not only women and the enrollment of women in

226 1. '7 t:: Sr:r `WtS1

220 MARGARET MODNEY,MARINI AND MARY C. BRINTON professional schools do Indicate significant and benefit from vocational education pro- changes over the past decade, although their grams by women is aidedby a 1980 report relationship to the implementation of leg- of the National Advisory Council on Voca- islation is unknown. The percentage of tional Education and the National Advisory bachelor's degrees awarded to women in- Council on Women's Educational Pro- creased from 44 percent in 1971-1972 to 55 grams, This report presents adetailed anal- percent in .1979 -1980. Comparablefigures ysis of enrollment data from 15 states that for master's degrees were 41 and 50 percent; together accounted for 55 percent of the na- for doctoral degrees, 16 and 30 percent; and tional enrollment in high school, postsec- for first professional degrees, 6 and 25 per- ondarjr, and adult vocational education pro- cent. Percentage increases in theenroll- grams. ment of women in professional schools dur- Vocationally oriented programs showed a ing 1972-1981 are particularly impressive: large overall increase in enrollment of 44 dental schools, 1,011 percent; law schools, percent between 1972 and 1978, while the 337 percent; medical schools, 296 percent; increase in the number of women enrolled and veterinary schools, 120 percent (Na- wrs even moredramatic -60 percent over tional Advisory Council on Women's Edu- the 6-year period. The percentage of women cational Programs, 1981). enrolled in traditional programse decreased These data indicate important changes in slightly (from 65 to 56 percent), and the per- the extent and ways in which women'are centage of women in mixed programs and participating in higher education. A study nontraditional prograMs increased by 6,and by Beller (1981) sheds somelight on changes 4 percent, respectively. The increases in in the actual advantages thateducation is women's enrollment in nontraditional pro- providing for women in terms of entrance grams occurred in courseswithout a strong into traditionally male occupations (defined masculine image, such as drafting, graphic as occupations in whichthe male share of arts, and law enforcement, rather than in employment exceeds the male share of the machine shop and Construction. experienced civilian labor force by at least Examination of data for states suggests 5 percentage points). She found that, in 1967, some identifiabledeterminants of changes increases in years of education resulted in in enrollments. Specifically, the greatest in- greater access to male occupations atabout creases in women'senrollment in nontra- twice the rate for men as forwomen but that ditional programs occurred in states where by 1977 this differential had narrowed sub- detailed plans were formulated, involving stantially for those with a college education specific goals and timetables. The more or more. Equalemployment policies were scrutiny to which schools were subjected by credited with these gains for college-edu- the state, the more action was taken. Schools cated women during the decade. Efforts to that were particularly active in attempting achieve equality of opportunity did not ap- to redress skewed ratios of males and fe- pear to have increased access tomale oc- males in vocational areas were those with cupations for those women with 12 orfewer higher nontraditional enrollments of women /ears of education,however. to begin with. A further finding wasthat The 1976 Vocational Education Amend- significantly greater increases in women's ments (VEA 1976) were intended toprovide a basis for thedevelopment of programs to eliminate sex bias, discrimination, and ster- 6Traditional vocational education programs were de- eotyping and to promote equal access of the fined as those made up of at least 75.1 percent women sexes to vocationaleducation. Assessment in 1972; mixed programs were made up of 25.1 to 75.0 of whether the implementation of VEA 1976 percent women; and nontraditional programs hid en- has accomplished more equitable access to rollments of 0.0 to 25.0 percent women. 22 7 SEX TYPING IN OCC PATIONAL SOCIALIZATION . 221, enrollment in nontraditional programs oc- development of attitudes and behavior pat- curred at the postsecondary and adult ed- terns in children. In examining the portrayal ucation levels than in high schools. This of males and females in the media, we limit finding is consistent with the lack of change our discussion to the role of television as the cited earlier in our discussion of tracking and principal medium by which the content and vocational education in high schools. importance of male and female roles are Several conclusions ean be drawn about communicated to children (Schramm et al., the effectiveness of federal legislation in 1961). There have also been studies of the providing a more sex-equitable environ- portrayal of males and females in'newspaper ment in the schools. First, as demonstrated stories (Forest et al.,1980), nonfictional by the differential success of various gates' magazine pieces (Hatch and Hatch, 1958; implementation of VEA 1976, the provision Clark and Esposito, 1966), magazine fiction of federal funds coupled with a broad com- (Franzwa, 1975), and other media sources. mitment will not effect change. Rather, ac- Many studies have indicated that males tive attention to the. monitoring of schools are disproportionately represented on 'tel- and, in particular, the administration of pre- evision, whether the framework be prime- and inservice courses for teachers and coun- time television programs, children's pro- selors seem to be important. Special atten- grams, or commercials (Courtney and tion needs to be focused on secondary schools, Whipple, 1974; Sternglanz and Serbin, 1974; where students are influenced to make de- Tedesco, 1974; Women on Words and Im- cisions that will have significant ramifica- ages, 1975a; Nolan et al., 1977). Wh_n sex tions for whether they continue on to col- differences in the frequencies of responses lege, receive postsecondary vocational made by characters are assessed, males education, or immediately enter the job dominate the verbal and nonverbal action market. Second, the provision of female role (Downes and Gowan, 1980): with respect models in courses with a male or a "mixed" to occupational representation, an early study image and of male role models in courses (DeFleur, 1964) found that less than 20 per- with a female image may be a way to en- cent of the occupational roles depicted were courage both sexes to consider broader ca- filled by women. More recent studies con- reer options (Rieder, 1977).Third, state ducted in the mid-1970s also indicate nar- programs that were the most successful id rowness and sex stereotyping in the roles diverting high female enrollments in tradi- assigned to women. In a study of prime-time tional courses to less traditional specialities television programs receiving high Neilsen had established a broad base of support for ratings in 1973, Women on WL:ds and Im7 women who took this route, by setting up ages (1975a) found that the range of occu- orientation programs and providing connec- pations was nearly twice as broad for major tions with potential employers. This type of male characters as for major female char -. comprehensive planning, including follow-_ acters (although it was not clear whether the up support for students making nontradi- number of characters was held constant when tional choices, appears to be successful in such a comparison was made) and that there changing sex ratios (Evenson and O'Neil, was little occupational overlap between the 1978). sexes. Findings were even more extreme in the case of commercials. In another study of prime-time programs, Kaniuga et al. (1974) Mais Meck4 Effects: Television Portrayal found that among the women depicted as of Male and Female Occupational Roles workers the most common occupational roles .In recent years, a good deal of attention were those of secretary, nurse, and educa- has been directed to the influence of the tor, and in only 10 percent of the cases were mass media, particularly television, on the the working women married. Especially no- 228 222 MARGARET MOONEY MAPINI ANDMARY C. BRINTON ticeable, therefore, was the predominance 'studies indicate that females are someivhat of white-collar employment for women and less likely to be employed than males and the implication that it Was not common to that when they are employed they tend to combine work outside the home with mar- work fewer hours per week. The distribu- riage and family responsibilities. tion of students across jobs is also signifi- Most studies of sex-role stereotyping on cantly different for the two sexes. Thus, even television have involved content analysis and when work is a secondary activity and both have not attempted to examine actual effects sexes are employed in low-skillnonspecial- on childsen. Fruehand McGhee (1975), ized jobs, as is the case in the adolescent however, examined the relationship be- years, job segregation by sex emerges. tween the amount of time children spent In addition, as in the adult workovorld, watching television and their identification- adolescent females earn lower hourly wages with traditional sex roles. They found that than males, a pattern that holds across job high-amounts of television viewing were as- categories, ethnic groups (whites, blacks, sociated with stronger traditional sex-role Hispanics), and high school grade levels. development and that the relationship be- Hourly wages for adolescents are positively tween television viewing and sex-role atti- related to the degree to which a job is dom- tudes existed across sexes and age groups inated by males, again mirroring the adult (kindergarten, grades 2, V, and 6). The de- occupational environment. Based on data gree to which actualoccupational choices from the High School and Beyond survey of are conditionedby exposure to televisie sophomores and seniors in a national sample remains unknown. But given thepredomi- of U.S. high schools in 1980, Lewin- Epstein nance of television as amedia form for chil- (1981) found that sex was the most important dren and its role as a sRurce of information determinant of wages earned by teenagers. about the world, especially prior to the de- He also found that the sex difference in ac- velopment of reading skills and prior to en- tual wages was somewhat greater than the trance into the adult working world, it is sex difference in reservation wages, as meas- likely to have a significant impact. ured by the lowest hourly wage students said they would accept in high school. He argued that this pattern might help account Early Work Experiences for the lower labor force participation of fe- A variety of groups, such as the Presi- males, since females may have greater dif- dent's Science Advisory Committee, the ficulty than males in meeting their wage ex- National Panel on High Schools and Ado- pectations in the job market. Together, sex lescent Education, and the Carnegie Coun- differences in labor force participation, type cil on Policy Studies in Higher Education, of work experience, and earnings during ad- have advocated the participation of teen- olescence are indicative of yet another way agers in the work force as a meansof helping in which adolescent males and females de- them develop skills and attitudes that will velop different orientations toward the adult facilitate a smoother transition into full-time world of work. q, adult work roles (Lewin-Epstein, 1981). Re- ,sea'rch on the employment experiences of Summary youth prior to high school graduation is sparse, but two recent studiesprovide data Of the socializing influences in the lives portraying a sex-segregated occupational of children and adolescents that are likely world for adolescents that closely mirrors to produce sex differences in occupational the adult work world (Lewin-Epstein, 1981; orientation and preparation prior to entry Greenberger and Steinberg, 1983). These into the adult work force, the earliest and

229 SEX TYPING IN OCCUPATIONAL SOCIALIZATION 223 most pervasive ones arise within thefamily, ences are in the characteristicsof workers where'mothers and fathers not only provide prior to .entry into the labor market, com- information as role models and teachers but pared to the actions of employers and other also treat male and female children differ- legal and institutional barriers in the work- ently. School influences reinforce the effects place, in accounting for sex segregation in of family socialization. Among these are the the labor market. This question cannot be greater availability of same-sex rolemodels answered, however, because it does not dis- for males across a variety of fields at higher tinguish between the operation of two dis- levels of education, sex typing in the pres- tinct, but related, processes. one at the mi- entation of occupational roles in textbooks cro level and one at the macrolevel. and other educational materials,, sex bias in At the micro, or individual, level, it is the attitudes and knowledge of guidance Passible to examine the relative effects of counselors regarding the appropriateness of different types of influences on the occu- various occupations for males and females, pational outcomes of individuals in one or sex segregation indifferent vocational ed- more cohorts. Socialization is a processthat ucation programs, and sex differences in operates at the micro level, since it is the training in mathematics and science. Sex process by which individuals come tolearn typing in the portrayal of occupationalroles about the world in which they live and to In the mass media provides another source understand what is considered appropriate of information about the adult occupfitional and acceptable behavior for them. In a so- world, as do sex difference's in the actual ciety in which adult roles are differentiated employment experiences of adolescents prior by sex and' where the labor market is highly to leaving school. It is difficult, if not im- sex segregated,females and males develop possible, to estimate the effect of any single different expectations of their adult work lives socializing influence on the development of and the jobs appropriate for them via sex- sex differences inoccupational orientation role socialization. The effect of socializaton and job-relevant. skills. However, it is clear prior to entry into the labor market on the that, collectively, they teach children to as- occupational outcomes of individuals can be pire to and 'prepare for different occupa- examined by addressing the question: How tional roles in adulth000d. important are sex differences in occupational orientation and preparation prior to entry into the labor market (which arise pri- SOCIALIZATION AS AN EXPLANATION marily as a result of socialization) compared OF SEX SEGREGATION IN THE LABOR to subsequent labor market experiences MARKET (which are attributable at least fin part to the Since the purpose of this paper is to ex- actions of employers and other legal and in- amine sex differences in occupational ori- stitutional barriers) in accounting for the sex- entation and preparation prior to entry into segregated pattern of employment for in, the labor market as an outgrowth of the dividuals in particular cohorts'? On the basis process of socializationand to consider the of the evidence we have presented on the effects of these differences on subsequent degree to which sex segregation in occu- sex segregation in thelabor market, we will pational aspirations approximates sex seg- conclude by discussing the role of sociali- regation in employment, socialization prior zation as a cause of occupational segregation to entry into the labor market appears tobe by sex. In attempting to understand the im- an importantdeterminant of occupational portance of socialization as adeterminant of outcomes for individuals, although the ex- sex segregation inthe labor market, itis tent to which preemployment differences in reasonable to ask how important sex differ- worker characteristics account for subse- t: 230 Is 224 MARGARET MOONEY MARINI AND MARY C. BR1NTON.'

quent sex segregation in the labor market tially a transmission process has implications remains to be estimated precisely using lon- for the conclusions to be drawn from our gitudinal data. findings regarding interventions for change Because socialization prior to entry into in sex segregationfin the labor market. Al- the labor market appears to play a large role though our findings indicate that socializa- in the determination of occupational out- tion plays an important role in the deter- comes for individuals, isit reasonable to mination of occupational outcomes for conclude that it is an important determinant individuals, it should not be inferred that of sex segregation in the labor market? Only interventions for change should focus pri- if one is referring to its predictive poWer in marily on socialization practices. Because accounting for the occupational outcomes of socialization is a process whereby existing individuals. Socialization cannot explain why cultural practices, including employment a. sex-segregated labor market emerged, why patterns, are transmitted, a reduction of sex each sex' is allocated to particular types of segregation in employment affects what is occupations, and why the sex typing of oc- transmitted via socialization and thereby ul- cupations changes in particular ways over timately reduces sex differences in occu- time. These characteristics of the labor mar- pational orientation and preparation. Inter- C' ket are outcomes, of macro-level processes ventions directed at changes in employment in which such factors as the supply and de- practices and in laws that affect sex segre- mand for particular types of workers, the gation therefore can bring about both im- structure of work 'organizations, cultural be- mediate change in employment patterns and liefs and practices, legal arrangements, and eventual change in the messages about the the actions of employers play a dominant occupational world that are conveyed to new role. To explain the existence of sex segre- generations. gation in the labor market, it is necessary Throughout our discussion of socializing to address the question: Why did sex seg- influences, we have commented on inter- regation in the labor market emerge and ventions that might be undertaken to change take the particular form it did? The answer socialization practices. Such changes are to this question is to be found by analyzing needed, and would undoubtedly effect some variation at the macro level, including dif- change in the occupational orientations and ferences among organizations and societies .preparation of the two sexes. However, and changes in these structures over time. changes in socialization practices must go Thus, although the maintenance of a sex- hand in hand with changes in employment segregated labor market and changes in the practices. Because the actions of employers pattern of segregation over time occur via and the structure of work organizations are the actions of individuals at the micro level, known to affect sex segregation, a reduction the origins, or causes, of sex segrogation of sex differences in occupational orientation cannot be understood through analysis of would not necessarily produce a concom- micro-level processes such as socialization. mitant reduction of sex differences in em-

Socialization is a process whereby prevailing .ployment patterns. 'Moreover, for a major cultural practices are transmitted to new reduction of sex differences in occupational generations, and as such it plays an impor- orientation to occur, a major reduction of tant role in the determination of outcomes sex segregation in the labor market is nec- for individlials. However, the content of what essary, since existing employment patterns is transmitted via socialization is determined affect what is learned via socialization. That by factors operating at the macro level. is, children and young adults must observe Understanding that socialization is essen- less sex-segregated employment patterns

231 a SEX TYPING IN OCCUPATIONAL SOCIALIZATION 225 prior to labor force entry if sex differences sponses From the Nation's High School Sen- iors. Ann Arbor, Mich.: Institute for Social Re- , in occupational aspirations and expectations search, University of Michigan. are to be significantly reduced.. Bahr, S.J. 1974 "Effects on power and division of labor in the family." Pp. 167-85 in L.W. Hoffman and F.I. REFERENCES Nye (eds.), Working Mothers. San Francisco: Jossey-Bass. Aiken, L.R.; Jr. Bandura, A., and A.C. Huston 1970 "Attitudes toward mathematics." 'Review of educational Research 40:551-96. 1961 "Identification as a process of incidental learning." Journal of Abnormal and Social Psychol- "Update on,attitudes and other affective vari- 1976 ogy 63:311-18. ables in learning rhatheinatics." Review of Ed- ,ucational Research 46:293-311. Bandura, A., D. Ross, and S.A. Ross 1963 "A comparative test of the status envy, social Aiken, L.R., Jr.., and R.M. Dreger power, and secondary ;reinforcement theories 1961 "The effect of attitudes on performance in of identificatory learning," Journal of Abnor- mathematics." Journal of Educational Psy- mal and Social Psychology 67:527-34, chology 52:19-24. Barclay, L.K. Alexander, K. L. and B.K. Eckland 1974 "The emergence of vocational expectations in 1974 "Sex differences in the educational attainment Preschool children," Journal of Vocational Be- process." American Sociological Review 39:668- havior s. 82. BarBaruch, Almquist, E. M. 1972 "Maternal influences upon college women's at- 1974 "Sex stereotypes in occupational choice: The titudes towards women and work." Develop- case of college women." Journal of Vocational mental Psychology 6:32-7. Behavior 5:13-21. Basow, S.A., ana K.G. Howe Almquist, E. M., and S.S. Angrist 1979 "Model influence on career choices of college 1970;Career salience and atypicality of .occupa- students." Vocational Guidance Quarterly tional choice among college women." Journal 27:239-43. of Marriage and the Family 32:242 -49. . Beardslee, D.C., and D.D. O'Dowd -101 "Role model influences on college women's ca- 1962 "Students and the occupational world." In N. reer aspirations." Merrill-Palmer Quarterly 17:- Sanfoid (ed.), The American College. New York: 260-79. Wiley. Altman; S.L., and F. K. Grossman Becker, C.S. 1977 "Women's career plans and maternal employ- 1964 Human Capital. New York: National Bureau ment." Psychology of Women Quarterly 1 :365- of Economic Research. 76. Beller, A.M. . American Institutes for Research 1981 "The impact of education on entry into non- 1980 "En: ollment and staffing patterns in vocational' traditional occupations." Revised version of pa- education." Pp. 27-78 in Education, Sex Equity per presented at the Econometric Society and Occupational Stereotyping. Washington,, meetings, Denver, Colorado, September 1960. D.C.: National Commission for Employment Policy. Bem, S.L. 1981 "Gender schema theory: A cognitive account Anderson, K. E. of sex typing. "Psychological Review 88:354-64. 1963 "A comparative study of student self ratings on the influence of inspirational teachers in sci- Bergmann, B.R. ence and mathematics in the developmentof 1971 "The effect on white incomes of discrimination intellectual curiosity, persistence and a ques- in employment." Journal of political Economy tioning attitude." Science Education 47 :429- 79:294-313. 37. 1974 "Occupational segregation, wages and profits when employers' discriminate by race or sex." Astin, A.W., anti R.J. Panos Eastern Economic Journal 1:103-10. 1969 The Educational and Vocational Development of College Students. Washington, D.C.: Berk, ILA., and S.F. Berk .1.* American Council on Education. 1979 Labor and Leisure at Home. Beverly Hills, Astin, H.S., and T. Myint Calif.: Sage. 1971 "Career development and stability of"young Bielby, D. D.V. women during the post high schoolyears." 1978a"Career sex-atypicality and career involvement Journal of Counseling Psychology 18:369-93. of college educated women: Baseline evidence Bachman, J.G., L. D. Johnston, and P.M. O'Malley from the 1960's." Sociology of Education 51:7- 1980 Monitoring the Future: Questionnaire Re- 28.

232 226 MARGARET MOONEY MARINI AND MARY C. BRINTON

el 1978b "Maternal unemployment and socioeconomic Clark, E,T. status as factors in daughter; career salience: 1967 "Influence of sex and *social class on occupa- Some substantive refinements." Sex Holes 4:249- tional preference and perception." Personnel. 65. and Guidance journal 45:440-44. Bingham, W., and E. House Clarke, P., and V. Esposito 1973 "Counselors view women and work: Accuracy 1966 "A study of occupational advice for women in of information." Vocational Guidance, Quar- magazines." Journalism Quarterly 43:477-85. terly 21:262-68. Courtney, A, E. , and T.W. Whipple Blau, F. 1974 "Women in TV commercials." journakof Com- 1977 Equal Pay in the Office. Lexington, Mass.: munication 24:110-18. Lexington. Cummings, S., and D. Taebel Blau, F, and W. Hendricks 1980 "Sexual inequality and the reproduction of con- 1979 Occupational segregation by sex: Trends and sciousness: An analysis of sex-role stereotyping prospects..: Journal of Human Resources 12:197- among children," Sex Roles6:631-44: 210. Davis, J.A. Blau. F., and C.L. Jusenius 1965 Undergraduate Career Decisions. ChiCago: Al- 1976 "Economists. approaches to sex segregation in dine.' the labor market: An appraisal." Pp. 181-99 in -.Daymont, TN., and P. L. Tsai M. Marra and B. Reagan (eds.), Women in 1981 "Sex inegtiality in the labor market: Amult i- the Workplace: The Implications of Occupa- disciplinary approach." Paper presented at the tional Segregation. Chicago: Univerlity of Chi- American Sociological Association's Annual cago Press, Meeting, Toronto,' Ontario, Canada, August. Blau, P.M., and O. D. Duncan DeFleur, M,L, 1967 The American Occupational Structure. New 1963 "Children .s knowledge of occupational roles and York: Wiley. prestige: Preliminary report." Psychological Block, J.H. Reports 13:760, , 1973 "Conceptions of sex role: Somecross-c ulbiral and longitudinal perspectives." American Psy- 1964 "Occupational roles as portrayed on television." Pyblic Opinion Quarterly 28:57-74. chologist 28:512-26. 1976 "Issues, problems, and pitfalls in assessing sex Doeringer, P. B., and M.J. Piore differences: A critical review of 'The Psychol- 1971 Internal Labor Markets and Manpower Anal- ogy of Sex Differences.' Merrill-Palmer ysis. Lexington, Mass.: D.C. Heath. Quarterly 22:283-308. Donahue, T.J., and J.W. Costar Brenner, O.C.. and J. Tomkiewicz 1977 "Counselor discrimination against young women 1979 "Job orientation of males and females: Are sex in career selection." Journal of Counseling Psy- differences declining?" Personnel PsYchology chology 24:481-86. 32:741-50. Downes, A.C., and D.C. Gowan Brief, A. P.. and L. J. 'Aldag 1980 "Sex differences in reinforcement and punish- 1975 "Male-female differences in occupational atti- ment on prime-time television." Sex Roles tudes within minority groups," Journal of Vo- 6:683-94. cational Behavior 6:305-14. Douvan, E. Brito, P. K., and C. L. Jusenius 1976 "The role of mOdels in women's professional 1978 "Sex segregation in the labor market; An anal- development."/ Psychology of Women Quar- ysis of young college women's occupational terly 1:5-20. preferences... Pp. 57-75 in F.L. Mott 1,ed.), DeSu. van, E.. and J. Adelson Women. Work, and Family: Dimensions of 1966 The Adolescent Experience. New York: Wiley. ( ;hang' in American Society. Lexington, Mass.: Lexington. Dreger, R. M. , and L. R. Aiken. jr. 1957 'The identification of number anxiety in a pot- Burble F. lege population." Journal of Educational Psy- 1976 "Sex-role stereotypings Occupational aspira- chology 48:344-51. tions of female high school students." The School Counselor 24:102-8. Dwyer, C.A. 1973 "Sex differences in reading: An evaluation and Callahan, W. J. a critique of current methods... Review ofEd- 1971 -Adolescent attitudes toward, mathematics." Mathematics Teacher 64:751-3. ucational Research 43;455-61, Engelhard, P.A., K.O. Jones, and R. R. Stiggens (;hodorow, N. "Trends in counselor attitudes about women's 1978 The' Reproduction of Mothering: Psychoatud- 1976 )sis and the Sociology of Gender,Berkeley. roles." journal of Counseling Psychology 23.365- University of California Press. 72. kV., and JKituse Englaud, P. 1961 The. Educational Decision-Makers. Indianap- 1981a"Assessing trends in occupational sex segre- olis llobbs-Merrill. gation, 19(X)-1976." Pp. 273.95 in I. Berg red.),

233 SEX TYPING IN OCCUPATIONAL SOCIALIZATION 227

Sociological Perspectives on Labor Markets. of time spent watching television." Develop- New York: Academic Press. mental Psychology 1;109. 1981b "Wage appreciation and depreciation: A test of Garrison, H.H. neoclassical explanations of occupational sex 1979 "Gender differences in the career aspirations segregation." Paper presented at the annual of recent cohorts of high school seniors," Social meeting of the American Sociological Associ- Problems 27:170-85. ation, August 1981. Ginsberg, E., S.W. Ginsberg, S. Axelrod, and J.L. 1982 "The failure of human capital theory to explain Herma occupational sex segregation." Journal of Hu- 1951 Occupational Choice: An Approach to a Gen- man Resources 17:358-70. eral Theory. New York: Columbia University Ernest, J. Press. 1976 Mathematics and Sex. Santa Barbara: Mathe- Goldstein, E. matics Department, University of California. 1979 "Effect of same-sex and cross-sex role models productivity of Evenson, J.S., and M.I. O'Neill on the subsequent academic scholars." American Psychologist 34:407-10. 1978 Current Perspectives on the Role of Career Education in Combatting Occupational Sex-Role Good, T.L., J.N. Sikes, and J.E. Brophy Stereotyping. Washington, D.C.: National In- 1973 "Effects of teacher sex and student sex on class- stitute of Education. room interaction." Journal of Educational Psy- chology 65:74-87. Farmer, and T.E. Backer 1977 New Career Options for Women: A Counse- Gottfredson, L.S. lor's Sourcebook. New York: Human Sciences 1978 "Race and sex differences in occupational as- Press. pirations: Their development and consequences for occupational segregation."Report Fennema, E. No. 254, Center for Social Organization of 1974a"Mathematics learning and the sexes: A review." Journal for Research in Mathematics Schools, johns Hopkins University; Education 5:126-39. Grasso, J. "The effects of school curriculum on young 1974b "Sex differencesinmathematics learning: 1980 Why ? ??" The Elementary School Journal 75:- women." Pp. 79-114 in Education, Sex Equity and Occupational Stereotyping. Washington, 183 -90. D.C.: National Commission for Employment Fennerna, E., and J. Sherman Policy. 1977 "Sex related differences in mathematics achievements, spatial visualization and affec- Greenberger, E., and L. D. Steinberg 1983 "Sex differences in early labor force experi- tive factors." American Educational Research come." Social Journal 14:51-71. ence: Harbinger of things to Foreit, K.G I. Agar, J. Byers, j. Larue, H. Lokey, Forces 62:467-86. M. Pallazini, M. Patterson, and L. Smith Gross, E. . .? The sexual structure of 1980 "Sex bias in the newspaper treatment of male- 1968 "Plus ca change centered and female-centered news stories." occupations over time." Social Problems 16:198- Sex Roles 6:475-80. 208. Fox. (:.L. Harmon, L.W. 1974 "Some observations and data on the availability 1971 "The childhood and adolescent career plans of of same-sex role models as a factor in under- college women." Journal of Vocational Bthav- graduate career choice." Sociological Focus 7: 15- ior 1:45-56. 30. Harnischfeger, A., and D.E. Wiley Fox, I.. H., D. Tobin, and L. Brody 1980 "High school tracking and vocational stereo- 1979 "Sex role socialization and achievement in typing: Means of socioeconomic placement." mathematics." Pp. 303-32 in M.A. Wittig and In Education, Sex Equity and Occupational A.C. Peterson (eds.), Sex-Related Differences Stereotyping. Washington, D.C.: National in Cognitive Functioning: Developmental Is- Commission for Employment Policy. sues. New York: Academic Press. Harren, V.A., R.A. Kass, H.E.A. Tinsley, and J.R. Franzwa, Ii. H. Moreland 1975 -Female roles in women's magazine fiction, 1979 "Influences of gender, sex-role attitudes, and 1940-1970." Pp. 42-53 in R. K. Unger and F.L. cognitive complexity on gender-dominant ca- Denmark (eds.), Women: Dependent or In- reer choices." Journal of CounselingPsychol- dependent Variable? New York: Psychological ogy 26:227-34. Dimensions, Inc. Harris, S.R. Frieze, 1. H.. J. E. Parsons, P.B. Johnson, D. N. Ruble, 1974 "Sex typing in girls' career choices: A challenge G.I. Zellman to counselors." Vocational Guidance Quarterly 1978 Women and Sex Roles: A Social Psychological 23:128-33. Perspective. New York. W.W. Norton. Hartley, R. E. Frueh. T.. and P.C. McGhee 1966 "A developmental view of female sex-role identification." In B. j. Biddle and j. Thomas (eds.), 1975 -Traditional sex-role developments and amount 234 Ft)

228 MARGARET MOONEY MARINI AND MARY C. BRINTON

Role Theory: Concepts and Research. New York Hoffman, L.W. Wiley. 1974 "Effects on child." Pp. 126-68 in L.W. Hoff- Hartley, R., and A. Klein man and F.L. Nye (eds.), Working Mothers. 1959 "Sex-role concepts among elementary school San Francisco: Jossey-Bass. age girls." Journal of Marriage and the Family 1977 "Changes in family roles, socialization, and sex 20:59-64. differences." American Psychologist 32:644-57. Hatch, M.G., and D.L. Hatch Hoffman, L.W., and F.I. Nye 1958 "Problems of married working women as pre- 1975 Working Mothers. San Francisco: Jossey-Bass. sented by three popular working women's Holland, J. L. magazines." Social Forces 37:148-53. 1959 "A theory of vocational choice." Journal of Hauser, R.M., W.H. Sewell, and D.F. Alwin Counseling Psychology 6:35-45. 1976 "High school effection achievement." Pp. 309- 1966 The Psychology of Vocational Choice: A The- 41 in W.H. Sewell, R.M. Hauser, and D: L. ory of Personality Types and Model Environ- Featherman (eds.), Schooling and Achieve- mants. Wv!tham, Mass.: Blaisdel. ment in American Society. New York: Aca- 1973 Making Vocational Choices: A Theory of Ca- de.* Press. reers. Englewood Cliffs, N.J.: Prentice-Hall. Heilman, M.E. Hornvy, K. 1979 "High school students' occupational interest as 1932 "The dread of women." International Journal a function of projected sex ratios in male-dom- of Psychoanalysis 13:348-60. inated occupations." Journal of Applied Psy- Huston, A.C. chology 64:275-79. In Hershenson, D,B., and A.M. Roth press "Sex-typing." In P.H. Mussen (ed.), Carmi- 1966 "A decisional process model." Journal of Coun- chael's Manual of Child Psychology,. 4th ed. seling Psyclology 13:368-70. New York: Wiley. Herzog, A.R. Kaldor, D.R., and D.G. Zytowski 1982 "High school seniors' occupational plans and 1969 "A maximizing model of occupational decision- values: Trends in sex differences 1976 through making." Personnel and Guidance Journal 1980." Sociology of Education 55:1-13. 47:781-88. Heshusius-Gilsdorf, L.F., and D.L. Gilsdorf Kaniuga, N., T. Scott, and E. Cade 1975 "Girls are females, boys are males: A content 1974 "Working women portrayed on evening tele- analysis of career materials." Personnel and vision programs." Vocational Guidance Quar: Guidance Journal 54:207-11. terly 23:134-37. Hewitt, L.S. Karpicke, S. 1975 "Age and sex differences in the vocational as- 1980 "Perceived and real sex differences in college pirations of elementary school children." Jour- students' career planning." Journal of Coun- nal of Social Psychology 96:173-77. seling Psychology 27:240-45. Heyns, B. Kaufman, D.R., and B.L. Richardson 1974 "Social selection and stratification within 1982 Achievement and Women: Challenging the schools. American Journal of Sociology 78:1434- Assumptions. New York: Free Press. 51. Kepner, H.S., and L.A. Koehn Hilton, T.L. 1977 "Sex roles in mathematics: A study of sex ster- 1962 "Career decision-making." Journal of Coun- eotypes in elementary mathematics texts." The seling Psychology 9:291-98. Arithmetic Teacher 24:379-85. Hilton, T.L., and G. W. Berglund Kimmel, E. A. 1974 "Sex differences in mathematics achievement: 1970 "Can children's books change children's val- A longitudinal study." Journal of Educational ues?" Educational Leadership 28:209-14. Research 67:231-371. Klein, M. Hind, R.R., and T.E. Wirth 1957 Envy and Gratitude. New York: Basic Books. 1969 "The effect of university experience on occu- Klemmack, D. L., and J.N. Edwards pational choice among undergraduates." So- 1973 "Women's acquisition of stereotyped occupa- ciology of Education 42:50-70. tional aspirations.". Sociology and Social Re- Hofferth, S.L. search 57:510-25. 1980a"High school experience in the attainment Kohout, V. A.,and J. W. M Rothney process of non-college boys and girls: When 1964 "A longitudinal study of vocational prefer- and why do their paths diverge?" Working Pa ences." American Educational Research Jour- per 1303-01, The Urban Institute, Washing- nal 1:10-21. ton. D.C. Kolberg, L. 1940b"Long-term labor market effects of vocational 1966 "A cognitive-developmental analysis of chil- education on young women:" In Education. dren's sex-role concepts and attitudes." Pp. 82- Sex Equity and Occupational Stereotyping. 172 in E. E. Maccoby (ed.). The Development Washington. D.C.: National Commission for of Sex Differences. Stanford, Calif.: Stanford Employment Policy. University Press. 235 SEX TYPING IN OCCUPATIONALSOCIALIZATION 229

Kruinholtz, J. D., A.M. Mitchell, and G.B. )Gees and stereotyping in children." Child Devel- 1976 "A social learning theory of career selection." opment 52:1119-34. Counseling Psychologist 6V-80. Mason, K.O., J.L. CzaJka, and S. Arber Kuvlesky, W.P., and R.C. Bea ler 1976 "Change in U.S. women's sex role attitudes, 1054-1974." American Sociological Review 1967 "The relevance of adolescents' occupational aspirations for subsequent Job attainments:" Ru- 41:573-96. ral Sociology 32:290-301. McClelland, D.C. Lauver, P.J., R.M. Castellum, and M. Sheehey 1976 Power: The Inner Experience. New York: 1975 "Bias in 00H illustrations?" Vocational Guid- Halsted. ance Quarterly 23:335-40. Mead, M. Lerner, H.E. 1935 Sex and Temperament in Three Primitive So- 1974 "Early origins of envy and devaluation of women: cieties. New York: Dell. Implications for sex role stereotypes." Bulletin Medvene, A.M., and A.M. Collins of the Menninger Clinic 38:538-53. 1976 "Occupational prestige and appropriateness: 1978 "Adoptive and pathogenic aspects of sex-role The views of mental health specialists." Journal stereotypes: Implications for parenting and of Vocational Behavior 9:63-71. psychotherapy." American Journal of Psychia- Miller, S.M. try 135:48-52. 1975 "Effects of maternal employment on sex role Levy, B. perception, interests, and self-esteem in kin- 1972 "Sex role socialization in schools." Today's Ed- dergarten girls." Developmental Psychology 11: ucation 61:9,27-29. 405-6. Lewin-Epstein, N. Milnar, J. 1981 Youth Employment During High School. Re- 1973 "Sex Stereotyping in Mathematics and Science port prepared for the National Centerfor Ed- Textbooks for Elementary and Junior High ucation Statistics under contract OE- 300-78- Schools." Pp. 56-59 in Report on Sex Bias in 0208 with the U.S Department of Education. the Public Schools. New York: Education Chicago: National Opinion Research Center. Committee, New York chapter of the National Organization for Women. Loot, W. R. 1971a"Sex differences in the expression of vocational Mischel, W. aspirations by elementary school children." 1970 "Sex typing and socialization." Pp. 3-72 in P. H. Developmental Psychology 5:366. Mussen (ed.), Carmichael's Manual of Child Psychology, 3d ed., Vol. 2. New York: Wiley, 197 lb"Vocational aspirations of second-grade girls." Psychological Reports 28:241-42. Mischel, W., and J. Grusec 1966 "Determinants of the rehearsal and transmis- Lueptow, L. B. sion of neutral and aversive behaviors." Journal 1980 "Social change and sex-role change in adolescent orientations toward life,work, and of Personality and Social Psychology 3:197-205. achievement: 1964-1975." Social Psychology' Mitchell, A.M., G.B. Jones, and J. D. Krumboltz (eds.) Quarterly 43:48-59. 1975 A Social Learning Theory of Career Decision Making. Palo Alto, Calif.: American Institute 1981 "Sex-typing and change in the occupational choices of high school seniors: 1964-1975." So- for Research. ciology of Education 54:16-24. Money, J., and A.A. Ehrhardt Man and Woman, Boy and Girl. Baltimore: Maccoby, E. E. , and C. N .Jackl in 1972 Johns Hopkins University Press. 1974 The Psychology of Sex Differences. Stanford, Calif.: Stanford University Press. National Advisory Council on Vocational Education and National Advisory Council on Women's Educational Marini, M.M. 1978a"Sex differences in the determination of ado- Programs lescent aspirations: A review of research." Sex 1980 Increasing Sex Equity: The Impact of the 1976 Vocational Education Amendments on Sex Roles 4:723-53. Equity in Vocational Education. Washington, 19786"The transition to adulthood: Sex differences D.C.: U.S. Department of Education. in educational attainment and age at marriage." American Sociological Review 43:483- National Advisory Council on Women's Educational Programs 507. 1981 Title IX: The Half Full, Half Empty Glass. 1980 "Sex differences in the process of occupational Washington, D.C.: U.S. Government Printing attainment: A closer look." Social Science Re- Office. search 9:307-61. Nelson, R.C. Marini, M.M.. and E. Greenberger 1963 "Knowledge and interests concerning sixteen 1978 "Sex differences in occupational aspirations and occupations among elementary and secondary expectations." Sociology of Work and Occu- school students." Educational and Psycholog- pations 5:147-78. ical Measurement 23:741-54. Martin, Cl.., and C.F. Halverson, Jr. 1981 "A schematic processing model of sex typing MARGARET MOONEY MARINI AND. MARY C. BRINTON

Nieva, V.F., and B.A. Gutek Plomin, R., and T.T. Foch 1981 Women and Work: A Psychological Perspec- 1981 "Sex differences and individual differences." tive. New York: Praeger. Child Development 52.346-88. Nilsen, A.P. Poffenberger, T.M., and D.A. Norton 1977 "Sexism in children's books and elementary 1959 "Factors in the formation-of attitudes toward teaching materials." Pp. 16180 in A.P. Nilsen, mathematics." Journal of Educational Re- H. Bosmajian, H. L. Gershuny, and J.P. Stan- search 52:171-76.. ley (eds.), Sexism and Language. Urbana, Ill.: Polachek, S.W. National Council of Teachers of English. 1976 "Occupational segregation: An alternative hy- Nolan, J. D., J.P.Geist, and M.A. White pothesis." Journal of Contemporary Business 1977 "Sex bias on children's television programs." 5:1-12. Journal of Psychology 96:197-204. 1979 "Occupational Segregation among women: O'Bryant, S. L., M.E. Durrett, and J.W. Pennebaker Theory, evidence and a p eognosis.Pp. 137- 1980 "Sex differences in knowledge of occupational 57 in C.B. Lloyd, E.S. Andrews, and C.L. dimensions across four age levels." Sex Roles Gilroy (eds.), Women in the Labor Market. 6:331-37. New York: Columbia University Press. 1981 "Occupational self-selection: A human capital Oerzel, R. M. 1966 "Annotated bibliography." Pp. 223-51 in E. E. approach to sex differences in occupational Maccoby (ed.), The Development of Sex Di- structure." Review of Economics and Statistics ferences. Stanford, Calif.: Stanford University 63:60-69. Press. Poloma, M. M., and T. N.Garland 1971 "The myth of the egalitarian family: Familial O'Hara, R. P. roles and the professimially employed. wife." 1962 "The roots of careers." Elementary School Pp. 741-61 in A. Theodore (ed.), The Profes- Journal 62:277-80. sional Woman, Cambridge, Mass,: Schenk- Oppenheimer, V. man. 1970 The Female Labor Force in the United States. Porter, R.I. Population Monograph Series, No. 5. Berke- 1954 "Predicting vocational plans of high school sen- ley: University of California. ior boys.'Personnel and Guidance Journal Papalia, D.T., and S.S. Tcnnent 33:215-18. 1975 "Vocational aspirations in preschoole. s: A man- Project on Equal Education Rights ifestation of early sex role stereotyping." Sex 1978 Stalled at the Start: Government Action on Sex Roles 1:197-99. Bias in the Schools, Washington, D.C.: Na- Parelius, A.P. tional Organization for Women Legal Defense 1981 "bender differences in past achievement, self- and Education Fund. concept, and orientation toward mathematics, Rebecca, M., R. Hefner, and B. Oleshansky science, and engineering among college fresh- 1976 "A model of sex-role transcendance." Joutna, men." Paper presented at the Ameri So- of Social Issues 32:197-206. ciological Association's Annual Meeg, To- Reese, H.W., and W.F. Overton ronto, Ontario, Canada, August. 1970 "Models of development and theories of de- 1982 "Recruitment to mathematics, science, d ent- velopment." Pp. 115-45 in L.R. Goulet and gineering: Gender differences and similarities P.B. Bakes (eds.), Life-Span Developmental in the experiences of college freshmen." Paper Psychology: Research and Theory: New York presented at the American Sociological Asso- Academic Press. ciation's Annual Meeeting, San Francisco, Cal- Rehberg, R.A., and L. Hotchkiss ifornia, September. 1972 "Educational decision-makers: The school Parsons, J. E. guidance counselor and social mobility." So- 1978 "Classic theories of sex-role socialization." Pp. ciology of Education 45:339-61. 95-113 in I. Frieze, J. Parsons, P. Johnson, D. Rieder, C. H. Ruble, and G. Zellman (eds.), WOmen and Sex 1977 "Work, women and vocational education." Roles. New York: Norton. American Education 13:27,32. Peng, S.S., W.B. Fetters, and A.J. Kolstad Robinson, J. 1981 High School and Beyond: A Capsule Descrip- 1977 How Americans Use Time: A Social-Psycho- tion of High School Students. Washington, logical Analysis. New York: Praeger. D.C.: National Center for Education Statistics. Rodman, H., P. Voydanoff, and A.E. Lovejoy Perrucci, C.C. 1974 "The range of aspirations:. A new approach." 1970 "Minority status and the pursuit of professional Social Problems 22:184-98. careers." Social Forces 49:245-59. Rogers, M.A. Pleck, J. H. 1975 "A different look at word problems even 1975 "Masculinity-femininity: Current and alterna- mathematics texts are sexist." Mathematics tive paradigms." Sex Roles 1:161-78. Teacher 68:305-7.

237 SEX TYPING IN OCCUPATIONAL SOCIALIZATION 231

Rosenbaum, J.C. Smith-Lovin, L., and A.R. Tickamyer 1978 "The structure of opportunity in school," Social 1978 "Nonrecursive models of labor force .partici- Forces 57:234756. pation, fertility behavior, and sex role attitudes." American Sociological Review 43:541- 1980 "Track misperception and frustrated college plans: An analysis of the effects of tracks and 57. track perceptions in the National Longitudinal Spitze, G., and J. Huber, Survey,Sociology of Education 53:74-88. 1980 "Changing attitudes towards women's nonfam- ily roles: 1938 to 1978." Sociology of Work and Rossi, A.S. . Occupations 7:317-35. 1965 "Barriers to the career choice of engineering, medicine, or science among American women." Steele, 13. Pp. 51-127 in J. A. Mattfeld and C.C. Van Aken 1977 "Sexism in math texts." Edcentric (Spring/ (eds.), Women and the Scientific Professions. Summer):17-19, 60. Cambridge, Mass.: MIT Press. Stephenson, R.M. Schafer, W.E., and C. Olexa 1957 "Mobility orientation and stratification of 1,000 1971 Tracking and Opportunity. Scranton, Pa.: ninth graders," American Sociological Review Chandler. 22:204-12. Schmidt, J. L , and J.W. Rothney Stern, R.H. 1955 "Variability of vocational choices of high school 1976 "Review article: Sexism in foreign language students." Personnel and Guidance Journal textbooks." Foreign Language Annals 9:294- 34:142-46. 99. Schlossberg, N. K., and J. Goodman Sternglanz, S. H., and L.A. Serbin 1972 "A wolnan's place: Children's sex stereotyping 1974 "Sex role stereotyping in children's television of occupations." Vocational Guidance Quar- programs." Developmental Psychology 10:710- te;ly 20:266-70. 15. Schramm, W., J. Lyle, and E.B. Parker Stright, V.N. 1961 Television in the Lives of Our Children. Stan- 1960 "A study ofre attitudes toward arithmetic of ford, Calif.: Stanford University Press. students anteachers in the third, fourth and sixth grades. The Arithmetic Teacher 7 :280- Sewell, W.13., and A.M. Ornstein 86. 1964 "Community of residence and occupational choice." American Journal of Sociology 70:551 - Super, D.E. 63. 1953 "A theory of vocational development." The W.H., A.O. Haller, and A. Portes American Psychologist 8:185-90. 1969 "The educational and early occupational at- 1957 The Psychology of Careers: An Introduction to tainment process." American Sociological Re- Vocational Development. New York: Harper. view 34:82-92. Sweet, J.A. Sewell, W.H., A.O. Haller, and C. Ohlendorf 1973 Women in the Labor Force. New York: Serb- 1970 "The educational and early occupational at- inar Press. tainment process: Replications and revisions." Tangri, S.S. American Sociological Review 35:1014-27. 1972 "Determinants of occupational role innovation Sewell, W. H. , R.M. Hauser, and W.C. Wolf among college women." Journal of Social Is- 1980 "Sex, schooling, and occupational status." sues 28:177-99. American Journal of Sociology 86:551-83. Tavris, C., and C. Of fir Shepard, W.O., and D.T. Hess 1977 The Longest War: Sex Differences in Per- spective. New York: Harcourt Brace Jovanov- 1975 "Attitudes in four age groups toward sex role division in adult occupations and activities." ich. Journal of Vocational Behavior 6:27-39. Tedesco, N.S. Sherman, J., and Fennema, E. 1974 "Patterns in prime time." Journal of Com- munication 24:119-24. 1977 "The study of mathematics by high school girls and boys: Related variables." American Edu- Terman, L. M., and L. E. Tyler cational Research Journal 14:159-68. 1954 "Psychological sex differences." Pp. 1064-1114 Siegel, C.L. F. in L. Carmichael (ed.), Maflual of Child Psy- chology, 2d ed. New York: Wiley. 1973 "Sex differences in the occupational choices of second graders." Journal of Vocational Behav- Thomas, A.11., and N.R. Stewart ior 3:15-19. 1971 "Counselor response to female clients with de- Siegel, E., and E.A. Curtis viate and conforming career goals." Journal of Counseling Psychology 18:352-57. 1963 "Familial correlates of orientation toward future employment among collegewomen." Thompson, C. L. , and J. L. Parker Journal of Educational Psychology 54:33-37. 1971 "Fifth graders view the work world scene." Singer, S. L., and B. Stefftre Elementary School Guidance and Counseling 1954 "Sex differences in job values and desires." ..54:281-88. Personnel and Guidance Journal 32:483-84.

238 /

232 MARGARETMOONEY MARINI AND MARY C. BRINTON

Thoniton, A., and D. Freedman Waite, L.J., and P.M. Hudis 1979 "changes in the sex role attitudes of women, 1980 The Development and Maintenance of a Seg- 1962-1977: Evidence from a panel study." regated Labor Force: Review, Synthesis, Cri- American Sociological Review 44:831-42. tique of Recent Research. Report prepared for Thurow, L. the National Commission for Employment Pol- 1975 Generating Inequality. New York: Basic. icy. Tibbetts, S.L. Walker, K.E., and M.E. Woods Time Use; A Measure of Household Produc- 1975 "Sex-role stereotyping in the lower grades: Part 1976 of the solution." Journal of Vocational Behavior tion of Family Gooch and Services. Washing- 6:255-61. ton, D.C.: Center for the Family, American Home Economics Association. Tidball, M.E. "Perspective on academic women and affirm- Weitzman, L.J., and D. Rizzo 1973 Biased Textbooks: A Research Perspective. ative action." Educational Record 54:130-35. 1974 Washington, D.C.: The Resource Center on Tittle, C. K. Sex Roles in Education, National Foundation 1981 Careers and Family: Sex Roles and Adolescent for the Improvement of Education. Life Plans. Beverly Hills, Calif.: Sage. L.J.D.' Miler, E. Hokada, and C. Ross Turner, R. H. .,1972 Sex role socialization in picture books for pre- 1964 "Some aspects of women's ambition." Ameri- school children." American Journal of Sociol- can Journal of Sociology 70:271-85. ogy 77:1125-50. Tyler, 1. E. Welch, F. 1965 The Psychology of Human Differences. New 1979 "Comment." Pp. 168-70 111 C.B. Lloyd, E.S. . York: Appleton-Century-Crofts. Andrews, and C.L. Gilroy (eds.), Women in Umstot, M.E. the Labor Market. New York: Columbia Uni- 1980 "Occupational sex-role liberality of third-, versity Press. fourth-, fifth-, and seventh-grade females." Sex Weds, C.E. Roles 6:611-18. 1966 "Social class and initial career choice of college U.S. Commission on Civil Rights freshmen." Sociology of Education 39:74:85. 1978 Social Indicators of Equality for Minorities and White, K. Women. Washington, D.C.: U.S. Govern- 1967 "Social background variables related to career ment Printing Office. commitment of women teachers." Personnel U.S. Department of Labor, Bureau of Labor Statistics and Guidance Journal 45:649-52. 1977 United States Working Women: A Databook. Williams, G. Washington, D.C.: U.S. Government Printing 1976 "Trends in occupational differentiation by sex." Office. Sociology of Work and OcCupations 3:38-62. U.S . Department of Labor, Women's Bureau 1979 'The changing U.S. labor force and occupa- 1975 1975 Handbook on Women Workers, Bulletin tional differentiation by sex." Demography 297. Washington, D.C.: U.S. Government 16:73-88. Printing Office. Witty, P.A., and H.C. Lehman U.S. President's Council of Economic Advisors 1930 "Some factors which influence the child's choice 1973 Economic keport of the President. Washing- of occupations." Elementary School Journal ton, D.C.: U.S. Government Printing Office. 31:285-91. Vanek, J. E. Women on Words and Images 1974 "Time spent in housework." Scientific Amer- 1975aChanneling Children: Sex Stereotyping in ican 231:116-20. Prime-Time TV. Princeton, N.J.: Women on Words and Images.' Verkiden-Hilliard, M.E. 1975bDick and Jane as Victims: Sex Stereotyping in 1977 "Counseling: Potential superbomb against sex- , ism." American Education (April):12-15. Children's Readers. Princeton, N.J.: Women on Words and Images. Vetter, L. Wood, P.S. 1973 "Career counseling for women." Counseling 1980 "Sex differences in sports." New York Times Psychologist 4:54-67. Magazine, May 18:30-33, 38, 98-104. Vogel, S.R., I. K. Broverman, D. M. Broverman, F. E. Zellner, H. Clarkson, and P.S. Rosenkrantz 1975 "The determinants of occupational segrega- 1970 "Maternal employment and perception of sex tion." Pp. 125-45 in C.B. Lloyd (ed.), Sex, roles among college students." Developmental Discrimination, and the Division of Labor. New Psychology 3:384-91. York: Columbia University Press. Vroom, V. H . Zinberg, D. 1964 Work and Motivation. New York: Wiley. 1974 "When the future becomes the present." In. Waite, L.J. R. B. Kundsin (ed.), Women and Success: The 1976 "Working wives: 1940-1960." American Soci- Anatomy of Achievement. New York: William ological Review 41:65-80. Morrow.

239 12Commentary WENDY C WOLF

In Chapter 11, Marini and Brinton pro- tional choices. Many have tried small ex- vide a very good review of the literature on periments to change one aspect of a girl's the multitude of factors that influence the educational experiencelo and behold, they occupational choice oirgirls prior to entrance don't find any impact. This is, not very sur- into the labor market. My perspective is that prising. Even if there is an intervention in the goal of such a paper should be to identify one or two areas, there are always other the most critical or most powerful forces that factors that in fact reinforce occupational influence girls' job choices, decide which of choices that are sex typed or views about these are amenable to policy intervention, appropriate roles for girls. and discuss what strategies are effective to The second point relates to the issue of intervene in these processes. In light of that premarket versus market forces. It may be goal, a few general comments are in order. a fallady to neatly separate all forces into two The forces that impinge on job choice for types. There are a number of ways thatthe girls are may, strong, and cumulative. Their labor market feeds back information to young cumulativeness is important to remember girls about appropriate roles for females. It when cowidering points of intervention. If is important, therefore, not make such a one intervenes early in a girl'slife, there are rigid distinction. myriad other forces that act on her before One area of special importance in Marini she gets to the point of making a job choice. and Brinton's review relates to the math/ Changing one aspect of the system rarely science issue in high school. Despite recent has, or for that matter should be expected. news articles suggesting that the difference to have, marked impact on the ultimate job in math/science ability between girls and choice. For example, we know that text- boys may be due to girls' lack of testoster- books affect girls in some way. So do coun- one, there is evidence that girls and boys selors. But changing one textbook, or text- with comparable levels of achievement in books in one course, or changing counselors math hnd science at the end of the eighth in one school in,one year, is not likely to grade take different amounts of math and have a significant impact on girls' occupa- science in high school. Enrollment in high

233 240 234 WENDY C WOLF school math and science courses has fairly There is anecdotal evidence. (and we need substantial consequences on the kinds of job more concrete empirical evidence) that at choices that are made at later points. This least for women with high 'income needs, is one area in which concrete steps can be one can stimulate the demand for nontra- taken to allow girls to make schooling choices ditional training and employment, despite that will not limit their access to jobs later. years of socialization. Nevertheless, there is Generally, the cumulativeness of forces variation in aspirations for nontraditional oc- makes me a bit skeptical about the success cupational choices -(and the ability to stim- of any one intervrntion,Qespecially if it is too ulate them). Older women tend to be more distant from the time of the actual job choice. likely to have such aspirations, and Hispan- The closer to the time of job choice that the ics less, than black and white women. In intervention is made, the more likely it is short, there is some promise, despite the to change the job aspirations of girls. evidence that high school g rls are likely to What prompts a statement such as this? be disinterested in nontraditional careers. Looking at a number of nontraditional, pro- One final comment is in order. It is a grams for women makes it clear that one can shame that we, as researchers and policy stimulate the.demand for such programs. If, makers, often focus so much on individual told about the advantages and disadvantages factors that affect individual women and their of men's jobs, women with high income choices and neglect employers and how they, needs, who are older, who have been out through subtle and some not so subtle in the labor force, or who have been in tra- mechanisms, influence occupational sex ditionally female jobs that are low paying segregation. will respond with into est. At least enough This neglect occurs for a number of rea- persons will respond to fill openings. The sons. First, it is easier to study women than demand for nontraditional work can be stim- it is to study employers. Second, in a pe- ulated. This may he one of the reasons that, culiar way, it is felt that we may have more looking at college students or young girls, control or, effect on individuals than we do one doesn't observe much of a change over on institutions and businesses. It should be time in sex-typed occupational aspirations. obvious that individual factors and often in- Once women have been working for a while dividual forces are but one part of the spec- and have had some negative experiences, trum. There are numerous factors related they are more willing to entertain nontra- to behaviors of employment that affect women ditional career options than at younger ages, once they enter the labor force. And, in fact, when peer group pressures are important there are labor market factors that influence forces reinforcing traditionally female aspi- their choices prior to entry into the labor rations. market.

f!'

241 Institutional Factors Contributing to 13Sex Segregation in the Workplace PATRICIA A. ROOS and I t BARBARA F, RESKIN

Researchers have frequently attempted to The effect of informal processes on women's explain sex segregation in the workplace by employment prospects has been the topic of invoking either workers' or employers' pref- much workin recent years (e.g., Epstein, erences. In economic terms,the former em- 1970a, 1975; Coser and Rokoff, 1971; Kanter, phasizes the characteristics and choices of 1977). For example, women's exclusion fry the labor supply; the latter claims gender or marginality in work groups,which oftin discrimination in 'the labor market. Re- extends into nonworking hours (Martin, 1980), search guided by each perspective has shed has been found to impair their performance light on the causes of the unequal distri- on the job (Kanter, 1977;Schafran, 1!.: 1). butions of the sexes across occupations, but Women may also lack, access to necessary in- neither workers' nor employers'references formation or be overlooked by senior people systematically assess how the organization who could facilitate their career advancement of labor markets and the way work is carried (Epstein, 1970a). Sometimes coworkers try to out within establishments-constrainthe sexes' sabotage women's entry into positions men occupational outcomes. As Granovetter (1981) customarily hold by harassing them or refus- persuasively argued, to understand the oping to provide help or instruction'(O'Farrell eration of the labor market, one must ex- and Harlan, this tqWme). Such informal bar- t. amine the processes through which jobsand riers have been ibund to hamperwomen's workers are matched. Scott (1981:186) di- employment prospects in such diverse occu- rects our attention to internal(or organi- pations as blue-collar jobs (Walshok, 1981a; zational) processes and workplace mecha- Harlan and O'Farrell, 1982), police forces nisms, that result in people beingrecruited, (Martin, 1980), forestry (Enarson, 1980), con- allocated, and retained in particular jobs. struction (U.S. Department of Labor, Em- Their work and others (e.g., Kanter, 1977; ployment Standards Administration, 1981),law Kelley, 1981) focus on both formal and in- (Epstein, 1981), medicine (Freidson, 1970), formal processes existing within the work- science (Reskin, 1978), and management place that constrain the free operation of the (Kanter, 1977). labor market. In contrast to these informal processes,

235 242 236 PATRICIA A. ROOS AND BARBARA F. RESKIN the segregative effects offormal barriers that to training, job assignment, and mobility oc- are institutionalized in labormarkets and cur within labor markets, and ourdiscussion firms' personnel practices have been inves- considers relevant labor market theories. In tigated less thoroughly. The effects of these synthesizing these theories, we draw heav- institutional mechanisms constitute the fo- ily on Althauser and Kalleberg's (1981) con- cus of this paper. We defineinstitutional- ceptual analysis of firms, occupations, and h.:A factors as those that are either embed- labor market structure, although we deviate ded in or stem from the formal procedures from their nomenclature and distinctions to or rules of firms and other labor market en- highlight what is relevant for our purposes. tities. These processes include recruitment It is useful first to distinguish between ex- and job assignment practices, promotion ternal and internal labor markets. External systems, administrative regulations regard- labor markets include traditional, competi- ing job transfers, stipulations regarding par- tive markets through which employers fill ticipation in training programs, and barriers entry-level jobs on ladders and qther jobs to information about certain labor market that are not on ladders (including Althauser opportunities. Some factors, such as senior- and Kalleberg's "secondary labor market"). ity systems, are by-products of administra- The entry-level and job assignment barriers tive procedures established for other pur- we identify occur in this market. poses.Othersrepresentdeliberateseg- Internal labor markets, according to Alt- regative practices in keeping with laws no hauser and Kalleberg (p. 130), have three longer on the books (as in the assignment of defining characteristics: a job ladder exists, men to "heavy" work and women to"liolit" entrance is restricted to the lowest level, work; Bielby' and Baron, this volume). In and movement up the ladder is accom- this paper, we examine how these institu- panied by the progressiVe acquisition of job - tionalized mechanisms within establish- related skills and knowledge. The barriers ments, and other organized entities in labor to mobility we identify reside primarily in marke s such as unions and federally ad- firms' internal labor markets. 'Althauser and ministered training programs, contribute to Kalleberg distinguish "firm internal labor sex segregation by limiting the accessof markets" from "occupational internal labor workers of one sex to certain occupations markets" and "occupational labor markets," and channeling them into others. The ein7 the last of which lacks the internal labor phasis in this paper is on institutionalized market characteristics described above. In- barriers to women's employment in sex- dividuals in the latter two markets have spe- atypical jobs (e.g., construction and police cialized skills and knowledge, acquired work). Of course, occupational sex segre- through extensive education or training, ac- gation is also maintained in part by insti- companied by practice, which may culmi- tutional barriers limiting men's employment nate in licensing, certification, or registra- in female-typical jobs (e.g., secretarial and tion. Although these occupational markets nursing) and mechanisms fostering both sexes' may exist within firms, they often span F':v- employment in sex-typical occupations (e.g., eral enterprises, and mobility among firms women in clerical orlibrarian jobs and men is common. The occupational labor market in engineering or firefighting work), and we is relevant here because it is the locus of discuss these processes, albeit less exten- institutionalized rules or policies restricting sively. The barriers we consider occur at access to training. four points: preemployment training, job ac- Institutional barriers to the retention of cess and assignment, job mobility and re- workers in sex-atypical jobs tend not to be tention. located in labor markets as commonly con- Most of the mechanisms that affect access ceived but in the way that tasks are orga-

2 4 3 INSTITUTIONAL. FACTORS CONTRIBUTING TO SEX SEGREGATION 237 nized or in the informal organization of the labor markets is available in a variety of set- workplace. Flintily,Arne constraints on the tings. High schools offer some vocationally sexes' free access to jobs operate outside specific classes, and other courses (for ex- labor markets in other, ,institutional arrange- ample, on electronic equipment repair or ments (examples include communication bartending) are available commercially to networks and child care facilities). We should anyone who can afford the tuition. The, higher note that these functions could be institu- education system selects and trains persons tionalized within labor markets (including for most professional occupations. Sex dis- the workplace) to expand women's occupa- crimination in training, along with institutional options. (Kanter, 1977, offers some tional barriers that discourage women's par- useful strategies along these lines.) ticipation in "male" professions, has been We note at the outset that because the thoroughly addressed elsewhere (Epstein, practices we consider are institutionalized 1970a,b; Theodore, 1971, Hochschild, 1975).1 their effect is net of employers' intentions In comparison, the effects of formal voca- and workers' preferences. However, they tional and technical training programs op- neither emerged nor exist in a vacuum, in- erated individually or cooperatively by dependent of social or cultural factors that union:.;, employers, and public agencies have define certain kinds of jobs as appropriate received less attention. Because such train- for one sex only. Widely held, deep-seated ing is the route to and indeed is some- stereotypes about differences between the times requisite for many predominantly sexes and assumptions about their proper male occupations, we first consider insti- roles provide an often invisible foundation tutional barriers in suCa training. for many of these organizational practices and encourage sex "traditional" decisions by Apprenticeship Programs individuals in the labor market (Reskin and Hartmann, 1984). In sum, these institution- Apprenticeship programs are an impor- alized factors have a life of their own in terms tant avenue for entry into skilled blue-collar of their segregative consequences. How- jobs, especially those in union-dominated ever, they persist in part because they are occupations and industries (since appren- reinforced by sex-role norms and cultural ticeship is often the simplest way to enter beliefs that shape the preferences of employ- a union; Briggs, 1981). These programs pro- ers, workers, and consumers. vide a formal mechanism whereby skilled workers pass on their knowledge to new workers through classroom and on-the-job INSTITUTIONAL BARRIERS TO JOB training. Such programs are seldom avail- TRAINING able to women. Despite progress in the For many occupations, workers acquire number of occupations in which women are the necessary skills on the job, and barriers now apprenticed (up from 17.5peticent in to training for women in sex-atypical jobs 1973 to 44.4 percent in 1977), in 1978 women reside primarily in the resistance of male constituted only 2.6 percent of the more coworkers. However, many occupations that than 250,000 apprentices registered with the exist in what Althauser and Kalleberg (1981: U.S. Department of Labor (Ullman and 134) call "occupational labor markets," require substantial preemployment training. Here training has the same function as "port We do not consider training that takes place in pub- of entry" positions in internal labor markets lic schools and institutions of higher education. For an in providing access to a job ladder. The analysis of the impact of educational institutions on sex training that permits entry into occupational segregation, see Marini and Brinton, in this volume.

244 238 PATRICIA A. ROOS AND BARBARA F. RESKIN

Deaux, 1981).2 In addition to their small Often apprenticeship requirements are numbers, women are located disproportion- harder for women to meet. The upper age ately in certain apprenticeship categories: limit (as low as 24 to 27 in some trades) is they constituted 54 percent of the barber the most significant obstacle for women, given and beautician apprentices in 1975 but only the typical timing of child-bearing. Because 6percentofcraft-workerapprentices of their socialization, few young women even (O'Farrell, 1982) and 1.9 percent of con- consider skilled blue-collar work before they struction apprentices (Ullman and Deaux, have spent several. years in sex-traditional 1981; Briggs, 1979:225-226). jobs, at which point economic exigencies often. The presence of women is rare in ap- force them to seek better-paying work (Kane prenticeship programs because they are less and Miller, 1981:90; Waite and Hudis, 1981; likely than men to learn about openings, to Walshok, 1981a; O'Farrell, 1982). meet their requirements, and to be se- Apprenticeship programs are geared to lected. Information about apprenticeship young unmarried men who can sustain un- programs, usually transmitted by friends or paid work or low wages and the uncertainty relatives (Sexton, 1977), outreach programs, of immediate employment. These condi- and publicity, is less likely to reach women tions constitute particular obstacles for eco- who seldom belong to the networks in which nomically disadvantaged women, who are such information is circulated (Waite and likely to have families to support. In the Midis, 1981). Even in the female-domi- construction trades, for example, high ap- nated occupation of hairdressing, Allison plication or union induction fees and the (1976:390) found that women were less likely long wait between application and accept- than men to have semiinformal apprentice- ance are hardships for women with depend- ships, which were associated with subse- ents and may deter them from seeking ap- quent employment in better shops and at prenticeships (U.S. Department of Labor, higher wages. The few women who had been Employment Standards Administration, 1981: apprentices had male relatives in the in- Walshok, 1981a). dustry and thus presumably were better in- Women who apply are at a disadvantage formed about their availability and/or value. in the selection process, Unions practice Many labor-management agreements nepotism or require sponsorship by a mem- stipulate that apprenticeship openings be ber in awarding apprenticeships (Simmons advertised only within the plant, where few et al., 1975:119; U.S. Department of Labor, women have worked. Moreover, they may Employment Standards Administration, 1981). he posted in areas inaccessible to women In the construction industry, for example, employees such as men's restrooms (Briggs, where unions often have absolute control 1974:13). When unions go outside the plant over the certification and supply of labor, toecruit, they might consult high school labor unions have been particularly resistant industrial at ts teachers, who are unlikely to to accommodating women. As a conse- nominate female candidates since only a small quence, women must generally rely on out- number of women take such courses.3 reach agencies to place them in jobs or cer-

2 This figure is even smaller than the number of women ing their students for traditionalls lower-paid employworkers in craft jobs in 1978, 5.6 percent of craft ment than the predominantly male schools. The study workers were women (Ullman and Deaux, 1981). cited sex-biased admission tests, guidance con. st-lors ' One recent study of New York City's vocational steering students to sex-typical fields, and male antag- education system found that cif 21 job-training high onism to female students in traditionally male fields as schools 12 were primarily male and 5 were primarily the immary explanations fur the large sex difference in female. The 5 predominantly female schools were train- vocational training (Neu York Times, 911,3

24? INSTITUTION4r6CTORS CONTRIBUTING TO SEXSEGREGATION 239 tification programs in the construction trades. developed to address the problems facing For example, one study found that women women apprentices. Walshok (1981b) claimed who had not been apprentices were aided that "competency-based testing" of appren- by family ties in gaining access to the con-- tices in a pilot program at General Motors struction trades (U.S. Department of La- offered apprentices feedback on expecta- bor, Employment Standards Administration, tions and performance, while reassuring 1981:34, 41). journeymen that standards had not been re- Another factor hampering women's selec- duced for women and minority apprentices. tion into apprenticeships is that they are Instituting placement services has also proved unlikely to have completed vocationally rel- essential for women, who often encounter evant programs in high school and are often discrimination from employers (U.S. De- unfamiliar with the tools, procedures, and partment of Labor, Employment Standards terminology used in blue-collar work. The Administration, 1981:33; Walshok, 1981a). current unstable nature of the economy un- Preapprenticeship training in certain con- doubtedly works against women being ac- struction programs enhances women ap- cepted in apprenticeship programs, since the prentices' chances for success. In addition, availability of apprenticeships declines with efforts to match female apprentices with rising unemployment (Briggs, 1979).4 journeymen can reduce friction on the job The structure of apprenticeship programs and thus contribute to more effective train- hinders women's ability to complete them ing (U.S. Department of Labor, Employ- and find craft jobs. Many female apprentices ment Standards Administration, 1981:57). whom Walshok 11981b:177) interviewed However, apprenticeship is not the pri- complained that they lacked the opportunity mary entry channel into the trades. Even for hands-on experience and that hostile in the unionized sector, which represents journeymen previnited their learning nec- 40 percent of the industrial work force, only essary skills. Some (New York Times,1982b) one-fifth of workers enter trades through ap- argue that unions may provide "separateand prenticeship programs (U.S. Department of unequal" i:pRrenticeship training for work- Labor, Employment Standards Administra- ers who aa. not white males. Oneexample tion, 1981:23). We now turn to other paths led the New York State Division of IIinnan by which workers enter jobs. Rights to find a construction union local guilty of unlawful discrimination for requiring non- Federal Job Training Programs white apprentices to work more than twice as long as whites to reach journeyman sta. The federal government sponsors training tus. Moreover, nonwhite trainees had ob- programs to provide an avenue for un- solete textbooks and were denied a fifth year skilled, economically disadvantaged work- of classroom training. Whether the same in- ers to move into more skilled blue-collar adequate training also affects women's ap- work. Recent investigations of the effects of prenticeship experience requires investi- thew federal programs on women (Harlan, gation. 1980, 1981; Berryman and Chow, 1981; Wolf, Several experimental programs have been 1981; U.S. Commission on Civil Rights, 1981; Waite and Berryman, this volume) indicate sex inequa:ay in training, employment, Oc- cupational placement, and wages. 4 O'Farrell and Harlanthis volume), for example, The Comprehensive Employment and finind evidence that women are more likely to make Training Act (CETA), enacted in 1975 and progress at integrating traditionally male employment in rapuil, expanding firms thur in companies experi- amended in 1976 and 1978, is the largest encing retrenchmv!it. federal program designed to increase the

246 240 PATRICIA A. ROOS AND BARBARA F. RESKIN employability and earnings of the disadvan- ployment.5 Second, CETA programs are taged (U.S. Commission on Civil Rights, targeted to families rather than to individ- 1981:28 -29). As, paraphrased byWolf uals and to the single person within the fam- (1981:87), the amended law recommends that ily who has "primary" support responsibil- (:ETA sponsors: ity. These features of the CETA legislation hinder the participation of married women artificial bar- . overcome sex stereotyping and because any man in the home is assumed to riers to employment . . . (presumably] by at- bear the support obligations for the family tempting to (1) expose women to nontraditional (Harlan, 1981:37). Third, veterans' prefer- additional bar- career options, and (2) overcome women's participation riers to the employment of women (such aschild ence policies reduce care, transportation to work). in CETA. Prior to 1973, this preference was explicit: President Carter directed that 35 Despite the fact that the law requires prime percent of those assigned to public service sponsors to include eligible groupsequitably, employment be Vietnam veterans. Al- other regulations favor other groups such as though this directive was rescinded, some Vietnam veterans or youth (Wolf, 1981:109). claim the' preference persists (Wolf, As a result,in 1977 women wereunder- 1981:109). Fourth, while the CETA author- represented in CETA programs relative to ization empowers prime sponsors to provide their numbers in the eligible populations (Na- support services (such as child careand tional Commission on Manpower Policy, as transportation costs) to those otherwise un- cited in U.S. Commission on Civil Rights, able to participate, the standard in-program 1981:29). More important here are the seg- ervaluation of a high ratio of trainees to ex- regative implications of their uneven partici- penditures discouraged sponsors from using pation in individual programs. Forexample, futxds in this way (U.S. Commission on Civil women were less likely to beassigned public Rights, 1981:30-32; Wolf, 1981). service jobs or on-the-job training (which more often leads to unsubsidized jobs) and more INSTITUTIONAL FACTORS ASSOCIATED likely to be in classroom training usually WITH ACCESS TO SEX - TYPICAL. AND 0'- preparation for clerical jobs (Harlan, 1980; SEX-ATYPICAL JOBS Neill and Braun, 1981:102; Wolf, 1981:94). In fiscal year 1980, for example, 56 percent of This section considers the institutional those in classnxmi training were female. com- factors affecting women's access to entry- pared with :36 percent of those in on-the-job level jobs, particularly barriers to sex-atyp- training programs (Bendick. 1982:259). Inad ical jobs. Typically these barriers reside in dition, while CETA women expressed in- external labor markets through which entry- re asing interest innontraditional jobs be- level positions are filled. They are of two tk ven 1976 and 1978. theirplacement in such types: firm-based limitations fin certainkinds jobs dropped (Wolf, 1981:98). of jobs and restrictions on women's access Several features of federally sponsored to certain occupational labormarkets. Sev- training programs impedewomen's access eral firm-based limitations restrict the sexes' to notrtraclitontal jobs. First,these programs put priority on quick, low-costplacement in order to reduce welfare dependency. which This problem affects all workers Schiller (1980.197) noted the general tendency of federally sponsored job does little to ensure long-run financial in - training programs to -cream- the rims?jobready pro. depein knee. 'Phis emphasis On placing the gram applicants for job placement, thusenhancing pro- most -job-ready'. individuals encourages gram success ratios at the expense of the mostneedy placing % i omen itt tiaditionalIv female em- job seekers 6

2-17 O

INSTITUTIONAL FACTORS CONTRIBUTING TO SEX SEGREGATION 241

access to sex-atypical entry-level jobs, in- differ for blue- and white-collar workers, we cluding employers' and workers' prefer- discuss them separately. ences and beliefs, entrance requirements, and organizational practices regarding job (Blue-Collar Workers assignment. Except when they are institutionalized in personnel practices, employ- Access to Information Regarding Job ers' discriminatory preferences fall outside Opportunities Women are unlikely to learn the scope of our essay. However, we men- about predominantly male blue-collar jobs tion their manifestation in statistical dis- for several reasons. First, the common as- crimination because of their special impor- sumption that women are not interested in tance in excluding women from entry-level craft employment is reflected in brochures positions.In employment, statistical dis- and publicity oriented toward men (Steven- crimination involves treating individuals son,1977; Briggs,1981). Such materials based on beliefs about group differences in generate little response from women. In an relevant characteristics (Phelps, 1972). With ingenious study to determine the impact of respect to women it is most often manifest such materials, Bern and Bem (1973) found in employers' reluctance to hire any woman that sex-biased wording in job advertise- for jobs that require appreciable on-the-job ments and the placement of ads in sex-seg- training, because they believe many young regated newspaper columns discouraged women leave the labor force to have chil- women's interest in traditionally male jobs: dren. As a result, newly hired females are only 5 percent of the women surveyed ex- often assigned to low-skilled dead-end jobs pressed interest in linemen and framemen (Drinker et al., 1970). Because transferring jobs when they were written in sex-biased across internal labor markets is very diffi- language, but 25 percent were interested cult, if not impossible (see the next section), when the language was sex-neutral, and 45 statistical discrimination has long-lasting im- percent expressed interest when the ad was plicationsfOr women's occupational out- written to appeal specifically to women. comes. Traditionally, blue-collar employees in the "With regard to the second harrier, two crafts have been recruited from secondary processes restrict the occupational labor schools and through employee referrals, markets in which women can seek jobs. The methods unlikely to elicit female recruits first we have already discussed mecha- (Colladay and Wulfsberg, 1981:78). To the nisms limiting their chance to train for cer- extent that employers rely on employee retain occupations. Women also lack access to ferrals, new recruits will tend to reps °duce selected labor markets because they have the existing sex-segregated work fOre.6 This insufficient infiirmation about their very ex- is especially true in certain industries such istence. Few methods ofjob recruitment are as construction, where referral and hiring fully public, state employment services and are often nnplished via nepotism and word classified advertisements in mass irmlation of mouth recruitment (U.S. Department of newspapers are notable exceptions. Instead, Labor. Employment Standards Administra- personal ties throtigh which job seekers learn tion, 1981:22). of possible job opportunities and employers of possible _pplicants are important in ele- temining who is hired at the entry level. " Ilarkess (1980) limnd that empliisers Often bold their employees responsible for the job candidates they noni Below We show hOW Sex- segregated per- so es en unbiased workers 111,. sheSitatr before sonal networks foster sex-typed occupa- taking the risk of recommending someone whose sex tional outcomes. Because these mechanisms or race does not match the work group

248 242 PATRICIA A. ROOS AND BARBARA F. RESKIN

Employers' reliance on traditional re- 40 percent of the women in nontraditional cnrinnent techniques reflects their belief that blue-collar occupations whom NA'alshok a homogeneous labor force will facilitate the (1981a) interviewed had direct contact with transfer of craft knowledge via on-theob special recruitment and counseling agencies training (Stevenson, 1977). Whether or not that were specifically designed to link in- they are right, the' practice perpetuates sex terested women with job opportunities in segregationIn general, to the extent that nontraditional fields. Almost none of these the recruitment process involves parties who women found their jobs through advertise- hold sex-typed notions about who should ments. Thus, while men can he recruited hold certain jobs whether they be re- through existing recruitment channels, cruiters, training program administrators, placing women in heavily male jobs appears current employees, or jot) seekers formal to require specialized intermediary place- mechanisms such as outreach programs are ment agencies or uther outreach efforts.' necessary to ensure that women are trained and recruited. As we noted with respect to Employer Practices Regarding Entrance apprenticeship programs, information about RequirementsSeveral kinds of rules or re- most typically male jobs is circulated in all- quirements employers impose restrict wom- male informal netw(,;ks. For example, a 1966 en's access to a variety of jobs. While insti- study by Sheppard and Belitsky (as cited in tuted to help returning veterans, veterans' Folk, 1968) noted that 77 percent of the preference rules also limit women's access blue-collar workers surveyed found their jobs to several occupations that have been la- through friends and relative., In our dis- beled male. For example, 65 percent of the cussion of access to information about white- government agencies 'surveyed gave some collar jobs (see below), we consider in more form of preference to veterans' in selecting detail the segregative implications of infor- police officers, an occupation women have mal networks and review several empirical had considerable difficulty entering (Eisen- studies. berg et al., 1974, as cited in Martin, 1980:47). Employers often claim that they cannot By restricting competition, veterans' pref- comply with federally mandated affirmative erence rules serve the latent function of re- action requirements because the pool of el- serving such occupations for men. Interest- igible women is too small (U.S. Department ingly,some'stateshave exempted tradi- of Labor, Employment Standards Admin- tionally female occupations from veterans' istration, 1981), while women interested in preference (Personnel Administration of nontraditional jobs contend that there are Massachusetts v. Feen(' y, 1979), so these too few openings to accommodate all those policies do not increase male access to tra- seeking blue-collar employment (Westley, ditionally female occupations. Despite their 1952). Kane and NI flier (1981:88) argued that segregative effect, the Supreme Court al- 1)0111 views are accurate: while the numbei lowed veterans' prekrence rules to stand in of women who want to participate in out- its 1979 decision iii Feeney. reach programs and the number of emplo For much of this century, protective labor s requesting referrals have increased, the laws ruled out many occupations to women resources for training have remained con- and provided an excise to employers who stant. As a consequence, the supply of trained women those programs are able to produce is serelv Svc U. Dpartment of labor.itiplo!. taunt

Special programs often succeed in placing Standard, AdministimumItbtil :12% for11U himmethod women in nontraditional jobs. For eample. 1)1 rtiltmnt oprats lit the totistnition itidihtr!.

2-19 INSTITUTIONAL FACTORS CONTRIBUTING TO SEX SEGREGATION 243 did not want to hire women for other jobs. dexterity) may inhibit men's employment. Under the reguLtions interpreting Title VII Employers have sometimes claimed that the of the 1964 Civil Rights Act, such laws can- cost of adapting machinery for women is not be applied to only one sex, Recently, prohibitive, This problem is highlighted in however, employers in some industries (e.g., the military. In discussing the costs of rede- rubber, lead, metal, and chemical) have re- signing special clothing and equipment to fused to employ women of child-bearing age accommodate the increasing the number of in jobs that expose them to toxic substances women in the Armed Forces, Binkin and (e.g., lead, vinyl chloride, carbon disulfide, Bach (1977:54) noted: pesticides), rather than develop standards that would protect both male and female In particular, the assignment of women to traditionally-male occupations could require exten- workers. Bell (1979) and ,Wright (1979) sive changes. . . . In a number of critical di- pointed out that employers ignore toxic haz- mensionsweight, stature, sitting height, . . . ards in traditionally female jobs (e.g., op- the average woman measures significantly less P erating room nurses' exposure to waste an- than the average man." aesthetic gases, beauticians' to hydrocarbon hairspray propellants, flight attendants' to above-average levels of radiation). Organizational Practices Regarding Job In Grigg v. Duke Power Company (1971), AssignmentIn establishments with only a the Supreme Court construed Title VII of few occupations, decisions regarding hiring the 1964 Civil Rights Act to prohibit job and job assignment may be one and the same. requirements that disproportionately ex- However, for large establishments that are clude members of protected groups unless continuously hiring for a variety of occu- they were demonstrably job related, This pations, it helps to examine separately the ruling was applied in 1975 to strike down factors associated with the kinds of jobs to the height and physical agility requirements which workers are assigned the focus of that barred nearly all women from being this section. Sex differences in initial job police officers in San Francisco (Gates, 1976; assignments reflect sex stereotypes about 'Martin,1980:44).Reflecting the lag be- appropriate work roles fin: men and women. tween court rulings and changes in prac- Certain jobs have been historically sex typed tices, many police departments continue to as male or female (Oppenheimer, 1968). Sex use height and agility requirements, with typing persists in part because men and the result that women are underrepresented women learn to "prefer" jobs that society among those eligible' to apply for positions deems appropriate for their sex. However, (Martin. 1980:47). A maximum-age restric- the persistence of occupational sex segre- tion for police recruits adversely affects g cannot be reduced to sex differences women's chances to become police officers in employees' preferences. in the case of for the same reasons it limits thi.ir partici- initial job assignments, employers' organi- pation in apprenticeship programs. Access to traditionally male jobs is also impaired by what Newman (1976:272) char- In the saine vein. iinlitar authorities have argued acterized as -sex bias in machinery design... that the number of women Who can be assimilated into Because of sex differences in physical size, the Armed Forces is limited 1, 'he cost of adapting some women may find it difficult to use ma- hying and working facilitie . .zse For example. the Department of Nay?. h.. . -d that the total chines designed for men. Similarly, Ma- cost lOr adapting all actv N.r! sow, would range from chinery used in traditionally female cm- $% to $132 million, depending 0,1 how mans women ph)vmeut such as fine work requiring finger needed to be avonumxIated tHinkin -111(1 Bach. 1977 .541

ION =EMMEN..NM 244 PATRICIA A. ROOS AND BARBARA F. RESKIN

zational practices contribute to the perpet- tional specialties were open to women; cur- uation of sex segregation, and these insti- rently, all but combat-related assignments tutional practices reflect, and are reinforced (about 42 percent of all enlisted positions in by, societal norms.9 the Armed Forces in 1977) are available to Based on their sex, workers are often as- women (p. 17)."Although the percentage signed to so-called light or heavy work. These of women working in male sex-typed mili- initial assignments are often due less to job tary specialties (e.g.,infantry, electronic content than to stereotypical notionsabout equipment repair) increased from 9 percent what kinds of work are compatible with fe- in 1972 to 40 percent in 1976 (p. 19), most male and male workers' alleged strengths women in the military stillwork as medical and weaknesses. For example, lifting one and dental specialists and administrative heavy object a day has justified restricting specialists and clerks. a job to males.Under Griggs (1971), Ala- The U.S. General Accounting Office (176) bama height and weight minima for prison identified three reasons for the persistence guards that excluded almost all women were of sex-segregated occupational assignments struck down (Dothard v. Rawlinson, 1977).1° in the military. First, women lack infor- Formal policies assigning women to light mation regarding the full range of job op- work would probably not survive legal chal- portunities. For example, over half of the lenge, but litigation always expensive and female Army recruits interviewed in 1974 slow is not a viable option for many women. reported that their recruiters had not in- Even women who are employed in such formed them of various assignments for which

14:2 nontraditional sectors as the military ti/pi- they were eligible (a comparable percentage rally work in what are traditionally female was not provided for men; p.10). Second, jobs outside the military. Considering how many women reportedlypreferred admin- this comes about is instructive. After the istrative or medical jobs, perhapsbecause 1970 decision to end the draft, the U.S. De- young women who choosemilitary careers partment of Defense began to increasethe may wish to avoid beingdoubly unusual in number of women in the Armed Forces. selecting specialties with few or no women. Within toter years, the proportion of women In addition, because military pay is deter- had more than doubled to approximately 5 mined by rank and time in service (and not percent of all Armed Forcespersonnel by occupation), women lack the financial in- (Binkin and Bach, 1977:14). Prior to 1972 centive to pursue jobs that in civilian life only 35 percent of the. military's occupa- are both higher payingand held predominantly by men. Third, and most important, women are automatically excluded from both combat- " While certain jobs have been historically labeled related occupational specialties and posi- male or female. this does not mean that jobs never hange their sex type. Clerical jobt'. for example, have tions set aside for men who return to the shifted from a male to a female sex type (Tilly and Scott, United States from male-only overseas and 197S 1571. as has public school teaching (Tyack and sea-duty jobs (estimated to be another 9 per- 1981) Carter and Carter (1981) argued, with Strohm. cent of all enlisted positions, yielding atotal respect to the professions at least, that thisshifting of sex ts pes derives front thedcskilling of occupations and that women etitertng jobs previously identified as mn's employment move into the most routinized sectors of these occupations " The definition of combat-related occupations has 1" The Supreme. Court, however, permitted the state recently been broadened to include 23 additional mil- to (fens winners jobs as prison guards in inale maximum itary occupational specialties. The S. Army currently se( oats prisons where their safety wasallegedly in bars women from a total of 61 (or 17 percent) of its job le opatils specialties +Nen. )(IrkrInW1,I9h2,0

251 INSTITUTIONAL. FACTORS CONTRIBUTING TO SEX SEGREGATION 245 of 51 percent of positions not open to women; Granovetter (1974) explored the role of per- Office of the Assistant Secretary Of Defense, sonal contacts in securing employment among 1977:Table 11).12 Each of the services has professional, technical, and managerial additional restrictions on the entry of women workers. He conpluded that the key to the that further limit women's job options. Ttms, process by which a worker with certain char- according to the U.S. General Accounting acteristics gets "matched" to a particular job Office report, while nearly all military oc- lies in large measure in the dynamic process cupational specialties were open to women, whereby job information flows through in- once all restrictive factors were taken into formal personal networks. Those outside account, only 26 percent of all enlisted po- networks (e.g., young labor force entrants, sitions were available to them.13 Not only recent immigrants) must rely on formal means do these restrictions inhibit job access at the of finding employment, such as intermedi- entry level,. but they also limit women's later ary agencies. With respect to sex segrega- mobility, since combat and other male-only tion, the questions of interest are whether jobs are the main route to upward mobility the sexes have equal access to personal net- in the military. works, whether they are equally likely to use them, and whether networks are equally effective for women .and men. White-Collar Workers In holding most professional and mana- Sex differences in access to information° gerial jobs, men 'enjoy personal and work and recruiting networks, entrance restric- associations that facilitate learning of other tions, and the allocation of men and women opportunities in those fields..Women, con- to sex-typical entry-level jobs also contri- centrated in clerical and service jobs, nor- bute to sex segregation, among white -coiar mally find themselves outside that network. workers. Instead they share information with same- sex friends and coworkers. Thus, men's and Access to Information and Recruitment women's positions in the occupational struc- NetworksOccupational sex segregation ture themselves contribute to continued sex persists in white-Collar jobs in part because segregation in occupational allocation. Sev- infM.mation networks are sex segregated. eral' network studies elucidate these sex differences., Langlois (1977:Table 1) found, for a small sample of government workers, that " This total estimate varies by military service. In men were slightly more likely than females 1977 the percentage of all positions unavailable to women to secure their initial employment via per- because of a combat restriction was 50 percent in the sonal contacts, while women more often ac- Army, 60 percent in the Navy, 7 percent in the Air quired their jobs through direct application. Force, and 73 percent in the Marine Corps (calculated from (Mike of the Assistant Secretary of Defense, Moreover, sex differences in the use of per- 1977.Tahle 11). sonal contacts were largest in the two oc- " This varies substantially by military service. In the cupational sate'-,1-.Atli the fewest women Army only h percent of all positions were available to (administrativt. ..:rs and laborers/serv- women, compared with 8, 76. and 5 percent for the ice workers). Navy, Air Force, and Marine Corps, respectively. The greater proportion of open jobs in the Air Force was Ensel and Lin (1982:8) found that men due to the small number (7 percent) of all positions were more likely than females to have used that are classified as combat related (all figures calcu- personal contacts in their initial employ- lated from Office of the Assistant Secretary of Defense, ment search, whereas women were more 1977 Table II). These figures are from 1977. As noted likely to have relied on formal job-search in note 11, the estimates for the Army will he reduced with the new restriction in the number of specialties methods. Interestingly, although each sex open to women relied predominantly on same-sex contacts

252 246 PATRICIA A. BOVS AND BARBARA F. RESKIN in searching for their jobs, men wereless managerial occupations, often involve sub- likely to use cross-sex contacts, which would jc.ctive evaluations of whether the applicant reduce their chances of learning about job will "fit in." Whilk,the problem of fitting in opportunities in female-dominated occupa- also contributes to *pawn's underrepresen- tions. (Of course, they may haverefrained tation in craft andother blue-collar employ- from consulting female informants because ment (Martin, 1980; Walshok, 1981a),hiring they did not want low-paying female jobs.) an "outsider" for a prestigious,illite-collar Finally, women who found high-status jobs occupation has greater organizational im- were more likely than men tohave used plications, given the higher levels of uncer- mere acquaintancesand indirect contacts tainty in these jobs as well as their greater ("weak ties") with males. This finding is con- rewards (Kanter, 1977). Because workers in sistent with the conventional wisdomthat these occupations have more control over women who want to progress intheir ca- their work, organizational elites must en- reers in male occupationsneed men's help sure that those hiredwill not disrupt the either as sponsors or at least as intermedi- ongoing work or challenge the nature of the aries (liennig and Jardim, 1977). organization. They do this by hiring those Caplette (1981:176) found that identical whose socialization and backgrounds resem- percentages of male and female employees ble those of members of the organization in a variety of publishing settingsused per- and by fostering unobtrusive controls that sonal contacts to break into the book pub- structure people's work attitudes and be- lishing field. However, men relied more on havior (Perrow, 1979:152; Smith and Gren- same-sex contacts thandid women. Because ier, 1982).14 Finally, firms with highly struc- men hold the more prestigiousjobs in pub- tured internal labor markets face more serious lishing, as in other industries, male contacts potential consequences of a hiring error. As are likely to be more.useful. Here too the'a result, exceptduring periods of labor women relied more onformal job search shortage, employers may prefer to err by methods: they were twice as likely as men failing to hire a qualified worker rather than to have used intermediary agenciesand half hiring an unqualified worker (Stevenson, s likely as men tohave "knocked on doors" 1977).15 Apparently, qualified applicants to secure their first job inpublishing. whose personal characteristics make them Taken in sum, these studies suggest that suspect are particularly likely to fall victim the use of personal contacts in securing ini- to such decision strategies. tial employment is more effective for men, Sexton (1977:26) argued that the use of which may explain men's greater propensity highly subjective evaluations unrelated to to use them. This differencereflects both job content tends to favor hiring men for the sex-stratified occupational structure and jobs they customarily hold (see also Epstein, employers' conservatism in hiring. 1975). There is also limited evidence that subjective evaluations based on interviews Entrance RestrictionsEmployers' entrance requirements operatesomewhat differently in white- than blue-collar occupa- Perrow (1979:152) labeled these unobtrusive con- tions because of the different nature ofthe trols "premise-setting." While premise controls are not jobs for which entry is sought. In blue-collar a direct example of entrance restrictions,by structuring jobs they often consist of formal rules (such people's behaviors and attitudes once they are hired they help ensure that any recruits with -deviant" back- as veterans'preference, for example), grounds (e.g., women or members of minority groups) whereas restrictions on white-collar entry are kept "in line'' within the organization operate more subtly. Hiring decisions, es- 15 The caution exercised by academic tenure com- pecially those for prestigious professional and mittees often illustrates this principle.

253

MEM INSTITUTIONAL FACTORS CONTRIBUTING TOSEX SEGREGATION 247 are detrimental to womenseeking sex-atyp- ditionally been assigned to jobs: Caplette's ical employment.16 (1981) study of the publishing industry and Finally, nepotism rules are also important Epstein's (1981) study of the male-domi- in restricting women's access towhite-collar nated legal profession. Although Caplette employment; the traditional exclusion of investigated an entire industry (which in- women from teaching atuniversities that cludes management, sales, editorial, and employ their husbands (although no longer clerical occupations) and Epstein an occu- the case in most universities) is ..n obvious pation, similar processes apparently oper- example. A recent example appeared in a ated to allocate men and women to sex-typ- news story in anational news magazine in ical employment in both settings. We should which the chairman of a major corporation note that, while the processesCaplette and was quoted asremarking that "we have a Epstein document still prevail, they are policy at this company that we don't hire nevertheless changing in the direction of the wives" (Newsweek, 1982a,b). In a letter to relaxation of sex typing. the editor, he corrected himself saying that Although publishing, which is one-half to the policy applied to spouses of corporate two-thirds female, is viewed as a women's officers, not wives in general. However, in business, it has always been controlled by reality women have probably borne most of men (Caplette, 1981:71).Having entered the brunt of such rules." book publishing in increasing numbers in recent years, women are stillconcentrated Organizational Practices Regarding Job in advertising and publicity, art design, pro- AssignmentWhile sex segregation in en- duction, and editing children's books and try-level occupations has been well docu- manuscripts (p. 75). Men dominate mar- mented, our understanding of the processes keting, sales, and management. Sex typing whereby differential allocation occurs is becomes even more apparent when these largely speculative. Two recent casestudies specialties are broken down further: women provide insights into how workers have tra- predominate as manuscript editors, editorial assistants, design directors, and secre- Sr, taries; men are editors, publishers,sales representatives, and marketingdirectors. Dubeck (1979) found that gender affectedhow in- Differences in background and credentials evaluated. terviewes for jobs in management were could not account for these sex differences Among applicants whom interviewersconsidered qualified men were significantly more to he rec- (p. 155). Women traditionallyentered pub- ommended for a job. Dubek suggested thatbeing lishing through secretarial jobs, whereas men -qualified- is determined differently for men and women. entered as sales representatives or editorial She found that the four most important factors affecting assistants. College textbook publishingil- the decision to hire males were (in order) anevaluation lusirates the general pattern of sex difler- of their qualifications, leadership experience. interest in the jot), and academic performance. In contrast,the enes in initial job assignment.According most important factors for women werejob interest, to Caplette (p. 205), those inthe field tacitly academic perfOrmanc, race, and qualifications.Thus, understand that sales, particularly the col- the criteria for evaluating female and maleapplicants lege traveler job,is the route to upward 97) concluded that *were ranked differently. DAM( (p. mobility: approximately two-thirds of the men the primacy of job interest ,whether the applicant was rated as interested inthe job) over qualifications began their careers in this position but only for female applicants reflects employer concernabout 9 percent of the women did so, all ofwhom female career orientation. were hired in asingle year. Women were '7 The Supreme Court declined to cimsklr this issue automatically excluded from these posi- Circuit in refusing to res iw a derision by the 7th U.S tions, on the assumption that the extensive Court of appeal, that fonnd 1,11,1)-OwensFords' oep- othro polic to he col, related traveling required would conflict with their

254 248 PATRICIA A. ROOS AND BARBARA F. RESKIN

present or future domestic responsibilities their occupational status. It is only for entry- (p. 208). level positions that workers must compete Epstein (1981) also found sex typing in in a relatively open labor market. Following lawyers' entry-level positions. First, a large labor market entry, job shifts both hori- number of women lawyers work for the gov- zontal and vertical are restricted largely ernment: in 1970, 37 percent did, compared by the job family in which a worker is lo- with 19 percent of the men. By 1977 the cated. An establishment's internal labor percentage of women lawyers working for market primarily consists of its promotion the government had decreased to 22 per- practices that define mobility opportunities cent, compared with 17 percent for men across and up job ladders: These may be (Epstein, 1981:112). Second, women law- firm, union, or civil service rules about con- yers are overrepresented in certain special- trolling movement across jobs. Several re- ties (including trusts and estates, domestic searchers (e.g., Stevenson, 1977; Osterman, relations, and tax law) and ill particular types 1979; Harlan and Weiss, 1981; O'Farrell and of jobs (research, writing of briefs, and pro- Harlan, this volume) have noted the role ,viding legal assistance). Not surprisingly, that internal labor markets play in limiting these specialties and positions rank lower women's mobility into and advancement in than those in which men are concentrated. traditionally male jobs. Internal labor mar- Third, women have generally been accepted kets do not rob personnel officers and su- in nonpublic positions but not in the more pervisors of all discretion in employment public corporate and litigation specialties. decisions. Both conventional and statistical Epstein's (p. 107) interviews revealed that sex discrimination still occur, although they firms were reluctant to hire women because are more likely to occur with respect to pro- they feared clients' reactions. Women were motion, where merit is often a legitimate also considered not tough enough for ne- consideration, than in layoff decisions, where gotiations and less able to participate in the seniority usually governs (Althauser and camaraderie between lawyer and client. In Kalleberg, 1981). However, discrimination sum, specialty assignments for lawyers re- may be less common than at the entry level. flected the norms governing the sex appro- First, to the extent that hiring at the entry priateness of jobs. This is due both to self- level is segregative, job groups will already selection of women into jobs in which they he relatively homogeneous with respect to are accepted and to employer biases that gender. Second, insofar as job ladders struc- derive from sex-role stereotypes. turally constrain promotion decisions, they should override biases of individuals. In this section we focus on how seniority systems INSTITUTIONAL FACTORS AFFECTING constrain mobility into sex-atypical jobs. We MOBILITY INTO SEX-TYPICAL AND SEX- then examine how the organization of two ATYPICAL JOBS particular promotion systems contributes to The recognition that workers' mobility sex segregation. opportunities are governed by internal Libor markets that exist within firms or occupa- Seniority Systems and Mobility tions (or both; Althauser and Kalleberg. 1981) Opportunities is critical for identifying institutional bay: Hers to women's mobility into and advance- Seniority systems consist of fOrmal rules ment within sex-atypical jobs. Internal labor within organizations in which a worker's length market theory identifies the job ladder as of employment must be given weight by de- the primary path by which workers improve cision makers in promotions, transfers, lay-

255 INSTITUTIONAL FACTORS CONTRIBUTING TO SEXSEGREGATION 24.9

offs, and benefits. Along with proceduresreg- hampers the mobility prospects of thoseout- ulating job posting and bumping and bidding side the seniority unit includes ambiguous rights, seniority systemsstructure workers' eligibility requirements for transfers. !nobility prospects. These formalizedproce- Adopting plantwide senioritysystems would dures ensure a stable work force andtrans- eliminate much of their disparateadverse ferability of skills from senior tonew employ- -^ women and minorities in one- ees, while providing workers with job security union firms, but, where employers have col- by protecting them against being replaced by lectivebargaining agreements withmore than younger workers (Stevenson, '1977). How- one union, opportunities for mobility canre- ever, they also limit minorities' and women's main limited (Steinberg and Cook, 1981:69). access to many jobs (Kelley, 1981). Recent Kelley's (1981) study ofan electric prod- court decisions (U.S. v. Teamsters et al., 1977; ucts manufacturing company (an industry that American Tobacco Companyv. Patterson, traditionally employsa substantial number 1982) have permitted seniority systemsas long of women) illustrates thatseniority reform as they were not adopted with discriminatory alone is not necessarily sufficientto enhance intent, despite their demonstratedsegrega- women's mobility prospects. All workersin tive effect. the plant were subject toone collective bar- Organizations vary in the units across which gaining agreement and senioritywas plant- a seniority system operates. They can gov- wide. In 1967 prohibitions againsttransfer- ern an entire plant, a department, or a job ring across seniority unitswere lifted and sequence. In job sequence systems the sen- plantwide bidding and posting procedures iority unit is a cluster of occupationallyre- established. Yet 9 years later, despite the lated jobs that representsa skill ladder even absence of exclusionary language in the col- lithe jobs are in different departments (e.g., lective bargaining agreement anda strong all painters in a firmmay constitute one sen- seniority clause that should have ensured iority unit). When paired with restrictions that seniority wouldgovern transfers and or penalties for transferring across units, more upgrading, few cross-system transfershad narrowly defined seniority systemsare occurred. Sixty percent of recentopenings problematic for women because theycon- that could have been filled by transferswere stitute a barrier to mobility for those outside filled by new hires, mostly whitemen. While the department or job family. to Workers some men helped integrate the plant by transferring across seniority unitsmay lose moving into some traditionally female fields, their seniority and are vulnerableto layoffs this did not enhance women'sopportunities. (Kelley, 1981:5). In keeping withnarrow Men continued to be concentrated inre- seniority units, rules may limit job bidding strictive high-wage job ladders with few to members of the unit and job openings workers. Thus, continued discriminationin may 1w posted only in the department or hiring undermined formalreorganization ef- work areas frequented by workers in the job forts, suggesting thata more active set of sequence. Hence, outsiders lack informa- reforms would be needed to reduceoccu- tion about such jobs. Another practice that pational segregation bysex (for documen- tation of this point, see O'Farrell and Har- lan, this volume). As the Conference Board (Shaeffer and Lynton, 1979)survey implied, 14 Of l'01111e, more narrowly definedseniority sys the right to transfer laterallymust he ac- tem% are preferable for incumbents of high-wage amts. companied by mechanisms to publicize job since the% insulate job% from the competition of other openings, encourage and preparewomen to workers pursue them, and provide support for those

256 250 PATRICIA A ROOS AND BARBARA F. RESKIN women who transfer intotraditionally male mobility track," while others had to shift jobs. 19 "tracks" to move up. A case study of promotion under NewYork State's civil service system by the New York The Structuring of Opportunity:Other State Commission on Management and Pro- Organizational Practices ductivity in the Public Sector (1977) showed O'Farrell's (1980) case study of a local union how career ladders perpetuate sex segre- in a large Northeasternindustrial plant il- gation. Women and minority workers, con- lustrates ways other than seniority systems centrated in the lowest-level jobs with short that mobility opportunities arestructured career ladders,had few advancement op- for blue-collar jobs. Employment inO'Far- portunities. Of the 43 different career, lad- rell's plant was highly segmented: of the two ders, women generally filled the low-floor/ company plants the unionlocal represented, low-ceiling ladders, while men predomi- the smaller plant was historicallyfemale nated in the higher, longer ladders (p. 30). dominated, while the more modern plant In a similar study in four other New York was maledominated. The two plants re- State agencies, Peterson-Hardt and Peri- mained sex segregated partly becausejobs . man(1979) found that in over 90 percent of were postedseparately within each. Fur- the career ladders, the incumbents were at thermore, jobs that opened up when aworker least 60 percent one sex.2° Moreover, in all changed jobs within the plant were not posted four agencies, female-dominated career lad- but instead were filled at the managers'dis- ders began at lower entry levels and offered cretion. Thus, workers were unawareof op- fewer opportunities for advancement: fewer portunities available in the otherplant. Even than 14 percent of the female ladders ranged if they learned of transfer opportunities, into high civil service grades,compared with workers lacked cross-plant biddingrights. 31 to 41 percent of the maleladders, de- Given the greater number of jobs inthe pending on the agency (p. 57). Not only larger mostly male plant, thedetrimental were women in NewYork State government effects of posting and bidding restrictions more likely than men tobe on truncated fell primarily en the mostly femaleworkers ladders in essentially dead-end jobs but at the smaller plant.Treating the two sex- their job ladders were also harder to climb segregated plants as separate organizational because the educational and experience re- entities ensured the persistenceof sex seg- quirements for promotion wereharder to regation. satisfy than in male-dominated career lad- Recent studies of state civil service em- ders (p. 78). Smith's (1979) findings replicate ployment focused Qn structural barriers in these results. In 13 job "chains" with at least white-collar jobs. Ilere researchers exam- three steps, high-opportunity chains (de- ined "career" or promotion ladders associ- fined as those in which at least 15 percent ated with particular entry-leveljobs. Work- of the jobs were at or above the entry man- ers in certainentry-level jobs were "on the

19 The Conference Hoard report also revealedthat barriers, expected opposition from coworkers or work- plantwide seniority had not been successful in musing ers' sense of achievement at haying escaped blue - collar women into higher-skilledblue-collar jobs in some origins. companies. where blue-collar womenworkers who had "4" Since women comprised about 45 percent of the accumulated enough seniority hid into clerical jobswhere civil service work force in New York in 1977, the results tbev preferred to remain (Shaeffer and I.y nton,1979.70). would has + been more useful had the researchers set Mobility may also 1135e fie -en illnited by moreinformal a higher k Alm. todefine sexdoromated jobs

257 INSTITUTIONAL FACTORS CONTRIBUTING TO SEX SEGREGATION 251 agtrial level) were filled predominantly by promote women (Kanter, 1977; Shaeffer and men, while low,opportunity job chains were Lynton, 1979). held mainly by women. Informal networks in the workplace also Research currently under way at the Cen- differentially affect the sexes' !nobility pros- ter for Women in Government (Ratner, 1981; pects. Epstein (1970a,b, 1975,1976) has Haignere et al., 1981) extends that of Peter- identified several informal mechanisms that son-Hardt and Perlman (1979) by investigat- restrict women's mobility prospects: women ing the differential impact of personnel prac- are likely to be less connected to commu- tices on women's and minorities' prospects for nication networks, less involved in sponsor- promotion to management positions. In New protege relationships, and less likely to have York State government, promotion involves access to the clublike relationships charac- several steps: setting criteria fir eligibility to teristic of many of the professions. (While compete for the promotion, a competitive ex- Epstein concentrated on the professions, the amination, and selecting the successful 4- logic of her argument holds for blue-collar didate from the top three who pass the exam occupations as well.) Kaufman (1978) found (Ratner, 1981:3). Eligibility is limited to em- t'iat female professors (especially those who ployees in -feeder" jobs specified in the job were not married) had fewer males in their posting. Although women made up 53 per- collegial networks than did their male col- cent of all state employees in 1979, they con- leagues. Because men dominate the upper stituted only 12 percent of those in designated levels of academe, women had less access feeder jobs for management positions. Fe- to those in authority positions. Contrary to male and male applicants for the New York these results, Strober (1982:32) found no management jobs were equally likely to pass significant sex differences in access to men- the exam, and, when women got into the three- torships for a sample of Stanford MBAs four person pool, their chances of being chosen years after graduation. Strober's study and were good. However, the consequence of bas- that of Harlan and Weiss (1981) suggest that ing eligibility on a feeder system composed more investigation is needed to provide a of jobs held disproportionately by men was definitive answer on the role of mentorship that over 70 percent of the three-person pools in female and male mobility. were all male. This system ensured that men Some companies have restructured their would hold almost all managerial jobs which internal labor markets to increase women's was the case. employment opportunities,partly in re- In less bureaucratized promotion sys- sponse to federal enforcement efforts (Shaef- tems. recommendations play a larger role fer and Lynton, 1979:34; O')H arrell and Har- than the tOrmal procedures we have &- lan,this volume). Changes include devel- scribed above. This too can contribute to oping strategies by employers, support groups, sex segregation because female clerical jobs and managers to make job openings known are more likely than male jobs to provide to women and minorities; analyzing jobs to direct services to one's immediate super- retain only those qualifications that are truly visor (a reflection of the institutionalization necessary; enacting safeguards against man- of women in helper or assistant roles; Ep- agers' biases for fair evaluation of candidates; stein. 1976:191), Because supervisors may and monitoring the promotion process. In be reluctant to recommend very effective recognition of the Fact that existing career lad- assistants for promotion, relying On super- ders curtail women's chances for !nobility, some visors' recommendations of candidates for firms have analyzed and revised job families promotion front clerical to managerial jobs to create new career lines for women into IllaV undermine organizational effortsto managerial positions.

258 2.52 PATRICIA A. RUGS AND BARBARA F. RE.SK1N

affect retention. Meyer and Lee (1978) in- INSTITUTIONAL MECHANISMS terviewed 164 women, and their supervi- ASSOCIATED WITH RETAINING sors, peers, andsubordinates, regarding the WORKERS IN SEX- ATYPICAL JOBS effectiveness of special programs devised to This section focuses on institutionalized move women intonontraditional jobs. The features of the workplace that affect the re- interviews suggested that informing female tention of workers employed insex-atypical applicants about the characteristics of jobs jobs. Since most workers are in sex-typical for which they were applying reduced turn- jobs, institutionalized factors that facilitate over. Nowhere is this moreevident than in the retention of workers in these jobs are the comparison of the typical experiences of probably more important in maintaining sex professional/managerial and blue-collar segregation; kor example, the availabilityof workers. Women selected for professional part-time clerical work enables women to and managerial jobs usually underwent ex- combine paid employment with child-rear- tensive screening, whereas pi ogramadmin- ing, thus contributing to the highly segre- istrators often had to persuadeblue-collar gative character of clerical employment.The women to apply fortraditionally male jobs. compatibility of short working days and free The dropout rate for women recruited for summers withchild-rearing attracts moth- blue-collar jobs in this manner was very high, ers to publicschool teaching. That these especially in jobs that required heavy phys- same features areaccompanied by smaller ical labor or had undesirable working con- salaries is likely to discourage the retention ditions.(p. 17). Shaeffer and Lynton (1979) of men. Many mechanisms that encourage also found that the more information firms workers to remain in sex-typed jobs have provided women entering traditionally male evolved hand in hand with the development blue-collar jobs, the higher the retention of these jobs and may have been influenced rates. Particularly useful wereclear descrip- by workers' sex. Unfortunately, space con- tions of job demands, slides, tours, oppor- straints preclude examining themhere. With tunities to talk with workers(especially fe- respect to sex-atypicaljobs, we note two male workers), and a chance to try out various recent studies that show aconsiderable aspects of the job. See O'Farrelland Harlan amount of mobility byworkers of both sexes (in this volume), fbr additional recruitment into and out of sex-atypical jobs:Rosenfeld practices found to be successful in retaining (this volume), and Jacobs (lfM3). Obviously, women in nontraditionaljobs. segregation results from both entrybarriers to sex atypical jobs andmechanisms that Training discourage workers who hold such jobs from remaining in them. We haveconsidered some Training both prior to beginning a non- of these factors in describing themecha- traditional job and on the job may be the nisms associated with access tosex-atypical most important determinantof retention. jobs and will return only to those that affect The study of women in public utilities cited retention differently. above as well as two surveys of women in construction stressed the value ofspecial pretraining for women entering traditionally male blue-collar jobs (Meyer and Lee, Recruitment Practices and InfOmation 1978:18; U.S. Department of Labor, Em- About jobs ployment Standards Administration, 1981; recent study of women innontraditional Westley. 1982). By exposing women to the Jobs in10 public utility companies sug- tools and techniques with which most men gestd that recruitment methodsstrongly become familiar while young, pretraming

259 INSTITUTIONAL FACTORS CONTRIBCITING TO SEX-SEGREGATION 253 puts women on a more equal fboting with quired shift work, overtime, acid extensive new male recruits (see O'Farrell and Harlan travel may discourage women from staying in this volume for additional evidence on in certain jobs. this point). Walshok's (1981a) interviews with women The Role of UnionsWhile unions can in nontraditional blue-collar employment in- facilitate women's entry into nontraditional dicated that unstructured on-the-job training employment, they can negatively affect in which apprentices depend on a single jour- women's retention (O'Farrell, 1980; New- ,neyman is problematic for ,women, since it man and Wilson, 1981; Steinberg and Cdok, makes them vulnerable to their trainers' biases. 1981). Lack of female leadership may limit Some formal on-the-job training enhances the the effectiveness of unions. Programs that chance that women will obtain necessary skills. would enhance women's retention in jobs But Walshok also stressed the importance of (such as child care) are more expensive than hands-on experience during formal training. the bread-and-butter issues unions have tra- Women whose preemployment training in ditionally addressed (Steinberg and Cook, male occupations included actual work ex- 1981:63). Women as a group are only one perience were more likely to find jobs, to constituency of unions and, given their un- learn how to perform them well, and to suc- derrepr6entation in leadership positions, ceed in them. not a particularly powerful or vocal one. Without female leadership to press for such programs, they are often bargained away in Organizational Mechanisms That negotiated agreements. Influence 1,Vomen's Retention in Sex- Atypical Jobs Lack of Standards for EntryLack of Certain organizational arrangements fa- specified standards for job performance has cilitate women's success in nontraditional limited women's ability to perform on the jobs. Of particular importance are commit- job hence, their retention in blue-col- ment by top management to improve wom- lar jobs.obs. The U.S. General &counting Of- en's employment opportunities (Shaeffer and fice (1976) found that some women in tra- Lynton, 1979:21) and a full-time equal em- ditionally male military jobs had been as- ployment opportunity staff (Meyer and Lee. signed to jobs for which they were not phys- 1978:4). O'Farrell (1980:124) identified lack ically suited, and their lack of strength con- of organizational support as an important tributed to their inability to complete re- barrier to women's employment in nontra- quired tasks. The report recommended that ditional work: missing at the industrial plant time military services develop physical and she studied were any special recruitment operational standards required for job per- programs to 'inform women about the nature formance as well as memi.sures of men's and and advantages of nontraditional jobs, tran- women's ability to satisfy established stand- sition programs to ease the shift into non- ards. Of course, the danger of instituting traditional employment, and support Oil the such standards is that they may be used to job f'or wonmen experiencing difficulty. keep all women out of traditionally male Other organizational practices, which employment rather than to ensure that only cannot be thoroughly examined here be- those men and women physically suited to cause of space constraints, foster orhinder the job will be hired. women's retention in sex-atypical jobs. Pregnancy leave, opportunities fin flextime SeniorityWe considered seniority sys- or part-time work, and child care can make tems in some detail earlier in our discussion jobs more accessible to women, whereas re- of mobility. Here we simply stress that sen-

2E0 254 PATRICIA A. ROOS AND BARBARA F. RESK/N ionty systems organized for units smaller into play when work is organized such that than an entire plant have predictable neg- women's ability to do tlivir jobs depends on ative consequences for women's retention. male cooperation. By assigning women jobs Department or job sequence systems ren- they can do alone or by providing female der women who transfer to male jobs in dif- work partners, organizations can retain ferent seniority units vulnerable to layoffs women working ill nontraditional jobs in an economic downturn. 21 Steinberg and (Shaeffer and Lynton, 1979; \Valshok, Cook (1981:68i noted that seniority systems 19)1 1a;.' also inhibit women's retention by reducing Some intrinsic aspects of the work itself, the likelihood that work-sharing systems the tools, and the typical division of labor might be implemented as an alternative to also influence the sueess and retention of layoffs. When narrowly structured seniority women working in nontraditional blue-col- systems are )sed to determine spilt assign- lar jobs. With training, women can learn ments and overtime allocation. as in the steel how to use unfamiliar tools, but as we noted industr,. womenwith low seniority who are above. equipment and tool design may oc- assigned to night shifts or required to work casionally interfere with successful job per- 0\ vomit. may have to (1111tif they cannot firrinane. Pioneers in nontraditional jobs ammo. adequate child care iValshok, 198 have fOund it difficult to obtain proper work seittoi Ih s\ Stems that guaranteebumping clothes (Business IV' eck, 1978;90). At AT&T, rights in the case of layoffs including, the for example, the higher accident rate of right to hump back into sex traditional jobs women in outdoor jobs spurred the com- the. left may facilitate womn's willing- pany to introduce lighter-weight and more ness to enter and their retention in sex-atyp- mobile equipment. Bales and White (1981) If al(1111)11)1 Not only art' their tolis of the Coal Employment Project found that Hum, secure in the event of an economic over one-half of their sample of women min- do.. nturn. but %%omen gain expertise III more ers feared for their safety becauseof im- ink in the plantthereb. enhancing their properl fit ing protective equipment. Boots tutu! e Iol, prospects and hard hitt', Were tolarge. and oversized glo. cs got stuck in moving machinery. studies of women in forestry, mining. ()1://11:://t/t)/1 (1I %V (111(1 1 hpl(P 1 construction. and other outdoor jobs rn-

\\ 11(11 11AN, \ (irk 111 14111)111 arson. 19.id, Bales and White, 1S).51, .11111 111,111' 111\1t111.11 s art' not suppoiti%e. the Department of Labor. Employment Stand- aninunt of inteidependence ilecssar\ to at ards Administration. 1()S1. \\*Ashok. PPIlio omplish ,1task is unpoitant of the has(' pointed out the s et.. real problem that 1111111'11 alslinkI 91. laIntel cd 111(11 rrNS to adequate tiallltar\finiittICS e- .It thatI halt' ti ht'! S11111.4,1111111411(111 to NriltN 'file !II,Sent eofsm h facilitiestin %k 1,1k '61011 I111'11) 1)11111111t1p1 rs1111 1,1\1.\ 11,1111 111'11'11 then,Ii ilit then wit( inh dune, ulturiatels ohs, onkiv.erl then, III 11,.. 11.1,uYllr,l lll.lt nutt''tttlr, 11111f -If \lrfitifit.r..b theI 'NI, 111 ('nImale ti wl-sr(S :If 41111MUM 111).1tP,111tutu 411101 Itt)t 11.1111r1 itt !1,1,/,,/t 1.1)),%dill/ OW 1.91)11i) )1111)111 flfpt this ../i /,%1.(11(.1 1W/1111111.1M e 11;1 IV I, sljlIt rIlltDI I

O. II Iltt 111 1,,.115., it,!Mtttst Itt I.11 %)111111 III 1,11, to .111 i lite 11 III...lf% t.) i..1:;plett .1

:t .., 0/ .1,i a 1,1.1.1.11 ! II'' 111.11. .111 Iltf kit 1,1.,1. ItIftt %, t 'II), f.ttfItti, ',t1.111f

, 1. f :I II 111.,

1) r 1,',) A. msTi'TioNAL FACTORS CONTRIBUTING TO SEX SEGREGATION 255 women exposes them to health risks and child care arrangements limit women's em- sexual harassment (see Enarson, 1980; and ployment prospects. Shortlidge (as cited in White et al.,1981). While costly (Bethle- U.S. Commission on Civil Rights, 1981:10) hem Steel spent over $10 million to outfit estimated that approximately 20 percent of its mills. shipyards, and mines with wom- currently nonemployed women do not work en's lockers, restoorns, and showers over a because of unsatisfactory child care arrange- 5-year period; Business Week, 1978:90), these ments. Presser and Baldwin (1980) reported facilities are essential for making tradition- a similar figure: 17 percent of nonemployed ally male blue-collar jobs accessible to many women would look for work if satisfactory child care were available and 16 percent of currently employed women would work more hours given suitable and reasonably priced INsTttioNAL FAcolis OUTSIDE THE child care. 1,1' otiKPLACE It has been suggested that inadequate child The focus of this paper is 011 institution- care may contribute to higher accident rates alized fame us that exist within the work- among women assembly line workers. Cuth- place. However, many factors outside the bertson (as cited in U.S. Commission on workplace ailed women's occupational and Civil Rights, 1981:12) suggested that stre.is advancement prospects indirectly by affect- is a significant factor in industrial accidents, ing their labor force participation. and worry about inadequate child care may The lack of adequate child care affects contribute to stress. However, more re- women's access to and retention in jobs. The search is needed before we can accept this rcognition that inadequate child care con. contention. %tames a harrier to women's employment is Federal laws contribute to sex segregation recent. at least among lawmakers. Not until by affecting women's labor force participa- 1978 did Congress. in an amendment to Ti- tion. Income tax laws discourage secondary tle \11 of the 1964 Bights Act (U.S. family earners usttil,. women from Commission on Civil nights. 1981:5). spe- entering the labor force, since additional ifically recognize that %voniens childbeat earnings are taxed at a progressively steeper mg role constrained educational and rate t Blumberg, 1979; Cordon, 1979a).2-' So- employment opportunities. Inadequate child cial security laws have a similar effect. since care idle( t1 %V0111ellS employment options in dual-earner families get a lower return to seeral wa% I) by limiting their entry ihtei their social security investment than do sin- the labor f nrc, t21 by restricting their par- gle-earner couples, and married women arc ticipation in federally sponsored education often entitled to higher benefits as their hus- and training programs, 131 by reducing the bands dependents than as retired workers amount of time tIucan devote to their jobs in their own right (Blumberg, 1979; (;or- and (meow-aging their retention in pat-time don. 1979b). The federal polities underlying lobs. A IP, restricting their ability to work these. laws discourage women's continuous crtin butts. b preventing them from labor force attachment, which in turn has a being to take advantage of training for strum; negative impact on their access to more demanding lobs for whichtho. are high-wage employment and mobility op- qualified. and 16, by constraining then par portunities. ti( 'nation in lobs that require tr:o.lingIt probably also contributes tee womens lesser - .- Illunion activities and ulti partiewation fir(ttit 111( ta1.1\kIt.ttitr, \ I.. it n..ol union leadership P)52 tti ,ilm.hinatfll .1 Li)t.tS Hereto itnr, tut `,11, es slea\ r melee ated him madc(illatt tltoIC(110 t tilt ( hum? !MO' 11C11.111\

262 256 PATRICIA A. ROOS AND BARBARA F. RESKIN

Finally, economic factors also reduce segregation. As the paper by O'Farrell and women's opportunity to go into business fOr Harlan (in this volume) shows, many of the themselves. One harrier to entrepreneur- mechanisms that we identity' as barriers have ship among women is difficulty in obtaining been manipulated by organizations attempt- financing.Financial institutions prefer to ing to improve women's employment op- support larger and less risky enterprises, and portunities. What remains to be explored women are prone to start small businesses more fully is what functions these institu- in low- profit, labor-intensive industries (U.S. tional mechanisms serve within organiza- Department of (:ommerce, 1978:5). tions and labor markets and for whom. Edwards (1975, as cited in Tolbert, 1982) noted that when bureaucratic control CONCLUSION emerged in labol..markets at the beginning This paper reviews workplace mecha- of the century, large firms could no longer nisms that act as harriers to women's em- personally manage employees. In response, ployment in traditiowilly male jobs. These ihey instituted administrative regulations mechanisms, institutionalized in the labor regarding qualifications for employment, market and in firms' personnel practices, are wages, criteria for promotion, and so forth. less well understood and less studied than In unionized industries, many of these pro- those factors more often cited as explana- cedures became union as well as company tions pier occupational sex segregation: char- policy. At that time, the exclusionary im- acteristics and choices of the labor supply, plications for women of bureaucratic pro- on the one hand, and genderdiscrimination cedures were not viewed as problematic in b employers, on the other. Using internal light of prevailing social valuek. By the time 'labor market theory as our theoretical law and social opinions challenged their dis- framework, we argue that such workplace criminatory effects, resistance to modifying mechanisms act as barriers to women's em- long-established personnel procedures would ployment prospects at four points in the job be expected from those with a stake in their allocation process: preemployment training. administration. For example, seniority sys- aceess and assignment to jobs,mobility, and tenis and other job ladders that structure retention. promotion opportunities are economically Investigating barriers to women's job op- advantageous to both employers and work- portunities that are institutionalized in the ers well positioned in firms'internal labor labor market and the organization of work markets. Fuller analysis of other functions is valuable in identifying useful areas for fu- these segregative mechanisms fulfill is nec- ture inquiry and essential fUr developing in- essary to devise nondiscriminatoryalterna- tervention strategies. Empirical studies that tives. document sex differences in access to employment information. the allocation of the NCKNOVI,VDC,NIENT,, sexes to "sex-appropriate- employment,the sews' differential location in job Inters, This paper was supported in part by a harriers to :yomen's access to entry-level po- UAI) Faculty Research Fellowship and Grant- sitions on high-prestige Job ladders, and so (#431-75:31-.V. 1982 from the State forth will help us better undei stand how t'iliversity of :tiew 'York at Stony Brook. NVe %void(' like to thank the following people for interna.1 1.aoor1 nuirkets operate andhow they might be modified to work to women's ad. their helpful comments on earlier versions: vantage. Our analysis also suggests the kind, Lee Clarke. ( ;arol%n Ellis. Cynthia Epstein. of orgainiational changes that might reduce !Orr\ I l'int2;rutd, Arne Kalk-berg. Mar\ ellen FACTORSCONTRIBUTING TO SEX SEGREGATION 257

Kelley, Edward Royce, Jo Shuchat, Robert retention of women in traditionally male skilled Smith. Glenn Yago, and members of the blue-collar trades in Wisconsin." Pp. 106.31 in Vernon M. Briggs, Jr., and Felician Rittman Sociology Workshop at S UNY, Stony Brook. (eds.), Apprenticeship Research: Emerging Findings and Future Trends. Ithaca,' N.Y.: New York State School of Industrial Relations. REFERENCES .Business Week 1978 'The hardships that blue-collar women face." Allison. Elizabeth K. Vol. 87, August 14, 88-90. 1976 "Sex-linked earnings differentials in the beauty Caplette, Michele K. industry." Journal of human Resources 11 I981 Women in Book Publishing: A Study of Ca- (Summer):383-90. reers and Organizations. Ph. D. dissertation, Althauser. Robert P., and Arne L. Kallberg State University of New York, Stony Brook. 1981 'Firms, occupatiems, and the structure of labor Carter, Michael J., and Susan Boslego Carter markets: a conceptual analysis." Pp. 119-49 in 1981 "Women's recent progress in the professions Ivar Berg (ed.). Sociological Perspectives on or, women get a ticket to ride after the gravy Labor Mai isets. New York: Academic Press. train has left the station." Feminist Studies Bales, Jeanne. and Connie White 7(E,a11):477-504. 1981 "Persinial protective equipment in the mining Omer.Rose L., and Gerald Rokoff industry. a survey of women miners." Unpub- 1971 "Women in the' occupational world: social dis- lished paper, Coal Employment Project, Oak ruption and conflict." Social Problems Ridge, Tenn. 18(Spring):535 -54. Bell Carob Dubck, Paula J. 1979 "Implementing safety and health regulations 1979 "Sexism in recruiting management personnel for worsen in the workplace." Feminist Studies for a manufacturing firm." Pp. 88-99 in Rudolf() r(Summer):286-301. Alvarez, Kenneth C. Lutterman, and Associ- Beni. Sandra L., and !)aryl J. Rem ates (eds.), Discrimination in Organizations, 1973 -Does sex-biased job advertising 'aid and abet' San Francisco: Jossey-Bass. sex discrimination." Journal of Applied Social Enarson, Elaine Pss cholog,y 3(11:6-18. 1980 "Sexual relations of production: women in the Mare U.S. Forest Service." Paper presented at the 1982 Employment. training, and economic dvel- annual meetings of the Pacific Sociological As- opment." Pp. 247.69 in John L. Palmer and sociation. Isabel V. Sawhill (eds.), The Reagan Experi- Ensl, Walter M., and Nan Lin mnt. An Examination of Economic and Social 1982 "Social resources and strength of ties: gentler hilicis Under the Reagan Administration. differences in status attainmeut.- Paper pre- ashington, 1).0 : Urban Institute Press. sented at the annual meetings of the Ainerican Bern man. Sue E.. and Winston K. Chow Sociological Association, Toronto, August 1981. 1981 ( :ETA. Is It Equitable fur Women? Rand Note, Epstein, Cynthia F. N1683.N(:EP Santa Monica: The Rand Corp. 1970a Woman's Place: Options and Limits in Profes- Binkin. Mai tin, and Shirley J. Bach sional Careers. Berkeley: University of Cali- 1977 Women and the MilitarsVashington, D.C.. fornia Press. The Brookings Institution. 1970b"Encountering the male establishment: sex- lilaloek. Hobert M. status hmits on women's careers in the prol- 1412 'Occupational discrimination some themeti sions.American jounial of Sociology 75(2).965- calpropositions. Sexual Problems 82. ter) 240.47 1975 "Insfitutiolial barriers: what keeps women out filumbig. Grace C of the executive suit'" Pi). 7.21 in Francine 1979 -Federal income tax and social serums law K Gordon and Myra 11. Strobr (eds.). Bring. III Ann Foote Calm id Women in the t' S ing Women Into Management. New York. Libor Fore e Ness York Praeger. McGraw -If ill. Ili iggs. Norma 1976 'Set role stereots ping. occupations and social I 971 Women in Applenticeship h. Not Man exchange." Womn's Studies 3.185-94 power fiesearh Monovjapli. \o. 3:3Wash I 98 I WomenIII IAlw.New York. Basic Books. ington. 1) t:lS Cosernmnt Printing Of Folk. II ugh be 1968 The problem of south unemploy went." Pp 1979 kpinnticeship Pp 225 33 in Ann Foote 76 107 in Princeton Manpower Symposium, calm .ed Women in the t'S(AbutForce The Trans.olon From Scheml to Work Princ- yo d,Pr.(tgt.r ton. N. JIndustrial Relations Section. Pr ince )sil«imoig !.caucus to siiteessfiil riar%ancf ton Unix ersits

264 2.58 PATRICIA A. ROOS AND BARBARA F. RESKIN

Freidsun, Eliot Harlan, Anne, and Carol Weiss 1970 Profession of Medicine. New York: Harper & 1981 Moving Up: Women in Managerial Careers. Row. Final report. Working Paper No. 86, Center for Research on Women, Wellesley College, Gates, Margaret Wellesley, Mass. 1976 "Occupational segregation and the law." Signs 1:3(Part 2, Spring):335-41. !felling, M., and A. Jardim Golladay, Mary A., and Rolf M. Wulfsberg 1977 The Managerial Woman. New York: Anchor Press. 1981 The Condition of Vocational Education. Wash- ington, D.C.: U.S. Government Printing Of- Hochschild, Arlie R fice. 1975 "Inside the clockwork of male careers." Pp. 47- 80 in Florence Howe (ed.), Women and the Gordon, Nancy M. Power to Change. New York: McGraw-Hill. 1979a"Institutional responses: the federal income tax system." Pp. 201-21 in Ralph E. Smith (ed.), Jacobs, Jerry The Subtle Revolution: Women at Work. 1983 The Sex Segregation of Occupations and Wom- Washington, D.C.: The Urban Institute Press. en's Career Patterns. Ph.D. dissertation, Har- vard University, 1979b"Institutioni,1 responses: the social security system." Pp. 223-55 in Ralph E. Smith (ed.), Kane, Roslyn D., and Jill Miller The Subtle Revolution: Women at Work. 1981 "Women and apprenticeship: a study of pro- ay. Washington. D.C.: The Urban Institute Press. grams designed to facilitate women's participation in the skilled trades." Pp. 83-105 in Ver- Granovtter, Mark S. ti non M. Briggs, Jr., and Felician Foltman (eds.). 1974 Getting a Job: A Study of Contacts and Ca- reers. Cambridge, Mass.: Harvard University Apprenticeship Research: Emerging Findings and Future Trends. Ithaca, N.Y.: New York Press. State School of Industrial Relations. 1981 -Toward a sociological theory of income differences." Pp. 11-47 in Ivar Berg (ed.), Soci- Kanter, Rosabeth M. ological Perspectives on Labor Markets. New 1977 Men and Women of the Corporation. New York: York: Academic Press. Harper & Row. (;rinker, William J., Donald D. Cooke, and Arthur W. Kaufman, Debra R. 1978 "Associational ties in academe: sonic male and Kirsch. female differences." Sex Roles 4:9-21. 1970 Climbing the job Ladder: A Study of Em- ployee Advancement in Eleven Industries. New Kelley, Maryellen R. York: E. F. Shelley and Co. 1981 "Legal and empirical issues in determining when liagstrom. W. 0. seniority systems discriminate." Unpublished The Scientific Community. New York: Basic papist-. Alfred P. Sloan School of Management, 1965 Massachusetts Institute of Technology. Books. Haignere, Lois V., Cynthia IL Chertos, and Ronnie J. Langluis, Simon "Les rseaux personnels et la diffusion des in- Steinberg 1977 "Managerial promotions in the public sector: fiirmations stir les emplois." Recherches So- 1981 ciographiques 2:213-45. the impact of eligibility requirements on women and minorities.Albany: Center for ' omen in Martin, Susan E. Government, State University of New York. 198(1 Breaking and Entering. Policewomen on Pa trol. Berkeley: University of California Press. Ilarkess, Shirley 1980 -Hiring women and blacks in entry-level man- Meyer, Herbert II., and Mary Dean Lee ufacturing jobs in a simthern city. particularism 1978 Vomen in Traditionally Male Jobs: The Ex- and affirmative actionPaper presented at the pe'rienc'es of Ten Public Utility Companies. annual meetings of the Society for the Study Washington, D.C.: U.S. Government Printing Social Problems, New York. Office. Harlan. Sharon L. New York State Commission on Management and Pro 1980 -Sex differences in access to federal employ- duch vity irr the Public Sector ment and training resources under (:ETA an 1977 Job Promotion Under New York State's 0%11 overviewWorking Paper No. 58, Center for Service System: A Case Study of the Office of Research on NVOIOVIL Vellele.College, General Services. New York: Commission on Mass Management and Productivity in the Public 1981 -'The social context ut employment choice. Sector viomen 111 C ETA New England Sociologist New York Times Iii32.45 I 982a -Army to enlist more women but in limited job categories.- August 29. Hai lan, Sharon I.. and Brigid O'Farrell lte,2. -After the pioneers. prospects for women 111 19821"Labor's enter. labor.- September 5. nonhaditional blue collar jobs.- Vork and Or 1983 Desex schooling for job, arit 9 upahons 9i August) :16386

235 INSTITUTIONAL FACTORS CONTRIBUTING TO SEX SEGREGATION 259

Newman, Winn Reskin, Barbara F. 1976 "Combatting occupational segregation; pres- 1978 "Sex differentiation in the social organization entation III." Pp. 265-72 in Martha Blaxall and of science." Sociological Inquiry 48(3/4):6-36. Barbara Reagan (eds.), Women and the Work- Reskin, Barbara F., and Heidi I. Hartmann (eds.) place: The Implications of Occupational Seg- 1984 Women's Work, Men's Work: Sex Segregation regation. Chicago: University of Chicago Press, on the job. Washington, D.C.: National Acad- Newman, Winn, and Carole W. Wilson emy Press. 1981 'The union role in affirmative action." Labor Sihafran, Lynn Hecht Law Journal (June):323 -42. 1981 Removing Financial Support From Private Newsweek Clubs That Discriminate Against Women. New 1982a"Did Bill and Mary blunder?" October 11:94. York: Women and Foundations Corporate Phi- 19826"Hiring spouses." November 29:6. lanthropy. O'Farrell, Brigid Schiller, Bradley R. 1980 "Women and nontraditional blue collar jobs: a 1980 The Economics oaoverty and Discrimination, case study of local I." Report prepared for the 3d ed. Englewood Cliffs, N.J.: Prentice-Hall. Employment and Training Administration, U.S. Scott, W. Richard Department of Labor. 1981 Organizations: Rational, Natural, and Open 1982 -Women and nontraditional blue collar jobs in Systems. Englewood Cliffs, N.J.: Prentice-Hall. the 41980s: an overview." Pp. 135-65 in Phyllis Sexton, Patricia C. A. Wallace (ed.), Women in the Workplace. 1977 Women and Work. Washington, D.C.: U.S. Boston, Mass.: Auburn. Government Printing Office. Office of the Assistant Secretary of Defense Shaeffer, Ruth Gilbert, and Edith F. Lynton 1977 Use of Women in the Military. Washington, 1979 Corporate Experiences in Improving Women's D.C.: Office of the Assistant Secretary of De- Job Opportunities. Report No. 755. New York: fns. The Conference Board. O'Neill, June, and Rachel Braun Simmons, Adele, Ann Freedman, Margaret Dunkle, 1981 -Women and the labor market: a survey of and Francine Blau issues and policies in the United States." 1975 Exploitation From Nine to Five. Report of the Washington, D.C.: The Urban Institute Press. Twentieth Century Fund Task Force on Women Oppenheimer, Valerie K. and Employment. Lexington, Mass.: Lexing- 1968 ''The sex-labeling of jobs." Industrial Relations ton Books. 7( M ay ):219-34. Smith, Catherine Begnohe Osterman, Paul 1979 "Influence of internal opportunity structure and 1979 -Sex discrimination in professional employ- sex of worker on turnover patterns. Admin- ment: a case qudy." Industrial and Labor Re- istrative Science Quarterly 24(September):362- lations Review 32 (July):451-64. 81. Perrow, Charles Smith, Howard L.. a;id Mary G render 1979 Complex Organizations. A Critical Essay, 2d 1982 "Sources of organizational power for %%omen: NI Glenview, Scott, Foresman and Co. overcoming structural obstacles." Sex Roles 8:733-46. Peterson Hardt, Sandra. and Nancy D. Perlman 1979 "Sex segregated career ladders in New York Steinberg, Rimini, and Alice Cook state government: a structural analysis of in- 1981 "Women. unions, and equal empkyment op- quality in mplo uncut. Albany: Center lilt portimity." Working Paper 3, Center for %Vomn in Government, State University of Women in Government, State University of New York, Albany. New York. Albany. Phelps, Edmund S Steynson, Mary II. 1972 "The statistical theory of raISIII Arid WX1s,111 1977 "Internal labor markets and the employment A 111CriCall Er0110MIC Ex'1, ,ew 62 (Septm- of women in complex organizatiims." Center 1)0-1.659-66 for Research on Women, NN'ellesly College, NVllesly, Mass. Presser, 11arrit B . and Wends Baldwin 1980 -Child care as a constraint on enploy mem. Strobr, Myra pr% alnce. correlates, mid Ih.aring on the work 1982 -The MBA: same passport to success for women and fertility nexus." American Journal of So- and men?" Pp. 2544 in Phyllis A. Wallace emit)* 85' larch) 1202-13. (ed.), Woinen in the Workplat. Boston, Mass.. Auburn Ratner, Bonnie Steinberg 1981 "Hai rirs to ad% ancetnent- promotion of women Theodore. Athena and unnontis into managerial positions in New 1971 The Professional NVonian Cambridge. Mass . York state gin:mm11C Albain,Center for Shnkman. VOIllt'llIII Goseminent. State Unwrsityof Louise A., and Juan W Sc nit New York, Albany 1978 Women. %Vork. and Family New lurk Holt. Rinehart & Norton.

268 260 PATRICIA A. ROOS AND BARBARA F. RFSKIN

Tolbert, Charles NI., Jr. Waite, Linda J., and Paula M. Hudis "The development and maintenance of a seg- 1982 "Industrial segmentation and men's career mo- 1981 bility." American Sociological Review regated labor force: review, synthesis, critique 47(August):457-77. of recent research." Prepared for the National Commission for Employment Policy, Wash- 'Tack, David B., and Myra H. Strober ington, D.C. 1981 "Jobs and gender: a history of the structuring of educational employment by sex." Pp. 131- Walshok, Mary L. 52 in Ptricia Schmuck and W. W. Charters 1981a Blue-Collar Women: Pioneers on the Male (eds.), Educational Policy and Management: Frontier. Garden City, N.Y.: Anchor. Sex Differentials. New York: Academic Press. 1981b "Some innovations in industrial apprenticeship Ullman. Joseph C., and Kay K. Deaux at General Motors: locally developed competency-based training as a tool for affirmative 1981 "Recent efforts to increase female participation 'in apprenticeships in the basic steel industry action." Pp. 174-82 in Vernon M. Briggs, Jr., in the midwest." Pp. 133-49 in Vernon M. and Felician Foltman (eds.), Apprenticeship Briggs, Jr., and Fe lician Foltman (eds:). Ap- Research: Emerging Findings and Future prenticeship Research: Emerging Findings and Trends. Ithaca, N.Y.; New York State School Future Trends. Ithaca, N.Y.: New York State of Industrial Relations. School of Industrial Relations. Westley, Laurie A. U.S. Commission on Civil Rights 1982 A Territorial Issue: A Study of Women in the Construction Trades. Washington, D.C.: Wider 1981 Child Care and Equal Opportunity for Women. Washington, D.C.: U.S. Commission on Civil Opportunities for Women. Rights. White, Connie, Barbara Angle, and Marat Moore U.S. Department of Commerce, Office of the Secretary 1981 "Sexual harassment in the coal industry: a survey of women ,miners." Unpublished paper, 1978 The Bottom Line: Unequal Enterprise in America. Report of the President's Ir,ter- Coal Employment Project, Oak Ridge, Tenn. agency Task Force on Women BusinessOwn- Wolf, Wendy C. ers. Washington, D.C.: U.S.Department of 1981 "The experience of women in federally spon- C1nmerce. sored employment and training programs." Pp. U.S. Department of Labor. Employment Standards 87-130 in National Commission for Employ- ment Policy (ed.), Increasing the Earningsof Administration Disadvantaged Women. Washington, D.C.: 1981 Participation of Females in Construction Trades. Washington, D.C.: U.S. Department of La- National Commission for Employment Policy. bor. Employment Standards Administration. Wright, Michael J. 1979 "Reproductive hazards and 'protective' dis- U.S. General Accounting Office crimination... Feminist Studies 5 (Sum- 1976 Job Opportunities for Women in the Military: Progress and Problems, Washington, 1).C.: U.S. mer):302-9. General Accounting Office.

2 G 7 Commentary: The Need to Study the 14Transformation of Job Structures MARYELLEN R. KELLEY

Research undertaken within an institutions: those that pertain to the allocation of tional framework attempts to explain labor workers to different kinds of employment market outcomes for different race and sex opportunities. Within that general category, groups (e, g. , relative wages, unemployment their analysis is further limited to rules that rates, and occupational status) as a function act as artificial barriers to the movement of of the efforts of trade unions, professional women out of the so-calledfemale domain associations, and employers to control the of work into male-!yped jobs. While their employment relationship. The arrange- efforts to identify all the exclusionary prac- ments that shape the work situations of dif- tices that have been uncovered by research- ferent groups have been described by John ers in recent years in the areas of recruit- Dunlop (1958) as a "web of rules," both for- ment, hiring, initial job assignment, training, mal and informal, that structure employ- promotion, and intrafirm transfer are valu- ment opportunities and allocate workers to able, I think this is an insufficient view of different segments of the labor market.1 the problem. The problem encompasses the In Chapter 13, Roos and Reskin have whole literature, and therefore my remarks shown that the phenomenon of sex-segre- should be taken to be constructive and mainly gated work can be analyzed, at least in part, directed toward future researchboth con- as a function of thisregulating of labor mar- ceptual and empirical. ket operations. They discuss institutional ar- First, I discuss some of the limitations of rangements that have been found to restrict the studies to which Roos and Reskin refer. women's entry into the higher-paying, more I then offer a brief criticism of the concep- stable jobs typically held by men. In.so doing, tual framework they themselves use in dis- the authors have focused on only one set of cussing the institutional arrangements that practices that regulate labor market opera- promote a sex-divided workplace. It is difficult to draw any general conclusions about the relative importance of the ' For a recent presentatm of segmented labor mar- many specific practices described by the' au- Le( them). see Gordon it ai. (19821. thors in explaining patterns of sex segr"a-

261 2 C8 262 MARYELLEN R. KELLEY tion in employment.This is because the re apprenticeship programs and hiring halls of search to which they refer is exploratory, its the building trades are no longer effective purpose being to identify different practices methods for controlling entry to construc- and to show how they act as barriers to the tion jobs, for either men or women (Mills, integration of women into male-typed jobs. 1980). For the most part, the research consists of Another problem with lists is that they case studies of parCcularestablishments or tend to imply that at least some increase in small sample surveys of individuals from se- equal treatment would be gained by the re- lected occupational groups. Some of the moval of any one of these barriers, holding practices the authors describe are well-known the others constant. But in any sort of com- and blatant examples of deliberate sex bias. plex work situation, that is unlikely. Rules But the more obvious barriers may not be interact. Let me illustrate by criticizing an the major impediment to the integration of aspect of my own recent work (Kelley, 1982). women into male-typed jobs. There is aneed In the case of complex seniority systems, for research that, in a systematic fashion, the absence (or even removal) of rules that examines the incidence of certain practices penalize mobility across sex-segregated job within and across industries, occupational ladders may not signal (or lead to) a mean- groups, and locales. ingful change or improvement in the op- I recognize that it is extremely difficult to portunities fbr women toThe promoted into do such research. It requires the coopera- male-typed jobs, if the rules governing the tion of managers and union officials, who selection of eligible workers permit the em- may not want to have aresearcher look too ployer to hire from the outside (rather than closely at the inequities in their practices. strictly promote from within) or if those Even so, such research is needed to dispel higher-paid jobs simply are not expanding. whatever misconceptions we may have about This leads me to my major concern with the relative importance of one or another the conceptual framework within which Roos kind of arrangement. Making checklists of and Reskin have placed their discussion of -source of bias" is not enoughcertainly how the regulation of labor market opera- not if' we ultimately care about formulating tions promotes sex segregation. They have an effective policy and strategyfin' change fiicused on those arrangements that inhibit that will result in the improvement of the the integration of women into the male do- economic position of large nilinbers of main of work. Res( arch on racial stratifica- women. tion tends to do the same thing: "White" One /Welder)) with lists is that they are jobs are the norm and the object of inquiry; static. Organizational rules and the institu- the probk:m is seen as one of how to reduce tional of laborl»arkets change the barriers to entry to those jobs fiir people over timeincluding exclusionary prac- who are not white. This association of an tices. Thus, for example. arbitrary and sex- institutionalist analysis almost exclusively biased entry requirements to union-con- with rules that restrict competition within trolled apprenticeship programs in the labor markets reflects an implicit theoretical building, trades were once effective barriers assumption: that the regulation of labor mar- to women's employment in the construction ket operations is primarily the result of ef- industry because the unions acted as "em- forts of limpidly organized groups of workers ployment intermediaries between their (in trade unions or professional associations) members and contractors" through hiring- to protect (or -shelter-) themselves from hall arratwements (Glover and Marshall, competition with each other, from different Iq7 it consequence of the increasing groups within the membership or from non- importance of the' tumunion sector in the members. There are, however, two other (1:o \ti 11( hull industryduring the 1970s. the anas of regulation that this pe'rspe'ctive

2 C COMMENTARY 263

glects. Both entail looking at the active, self- these structures are taken as given. Like so interested roles of personnel managers, in- many writers concerned about the problem dustrial engineers, and strategic corporate of sex-segregated work, these authors treat planners in structuring the employment re- the problem of the sex-typing of jobs almost lationship. I have in mind rules that channel as if it were a fact of nature. That is, because women into same-sex employment oppor- the sexual division of labor in some form is tunities in the first place, together with those evident in all societies, regardless of their that govern the creation of new, explicitly social or economic structures, and because "female" jobs.2 sex differences in treatment have a long his- There are a number of examples of re- tory in this country, it is assumed that the search examining how men are channelled separation of the sexes in. the workplace to- into what are thought to be appropriate ea- day has been a constant for a long period of 4 reer paths, but little research has been done time and is ultimately exogenously deter- to investigate that question for women. Os- mined by social and cultural forces outside terman's recent study (1980) of the early work the employment relationship. To Roos and experience of young men is the kind of re- Reskin, "traditional" sets of jobs are readily search that needs to be done on young identifiable as invariantly male or female. women. In that study, Osterman examines These distinctions are so apparent and the function of certain small-sized establish- thought to be so enduring over time that ments in the secondary labor market in pro- the authors do not feel that they even need viding training and experience for entry into to tell us what they mean by the categori- large organizations with characteristically zation "traditionally male" or "traditionally primary-sector jobs. Such linkages between female." young women's early-work experiences (by- Besides the implication that the compe- type of employer) and their future career tition between men and women A more im- paths within the so-called traditional domain portant than the conflict between workers of female-typed work need to be investi- and the managers who administer employ- gated to discover if there exist typical "feeder" ment systems, Roos and Reskin's depiction systems for regulating the flow of young of rigidly sex-segregated spheres of work ig- women into labor market segments in which nores the great changes in technology and W0111111 predominate and to understand in the occupational structure of the U.S. econ- what ways they are similar to or different omy that have taken place over the past 80 from those that seem to apply to young men. years and the shifts in the domains of wom- Roos and Reskin's focus on barriers gives en's work that have occurred at the same short shrift to those practices that structure time.3 Because affirmative action policy pre- the employment opportunities facing wo- scriptions motivate their analysis, the au- men e.g.. job design, wage setting for in- thors are concerned almost exclusively with dividual jobs, and location decisions pro- the set of practices by which people are mulgated unilaterally by managers. Instead, processed through a given structure of jobs

" For an exposition of the relationship between technological change and the growth of employment op- There is vet another approach. which Roos and Re- portunities for women in the tswertieth century, see skin ignore altogether. the radical feminist literature Baker (1964). For a less benign view on how the inthat focuses not so much on competition or other proc- troduction Of new work methods and machinery has esses within markets for wage labor as on the relation. affected the task structure of jobs and the demand for ship of such markets to nonmarket institutions. notably different kinds of workers, see Braverman (1974) and Imt r larchir al relations fit the household and the linkage . Edwards (1979). For a study that describes the charg- be paid and unpaid weak. For example's of each. ing sexual division of labor within new arras of work. we [(drumlin '19;9) and Power (1983). we Kraft (1979).

270 264 MARY ELLEN R. KELLEY and the reward systems to which they are REFERENCES attached. The importance of job design cri- Baker, Elizabeth Faulkner teria, job evaluation practices, and the lo- 1964 Technology and Women's Work. New York: Columbia University Press. cation of work in maintaining sex segrega- Braverman, Harry tion is hardly considered. But in fact we 1974 Labor and Monopoly Capital: The Degrada- have evidence that new jobs are often de- tion of Work in the Twentieth Century. New signed and valued explicitly in relation to York: Monthly Review Press. Dunlop, John the gender of the work force expected to be 1958 Industrial Relations Systems. New York; Holt. recruited to fill those positions. For exam- Edwards, Richard ple, according to one recent study on work 1979 Contested Terrain: The Tran.lf -nation of the organization and the location decisions of Workplace in the Twentieth Century. New York: located in Basic Books, managers, branch offices were Fuentes, Annette, and Barbara Ehrenreich communities in which large numbers of 1983 Women in the Global Factory. New York: In- married women could be expected to he in stitute for New Communications. need of employment (because of high un- Glover, Robert W and Ray Marshall 1977 The response of unions in the construction inemployment among male heads of house- dustry to antidiscrimination efforts. In Leon- holds). Their labor market choices were also ard J. Hausman et al., eds., &ia/ Rights and constrained by geographic immobility and Industrial Relations. Madison, Wis.: Industrial Relations Research Association. child care responsibilities (Teegarden, 1983). Cordon, David M., Richard Edwards, and Michael Barbara Ehrenreich's most recent mono- Reich graph (Fuentes and Ehrenreich, 1983) is one 1982 Segmented Work, Divided Workers. Cam- of a growing number of feminist studies of bridge, Eng.: Cambridge University Press. the ways in which electronics firms search Hartmann, Heidi I. 1979 Capitalism, patriarchy and job segregation by the globe for locations where they will be sex. In Zillah Eisenstein, ed., Capitalist Pa- able to continue to organize assembly work triarchy and the Case for Socialist Feminism. around the use of extremely low-paid young New York: Monthly Review Press. Kelley, Maryellen R. women. 1982 Discrimination in seniority systems: A case By limiting the analysis to only those rules study. Industrial and Labor Relations Review that act as harriers to or constraints on wom- 36(1):40-55. en's movement into and out of different types Kraft, Philip 1979 The industrialization of computer program- of work, the analyst can account only for ming: From programming to -software pro- differences in the ways in which women and duction.- Pp. 1-17 in Andrew Zimbalist, ed., through a given structure Case Studies on the Labor Profess. New York. men are processed Monthly Review Press. of jobs and system of rewards. To explain Mills, D. Quinn how jobs become sex-typed or indeed, even 198) Construction. Pp. 49-78 in Gerald D. Somers, resegregated, after having been integrated, ed., Collective Bargaining. Contemporanj?tier- haw the ican Experience. Madison. Wis.. Industrial Re- we need also to take into account lations Research Association. structure of work changes, i.e., how man- Osterman, Paul agers bundle tasks intojobs and him those 1980 Getting Started The Youth 141,E)r Marko . jobs are then linked to particular reward Cambridge, !slim.: !s1 IT 'ress. systems and opportunity structures. Powt.r. Marilsn liA3 From home production to wage labor Women as a reserve arms of labor. lirriw of Radical Political Economy I it 1) 71-91. e4ariti. Nil/41111 i9K3 Women's Labor ,tid Changes in the Ota.tipa tiotial Structure in the Office Inclic,trs ..k Case Studs of an Instirante Compans . Unpublished master's thesis. Department of City and lie unnnal Planning. liiisersits of Calitotnia. Ilei kyles

271 ..

Part III Reducing Segregation: The Effectiveness of Interventions

272 Job Integration Strategies: 15Today's Programs and Tomorrow's Needs

BRIGID O'FARRELL and SHARON L. HARLAN

Scores of private employers, including by men. By studying the experiences of these large corporations such as AT&T, General companies, we hope to learn what inter- Electric, and Ford, have experimented with ventions have been successful and howthey programs to reduce sex segregationin their can be more effective inthe future. work forces) Primarily in response to fed- This corporate perspective on job inte- eral equal employment opportunity (EEO) gration is important because the extremely enforcement activities and pressure from slow progress reflected in national statistics women desiringexpanded opportunities, masks both progress and problems in the some of these firmshave successfully in- industries and firms where enforcement ef- creased the number of women in nontra- forts have been targetea.2 The experiences ditional jobs, i.e., jobs predominately held

'It is not possible to identify or even to enumerate national origin, and religion as well as sex. In the same all the American companies that have activelytried to year EEOC reported receiving 47,454 newcharges of facilitate job integration. Although there has been no discrimination under Title VII (including ;.,462 filed systematic data collect ion. we estimatethat the n um} ,er concurrently under age discrimination (Jr equal pay is quite large based on the numberof company expe- acts). Because there are multiple bases (e.g., race, sex) riences discussed in this paperand on other literature and issues hiring. discharge), the Commission designed to help firms meet affirmative action require- analyzed a total of 77,802 charges against private em- ments (e.gPfeiffer and Walshok, 1981; Farley. 1979; ployers, 37,703 or 48 percent of which involved charges Hall and Albrecht, 1979, Stead, 1978, [Arwood, 1977, of sex discrimination (U. S. Equal Employment Op- Cunningham. 1976, Fox ley, 1976, Purcell, 1976, Gor- Port unity Commission, 1982). don and Strober. 1975, Jambe lli and Muczyk, 1975, 2 In part this is due to the inadequacy of available Hollander. 1975; Jungeword and Scott, 1973). The ac- data discussed in papers from this and other confer- tivities of the Equal Etophiyment OpportunityCom- ences (C. Brown, 1981; Wallace andIANIond, 1977). mission also provide an indexof corporate actisity in For example, the census data are not current enough. this area. The Commission reported that itis moni- occupational codes and EF.0-1 categories are too broad, toring 20 major aifirmatis e action agreements(personal the U.S. Department of Labor's Establishment Survey communication, 1982). In fiscal 1981 it settled16,7:30 does riot include sex (Brgmann. 1980. Barrett, 1978), charges of employment discrimination and filed 368 and none of these studies controls for program inter- lawsuits under Title VII of the 1964 (:isil Bights Act ventions. as amended These includedcharges based on race, 267 273 268 BR1G1D O'FARRELL AND SHARON L. HARLAN of these firms, reflected primarily through plish job integration and (2) a body of re- case studies and corporate surveys, arethe search consisting of case studies and cor- focus of this paper. More specifically, the porate surveys that enhances the understand- focus is On two important policy questions ing and interpretation of national employ- posed by the Committee on Women's Em- ment data and complements statistical anal- ployment. First, what kinds of interventions yses of company compliance with EEO laws. are likely to succeed orfail in reducing sex segregation in the workplace? Second, under what conditions are the chances fin success The Evaluation Framework enhanced or impaired? Firms that have entered the postpioneer The first section of this paper discusses era of job integration are thesubject of this two important issues in research on corpo- paper. Postpionecr companies haveagreed rate job integration policies: establishing an to initiate organizational changes that facil- analytic framework to identify separate di- itate the entry of more than token numbers mensions of change in corporate equal em- of women into nontraditional jobs and to ployment policy and assessing the quality of take an active role in their recruitment, hir- available data. The second and third see: ing, and training. In contrast, firms in the tions address the committee's concern with pioneer era have not undertaken such successful intervention programs. We ana- changes and may he trying to discourage lyze case examples of strategies used by women from following the lead of a few ex- companies to recruit, hire, and train women ceptional pioneers who have gained access for nontraditional entry-level jobs and iden- to men's jobs through personal initiative." tify the sources of problems encountered in promoting women. The fourth section addresses the committee's question about so- Pioneer is a term commonly used to describe the first svonien in nontraditional jobs (e.g.. Walsh 1k. cial and economic conditions that influence and several studies have doctitnented the experiences success. Several examples are givenof how of these %%nnen. It'SS is known about the progress external business conditions, internal man- and problems of %%Innen who came after the piotieers agement structure, and union involvement Epstein 111.0(11 a distinction betvi.cli old mid new affect the likelihood of reducing job segre- voiliii of the lasa hot did not tie it to organizational changes within firms. Our re% it IV Or the research. how- gation in a firm. In the' final section we offer eser, indicates that a pattern of transition fromthe eommendatiims fiir what the federal gov- pioneer to the postpioneer era of institutional change ernnient and private sector employers and is typical of matis firms For example. in our own lon- unions can do to increase the effectiveness gitudinal studs of a large manufacturing firm. the liar. of" programs for job integration. bur Cminpans illArl,ut and O'Farrell. 19821. distin. gnashed Itetsvemi these two fundarrientallsdiffirent phases. 1)iiritig the pioneer era, from the late 19fiOs to RSEMICII ISSUES the late 19705, .1 fess. %stimli of extraordinar mitiatise recognized disci Mini: ors practices and pressed their Some research has been undertaken du demands in light of the federal cisil rights legislation ing the past 10 years to assess the impact I losves tr. the hustilit \ \ laced Iron) management and niale cosvot kers and the isolation tiles experienced and to evaluate the' effectiveness of federal presented most ssannen from follosing their lead. The lavs and corporate policies aimed at reduc- postpionver eta began inI 97ss %%lien the (autumns .te.d ing occupational segregation. The joint ev- the federal go% ertmictit signed an allirmatis e action olution offeeleral and corporate policies over agieement This second phase of job integration %Anhui time has led to two important developments the firm has been rharartcrized bs the hiring of a in an% (.1 large nimilict of %%ine %%Innen into entry-les el \\inch bear directly on the analysis pre- lobs pies Musk reset ed for men and a reassessment sent,.( Ihere: 11) an evaluation frainexvork ofutimpans training progiamsPriscilla 1)ooglas and based on the experiences of companies that \larsellen Kells contributed to out earls thinking abont {hi\ e made Institutional changes to aceotn- this pumice' postpionet distinction

1-4, 4, fir 1 11

111 i\(:n.krvic S 269

all e'111- l'')stputtleet companies ale' not necessat11% ing personnel s stems and benefits for ts weal of most t.S firms, but the MCtilt 1)1(lyCCS, and the development of procedures anus h(1111\\ 111(11 \VC Call 'Part) the for monitoring the terms of the agreement. clic( tsif PallcfltI()11 StlatCtrj, Tilts ha\ e also specified intervention at four liCsCafell .111(1 program accounts of tub lit crniid points in the empio) ment imicess for empio)- ti 141 AIM) Ill'()(ssS S1114$2,CSt th.tttiltv.a nommtnagenn nt and management t(...ii(I in ,t titm's transition toIce ptistpi eev recruitment.hiring.tr-aining, and pro- ')meet era OITCNI)011(IS\\ Ith some sort of motion These parameters of the postpi(ncer federal pressure iThe' (' iden". agrements estabitsh Our analytic. frame- (Ic.S II(,t (aid( 115 to the alll()1111t work.' 01 lob uttev,r,ition resultiiii.; from unforcc but it does nniicitte that tile went art! \ int.% Quality 1)(thi fvfierii pretem..'is .1sitrinficiint moil\ ator ompaffics (11(qt, !Ii this regard. Iii mans \v,. 'wk.,.I (tVitiV,(i ,Ui(1 (1)111pa1'tfd many IIISLIU4 VS ItIS hrcilISC of directfederal in- case examples of experiences based scr discrwiiii,if Ioncomplaints on company slf-reports inpublicly avail- 114.onst Lu fieompanies. t14.. the ourt-oi abl sources, original reseal., li on single firms del misent decre of..V1.6(.1 (Vitilace. or ()11)(Katl employersdone by social si-

107() \mini up and Lifson. 19791 IA'SS 1 C11 entists, and surrey's that compare a large dal nniente(l IS the indirect eflect oil corn- !timber of eittpl,er.,.. primarily front the 'mines anticinatio)4federal action, e,14..01- perspective of personnel directors. These witars aceptan.of the steel industry C()11- data differ Vrciitly in quality and degree of sc II t(ICCI CC 1/s(III' III(IIIStr completeness. The principal strengths of the data rest fc . I () S3.Ir an of the stIgheS 55 C perSpeeti\ C I c lc \\ Cll. ,(55 arClIc.,Sof the federal la\s and on their ability to show it utlillue tegitlations dmi then- related finaniiti costs (111 the process of change. tintsomple- 'Slue ( Itc(1 1)1IllallaS4CIS ,111(1 \V(11'kI'S asIlli- menting and enriching the alit6 NIS of na- ptnt.Int 1,n tors hinidatinv change. tional data on occupational .segrgation. The consent devices re,iiiting from eolirt in contrast to the slosv rate ofchange pot.- .ttleinents of (list inninatiun complaints an t ed in the national data. our analysis shos cite fedeial action gliideline for Mein'anittioti across and within companies. ,m1)1(Acist Emplo !tient ()ppor hich permits a twin I(t7 1. II cod, of of both prog,iess and pr( Heins. At AT&T. ui § () 1()79 ha\ e taken a (nn for{"1/4"1111)1(' " 11110total "111)1("111"It piehnsie apiti(ta( h 'detail\ ing ',Niles and (.1Casc(1I)\(nil 1.:270 hchvern 1973 and Its eloping inteiSention strategies. .1.inong the the nurnl,er of Ss011(.11 in ow Alicia', 1,1(1.,..,.(1 Its 15.3(. Whist impoit.int poiblinsad(11CsSC(1 are Ow .011(1 atcgur t,t.thlt,lilm titif mechanism, fur intilatint; moll' duals Into iol)s unelnilini.; the ,151,01 plc .01101 Fin tit( thi+ 1).(1)11 v.t ,Its, littilt dtt1 111)V.1.111111! \ \ (1111cli's lulus. Chiang Lwirt.Ihics tott.in quid iiiploliwut 011)()! Unlit\ that ItINIciII11 II\tItiq.;11 SS the 1e1411 Linifft, 'it agrh. "iik.Nith 1)."r2.1.1111 nitre.' photo th..it 1.d1 %%ownthc,..t It 11 ,.,11 it iii, par.im.te, 55, hlit t ft'. (\ tythat tin, titwititm .y.111.:.(1 to 11..:111(.til oo.gilip IIII I .t IIt..114 ti pit "Mt littill "Olt 1 WI, NJ I II W.:41'11k it. .. :,..t t 1.)11. And the flit Hittit ftgittif isI:. , (t11,4I fulfill. tutu, 111.tt in, , /kit ; . .. ill d hut ii,itt tit II NII t 111111111.111111,111 nil I. 1%. I1.I. I1 111. /1.,/ 11114 \1.i.t 1. It, \\ \ it I Bli/(:11) O'FARRLL SHARON L. of 7).000 more than the increase for men done in cooperation either with manage- -.Northrup and Liiison, 1979). went or with a union. none offers a truly Second. the perceptions of managers and asse.ssment of problems and prog- %% ()risers about the effectiveness of fede.ral ress based on the views of ninag,eis and equal emplo% meta policy in reducing job workers. Most of these studies also provide It ation within companies is often much limited analysis of quantitative data, and, wore optimistic. than analyses of economic althougli all of the case study companies were impact based on tuitional data. rot example, targets of federal enfeirceent activity, most manager, re'por't that goals and timetaliles of the research does not systematically focus ale critical management tools for achieving on the impact of federal regulations on or- greater job integration. Evaluation research ganizational change. Studies by Northrup tit the. construction trades (Kane and Miller, and Larson( 1979) and Ichniowski (19831 NS1 and the maritime industry suggest there is considerable data available

(1al . l()7Sslimy that they proelue.e. from consent decrees in the public domain, stailhal change. The psychological itii)o- but it is expensive toCOPyand01.0211difficult ton( e of national 1.: t() la%%., in encouraging to IINC. There has been little public or pri- %%omen to 1). ess their demands forbetter vate interest in supporting more compre- lobs %%as% 'deuced in our researe.11 at the hensive analysis of the compliance data. I larl)(11 Cullom% i()'Fairell,1 980ii, [lariat) Finally, in addition to focusing mostly on and ( Warrell. NS-2). large' companies acting under government The principal weaknesses ()I' the data lie pressure., we are reporting almost exclu- in then selettx it% and uneven quality, %%inch sively on the experiences of white women. del 1% e hum the fat.t that available data Coll- There is very little information about ISI 111,11111 \ of \% hat companies choose to put 1101AV omen. hilt available research, pri- in the public domain. limiting theobjecti%- maily in the nonmanagement sector, sug- It\ (mess, and conit)iirabilify gests that they are moving to factory and lin both legal and (..oniptitive reasons. Se% - clerical jobstraditionally held by white .1.e1 ot the best re\ te\v, and case studies are .onmen (fictit)ous and lienbens.1 979, illust rah of the data limitations. Shaeffer Douglas, i9S1, Molvcaux. 1 9S21. Thus. the' and 1.5 loon s1479.1 soul% of 2(i5 companies negative eflects of runtand \ ate com- jut , p)`s(). 1,.\ in .12 ow_ pounded for both aod promotion of panics a l e both Iiiied on pet sutincl(lipart- tillItoItt\1,(-01111.11, mein sin(-% s and Intel% ievs. and neither s% shinati( teottipan% 1,0141am, \i()\ (i\I EN, j () not telate ino,41.(i Hokin\ es to I.N I able Inc isines 01 change Each has a selee- tion bios.sin( ,111(1lis loon stir Most ill the slievssinl inter% ention pro- I mil\ Lull.,t otporations and %loin(' t4la11iis dealt s.sithliam4es titonipan% to Ie\ 11,55 1.11 old% data on ilianau.eis ruitiiwiet. lining. and hawing, piogams for ( (Si studiossuch iis thou' \kW,.atid etltry leveljob.s. The famexyork for these

1()"L'i 11111)111111(IltdItICS.V0111.111)and programs \vas established ;Is an employer I1)71)oli\1i1. Dr(ll\add obligation in atilrinati%e action programs and

IT) stet I (1 1111)1,11drl.ofII.Il1.11i,111(1 supported to some e \tent b t,dl pro- O h.111111 I(Y").2 or. Milk)! CA)1111).111% arc grams such as the (:oiliprehensi%e. (1Iti.111.(1hi«IS (I mil\ une cmpali\ urNu\ mem and Training Act i(:ETAI. Ceradualk- titsIii \.11111 \ the\ empliier, hase begun to reach the in,reas- IIInut()Ilt.f 1,11,th I\k cr ing number of qualilied orqualifiable. Soilleie

iitiplritivnt.ttInii1): 1 haIitLt )1111e the\alt. intr. I- 1'1(I in and to tr\ nontrad;-

) c j013 1..\ IR S IIINTEGI 271 ,

tional positions in both blue-collar (('Ear- sources, reevaluating,job requirements in de, .1- re]) and I lariat), 1982; (YEarren, 1982; Kam. relation to job content, and N t oping or snp- and Miller. 1981: Valshok, 1981; Schreiber. plementing training progranis. Programs have I 97W and white-collar occupations been develop lk to overfnie sex differences I in, 1980, Shaeffer and Lynton, 1979)." in early socialization, education, and training There are still [Mint' %V(Ittleti WIN) (It) not and thus reduce sex segregation in initial job seek nontraditional jobs (Hoffman and Reid, assignments. 1981, Barrett, 1980) and many institutional barriers remain (Roos and lieskin, this vol- Recruitment: External and Internal ume), Init stereotypes aboutwomen's \York Outreach behavior and motivation are gradiaidk being discredited i(:rowley et al. 1973, Kanter. An issue faced by many firms is how to 1977, I:eldber.; and (lenn, 1979; Ileidrick expand .raclitional recruiting sources and and Sruggles.Inc..1979; (Y.Farrell and methods to reach \vonien A successful strat- 1982). 1.: .() policies that increase egy appears to combine very activerecruit- women's sub choices are being incorporated ing with greatly expanded external and in- into pct procedures, and there is a ternal recruitment sources (Shuchat et al., gnu W111$4 aceltaliCv these policies by cor- 1981; Shaeffer and Lynton, 1979). In screen- porat managers. In two recent surveysof ing and selecting women fOr new jobs, em- managers the majorityreported that equid ployers must also provide sufficient infor- ment was a major concern(Ileidrick mation to enable women to makeinformed and Struggles. Inc.. 1977) and that corpo- choices and must elicit sufficient infOrina- rate eseentic es support theguidelines oldie tion from the women to determineif they V.SEqual mplo mega Opportunity can perfOrm the necessarytasks. Increased tOr affirmative action tliarnhill- selectivity based on job-relevant criteria may I laces. I979,. In the liarnhill-Hayes survey result in fewer new hires but more long- the majority of executive respondentssaid ter success (Meyer and Lee. 1978,Mc- that affirmatie action had not resulted in a Lane, 1980). decline in employee productic ity. Theint 'I'o expand external recruiting sourcesfor port.ince of FF.() to business practice isfur- blue-collar .jobs. companies must go beyond ther cc idenced at theI larard high school shop class( s, trade schools, and StImo!. 55 lierc mr 1() (if the cases in the the military cervices, which traditionally a, 1 earingI linise nov address equal supply young men. Skilled trades training mploc went and allirmatix e ,ietion issues. programs, funded primarilythrough federal \\ nen turned into action, management government programs, such as the Com- concern and supportfi,te concentrated on prehensic r Education and Training Act intev.tatini2; 'lin.% le% elblue-collar and mai, 10E.TA1 outreach and training rrogam for agement jobs hl. expanding recruitment ri `")111(11 in aPPIt'IltIresiliPhavt'iwt'll all poi taut resource fur companrecruiters (Kane and \Niel, 1981. Ullman and 1)eaux 1981;. One Midwester., steel company re- t1 lltl 1.ittit 1- (In alp I I\If',ttill v.futitii tin tr(111111,d ,1111 spondd to the I97. industry consent (h- III hitt tillitititt establhing a training, school for mo- in mat', iHrsttImIN I,111,11111.114111V \ ere( of 111111.1'11 OW P.11111/11lIt%.% I/1111 II III1111;1111'11111'4 tor insdcturs that could pill\ 1(1(a source III fiii,!r .11111 :1?1,141.111IN "VI el craft-named women and minorities. In citch

I A . 1 1 1 1 1 ) 1 1' till 0 1 P P e I \ I lens, In 11 1 , I I I I 1 I.01.! 1."1 class halt the students \vert plant emplo cos II" I/1111II1111111 l')1/..1 1.111;11111111,l' 1'111,111,1'd and hallv..t.tt. rt.lcruals..1ppro\i III 01111110 IN11111.11 11.1 11 //I till match t \So thud, of tint Ion Icinalv motor \I IIII, l'i111

2 77 kW.17.) BRIGID OTARRELIAND SHARON L. HARLAN inspectors and millwrights were recruited 1981; Shaeffer and Lynton, 1979).7 For ex- from outside the firm through CETA, but ample, many unionized firms traditionally no .ftonale applicants were recruited from have departmental seniority systems under trade schools (1.111maii and 1)eaux, 1981). which years of service in a department en- Companies that have done successful re- title workers to promotion opportunities crinting recommend reaching out specifi- during prosperous times and to protection cally to rural women and ph!,'sicareducation from layoff's during slow periods. Workers majors because of a general interest in and cannot, however, move between depart- experience with physically demanding work ments without losing seniority, possibly some and to women in blue-collar community or- pay, and other benefits relatedto seniority. ganizations because of their general famil- If women have to give up seniority, they are iarit!,- with blue-collar jobs (Shaeffer and less likely to move when jobs in traditionally Lynton, 1979). There is also growing evi- male departments are opened to tin on re- dence that women interested in nontradi- gardless of other incentives such as higher tional blue-collar jobs are more likely to be pay or bonuses. When a large company in in their late twenties and thirties rather than the Northeast initiated a transfer program recent high scluml graduates (O'Farrell, 1982; based on company rather than departmental Kane Add Miller, 1981; U.S. Dept. of Labor, seniority, women were able for the first time 1978). This suggests reaching out to cur- to bid on craft jobs without loss of seniority. rently employed women or to older women Three years later. 101 women, almost all the now reentering the work force. women in one of the craft jobs, had trans- For entry-level management positions, ferred from other departments in the com- expanded external recruitment may involve pany,. bringing from 1to over 20 years of going to more i:ollegcsin different geo- company service (O'Farrell and Harlan, graphic regions or initiating recruitment 1982). programs at women's colleges (Shaeffer and (lunges in formal seniority provisions ap- I.!,uton. 1979). Nontraditional methods for peer to be an important first step in opening recruiting higher-level managers include nontraditional jobs to women currently em- searching in the public sector and on the ployed in firms. Additional incentives may East and West coasts (especially New York). he needed, however, to encourage women Retaining a consultant or search firm that (or minorities or white men) to transfer from specializes in recruiting women or specify- One department to another (Ichniowski, 198:i; ing that .he recruiter be a woman are also Kelley, 1982; Shaeffer and Lynton, 1979). effective (McLane, 1980). An example of a In the steel industry's consent di exc. fOr model lung -terns recruiting system for tech- example. rate retention (or ''redirling-) nical fields involving universities and com- was used as an incentive (available to all panies is the Program fin. Increasing Mi- workers) for transferring from one depart- nority Engineering Graduates (P1N1E(;) ment to imolier under the new plantsen- described by Hayes (1980). iority system. If the rate of pay for those Women already employed by a company transferring to an entry -level job in a nets in traditional female clerical and factory jobs department was lower than their current rate are an important source of candidates for of pay, the could keep their eurient rate nontraditional jobs in the skilled trades and of pay in the new department for a 2-year Internal interxention programs fin white- mid blue-collar jobs reiluire opening access systems to mine!), winch St it% mid 1) I.t%,h,t.t }rit thy siihicut of may inhule establishing job posting pro- 41 of (a.... mid imiclit decipt, (Intim! Ow cedues or changing seniority systems ii'mos 1.1%t I) %car, Fur,tIf.%ItM. tilt' Isstlys %tt and Iiiskin. this volume. Steinberg and Cook. 1943. Srllr. ,It)52. and kVallat t and Di iscoll19S I

27s 1013 INTEGRATION STRATEGIES 273 period. In other words, a woman would not isfactory records remaining in grade beyond have to take a cut in pay to enter a depart- the average time spans; companywide in- ment with much better long-term oppor- ternal searches; job posting of exempt po- tunities previously reserved for men. The sitions; a system for employee transfer and agreement also included a 45-day trial pe- promotion requests; a hotline for employees riod during which transfers could return to who believe they are' in dead-end jobs or their former jobs without any penalties (Leh- are being held back; a liberal,well-publi- niowski, 1983).8 cized tuition-refund program; analysis of job The Teamsters decision of 1977 greatly families to create new career paths; individ- reduced the legal pressure on companies ual career counseling on request; and group ao,l unions to change seniority systems. career counseling (Shaeffer and Lynton, Under the Supreme Court's ruling, senior- 197-9:52-53). ity systems are considered bong fide or legal Both external and internal recruitment if there was no intent to discriminate even ,require aggressive methods. This may in- though the system may perpetuate the ef- volve using newspapers, school and com- fects of previous discrimination (Kelley, 1982; munity groups, company bulletins, meet- Wallace and Driscoll, 1981). Changes in ings, job fairs, etc. (Shuchat et al., 1981) as- seniority systems, however, remain an im- well as providing more information about portant intervention strategy to reduce oc- job content. !Amy women are unfamiliar cupational segregation by race or sex. with nontradit:onal work. Some companies Opening careet opportunities in manage- have reported the need for clear job de- ment to current office workers may involve scriptions that include information on train- less formal procedures than changing sen- ing and promotion opportunities, hours, pay, iority systems established through collective and pressures. CoMpanies have found it ef- bargaining agreements, but it is nonetheless fective to involve women already doing the reported to "take_a lot of effort and a close work in interviews or presentations. It was critical look at the company's inner work- often important, particularly for internal re- ings." The personnel director of a major bank cruits, to have trial periods during which outlined the following internal mechanisms women can go hack to their previousjobs in use at the bank to facilitate the upgrading without loss if the nontraditional job Was not of office workers: a comput.9ized skills data acceptable. Tours of the work environment, bank; identification of employees with sat- some hands-on experience, and useof media presentations seemed particularly useful' for blue-collar jobs (Kane and 1981; Walshok, 1981; McLane, 1980; Shaeffer and Stqii(Y.y systems are often minplex nd extensive Lynton, 1979). changes nay yield only limited results. For example, More careful recruitment and screening AT&T adopted company but not systemwide seniority fur- that did not address the regional difference in oppor- may also reduce employee turnover, tunities for blue-collar workers generally recognized in thering the long-term goal of job integra- the nienagement job structure. In the steel industry, tion. Meyer and Lee (1,978) found that women plant seniority repiesented an expansion of opportu- were much less carefully screened forblue- nities that the union had tried and been unable to collar jobs than for white-collar jobs and that achieve through collective bargaining. Women in the' clerical and service jobs were still excluded. hokever, the turnover rates for blue-collar jobs were leaving then with limited opportunities (Kleiman and higher. McLane (1980) reported similar ex- Frankel. 1975). (:ompany seniority was in place at Har- ainiiles comparing entry-level and higher- bor Companx but was limited to specific geographic level management. Companies reported locations that wet e not changed by the affirmative ac- conducting much more extensive screening tiongre. meat This left sex-segregatedplants with different advancement opportunities intact Marian and fir higher-level jobs and better retention 0 Farrell. 19h2t rateslhan for entry-level management jobs. 279 274 BRIGID O'FARRELL AND SHARON L. HARLAN

Companies surveyed by Shaeffer and siderable disagreement about the effective- Lynton (1979) also reported that job analysis ness of related programs todevelop aware- is crucial to increasing the number of women ness, mentors, and networks. in nontraditional jobs of all kinds. job anal- The basic competence of women to do ysis means reevaluating the current quali- nontraditional jobs affects their initial ac- fications required for a job in light of the ceptance and the ultimate pace of integra- actual job content. Establishing the actual tion. Introducing unqualified women into skills and personal experiences necessary for nontraditional jobs reinforces male workers' adequate job performance can expand the stereotypes about women and strengthens range of jobs for women by identifying com- resistance to their integration (McLane, 1980; mon domains of skills across seemingly un- Meyer and Lee, 1978; O'Farrell, 1977). Be- related jobs and by recognizing past related cause of sex differences ineducational back- experiences. For example, a restaurant chain ground and experience, women entering recruiting recent college graduates for first- blue-collar jobs need additional training in line supervisory positions was unsuccessful such areas as tool familiarization and basic because many of the women were unable to electronics. In general, they have not learned meet the demands of the work and were too basic craft skills in high school shop courses, inexperienced to manage a work crew. Sub- trade schools, the military service, or just sequent management discussions led line from "tinkering around cars with their fath- managers to revise the job qualificationsby ers." Companies surveyed by Shaeffer and replacing the college degree requirement Lynton (1979) found that their ability to pro- with knowledge of the food industry (e.g., yid( training was critical. to opening these restaurant work),retail experience (with jobs. Preplacement training and supple- customer contact), or other experiences re- mentary courses available during on!the-job lated to supervisory ability (e.g., teaching training or between formal training sessions or home economics ) (Shaefferand Lynton, have been tried successfully. In the Meyer 1979). and Lee study, training directors frequently reported that (formal) training programs for craft jobs assumed that all of the trainees Training: Skills and Information ;lad mechanical interests and experience. Training is an established part of affirm- Therefore, terminology' such as open-end ative action programs. Two innovative ex- wrench, hexnut, or right-hand thread was amples arc the multimillion-dollar training used %vithout considering that many women fund for women employees established by were unfamiliar with such terms and were Bank America in lieu of back pay awards likely to be at a decided disadvantage. To and the 1978 General Electric Agreement overcome these difficulties, some compa- that provided more than $500,000 to train nies designed special vestibule training for managers to implement the affirmative ac- women. These special

by senior craftsmen or foremen (O'Farrell, level and in every location. For example, 1977; Ullman and Deaux, 1981). Equipment Continental Illinois Bank used a series of modifications may also be necessary for blue- brief film clips highlighting a variety of work- collar women (Meyer and Lee, 1978; North- related problems to remind people of po- rup and Larson, 1979). For example, the tentially discriminatory actions. Managers Bell System's equipment for pole climbers were cautioned, for example, not to assume (gloves, shoes, climbing apparatus) was all that a woman being considered for a pro- specifically designed for men. Early failure motion would not be free to travel (McLane, to modify the equipment contributed to 1980). higher accident rates for women. Shuchat Companies suggested holding meetings et al, (1981) found that many men also ben- and issuing publications to keep the staff up- efit from these sorts of changes. to-date on changes in the law and their or- Comparing the experience of nontradi- ganizational implications. They reported that tional female employees in 10 utility com- internal communications are improved if panies, Meyer and Lee (1978) found that provisions are made for opponents of EEO women in blue-collar jobs needed job-spe- to voice their concerns (McLane, 1980). The cific training but that women in white-collar development of information programs was jobs did not. Special training is generally also recommended for union officials, es- not considered necessary for women in man- pecially shop stewards, but there are few agement positions and is perceived by some models to follow (O'Farrell, 1980b; Shaeffer as detrimental, either by providing women and Lynton, 1979). with services not available to men or by ex- Training resources have also been allo- aggerating any sex differences (Shaeffer and cated to programs that focus on changing Lynton, 1979; McLane, 1980). It is neces- attitudes and interpersonal relationships often sary, however, to identify the critical, for- involving the development of mentors and mal and informal sources of male training networks. Both the corporate and the re- and to make sure that women are inte- search communities are divided on the use- rated. Formal classroom training for ex- fulness of such programs in general and on ecutives, specific patterns of job rotation, the effectiveness of company-sponsored and assignment to particular jobs or super- programs in particular (L. Brown, 1981; visors are examples of such training. (Bar- McLane, 1980; Shaeffer and Lynton, 1979). riers to managerial training are discussed Even to some who accept the idea that men- more fully in the next section.) tors and networks are important elements Providing information about EEO issues in achieving success, company-sponsored is another important area. In a 1973 study, networks seemed to be artificial construc- Lyle commented on the appalling lack of tions that may not produce the essential con- infOrmation available to line managers about ditionstrust, shared beliefs, commonality EEO matters, and there is increasing aware- of intereststhat foster the growth of these ness that laws and company policies are not personal relationships (McLane, 1980). Such automatically known or understood programs may ben,!fit women in some companies, but they do not appear to he a high . throughout an organization. Manager preparation was frequently mentioned in the priority among intervention 'strategies. Shaeffer and Lynton (1979) study as an activity critical to successful implementation PROBLEMS: PROMOTING WOMEN TO of EE() policies. The purpose of the prep- HIGHLY SKILLED AND EXECUTIVE- aration is to explain tint organization's over- LEVEL POSITIONS all EE() obligations and the basic responsibility of line managers. It is essential that In companies where WOHICTI have been the information go to all employees at every hired for entry-level management and blue-

281 276 13R1 (:ID O'FARRELL AN/3 SHARON L. HARLAN collar positions, there are no guaranteesthat have very limited opportunities for advance- they will progress to the higher-paying, more ment, but the nontraditional jobladders skilled positions. In fact, it appears that when women are enteringalso vary in the degree solutions to recruitment, hiring, and train- of opportunity they offer. In fact, the ad- ing problems begin to work (i.e.,there are vancement potential of women inlow-op- more women inthe organization), more portunity nontraditional jobs maybe no complex problems arise. Promotion, de- greater than their counterparts' intradition- fined as individual mobility through lines of ally female jobs. Several studies have found progression in internal (firms labormarkets, limited opportunities in management and is emerging as a critical issue inthe 1980s professional categories, for example, in legal for firms in the postpioneer era. or personneldepartments (Strober, 1982; Doeringer and Piore (1971) identified lim- Harlan and Weiss, 1981, 1982; Epstein, 1981; ited job progression lines, administrative Rosenbaum, 1979; Swanson and Milward, regulations for job upgrading systems, and 1979; Kanter, 1977, 1979) and in the blue- lack of training as mechanisms that perpet- collar categories of laborer or maintenance uate discrimination withinfirms. Based on work (Deaux, in press; Schreiber, 1979; several recent studies, Roos and Reskin (this Harlan and O'Farrell, 1982). Reubens and volume have described specific examples of Reubens (1979) and Briggs (1981) have sug- how these institutional barriers block tra- gested that newly hired women are becom- ditionally male promotion paths and prevent ing the majority in some of thesetradition- women from advancing inorganizations. Al- ally male jobs. though affirmative action agreements (es- Studies by Kelley (1982) and Harlan and pecially consent decrees) have tried to an- O'Farrell (1982) illustrate the diversity among ticipate some difficulties by imposingchanges nontraditional job ladders. Nontraditional on internalmobility structures, the evi- jobs, like traditional women's jobs, vary in dence suggests that these alterations have entry-level wage (floor), top wage (ceiling), been tqc) few and too limited to providelong- number of jobs within job progression lad- term relief from job segregation. ders (range), and number of people (den- Looking to the future, we believe thatthe sity).thus providing very different pro- promotion problems of both womenand men motion opportunities (Kelley, 1982). will be compounded by the growing number The Harbor Company (O'Farrell, I980a; of women in entry-level nontraditionaljobs, Harlan and O'Farrell, 1982) provides ade- the low advancement opportunities in some tailed example of women disproportionately job progression lines, the limited number entering traditionally male jobs withlimited of higher-level jobs, and low turnover. Suc- opportunities and the potential for resegre- cessful job integration efforts ultimately mean gation. During a period of rapidhiring (40 finding solutions to two related questions: percent of the work forcehad been hired in How can the newly integratedentry-level the Irevions 5 years), women's opportuni- positions be kept from resegregatingand be- ties were improved in theshort Mil. By coming new female "ghettos"?How ean op- l980, 9 percent of the blue collar workforce portunities fbr women and menbe ex- was female, and 66 percentof these women panded to facilitate advancement intothe had been hired since the EE() agreement highest-paying, prestigious positions? with the :;overnment was signed in 1978. However, no precautions were taken to recruit and train women at difierent Variations in Opportunity Structures the organization. Most resvarch on occupational segregi;- Relative to newly hired men, newt hired tion empha.iizes that traditionallyfemale jobs women of Loth races weredisproportion-

282 10B INTEGRATION STRATEGIES 277

ately represented in jobs at the bottom of comparison of the 1960 and 1970 census data, the job hierarchy. Hierarchies at the Harbor Reubens and Reubens (1979) found that Company were organized into 8 predomi- women had made the greatest breakthrough nately male job families that connect in male-intensive occupations in the clerical hundreds of blue-collar classifications based field: insurance adjuster, postal clerk, dis- on similarities in work tasks and increasing patcher, production controller, ticket agent. skill levels. Women's chances for moving up The skills required for these jobs are related were lower than men's because they had to those used in traditional female jobs, such been disproportionately hired into the job as filing, reporttritinb, and customer con- families with the lowest ceiling rates and the tact. In additimi, qualifications for the two fewest number of highly skilled jobs. Newly sets of jobs are not distinguished by edu- hired men were also placed in jobs above cation, training, or physical characteristics. them in the hierarchy, effectively closing off Despite these similarities, the women pi- those potential opportunities as well. Ad- oneers at the Harbor Company (O'Farrell, vancement opportunities were further lim- 1980a) found that it was just as difficult to ited for women by a recession, during which enter the higher-skilled clerical jobs as it hiring stopped and layoffs were threatened. was to enter the traditionally male factory The result may be a bottleneck where women jobs. After the first women succeeded, how- newly hired in nontraditional jobs find ever, others in the lowest-level clerical jobs ,themselves stuck in jobs that increasingly (and some in factory jobs) aspired to the become traditional for women. relatively small number of traditionally male An important question that bears on fu- clerical jobs such as production clerk. in ture promotion opportunities is how wom- fact, by 1980 that job was almost half female. en's first nontraditional job assignment is Without any intervention on the part of the determined by the company. Research at company (and none was in evidence), it ap- the I labor Company suggests that there are peared that these clerical jobs, similar to the differences in the education, training, and entry. -level factory jobs, were likely to be- previous experience of female and male ap- come predominately female.° plicants that may partially account for the assignment of most women to low-oppor- Assessing Qualifications and Potential for tunity job ladders. For example, men were Promotion more likely than women to have tradeschool experience that qualified them for entry-level A second set of issues concerns the cri- positions in the skilled trades. On the other teria and methods used to select individuals hand, many of the women interviewees be- for promotion and layoff and how these are lieved that the company selectively placed women and minorities in the least-skilled jobs regardless of qualifications, and they The future of wage rates in both the entry .1eY el gave specific and convincing examples ofhow factory and high-level clerical jobs at Harbor Company this was done. The Harbor Company's per- is .provisional. In both cases the wages are relatively high and better than what was available to women be. sonnel department ovulev used unsys- fore. It seems unlikely that there would be any attempt tematic screening ocedures with little to lower the wages in these job categories. However, analysis of job lacors. whether these jobs keep pace with the overall wage in searching fiways to increase women's increases in the skilled', predominately male jobs must be carefully monitored. Preston (1978) found that in access to job lad( rs with greater opportu- New England the wages of teachers did not go down nities fOr advancement, we should not over- when women entered the field in large numbers, they look highly skilled clerical jobs that have just did not continue to increase at the pace of men's traditionally been closed to women. In their lobs.

, 283 278 BRIGID O'FARRELL AND SHARON L. HARLAN applied to women, To ensure upward mo- Seniority systems may reinforce the lim- bility for women, companies and unions must ited opportunities and bottlenecks de- examine the ramifications of collective bar- scribed previously. Changes in seniority gaining agreements for job advancement, systems (discussed in the recruitment sec- reassess their current management quali- tion) that facilitate the advancement of women fying procedures, and expand opportunities from traditional to entry-level nontradi- fbr all workers. tional jobs also facilitate advancement within Individual qualifications and seniorityare the nontraditional job categories. two standard criteria used by managers in Another issue of particular concern to as- selecting candidates for promotion oppor- piring women managers is access to com- tunities. Although the ways in which these pany programs that develop and measure criteria are applied may vary greatly from personal qualifications for upper manage- company to company and across levels of ment. '° A survey of 2,000 n*mbers of the the occupational hierarchy, the basic prin- American Management Assolciation (AMA) ciples are deeply ingrained in industrial re- found that 70 percent of the employers used lations and are often formalized in collective formal programs to assess managerial po- bargaining agreements. It is not surprising, tential in their employees (McLane, 1980; then, that one of the most controversial pro- Burrows, 1978). Yet many companies ap- visions of the AT&T consent decree con- parently denied women access to these pro- cerned changes in the use of qualifications grams. Fifty-eight percent of the AMA re- and seniority to determine job advance- spondents considered only a limited number ment. Under AT&T's agreement, certain of women qualified for development, 12 circumstances permitted a less-senior but percent did not admit any women, and only basically qualified member of a protected 9 percent specifically recruited women for group to he promoted over more-senior, developmental activities. more-qualified white men in order to meet Managers alleged that women's lack of ed- affirmative action goals, Although disaggre- ucation, experience, motivation, and career gated statistics by sex and race are not avail- commitment were the principal reasons why able, AT&T reports that the override pro- they did not recruit women. AT&T's ex- vision was used a total of 35,479 times for perience with its extensive management as- blue-collar workers during the 6 years of the sessment center, however, contradicts these consent decree. There are no statistics cov- steeotypes about women workers. Once ering instances where management person- women were allowed to participate in the nel were passed over by less-senior, less- Center, they performed equally with men, qualified employees, but personnel execu- and assessment results correlated with later tives interviewed by Northrup and Larson management progress for men and women (1979) testified that it happened regularly. (Moses and Boehm, 1975). After the first year of the consent decree, Access to existing programs is one barrier AT&T achieved between 90 and 99 percent for women, but absence of development of its targets for hiring and promoting women and minorities, and the override provision appears to have contributed to that success. RI Extensive procedures have been developed for test Loomis (1979) and Northrup and Larson validation. Companies must establish that any tests they (1979) feel that the AT&T data actually un- use are both job related and free of ra;:ial, ethnic, or derrepresent the impact of the override. They sex bias (Guidelines on Eraployee Selection Procedures, 29 ( ;FR §1607, 1970, U.S. Equal Employment Op- estimate that it was used in close to 25 per- portunity Commission, 1974). For a discussion of test- cent of all promotions during the 6-year pe- ing at AT&T, see the chapters by Ash and by Lopez riod. in Wallace (1976).

234 JOB INTEGRATION STRATEGIES 279 programs in many companies is a serious the 1980s and 1990s. The pace of change concern for other women as well as men. and the level of effort required to implement Harlan and Weiss's (1981) study of managers it, however, will vary with selected business in retail companies that had no formal pro- conditions and the internal management grams for developing managerial potential practices of postpioneer-era firms. This sec- fOund that the majority of women and men tion examines recent evidence that suggests had unfocused career plans and did not in- how these two sets of factors operate singly itiate training or development activities. A and in combination to affect the degree of battery of psychological tests showed few six segregation in the workplace. First, it is significant sex differences in level of moti- widely recognized that national and indus- vation or aspirations. The authors concluded try-specific economic conditions are related that both women and men need more career to the rate of job integration. Norton (1981) development activities, including assessing has argued persuasively that future affirm- and measuring skills and competencies, ca- ative action efforts will be more effective if reer planning, and supervisory training. The employers coordinate their planning with need for better career development for men trends in labor force composition, regional and women has been further documented changes in economic opportunitile and by Kanter (1977, 1980) and Fernandez (1981). technological innovation. Second, program Given the resistance of management and interventions implemented under corporate unions to changing general qualification and affirmative action plans, such as those dis seniority rules and the courts' support of' cussed above, are major organizational bona fide seniority systems, itis unlikely changes that require sound management to that firms will adopt interventions like the be effective. AT&T override or change seniority systems without strong government pressure and Business Conditions continued litigation. The issue of modifying existing systems to accommodate women and Two studies that compared national data minorities, however, raises important ques- on industries and occupations have shown tions about how promotion systems worked that rates of job integration are positively fir white men in the past. For example, associated with growth in total industry em- Medoff and Abraham (1980,1981) have ployment and female participation in the work questioned the assumption that the more force. Shaeffer and Lynton's (1979) study of productive workers are rewarded. They found large corporations indicated that the inthat earnings were positively related to ex- crease in women's employment was greater perience but not to rated job performance for each nontraditional occupational cate- for white male managers in two major com- gory in the industries thatxperienced panies. The definition and use of qualifica- greater increases in total employment be- tions ,iiould be carefully examined within tween 1970 and 1975. Similarly, Reubens each firm to discover how the system might and Reubens' (1979 4analysis of' census data be improved for all employees. for f wcupational ca ories in 1960 and 1970 showed that women made greater gains in .) nontraditional occupations where men were CONDITIONS; MANAGING also increasing in number. They concluded INTERVENTION STRATEGIES that "the fortunes of men and women ride in The ilionlentinn for job integration will tandem, and for both sexes the white-collar be sustained by the pressure of large num- and higher paid occupations have shown the bers of %vol with 11...ightenecl expecta- greatest exKansion- (p.123).Shaeffer and tions continuing to enter the labor fo,ce in Lvnton also timid that the increase in wom-

285 280 BRIGID O'FARRELL AND SIIARON L. HARLAN en's nontraditional employment was propor- conflict between seniority-based layoffs and tionately greater in relatively female-intensive affirmative action efforts should eventually industries. They proposed several plausible be resolved'by the Supreme Court. Unions, and interrelated explanations, including a in the meantime, have made several alter- higher average employment growth rate in native proposals for changing seniority rules female-intensive industries between 1970. and that would lessen the disproportionate eco- 1975, a larger pool of women available in those nomic impact of layoffs on recently hired industries, managements' experience working women and minority workers, but fewhave with women in a wider variety of jobs, the been implemented." less technical nature of the jobs, and women's Economic growth in the Sunbelt and loss familiarity with industry working conditions. of job opportunities in the Northeast have It is not always the case, however, that important implications for future strategies employment growth and a female-intensive to reduce job segregation (Norton, 1981). work force will reduce negative stereotypes Northrup and Larson's (1979) analysis of the of women in an industry. For`example, in results of the AT&T consent decree showed a study of female andmale managers in two that the southwestern and western regions retail firms, Harlan and Weiss (1981) found of the company accounted for 38 to 50 per- that the level of perceived sex bias was higher cent of the new openings in each of five in the firm with a faster growth rate and a major nontraditional occupational cate- larger increase in the proportion of women gories between 1973 and 1979. These two managers between 1974 and 1979.This pe- regions alone accounted for more than half riod of high growth, a 46 percent increase the new female entrants to a nontraditional in women managers, was followed by a sharp job that had been particularly difficult to business decline and a period of economic integrate, and they reported more success- instability that increased the anxiety of male ful recruitment of Hispanic workers and managers about promotion opportunities. managers than any other region. Theshift Although progress in increasing the number in AT&T employment to the Sunbelt is ex- of women managers was slower in the sec- pected to he even greater in the future. Thus, ond firm, a 20 percent increase in women despite the bleak einpy ment picture io managers, it was based onmoderate growth many parts of the country, geographical var- followed by a period of Lousiness stability. I larlan and Weiss concluded that the higher level of sex bias in the first company could have been caused by uncertainty in the eco- The United Auto Workers Cnion has proposed ex- tending indefinitely the tune that minorities iind women nomic environment and that it could ulti- are laid off (and tns eligible for recall as well asbenefits mately result in poorer future job oppor- such as insurance and vested vacation pay) rather thai tunities for women managers despite their limiting it to the equivalent tine they were actually rapid short-term progress. employed. When disriminathinin hiring has oc- Economic recession, especially in male- curred. the UAW' recommends front pay, that is full the wages and benefits for tu.nor...I ft'es and women wild( intensive industries such as steel, has laid off Incentives for voluntary early retirement are anticipated negative effect. It quickly halts another possible program. An alternative approach is or eliminates recent progress inhiring and the recent California unemployment insurance law promoting women. Deaux and Ullman (1983) which allows workers to reduce their work week to report that fewer women were employed in avec or brit- days and collect unemployment benefits fOr the re aaining workdays (UAW Admini.strunre the steel industry in 1983 than before the ter, 1975, 1976; Steinberg and ('ook. 1981). These and 1974 consent decree. This was due to sen- other alternatives are in need of further research and iority-based layoffs and the closing of entire discussion, which should be given high priority iw plants where all workers lost their jobs. The unions. employers, and the gilt eminent.

2E6 1013 IVIEGBATION STRATEGIESt. 281 iations offer' some promise fOr increased in- technological jobs developing in high-tech tegration on a regional basis. industries will be integrated. The AT&T ex- Rapid technological changes have very ample supports the need for further inquiry important long-term implications for job in- into the consequences of technological change tegration, particularly in telecommunica- for female employment in particular indus- tions and other industries where automation tries. More broadly, it stresses the urgency is reducing the total number of jobs while for managers to treat technological innova- creating new jobs requiring new skills. New tions and affirmative action efforts as planned female "ghettos" are developing as a result interventions in the firm that must be more of women moving into declining occupations closely coordinated if job resegregation is to traditionally held by men (Hacker, 1979; he avoided, Reubens and Reubens, 1979). Reubens and Reubens fOund that, although women made Management Practice greater gains in jobs that were also expanding for men, the increase of women in 27 of During the 1970s EEO-related issues 53 nontraditional occupations between 1960 gradually moved from the domain. of minor and l970 was due to slow growth or decline personnel officials to the level of corporate in the number of men in those jobs. An...policy set by top management and carried occupational shift of this type often reflects out by line managers. Corporate managers the displacement of skilled men by semi- and government officials rate an effective skilled or nnskilled women due to techno- internal administrative structure for setting logical innovations. policy and carrying out programs as ex- Data presented by Hacker (1979) and tremely important to success in integrating Northrup and Larson (1979) suggest that the jobs (U.S. Equal Employment Opportunity simultaneous occurrence of technological Commission, 1974; Stead, 1978; Gordon and changes and affirmative action efforts at AT&T Strober, 1975; Meyer and Lee, 1978; Shaef- will gradually lead to some traditionally male fer and Lynton, 1979; McLane, 1980). Five jobs becoming newly segregated into a components of this structure include top- smaller 1111111ber of less - skilled jobs for level commitment, line responsibility, cen- women. Hacker (p. 550)described the tralized accounting and control, resource al- movement of won ten into traditionally male location, and union involvement. craft jobs that were becoming obsolete even Despite the rapidly' changing image of as the affirmative action agreement wasbeing corporate presidents from rugged entrepre- implemt rated: neurs to committee members, job integration appears to be an area where the indi- A. women learned to climb poles, AT&T was vidual chief executive officer (CEO) can and Nhilt:lig1%) microwase and laser (Fiber Optic) does make a difference. Since the average transmission systems. As women learned to in- tenure of CEOs is only about 6 years, how- stall telephones. -clip and take- customer in- ever, continued commitment cannot be taken stallatio and phone ,tort, were markedly re- for granted. Commitment of the CEO, re- (Imam; the need fir installers. Framework is a gardless of the motivation, is crucial to the lob %viler(' women I aye made the success of intervention programs and cat) be i4reatest inroads. Franwssark is slated for total demonstrated in several ways. Strong policy automation. .11.:11cctric switching systems can ot nails eliminate most switchwork end all statements are an initial step, but they must Imine.sork Framework went from 20 percent be followed by the allocation of resources telltale in 1972 to 32 percent in 1973. (staff, money, facilities), the direct review of results, reports to the board of directors, itrelliains to He' Well whether the new and concrete examples set by hiring NVOHIV/1

287 282 BRIGID O'FARRELL AND SHARON L. HARLAN into the highest levels of the organization. One midwestern company seems to have For example, in his initial address as chief succeeded in transferring responsibility to executive' of Equitable Life Assprance in line managers. The first 3 years of the af- 1975, Coy Eklund commented on the de- firmative action program were conducted as sirability if advancing women in the orga- a personnel department activity.Progess was nization, and the following month he hosted minimal until the CEO. delegated respon- a summit conferenceof women from sibility to line managers, down to the first throughout the company. An advisory panel line supervisor. Those who did not meet of women meets on a regular basis with the objectives got smaller bonuses. "We now chief executive, and affirmative action goals have 45 women department managers, com- are part of the executiveappraisal process. pared with one when the program started. Although the company traditionally had 1 . .. Nine management was the turning female officer, there now are nearly 30, and point" (McLane, 1980:23). Providing incen- women serve on Equitable'sBoard of Di- tives is an important motivation, and several rectors (McLane, 1980). companies in the McLane study reported Top-level policy is ultimately carried out tieing performance to bonuses. The impor- by line managers. in their very early as- tance of line involvement in successfully im- sessments of affirmative actionplans, Lyle plementing other kinds of organizational (1973) and Heard (1975) noted the lack of change is well documented (Whyte, 1969). involvement by line managers and the heavy Failure to enlist the support of foremen and concentration of responsibility for meeting other managers directly responsible for ;pals on personnel staffs. Line managers business operations has undoubtedly slowed should be involved in planning the goals and the progress of job integration. programs they will beresponsible for im- There is a fine balance, however, be- plementing. Most of the authors recom- tween line responsibility and centralized mending line responsibility for affirmative management and control, as demonstrated action, however, have not realisticallyad- by the AT&T case. Managers reported that dressed the problems of gaining managers' both are very important. Two critical com- support (Wallace and LaMonde, 1977; Meyer ponents in maintaining the balance arethe and Lee, 1978). AT&T provides an example person heading the EEO programand the where the line supervisors' powers were data management system. According to. curtailed as part of the affirmative action McLane (1980), the EEO post is now re- agreement. The personnel officehad the garded by managers as a very demanding power to override thedecisions of line man- job requiring knowledge of the legal regu- agers by holding up Fromotions,vetoing latory process, quantitative skills for increas- promotion and hiring decisions, and inter ingly complex analysis of the work force, and discipiinarty matters involving the ability to work with a variety of individ- women and minorities. Managers atthe lower uals both internal and external to the or- end of the hierarchy may feel more imme- ganization. diately threatened by affirmative action ef- Accurate inforn!atiem about current em- forts and consequently be less supportive ployees and projections of future needs is (Meyer and Lee, 1978). For example, O'- essential for the EE() director's effective- Farrell (1977) found that fOremen acquired ness. Approximatelytwo-thirds of the com- an increased workload(which was unrec- panies in. the Shaeffer and 1.,nton (1979) ognized and uncompensated fOr by the com- study reported establishing overall.results- pany) because women we've hired without orientedmanagement planning and control adequate training. At the same. time, women systems for affirmative action, and fully one- were being promotedabove the first-line thir' of those said it was the single most supervisors who were training them, successful LEO action they had taken. Pro- 28 JOB INTEGRATION STRATEGIES 283

cedures for setting goals and measuring structures, job posting and bidding proce- progress have become ifteredsingly sophis- dures, seniority systems, training programs, ticated. General Electric has experimented job qualifications, and dispute7resolution with a series of mathematical models to de- procedures are subjects of collective bar- velop realistic EEO goals (Hayes, 1980). gaining as well as affirmative action efforts Churchill and Shrank (1976) have developed (Wesman, 1982; Newman and Wilson: 1981; a flow model that requires management to Steinberg and Cook, 1981; Wallace and identify job ladders, measure the current Driscoll, 1981; O'Farrell, 1980b; Ratner, race and sex mix olemployees, specify a 1980; The Women's Labor Project, 1980; degired hiring mix for minorities and women, Leshin, 1979; Hausman et al., 1977; Stone and develop promotion probabilities based and Baderschneider, 1974). on anticipated hiring and turnover rates.All Under Title VII, unions, like employers, of these efforts have generated more real- are prohibited from discriminatory prac- , istic estimates of the amount of time needed tices. Unions are held responsible with em- to change a firm's employment profile, par- ployers for discrimination caused by provi- ticularly given the low turnover in the high- sions in collective bargaining agreements,

est-rated management and blue-collar jobs. , and the union duty of fair representation in Underlying the success or failure of in- grievanCe handling is well established tervention strategies is the allocation of cor- (Steinberg and Cook, 1981; Wallace and porate resources. The best strategy will fail Driscoll, 1981). Newman and Wilson (1981) if it is not sufficiently supported. Meyer and articulated several ways in which govern- Lee (1978) found that the public utility comment agencies not only do not cooperate panies had different patterns of resource al- with unions but, in fact, also discourage location for different types of jobs. Some unions from pursuing charges of disci imi- companies placed the major thrust on pro- nation by their members. fessional and managerial positions, while Unions, however, like employers, have others balanced their between blue- responded diversely to government en- and white-collar areas. No company, how- forcement activities.Unfortunately,little ever, reported giving a major priority to in- attention has been paid to the role of labor tegrating blue-collar jobs, and they directly unions in facilitating intervention programs linked this difference in priority to success- (Wesman, 1982; Wallace and Driscoll, 1981; ful results. Of the companies surveyed that O'Farrell, 1980b). Most research on unions employed blue-collar workers, fewer than and equal employment opportunity policy '10 percent placed their primary emphasis hal focused on the discriminatory behavior on the blue=collar area compared with 44 of unions toward black workers and the re- percent that placed primary emphasis on sulting legal developments (Wallace and the professional and managerial jobs. Cor- Driscoll, 1981). This in part reflects federal respondingly, far less success was reported policy, holding unions equally responsible in the entry-level, semiskilled, and skilled with employers for discrimination (O'Far- blue-collar jobs. It appears that more prog- rell, 1980b; Steinberg and Cook, 1981). ress has been made for women in manage- Changes in federal policy toward unions ment than for women in blue-collar jobs, at were initiated in the late 1970s. In 1980 the least in part because more company effort U.S.. Equal Employment Opportunity and resources have been allocated to inte- Commission developed a policy recognizing grating management jobs. and encouraging the efforts of unions in the Finally, an important part of the admin- area of equal employment opportunity pol- istrative structure affecting job integration icy. At the same time, the U.S. Department in unionized firms is the collective bargain- of Labor proposed changes in the guidelines ing agreement. Developing or changing wage of the Office of Federal Contract Compli-

2 8 9 284 BRIGID O'FARRELL AND SHARON L. HARLAN

ance Programs (OFCCP)that would have take a more active role'in identifying discrim broadened union participation. For exam- inatory practices, informing workers about their ple, unions would be notified if a compliance rights, providing financial and legal assist- review was to take place in a company where ance, and offering moral support.Newman the unions are party to a collective bargain- and Wilson also concluded that discrimination ing agreement. Currently, however, the cannot be corrected exclusively through col- Commission's policy has not been imple- lective bargaining and called for increased mented, and the proposed regulation changes government-union collaboration. have been withdrawn. Yet union officers and staff are potentially CONCLUSIONS AND as important for integrating nonmanage- RECOMMENDATIONS ment jobs as are CEOs and line managers. Recent studies have found women to be This review of intervention programs be- generally satisfied with their unions (Ko- gan by distinguishing betweenthe pioneer chan, 1979) and to have positive attitudes and postpioneer eras. We then focused on toward local union policies and practices the current postpioneer era during which (Bohlander and Cook, 1982). In O'Farrell's companies make formal agreements and (1980a) study of one union local, women re- commit resources to redress the institu- spondents were generally satisfied with the tional causes of job segregation. But many union, and even dissatisfied women feltthat companies, or their departments, remain in they needed the union to represent their the pioneer era. They have no women in positions to management. Women identi- certain jobs or departments, and pioneer fied such issues as job upgrading and ma- women are filing complaintsof discrimina- ternity leave that they thought the union tion. The pioneer and postpioneer eras exist 'should bargain on, and they used union pro- .simultaneously, and a continued federal cedures such as filing grievances, voting in presence is likely to be necessaryfor some elections, and establishing a union women's time to initiate and keep companies actively committee. recruiting, hiring, and training women for At both the national and local levels, this nontraditional entry-level jobs. At the same particular union had been excluded frormaf- time, new initiatives by the government, firmative action negotiations between the companies, and unions are needed to meet company and the governmentdespite a strong the challenges presented by barriers to record supporting EEO programs and pol- women's advancement in nontraditional jobs. icies, including filing sex discrimination suits There are 10 key findings from our anal- on behalf of women workers.Union officials ysis that bear further considerationand and women members were unhappy with scrutiny in future evaluations pf corporate the national affirmative action agreement, intervention strategies. and this led to 3 mgre years of litigation by the union on behalf of its women members. 1. Federal EEO laws have been impor- Ultimately, these activities resulted in a tant in producing substantial changes inthe stronger affirmative action program. work forces of targeted firms. There is a The experiences in European countries growing acceptance of EEO principles by (Ratner;--1980) and research in the United corporate managers and a demonstrated States (Steinberg and Cook, 1981; O'Farrell, psychological impact on women's .willing- 1980b) have recommended a much stronger ness, and ability to presstheir demands for EEO role for unions in the future. Newman nontraditional jobs. and Wilson (1981) argued that because of 2. Most of the successful corporate intheir knowledge of plant practices and ac- tervention strategies for increasing job in- cess to employer informationunions could tegration have been in the areas of recruit-

290 JOB INTEGRATION STRATEGIES 285 ing, hiring, and trai ng women for previously 9. An effective internal administrative all-male entry-leve structure for planning EEO policy and im- 3. An effective recruitment strategy plementing intervention pk grams is essen- combines active external recruitment from tial for ultimate success in job integration. nontraditional sources with innovative in- The important elements of administrative ternal recruitment efforts that usually in- effectiveness are commitment from top ex- volve changes in company seniority syg- ecutives and line managers (which can be terns. Careful screening of applicants results facilitated by staff EEO training), a skillful in lower turnover, which furthers the long- EEO' manager, an 'accurate data manage- term goal of job integration. ment system, and allocation of sufficient cor- 4. Preplacement training and supple- porate resources for implementation and mentary courses for women in blue-collar monitoring. jobs are effective in overcoming women's 10. In firms with collective bargaining lack of technical education and experience, agreements, the cooperation of union staff in helping them to perform better on the and officers in eliminating barriers to job job, and in increasing their acceptance by integration (e.g., changes in job posting, male coworkers. outreach, qualification assessment, training 5. Special training for women in man- and seniority systems) is essential for achiev- agerial positions is generally not necessary, ing a strong and effective EEO policy. but women need to be integrated into the There is little doubt that the policies of formal aria informal sources of training tra- the Reagan administration are negatively af- ditionally available to men. fecting progress on reducing job segrega- 6. The exclusive focus on integrating tion. Lack of leadership at the U.S. Equal men's jobs is not a sufficient long-term strat- Employment Opportunity Commission and egy for reducing job segregationthe evi- the U.S. Department of Labor, proposed dence suggests that it may provide only a changes in guidelines for federal contrac- temporary solution. The increasing number tors, and reduced commitment to employ- of women being hired into entry-level jobs, ment. and training programs diminish the combined with bottlenecks in promotion op- pressure and resources for change. Current portunities, may lead to resegregation of the economic policies and high unemployment loweit-paying, least-prestigious men's jobs, limit new opportunities and atct recent resulting in new female "ghettos." gains. These concluding recommendations 7. The most effective strategies to en- attempt to address the realities of today within sure that women have equal promotion op- the context of the long-term goal of achiev- portunities in an organization are to make ing equal employment opportunity for initial job assignments that place women on women and men. Future programs and pol- career paths with 'high opportunities for ad- icies should include the following: vancement, to make temporary modifications in qualifications and seniority provi- federal support for the development of sions to meet affirmative action goals, and alternative EEO monitoring systems, the to develop methods for individual qualifi- involvement of labor unions in EEO nego- cation assessment and career planning that tiations, and upgrading women's jobs; increase opportunities for men and women. federa; support for skills training, infor- 8. To increase the chances for ultimate mation dissemination, and leadership de- success in reducing job segregation, com- velopment; panies should implement their intervention corporate improvement of human re- programs in areas of projected corporate source planning; growth and in coordination with long-term union programs to develop women lead- plans for technological innovation. ers and to identify EEO problems;

291 286 BRIGID O'FARRELL AND SHARON L. HARLAN

joint union and company initiatives to velop or implement agreements at the na- improve the terms of collective bargaining tional and local levels. The U.S. Depart- agreements for women workers; and ment of Labor should revise the OFCCP 'cooperatively planned and executed Guidelines, under considelation since 1981 longitudinal comparative research by the (Federal Register, 1981), to require notifi- government, companies, unions, and re- cation of unions when compliance reviews searchers. are conducted in firms withcollective bargaining agreements and to enable their voluntary cooperation. Federal Equal Employment Opportunity Company experiences in the postpioneer Policy, era suggest that integrating entry-level tra- Federal enforcement activities have had ditionally male jobs is an important but lim- a positive effect on reducing sex segregation ited approach. Reducing sex segregation will within the work force of some firms, yet the require integrating women's jobs as well. government cannot and indeed shouldn't The Commission should continue to de- supervise employment practices within firms. velop current strategies for integrating Rattier, federal policy must strengthen the women's jobs: establishing goals and time- incentives for change at the firm level, and tables (e.g., the AT&T consent decree); lim- there are at least three areas where the U.S. ited upgrading (e.g., the GE consent de- Equal Employment Opportunity Commis- cree), and pursuing affirmative action sion and the Office of Federal Contract negotiations in the area of equal pay for work Compliance Programs need to initiate ac- of comparable worth. Upgrading women's tion: monitoring, union involvement, and jobs not only improves working conditions integrating.traditionally women's jobs. for women but also will facilitate the in- &targeted monitoring plan, similar to the crease of men into these jobs. existing targeted enforcement plan (U.S. Equal Employment Opportunity Commis- Federal Training and Education Policy sion, 1982), should be d weloped and implemented by the Commission. A limited An important issue for reducing job seg- number of companies could be identified, regation involves how much outreach and based on carefully developed selection cri- training are needed and who should pay for teria, and comparative analysis could be itwomen, employers, or government; The conducted to reduce and refine the type of burden is currently on women and employ- data needed to measure compliance, to de- ers. Government subsidy for skills training velop technical assistance materials for a wide and worker education is lower in the United range of firms, and to improvethe process States than in almost any other industrial of establishing goals and timetables. country (Woodcock, 1977). Ytthe few gov- At a minimum, enforcement agencies ernment programs for integrating women should facilitate the active involvement of into nontraditional jobs have been some- labor unions in efforts to reduce job segre- what successful and need to be continued gation. The Commission should reaffirm and and expanded. Congress should increase the implement its 1980 policy statement that funding for federal training subsidies, main- recognizes and encourages the voluntary ef- tain targeting for women and women's pro- forts of unions in the area of equal employ- grams, and emphasize training for nontra- ment opportunity. Unions should be brought ditiona'. jobs. The U.S. Department of Labor into negotiations that affect collective bar- should disseminate information about suc- gaining agreements, and women members cessful techniques used by public and pri- should be included on committees that de- vate organizations to train and recruit women

2 9 2 OB INTEGRATION STRATEGIES 287

for nontraditional jobs and should act on Research Agenda proposals to increase financial incentives (e.g., tag exemptions) for employers to de- All of the current EEO policies, pro- velop and expand apprenticeship training grams, and proposed new initiatives are in with targeting for women. Finally, feder- need of more information. To measure the ally aided programs for worker education and effectiveness of program interventions on leadership training in EEO policy could reducing job segregation, it is essential to be initiated, for example, through subsi- have longitudinal analysis of employee ad- dies to community colleges and land-grant vancement within firms coupled with infor- institutions. mation about the changes in firm activities and procedures (Kanter, 1979; Rosenbaum, Management and Union Initiatives 1979). How important is init.ika job assign- Managers reported that EEO policies have ment? Do existing training programs malce had a positive effect In improving overall a difference in career advancement? How personnel policies and procedures. Affirm- can we identify formal and informal oppor- ative action must now move beyond for- tunity structures in the organization? How ' malizing, clarifying, and modifying existing do these factors interact with the education procedures to developing new and innova- and training 'individuals bring to the job or tive practices that result in and are part of acquire on the job? What are the effects of larger organizational changes. Corporations external economic factors? and business schools should expand their All of these questions can be systemati- research and development programs for ex- cally addressed within an organization that empt and nonexempt employees in the fol- has reasonably accurate, computerized per- lowing areas: systematic career develop- sonnel files. The methodological issues are ment and training, performance appraisal complex, however, and few organizations systems, impro ed communication and in- have the internal capabilities to design and formation sharing, job redesign, job rota- carry out such a comprehensive evaluation tion, decentralization of decision making, and that would enable them to effectively inter- the planning and implementation of tech- vene-rin the mobility patterns of the orga- nological changes. nization. There are also serious legal and To adequately represent women, unions competitive constraints on such analysis. The must undertake internal affirmative actions necessary research might best be done such as increased support for leadership through a cooperative effort jointly funded training forewomen members, developing and implemented by government, compa- EEO training for union officers, appointing nies, unions, and researchers. and hiring women in staff positions, encouraging women to run for union office, ACKNOWLEDGMENTS and requesting and analyzing EEO compliance data now available through recent court We would like to thank the committee decisions. Collective bargaining between and workshop participants for their helpful management and labor should include the comments, especially Barbara Bergmann, following subjects with specific concern for Alice Ilchman, and Barbara Reskin. their impact on women and minority workers: job posting and bidding procedures, REFERENCES seniority systems, grievance procedures, job redesign, introduction of technological Alvarez, Rodolfo, Kenneth G. Lutterman, and Asso- ciates changes, job evaluation and comparable 1979 Discrimination in Organizations. San Fran- worth, and nontraditional job training. cisco: Jossey-Bass.

293 I

288 BRIGID O'FARRELL AND SHARON L. HARLAN

McLane, Selecting, Developing and Retaining Ash, Philip Women Executives.. New York: Van Nostrand 1976 "The Testing Issue." Pp. 203-216 in Wallace (ed.), Equal Employment Opportunity and the Reinhold Co., 1980. AT&T Case. Cambridge; Mass,: MIT Press. Calm, Ann F. (ed.) New York: Barnhill-Hayes 1979 Women in the U.S . Labor Force. 1979 Employer Attitudes Toived. Affirmative Ac- Praeger. tion. Milwaukee,.: Wis.: Barnhill-Hayes. Re- Churchill, Neil C., and John K. Shrank ported in Linda It. Brown, Mb, Woman Man- 1976 "Affirmative Action and Guilt-Edged Goals." ager in the United States: AResearch Analysis Harvard Business Review March-April, pp. 111- and Bibliography." Washington, D.C.; Busi- 116. ness and Professional Women'sFoundation, Crowley, Joan E., Teresa E. Levitan, and Robert P. 1981. Quinn .Barrett, Nancy , 1973 "Facts and Fiction About American Working 1978 "Data Needs for Evaluating the Labor Market Women:win Robert P. Quinn et al., The 1969- Status of Women." Census BureauConference 70 Survey of Working Conditions (Final Re- on Issues in FederalStatistical Needs Relating port). Washington, D.C.: Employment Stand- To Women. Washington, D.C.: U.S. Census ards Administration, U.S. Department of La- Bureau. bor. 1979 "Women in the Job Market: Occupations, Cenningham, John Earnings, and Career Opportunities." Pp. 31- 1976 "Avoiding Common Pitfalls in Affirmative Ac- 62 in Ralph E. Smith, The Subtle Revolution. tion Programs." Personnel Journal March, pp. Washington, D.C.: The Urban Institute. 125-128. Bergmann, Barbara R. Deaux, Kay K. 1980 "The Contribution of Labor Market Data in Iri "Blue Collar Barriers." American Behavioral' Combating Employment Discrimination." Pp. press Scientist. 443-457 in Ronnie Ratner (ed.), Equal Em- Deaux, Kay K., and Joseph C. Ullman ployment Policy for Women. Philadelphia: In "Hard-Hatted Women: Reflections on Blue- Temple University Press, pressCollar Employment." In J. Bernardin (ed.), Bohlander, George W., and Suzanne Cook Women in the Workforce. New York: Praeger. 1982 "Women Unionists: Attitudes Toward Local 1983 Women of Steel: Female Blue-Collar Workers Union Policies." Labor Studies JournalVol, 7 in the Basic Steel Industry, New York: Prae- No. 2 (Fall), pp. 142-157. ger. uriggs, Norma Doeringer, Peter B., and Michael J. Piore 1979 "Apprenticeship." Pp. 225 -238 in Cahn (ed.). 1971 Internal Labor Markets and Manpower Anal- Women in the U.S. Labor Form New York: ysis. Lexington, Mass.: D.C. Heath. .Praeger. Douglas, Priscilla H. 1981 "Overcoming Barriers to Successful Entryand 1981 "Black Working Women." Doctoral disserta- Retention of Women in Traditionally Male tion, Graduate School of Education, Harvard Skilled Blue-Collar Trades in Wisconsin." Pp. University. 106-132 in Briggs and Foltman (eds.), Appren- Epstein, Cynthia F. ticeship Research. Ithaca, N.Y.: ILR Publica- 1981 Women in Law. New York; Basic Books. tions Div., CornellCornell University, 1981. Farley, Jennie F. Foltman (eds.) Briggs, Vernon B. and 1979 Affirmative Action and the Woman Worker: 1981 Aprenticeship Research: Emerging Findings and Guidelines for Personnel Management. New Publications Future Trends. Ithaca, N.Y.: ILR York: AMACOM. Division, Cornell University. Federal Register Brown, Charles 1981 Vol. 46, No. 164, :Proposed Rules," August 1981. "The Impact of Federal Equal Opportunity Policies: What We Do and Do Not Know." 25. Rockefeller Foundation Conference on Equal Feldberg, Roslyn L., and Evelyn N. Glenn Employment Opportunity. New York: Rock- 1979 "Male and Female: job Versus Gender Models in the Sociology of Work." Social Problems Vol. efeller Foundation. 26, pp. 524-538. Brown, Linda K. "The Woman Manager in the United States: Fernandez, John P. 1981 Racism and Sexism in Corporate Life: Chang- A Research Analysis andBibliography." Wash- 1981 ington, D.C.: Business and Professional Wom- ing Values in American Business. Lexington, Mass.: Lexington Books. en's Foundation. Foxley, Cecilia H. Burrows, Martha G. Developing Women Managers: What Needs to 1976 Locating, Recruiting and Employing Women: 1978 An Equal Opportunity Approach. Garrett Park. Be Done? New York: American Management Associations, 1978. Reported inHelen J. Md.: Garrett Park Press.

294. a o

lOB INTEGRATION STRATEGIES 289

Gordon, Francine, and Myra Strober (eds.) Jongeword, Dorothy, Dru Scott, et al. 1975 Bringing Women into Business, New York: 1973 Affirmative Action for Women, A Practical McGraw-Hill. Guide. Reading, Mass.: Addison-Wesley. Hacker, Sally L Kane, Roslyn D., and Jill Miller 1979 "Sex Stratification, Technological and Orga- 1981 "Women and Apprenticeship: A Study of Pro- nizational Change: A Longitudinal Case Study grams Designed to Facilitate Women s Partic- of AT&T." Social Problems Vol. 26, pp. 539- ipation In the Skilled Trades." Pp. 83-105 in 557. Briggs and Foltman (eds.) Apprenticeship Re- Hall, Francine S., and Maryann H. Albrecht search: Emerging Findings and Future Trends. 1979 The Management of Affirmative Action. Santa Ithaca, N.Y.: ILR Publications Division, Cor- Monica, Calif.: Goodyear Publishing Co. nell University. Harlan, Sharon L., and Brigid O'Farrell Kanter, Rosabeth M. 1982 "After the Pioneers: Prospects for Women in 1977 Men and Women of the Corporation. New York Nontraditional Slue-Collar Jobs." Work and Basic Books. Occupations Vol. 9, No. 3 (August) pp. 363- 1979 "Differential Access to Opportunity and Power." 386. Pp. 52-68 hi Alvarez et al. (eds.), Discrimi- Harlan, Anne, and Carol L. Weiss nation in Organizations. San Francisco: Jos- 1981 Moving Up: Women in Managerial Careers sey-Bass. (Fistal Report). Center for Aesearch on Women, 1980 'The Impact of Organization Structure: Models Wellesley College. and Metitods for Change." Pp. 311-327 in Rat- 1982 "Sex Differences in Factors Affecting Mana- ner (ed.), Equal Employment Policy for Women: gerial Career Advancement." Pp. 59-100 in Strategies for Implementation in the United Wallace (ed.), Women in the Workplace. Bos- States, Canada and Western Europe. Phila- ton: Auburn House. delphia: Temple University Press. Hausman, Leonard J., et al. (eds.) Kelley, Maryellen 1977 Equal Rights and Industrial Relations. Madi- 1982 "Discrimination in Seniority Systems: A Case son, Wis.: Industrial Relations Research As- Study." Industrial and Labor Relations Review sociation. Vol. 36, No. 1 (October), pp. 40-55. Hayes, Harold P. Kleiman, Bernard, and Carl B. Frankel 1980 Realism in EEO. New York: John Wiley & Sons. 1975 "Seniority Remedies Under Title VII: The Steel Heard, Jamie Consent DecreeA Union Perspective." 28th Annual Conference on Labor. New York: New 1975 "Availability and Use of Information on Cor- porate Equal Employment Programs." Wash- York University. ington, D.C.: Investor Responsibility Re- Kochan, Thomas A. search Center, Inc. 1979 "How American Workers View Labor Unions." Heidrick and Struggles, Inc. Monthly Labor Review (April), pp. 23-31. 1977 Profile of a Chief Personnel Officer. Reported Larwood, Laurie in Helen J. McLane, Selecting, Developing and 1977 Women in Management. Lexington, Mass.: Retaining Women Executives. New York: Van Lexington Books. Nostrand Reinhold, 1980. Leshin, Geraldine 1979 Profile of a Woman Officer. Reported in Helen 1979 Equal Employment Opportunity and Affirm- J. McLane, Selecting, Developing and Retain- ative Action in Labor Management Relations, ing Women Executives. New York: Van Nos- A Primer. Los Angeles: Institute of Industrial trand Reinhold, 1980. Relations, University of California. Hoffmann, Carl, and John S. Reid Loomis, Carol J. 1981 "Sex Discrimination?The XYZ Affair." The 1979 "AT&T in the Throes,, of Equal Employment." Public Interest No. 62 (Winter), pp. 21-39. Fortune (January 15).-' Hollander, James Lopez, Felix 1975 "A Step-by-Step Guide to Corporate Affirma- 1976 "The Bell System's Non-Management Person- tive Action." Business and Society Review (Fall), nel Selection Strategy." Pp. 217-242 in Wal- pp. 67-73. lace (ed.), Equal Employment Opportunity and lacobelli, John L., and Jan P. Muczyk the AT&T Case. Cambridge, Mass.: MIT Press. 1975 "Overlooked Talent Sources and Corporate Lyle, Jerolyn R. Strategies for Affirmative Action." Personnel 1973 Affirmative Action Programs for Women: A Journal (November), pp. 575-578. Survey of Innovative Programs. Washington, 1chniowski, Casey D.C.: U.S. Equal Employment Opportunity 1983 "Have AngelstDone More? The Steel Industry Commission. Consent Decree." Industrial and Labor Re- Malveaux, Julianne lations Review Vol. 36, No. 2' (January), pp. 1982 "Recent Trends in Occupational Segregation 182-198. by Race and Sex." National Research Counc.:1,

295 290 BRIGID O'FARRELL AND SHARON L. HARLAN

Workshop on Occupational Segregation by Race O'Farrell, Brigid, and Sharon L. Harlan and Sex. 1982 "Craftworkers and Clerks: The Effect of Male Coworker Hostility on Women's Satisfaction Marshall, Ray, L. Knapp, M. Liggett, and R. Glover with Nontraditional Jobs." Social Problems'Vol. 1978 Employment Discrimination: The Impact of Legal and Administrative Remedies. Austin: 29, No. 3 (Febluary), pp. 252-265. University of Texas. Pfeiffer, Donald, and Mary L. Walshok 1981 'Sortie Innovations in Industrial Apprentice- McLane, Helen J. ship at General- Motors: Locally Developed 1980 Selecting, Developing and Retaining Women Executives: A Corporate Strategy for the Competency-Based Training as a Tool for Af- Eighties. New York: Van Nostrand Reinhold firmative Action:" Pp. 173-184in Briggs and Foltman (eds.), Apprenticeship Research:' Co. Emerging Findings and Frau% Trends. Ithaca, . Medoff, James L., and Katharine G. Abraham N.Y.: ILR Publications Division, Cornell Uni- 1980 'Experience, Performance, and Earnings." The versity. Quarterly Journal of Economics Vol. XCV, No. 4 (December), pp. 703-736. Preston, Joan 1978 Presentation on Women and Teaching in New 1981'"Are Those Paid More Really More Produc- England. Wellesley, Mass.: Center for Re- tive? The Case of Experience, '.The Journal of search on Women, Wellesley College. Human Resources Vol. XVI, 2, pp. 186-216. Purcell, Theodore V. Meyer, Herbert, and Mary D. Lee 1976 "How G,E. Measures Managers in Fair Em- 1978 Women in Traditionally Male Jobs: The Ex- ployment." In Equal Opportunity in Business. perience of Ten Public Utility Companies (R&D Harvard Business Review (Reprint Series No, Monograph No. 65). Washington, D.C.: Em- 21132), pp. 102-107. ployment and Training Administration, U.S. Department of Labor. Ratner, Ronnie (ed.) 1980 Equal Employment Policy for Women: Strat- Moses, Joseph L., and Virginia R. Boehm egies for Implementation in the United States, 1975 "Relations! ,ip of Assessment-Center Perform- Canada and Western Europe. Philadelphia: ance to Management Progressof Women," Temple University Press. Journal of Applied Psychology Vol. 60, No. 4. Reubens, B.C., and, E.P. Reubens Newman, Winn, and Carole W. Wilson 1979 "Women Workers, Nontraditional Occupa- 1981 "The Union Role in Affirmative Action." Labor tions and Full Employment." Pp. 103-126 in Law Journal (June), pp. 323-342. Cahn (ed.), Women in the U.S. Labor Force. Northrup, Herbert R., and John A. Larson New York: Praeger. 1979 The Impact of the AT&T.1E0 Consent De- Rosenbaum, James E. Polley Series cree, Labor Relations and Public 1979 "Career Paths and,,Advancement Opportuni- No. 20. Philadelphia: Industrial Reseakh Unit, ,.ties." Pp. 69-87 in Alvarez et al. (eds.), Dis- The Wharton. School, University of Pennsyl- crimination in Organizations. San Francisco: vania. Jossey-Bass. Norton, Eleanor Holmes Schreiber, Carol T. 1881 "Affirmative Action and Labor Force Trends." 1979 Changing Places: Men and Women in Transi- Rockefeller Foundation Conference on Equal tional Occupations. Cambridge, Mass.: MIT Employment Opportunity. New York: Rock- Press. efeller Foundation. Shaeffer, Ruth G. O'Farrell, Brigid 1980 Nondiscrimination in Employment and Beyond 1977 "Affirmative Action for Women in Craft Jobs: (Report No. 782). New York: The Conference Change in the Small Industrial Work Group" Board. (Working Paper). Wellesley, Mass.: Center for Shaeffer, Ruth C., and Edith F. Lynton Research on Women, Wellesley College. 1979 Corporate Experiences in Improving Women's 1980a Women and Nontraditional jobs: A CaseStudy Job Opportunities (Report No. 755). New York of Local I (Final Report). Washington, D.C.: The Conference Board. Employment and Training Administration, U.S. Shuchat, Jo, with Gerii Guinier and Aileen Douglas Department of Labor. 1981 The Nuts and Bolts of NTO: A Handbook for 1980b "Equal Employment Opportunity, Womenand Recruitment, Training, Support Services, and Unions: An Industrial Union Alternative." Placement of Women in Nontraditional Oc- (Working Paper No. 54). Wellesley, Mass.: cupations. Cambridge, Mass.: Technical Ed- Center for Research on Women, Wellesley ucation Research Centers. College. Stead, B. 1982 "Women and Nontraditional Blue-Collar Jobs 1978 Women in Management. Englewood Cliffs, N.J.: in the 1980s: An Overview." Pp. 135-166 in Prentice-Hall. Wallace (ed.), Women in the Workplace. Bos- ton: Auburn House.

296 LB INTEGRATION STRATEGIES 291

Steinberg, Ronnie, and Alice Cook Guidebook for Employers, Vols. 1& 2. Wash- 1981 "Women, Unions and Equal Employment Op- ington, D.C.: U.S.. Equal Employment Op- portunity" (Working Paper No. 3). Albany, N.Y.: portunity Commission. Center for Women in Government, State Uni- 1980 "Title VII and Collective pargaining: Proposal versity of New York. and Resolution to encourage Voluntary M.- Stone, Morris, and Earl R. Baderschneider (eds.) * fox's." Washington, D.C.: U.S. Equal Em- 1974 Arbitration of Discrimination Grievances, A ployment Opportunity Commission, April. Case Book. New York: American Arbitration 1982 16th Annual Report, FY 1981. Washington, Association. D.C.: U.S. Equal Employment Opportunity Strober, M.H. Commission. 1982 The MBA: Same Passport to Success for Vetter, B., E. Babco, and S. Jensen-Fisher Women and Men?" Pp. 25-44 in Wallace (ed.), 1982 "Professional Women and Minorities: A Man- Women in the Workplace. Boston: Auburn power Data Resource Service." Washington, Houie. D.C.: Scientific Manpower Commission Swanson, Cheryl, and H. Brinton Milward (AAAS). Reported in Women Today, Vol. 12, 1979 "The Impact of Organizational Structure, No. 8, April 18, 1982. Technology and Professionalism on the Policy Wallace, Phyllis A. (ed.) of Affirmative Action." Pp. 41-56 in Marian L. 1976 Equal Employment Opportunity and the AT&T Palley and Michael B. Preston (eds.), Race, Sex Case Cambridge, Mass.: MIT Press. and Policy Problems. Lexington, Mass.: Lex- 1982 Women. in the Workplace. Boston: Auburn ington Books. House!' The Women's labor Project Wallace, Phyllis A., and James W. Driscoll 1980 A Guide to Collective Bargaining Solutions for 1981 "Social Issues in Collective Bargaining." Pp. Workplace ProblenitThat Particularly Affect 199-254 in Jack Stieber et al. (eds.). U.S. In- Women. San Francisco: San Francisco Bay Area dustrial Relations 1950-1940: A Critical As- National Lawyers Guild. o sessment. Madison, Wis.: Industrial Relations UAW Administrative Letter Research Association. 197,7,Vol. 27, No. 3, August 12. Wallace, Phyllis A., and Annette M. LaMond (ed.) 1976 Vol. 28, No. 7, July 14. 1977 Women, Minorities and Employment Discrim- Ullman. Joseph C., and Kay K. Deaux ination. Lexington, Mass.: Lexington Books. 1981 "Recent Efforts to Increase Female Partici- Walshok, Mary L. pation in Apprenticeship in the Basic Steel In- 1981 Blue Collar Women: Pioneers on the Male dustry in the Midwest." Pp. 1..-149 in Briggs Frontier. Garden City, N.Y.: Anchor Books. and Foltman (eds.), Apprenticeship Research. Wesnian, Elizabeth C. Ithaca, N.YILR Publications Div,, Cornell 1982 "Public Policies at Loggerheads: The Effect of University. Equal Employment*A Opportunity Legislation U.S. Commission on Civil Rights on Unions." Doctoral dissertation, Cornell 1981 Child Care and Equal Opportunity for Women. University, Ithaca, N.Y. Clearinghouse Publication No. 67, June. Whyte, William F. U.S. Department of Labor 1969 Organizational Behavior. Homewood, Ill.: 1978 "Women's Guide to Apprenticeship" (Office of Richard D. Irwin Inc. and the Dorsey Press. the Secretary, Women's Bureau). Washington, Wooecock, Leonard D.C.: U.S. Government Printing Office. 1977 "Call for a Coalition to Establish a National U.S. Equal Employment Opportunity Commission Extension Service." Labor Studies Journal Vol. 1974 Affirmative Action and Equal Employment, A 1, No. 3 (Winter), pp. 281-285.

297 a

Occupational Desegregation in CETA Programs

LINDA J. WAITE and °SUEE. BERRYMAN

This paper examines the occupational dis- the wage differential (Chiswick et al., 1974)., tributions in the Comprehensive Employ- Even when labor force attachment is con- ment and Training Act (CETA) andthe wage trolled, women alio have much flatter life- implications of these distributions for men, time earnings profiles than do men (Sawhill, mid women of different racial and ethnic 1973). Theoretical arguments (Wolf and Ro- origins. Our data come from two separate senfeld, 1978) and fragmentary evidence projects on CETA, both conducted for the (Barrett, 1979) implicate occupational seg- National Commission on Employment Pol- regation in these profile differences. Male icy. One (Berryman et al., 1981) assessed but not female occupations seem associated the nature and equity of men's and women's with career paths that carry wage advance- experiences in CETA, a substantial part of ment with experience. the study being devoted to CETA's occu- Second, poverty in the United States is pational desegregation record for women. increasingly female poverty, pri- The second (Berryman and Waite, 1982) as- madly as the result of the rising number of sessed ethnic and racial differences in CETA female-headed households and the relation- experiences, focusing on whites, blacks,. and ship between households of this kindr.and Hispanics and on Hispanic subgroups. poverty.' Thus, from the economic per- CETA's occupational desegregation rec- spective, the issue of occupations and wages ord for women is impdrtant for several rea- for women is not transitory. sons. First, one ofCETA's legislated purposes is to improve the economic prospects of its clients. As we know, substantially nme From 1969 to 1979 the percentage of female-headed female than male occupations pay poverty- households of all races increased by a third. For whites level wages (Sawhill, 1976). Persistent oc- and Hispanics the increase was about 25 percent: for blacks, over 40 percent. Although the chances that a cupational segregation parallels the persist- household of this kind was poor declined slightly over ent male-female wage differential,and dif- the decade, in 1979 they were still very high: 30 per- ferences in male and female occupational cent for all races and almost 50 percent for black female- distributions account for over a quarter of headed households (Bureau af the Census, 1981).

292 298 OCCUPATIONAL DESEGREGATION IN CETA PROGRAMS 293

Third, CETA has represented a major II, and VI.2 The major services available federal lever for affecting occupational de- under these titles were basic skills, job train- segregation for women and women's wages. ing, and jobs, although, as we describe later, From FY 1974 to FY 1980, 19 million in- not all services are available in all titles. For dividuals entered CETA in job training or example, basic skills and job training are employment capacities, somewhat fewer than essentially restricted to Title I. The pur- half of these being women. Thus, over time poses of the jobs also vary by title. Most CETA has had the potential for affecting the Title I jobs, called work experience,_are fn- occupational preferences and skills of large came transfer jobs3 that are not intended as numbers of women. a bridge to unsubsidized employment. Jobs Finally, CETA flows from early federal in Titles II and VI, known as public service manpower programs of the 1960s and can employment (PSE) jobs, are expected to lead be expected to affect future federal training to unsubsidized employment, although the and employment programs. Thus, even if economic environments in which these jobs CETA is virtually dismantled under the are offered presumably vary. Title II jobs Reagan administration, its occupational de- are available in ,areas with high, structural segregation record for women is of more unemployment; Title VI jobs, in areas with than historic interest. As we show later, short-term, Cyclical unemployment.' women's occupational options in CETA are The eligibility by title varied, although, affected by how CETA is structured and by as we discuss below, titles overlapped in how men and women are funneled through their eligibility requirements. All of the ti- this structure. Our experience with CETA tles had eligibility criteria, of economic dis- has implications for designing future pro- advantage, underemployment, or unem- grams that would increase women's expo- ployment. For Title I, eligibility was sure to occupations currently held mostly restricted to those economically disadvan- by men. taged or unemployed or underemployed.5 The paper has five sections. The first briefly describes CETA's legal structureits titles, their legislated purposes, and eligibil- These are the title numbers before the 1978 reauthorization of CETA; they correspond to the postreau- ity rules. The second describes the data base thorization numbers of IIB, IID, and VI. This paper used in the two studies that underlie this does not discuss Title III because most slots in this title paper. The third.lhows how the CETA title (Title ILIA or the Summer Youthprogram) are jobs of under which individuals enter CETA and short duration, Untended as a mechanism of income their CETA activity (e.g., work experience) transfer, and without a training component. 3 By "income transfer jobs" we mean jobs used pri- affect their occupational options. The fourth marily as a means of allocating money to people, not documents CETA's occupational desegre- as bridges to private or public se jobs unsubsidized gation record for white, black, and Hispanic by CETA. women; and the final section shows the wage__ 4 Title II was targeted on regions with lingering un- cons., quences of women's occupational dis- employment. Title VI was designed to reduce the presumably short-term unemployment associated with the tributions in CETA. recession of the mid-1970s. However, as Mirengoff and Kindler (1978) observe, the unemployment rate used to define an area's eligibility for Title II was surpassed in most places by that used to define an area's eligibility DESCRIPTION OF CETA TITLES AND for Title VI funds. Thus. de facto the distinction be- ELIGIBILITY REQUIREMENTS tween the two titles was eliminated. s To receive one of the small number of PSE jobs in For reasons of simplicity and of data re- Title i the individual had to be unemployed or under- strictions, we deal only with CETA Titles I, employed.

299 294 LINDA I WAITE AND SUE E. BERRY AN

For Title II, individuals had to reside in the usual sources, the prime sponsor' re- areas of substantialunemployment. They also porting system. Second, the CLMS is in- had to be unemployed for at least 30 days tended to measure the effect of CETA pro- prior to application or underemployed. grams on participants, includingearnings and Before January 1977 individuals were el- labor force status.8-- igible for Title VI if they had been unem- The CLMS comprises an initial intake in- ployed for 30 days, oritthey resided in an terview, an activity record, and severalother area with excessiveunemployment and had interviews during and after the CETA en- been undereMployed or unemployed for at rollment. In the initial interview, the CLMS least 15 days. After January 1977 the eligi- determines what CETA service the enrollee bility rules became more complicated. received (e.g., public service employment) However, in general, individuals could en- and, if the service was a job or job training, ter if they were (1) unemployed orunder- the enrollee's occupation andwages. The employed; or (2) a member of an economi- CLMS also obtains information on the en- cally disadvantaged family and either a rollee's attitudes toward marpoWer pro- member of an Aid to Families With De- grams and servicesreceived, what service pendent qildren (AFDC) family, or un- and dccupation the enrollee wanted from employed and an unemployment insurance CETA, his or her trade or vocational train- recipient, or unemployedand ineligiblefor, ing before entering CETA, veteran status, unemployment insurance, or unemployed marital status, number of dependents, fam- and an insurance exhaustee. ily composition, receipt of ,government transfer payments (food stamps, subsidized' housing, AFDC, Supplemental Security In- DATA come, unemploymentbenefits, and other

public assistance), the enrollee's employT 4.4 Continuous Longitudinal Manpower ment/schooling history in the previous year, Stirvey (CLMS) wages or'salary in the last year,and personal Both studies on which thispaper is based and family income by source. The CLMS used Continuous LongitudinalManpower contains information on the highest grade Survey (CLMS) data. The Bureau ofthe or year of regular Schoolthe enrollee at- Census has conducted the CLMS quarterly tended, whether that grade had been corn- since January '1975, samplingrespondents from the.. previouS quarter's new enrollees in CE FA. Respondents aresampled from public serv- 7 The federal government administers decentralized four CETA functional activities: CETA programs through administrative units called ice employment, employabilitydevelop- prime sponsors. Federal funds for these programs are ment, direct referrals,6 and youthwork ex- allocated to the prime sponsors. State, county, or local perience (including summer programs). governments can he prime sponsors if they govern a Th;> CLMS has two main .objectives. First, minimum population of 100,(X00. State governnients of the tend to become prime sponsors for governmental units it is designed to obtain characteristics within the state that do not meet the minimum pop- they re- CETA 'participants and the services ulation requirement: ceived, thus providing data not available from 8 The CLMSsponsored by the-Employment and Training Administrationsamples mainly decentralized CETA,programs, i.e., programs operated by (:ETA prime sponsors. Thys, special-purpose programs such In a direct r-terral, CFTA refers theindividual to as the Job Corps (Title IV, reauthorized asTitle IVI3), a job %aancy. Theindividual does not receive itny Young Adult Conservation Corps (Title VIII), and sev- other CETA services and does not necessarily get the eral Title III (reauthorized as various Title IV) programs job to which he or she' is referrs.d. are not included in the CLMSfile.

300 OCCUPATIONAL DESEGREGATION I.N CETA PROGRAMS 295 pleted by the time of CETA enrollment, and aggregate the sample by sex and race/eth- whether the enrollee had a high school nicity. For the October 1975 to March 1979 equivalency certificate or General Educa- period, the CLMS' contains approitimately tional Development(GED) certificate., Our 42,000 initial interviews. analysis relies especially on detailed data on the enrollee's ethnic origin or descent and Analytic Strategy on whether theenrollee was limited in the amount or type of work he or shecould do We agsessed the impact of race and eth- because of problems in speaking English. nicity on enrollees' experience in CETA in We use all of this inforniatiOn in various two ways.° First, we estimated a general sections of our analysis, either as dependent linear model of each CETA outcome sepa- or as independentvariables. rately for men and women in which we con- In our analyses we use only data fromthe trolled for all cluacteristics of the enrollee initial questionnaire and activity record, 'since and the enr011ifient that were relevant for our purpose is to assessthe services pro- CETA assignment." This model included a vided within CETA and not to assess the series of dummy variables for race/ethnicity: impact of CETA services on later outcomes. white, black, and Hispanic. Second, we per- We included all CETA enrolleessurveyed formed an analysis of covariance for each by the CLMS during the period October CETA outcome in which we tested for dif- 1975 through September 1978 in order to ference between' race/ethnic groups in the (1) maximize the period covered by' our slope coefficients in the model. analyses and (2) maximize the number of cases available for analysis.9In each quarter CETA AS A SYSTEM OF OFPCPTUNITIES the CLMS sampled between 3,500 and 4,000 CETA enrollees and completed initial in- We can think of CETA as, a system for terviews with 3,300 to 3,600. Tohave suf- distributing opportunities of several kinds: ficient numbers of observations for race/eth- (1) participation in CETA; (2) a CETA serv- nic groups by sex we pooled informationfor ice or activity basic education, job train- all quarters in the October 1975 to Septem- ing in a classroom setting, on-the-job train- ber 1978 or March 1970 time period (Bar- ing, work experience, and public service rett, 1979). Pooling observations across time employment; (3) an occupation for those in periods provides large sample sizes thatal-- jobs or job training; and (4) a CETA wage low us considerable flexibility in the types for those in jobs or job training. Since thig of analyses we do and that allow us to dis- paper focuses on CETA'scontribution to oc- ,cupational desegregation for women, CETA occupations are the resource of primary c on- :cern. However, to interpretthe data or oc- 9 We begin. with October 1975 because the CLMS did not record CETA title until the second quarter of FY 1976 (OctOber 1975). The samplefol. the multivariate analysis ends with March 1979because CETA was reauthorized in October 1978 and regulations govern- 10 We follow census definitions; persons of Hispanic ing the revisel act were released to prime sponsorsin origin may be of either race. We divide enrollees into April 1979. Since those enrolled in CETA in thethird whites (non-Hispanic). blacks (non-Hispanic), and His- and fourth quarters of 1979 entered underrevised panics of both races. We omit those of other races who guidelines, the data for these quarters are not com- are not Hispanic. pletely comparable with early data, and we eliminated " These included age. marital status; poverty status. them to ensure comparability. The sample for the cross- labor force experience, educationalattainrrispt. desired tabular analysis ends with September 1979 because CETA services, and problems with the English lan- data to this date only were available at the time this guageall at the time of enrollment, plus, for males, analysis was done (Berryman et al.. 1981). veteran status.

f. 3o1 296 LINDA]. WAITE AND SUE E. BERRYMAN cupations it is important to understand the experience activities, and these services oc- process by which a CETA participant is as- cur only- i1i.Title I. In sum, Titles II and VI signed an occupation, either in the form of imply a. public service job; Title I, a 'basic a job or job training. education, job training, or work experience An individual enters CETA under a title activity. If a CETA participant is only eli- and a CETA activity that is authorized for gible for Titles. II or III, his or her CETA that title. If the activity is job training or a activity is determined. If the participant is job, the individual is assigned to an occu- only eligible for Title I, his or her activity pation and receives a wage in connection options are constrained but not determined. with it. Eligibility rules determine if an in- As Table 16-1 shows, each CETA service dividual can enter CETA, and under what has a different occupational distribution and title he or she may enter. Although these therefore different occupational assignment rules vary for different CETA titles, indi- probabilities. All of the occupations avail- viduals can be eligible for more than one able in CETA are available in each of the title, giving CETA prime sponsors some dis- services, but the occupational emphases dif- cretion in their title assignments. fer for each CETA service. Relative to the Titles affect CETA service or activity as- distributions for the other services, class- signments in that not all CETA services are room training has the highest percentage of available in all titles. Titles II and VI consist clerical openings; on-the-job training, the only of public service employment (PSE) highest percentages of ctafts and operatives jobs, and almost all of these jobs occur in options; work experience, the highest per- these two titles. Title I consists primarily of centage of service jobs; and public service basic education, job training in a classroom employment, the highest percentages of setting, on-the-job training (OJT), and work professional/technical and laborer jobs.

TABLE 16-1CETA's FY 1976-FY 1979 Occupational Structure by CETA Activity (percent) CETA's Occupationd Structure All Public CETA Classroom Work Service Occupational Category Activities Training OJT ExperienceEmployment Professional/technical 10.7 6.9 4.9 6.7 15.8 Managerial/administrative 2.3 0.4 2.9 0.9 3.6 Sales workers 1.0, 1.2 3.7 0.9 0.3 Clerical 27.2 38.0 16.3 32.1 23:5 Crafts 12.0 20.3 21.5 6.7 10.3 Operatives 7.5 . 14.9 28.0 4.2 2.1 Transportation equipment operatives 2.7 1.2 3.6 2.1 3.3 Laborers 15.2 1.2 8.6 13.8 22.0 Service 21.5 15.8 10.4 32.8 19.0 Total" 100.0 100.0 100.0 . 100.0 1(X).0 N (000) 2,7704 389 319 790 1,272 These are the 1-digit census occupational categories. They exclude three categories that do not occur in the CETA occupational structure: Farmers and Farm Managers. Farm Laborers and Supervisors, and Private House- hold Workers. Columns may not add to 100 due to rounding. SOURCE: Table 31, Berryman and Waite (1982). p. 79. 302 o 0

OCCUPATIONAL DESEGREGATION IN CETA PROGRAMS 297

The process by which a CETA participant were men to receive basiceducation, job is assigned an occupation varies across time training, and work experience services and for any given CETA office and across CETA less likely to get public service jobs. offices at any given time. Any specific oc- Although sex affected title assigninent, cupational assignment reflects several fac- Berryman and Waite (1982) found few ef- tors: (1) the participant's title eligibility and ,fects and no important effects of race/ the subsequent activity and occupational ethnicity on the CETA title under which constraints that are associated with each ti- enrollees enter CETA. Whites of both sexes tle; (2) the participaiit's activity and occu- entered CETA under Titles I and II slightly pational needs and preferences; (3) the ac- more often than did blacks or Hispanicswith tivity and occupational vacancies available similar characteristics. But these differences at any given time as a function of the local never excel ed about 3 percentage points labor market, the CETA office's efforts to and, although statistically significant, were develop particular opportunities; and the hardly stibstrintively so. amount of federal money then available for As noted, Title I consists of several CETA different CETA titles; and (4) judgments by services: basic education in a classroom, job CETA assignment personnel about what training in a classroom setting, OJT, work kinds of occupations are appropriate for 'Wh4t experience, and a small number of PSE jobs. kinds of people. Again, multivariate analyses showed that race Our analyses show that in FY 1976 to FY and ethnicity had no or only trivial effects 1978, relative to their eligibility, women 18 on assignment to CETA services.However, to 65 years of age were underrepresented relative, to males in Title I, women in this in all CETA titles for all three fiscal years title were placed more frequently in class- except Title I in FY 1978. The underrepre-, room training andwork experience jobs and

, sentation varied from 64 to 87 percent, .de- less frequently in OJT and PSE jobs. Al- pending on fiscal year and title. Thus, women though the percentages declined across fis- did not receive CETA resources, including cal years, eves! in FY 1978 a third of all occupational experiences, at rates commen- wotnen in CETA were in Title I classroom surate with their eligibility. W,he discrepancy training. between eligibility and participation was Thus, relative to men's occupational op- greater for Titles II and VI than for Title1.'2 tions, women's options were more apt to be When we looked at how female and male thoselassociated with classroom training and CETA participants distributed across titles, work experience. They were less apt to be a logistic regression showedthat relative to those associated with OJT and PSE jobs. men with the sameplacement-relevant We would like to use multivariate anal- characteristics (see note 11), women were yses to assess CETA's occupational sex seg- more likely to enter CETAunder Title I regation for racial and ethnic groups. and-less likely to enter CETA under Titles ever, Berryman et al. (1981)did not conduct H and VI. Thus, they were more likely than multivariate analyses of occupational segregation by sex and race and ethnicity. Ber- ryman and Waite (1982) conductmultivar- iate analyses separately by sex and by race 12AvAable eligibility estimates for this time period are by sex and by race andother ethnicity separately. and ethnicity and have no direct measure of Thus. we cannot assess racial and ethnic differences in the sex composition of occupations for these women's CETA participation, relative to eligibility. We groups. The occupational measure used in can note that, relative to eligibility,whites are under- Berryman and Waite was occupational sta- represented and blacks are overrepresented in all three directly bear titles, and Hispanics are overrepresented in Title I and tus, a measure that does not underrepresented in Titles II and IV. on occupational segregation.However, we

S'03 4$

298 LINDA J. WAITE AND SUE E. BEARYMAN can use our multivariate results for the ef- TABLE 16-3Effects of Race and fects .of race and ethnicity on CETA title, Ethnicity on Hourly Wage of CETA Job CETA service, occupational status, and Training and CETA Job CETA wages to draw tentative inferences Unstandardifed Regression Coefficients, Net of So6al, about these effects on occupational segre- Economic, and Demographic Characteristics of the gation in CETA by race and ethnicity. CETA Entrant' We have already noted that there are few, Males Females if any, effects of race and ethnicity on CETA Job Training title and service assignments. Our analyses White .0001 .0225 of the impact of race/ethnicity on the oc- Black .0149 .05556 cupational status of CETA jobs and job train- job ing showed mixed results. As Table 16-2 White .0065 .0149b Black .0396b .0285h shows, we found no differences among white, black, and Hispanic males in status of job ° These characteristics are listed in note 11 of this paper. training, but we found lower occupational < .05. This indicates effects that would appear status for white and black than for Vispanic by chance less than 5 times out of every 100 analyses. females, net of other characteristicFor job SOURCE: Tables 46 and 47, Berryman and Waite status we found lower scores for whites and (1982), pp. 105 and 107. blacks of both sexes than for Hispanic. As before, the differences tended to be statistically significant but substantively unim- forced the conclusions we reached for oc- portant. The largest coefficient for race/eth- cupational status. Table 16-2 presents re- nic groups appeared for black females in sults of the regression of the Duncan occupational status of job training and equaled Socioeconomic Index of CETA job training 5 points on a 100-point scale, the Duncan. or CETA job on assignment-relevant char- Socioeconomic Index. acteristics (see note 11) of the individual. Our results for race/ethnic difkrences in Since these models omit the variable for the CETA wages, shown in Table 16-3, rein- "Hispanic" race/ethnicity category, the coefficients show the deviation of white and black occupational status from that for His- TABLE 16-2Effects of Race and panics, controlling for the other character- Ethnicity on Occupational Status of CETA istics of the enrollee. Among males in job Job Training and CETA Job training, we found no differences in wages, but among males in jobs, black males re- Unstandardized Regression Coefficients, Net of Social, ceived wages 4 percent lower than those of Economic, and Demographic Characteristics of the CETA Entrant° Hispanic and white males with comparable characteristics. For females, we found very Males Females small differenceson the order of 1 or 2 job training percentbut those that did exist favored White .5086 -3.31oob Black .5447 3.9040h I ispanics. Job The analyses of covarance allowed us to White 2.4285b 3.3197h test the hypothesis that the process which Black = 2.7244h - 4.9850 determines CETA occupational status and ° These characteristics are listed in note 11 of this wages depends on race/ethnicity. We found paper. evidence of some rather minor differencA. b p < .05. This indicates effets'that would appear These analyses showed different effects of by chance less than 5 times out of every 100 analyses. SOURCE: Tables 42 and 43. Berryman and Waite the variables in the models for race/ethnic (1982). pp. 97 and 99. groups on occupational status of CETA job

304 OCCUPATIONAL DESEGREGATION IN CETA PROGRAMS 299

training for males but not for females and stereotyping in employment and training. for status of CETA jobs for both males and We only had data for October 1975 to Sep- females. But few sizable differences ap- tember 1978 for these analyses. Thus, we x peared in individual coefficients in any of can only describeCETA's occupational 54 these models. regation record prior to the introduction of In sum, when we considered males and the desegregation directive and cannot as- females separately, we found small or no sess CETA's responses tothis directive. effects of race/ethnicity on CET,.. title, ac- At the same time, even prior to CETA's tivity, occupational status, or wages. The 1978 reauthorization, CETA - especially differences that existed tended to favor His- Title I - was expected to improve the eco- panics over blacks and whites. These results nomic prospects of its clients. Since female- suggest that, given an appropriate measure dominated occupations command lower of occupational segregation, we would have wages than those of nkixed andmale-domi- found that the process of occupational seg- nated occupations, it ii-feasonable to look regation did not depend on race and eth- for evidence that CETA tried to train and nicity. employ women in mixed and male occu- The remainder of this paper focuses on pations. CETA's occupational distribution and its wage In describing CETA's occupational de- implications by sex, without regard to race segregation record, we use the CETA's def- or ethnicity. Theconclusions from our mul- initions. In a male-dominated occupation fe- tivariate analysis of the impact of race/eth- males constitute less than 2.5 percent of that nicity on CETA experiences argue for this occupation's labor force; in amixedoccu- approach. In addition, analysis of two sexes pation, 25 to 74 percent; and in a female; and three racial/ethnic groups becomes too dominated occupation, 75 percent or more. cumbersome for the resulting small gain in Table 16-4 shows the distribution of CETA analytic detail. jobholders among male, female, and mixed CETA jobs by sex and race. For FY 1976 to FY 1978, although only about 10 percent of OCCUPATIONAL DESEGREGATION IN the women in CETA jobs (work experience CETA or PSE jobs) worked inmale-dominated jobs, Since FY 1974 millions of adult women CETA placed about 25 percent in mixed have participated in CETA. In connection occupations. Data published elsewhere show with the reauthorization of CETA in Octo- that CETA's occupational desegregation ber 1978, CETA regulations directed state record for jobholders improved across the and local CETA administrators to reduce sex three fiscal years, the percentage of adult

TABLE 16-4Distribution of FY 1976-FY1978 CETA jobholders by Sex Composition of Occupation and Sex (percent) Sex and Race/Ethnicity Female Male Sex Composition of Occupation Total White Black Hispanic Total White 'Black Hispanic Male-dominated 10.8 10.9 11.9 6.2 71.1 71.4 70.3 70.1 Female-dominated 64.1 62.8 64.4 74.4 8.3 8.0 8.2 11.0 Mixed 25.1 26.3 23.7 19.4 20.7 20.6 21.4 18.9 Total 100.0 100,0 100.0 100.0 100.0 100.0 100.0 100.0 407.838 150,568 47,078 IV 401,176 256,073 115,261 29,642 605.484 SOURCE: Table 9, Berryman et al. (1981), p. 31. 305 O s

300 LINDA 1. WAITE AND SUE E. BERRYMAN

TABLE 16-5Distribution of FY 1976-FY 1978 CETA Trainees by Sex Composition of Occupation and Sex '(percent) Sex and Race/Ethnicity Female Male Sex Composition of Occupation Total White Black HispanicTotal White Black Hispanic t Male-dominated 11.6 12.6 9.5 9.1 65.9 68.1 63.3 57.1 Female-dominated 49.3 46.2 55.tf 55.8 4.1 3.0 6.8 6.5 Mixed 39.1 41.2 34.7 35.1 30.1 28.9 30.0 36.4

Total 100.0 100.0 . 100.0 100.0 100.0 100.0 100.0 100.0 N 56,264 38,030 13,359 13,792 104,828 74,169 16,867 4,875 SOURCE: To Lle 11, Berryman et al. (1981), p. 38. women employed in male-dominated CETA composition of the occupation in which the jobs increasing from 7 to almost 12, the per- person is trained. As noted earlier, the causal centage in female-dominated CETA jobs de- relationships between activity and occupa- creasing from 68 to 62, and the percentage tional assignments vary: an activity assign- in mixed jobs remaining stable (Berryman ment may precede an occupational assign- et al., 1981). Adult females showed slightly ment, or vice versa, and in some cases both more distributional change across time than may be simultaneously determined by a third that for adult males, but neither sex showed factor, such as title eligibility. Without ad- large change.. dressing causality, we can note that women CETA's occupational desegregation rein classroom training were 60 percent more cord in job training may be a better test of likely to be trained in a sex-typical occu- its desegregation success than is its record pation and about 60 percent less likely to for jobholders. Since clients in job training be trained in a mixed occupation than were presumably lack human capital in any spe- women in on-the-job training. Although cific occupation, CETA's occupational as- classroom training assignments reduced fe- signments should be less constrained by male chances of being trained in a male- clients' prior occupational investments. More dominated occupation, the effects were not importantly, training can provide women with as 'great for this as for the other two occu- skills and credentials to enter male-domi- pational types. nated occupations. The data reveal that women in on-the-job Table 16-5 shows the distribution of those training were more likely to be trained in in CETA job training among male-domi- mixed and male-dominated occupations pri- nated, female-dominated, and mixed occu- marily as a function of OJT's occupational pations. Although CETA trained about the mix. OJT contains much larger proportions same percentage of women in male-domi- of male-dominated and mixed occupations nated occupations as it employed in those than does classroom training. Although occupations, it did train higher percentages women were substantially overrepresented in mixed occupations, reducing the per- in the female-dominated occupations in OJT, centage in female-dominated occupations to the smaller numbers of female-dominated a little under 50 percent. occupational slots in OJT produced some As the data in Table i6-1 suggest and data occupational desegregation. These data in- published elsewhere (Berryman et al., 1981) dicate that if CETA increases women's OJT show, where training occurs (in a classroom participation, iit should simultaneously in- or on the job) is clearly related to the sex crease occupational desegregation for women.

306 OCCUPATIONAL DESEGREGATION IN CETA PROGRAMS 301

TABLE 16-6Occupation of Last Pre-CETA Job by Occupation of CETA job for Males and Females (FY 1976-FY 1978) (percent) Occupation in Pre-CETA job Female Male Occupation of Male- Female- Male- Female - CETA Job DominatedDominatedMixed DominatedDominatedMixed Male-dominated Job 37.6 6.8 9.9 84.0 39.4 54.0 Female-dominated job 43.4 75.8 44.8 4.2 37.2 7.6 Mixed job 19.0 17.4 45.3 11.9 23.4 38.4 Total' t 100.0 100.0 100.0 100.0 100.0 100.0 N (000) 24 138 83 279 27 114 SOURCE: Tables 13 and 14, Berryman et af. (1981), pp. 36-37.

Table 16-6 shows whether adult female type of proportionately more females than and male CETA jobholders stayed in the of males who had pre-CETA occupations same occupational type as their last pre- typical for their sex. For those with pre- CETA job or moved to a new one. Thus, CETA mixed occupations or occupations this table shows how much CETA changed atypical for their sex, CETA retained the participants' occupational patterns. same or a higher percentage of females than About 75 percent of adult females in fe- of males in CETA occupations of the same male-dominated pre-CETA jobs entered fe- type. However, CETA shifted only one- male-dominated CETA jobs. Of those who quarter of those females in female-domi- moved out of female-dominated pre-CETA nated pre-CETA occupations into mixed or jobs, more than two-thirds entered mixed male-dominated occupations. It did not re- CETA jobs. tain even half of those women in pre-CETA CETA retained less than 40 percent of mixed or sex-atypical occupations in occu- adult females whose pre-CETA job was in pations of the same type and placed most of a male-dominated occupation intheir pre- the changers in female-dominated occupa- CETA occupational type and placed more tions, not mixed or male-dominated occu- than 40 percent in female occupations. For pations. females who had pre-CETA mixed jobs, Finally, we can ask about CETA's record CETA retained 45 percent in the same oc- in meeting clients' occupational prefer- cupational type and placed more than 40 ences,as expressed in terms of its sex com- percent in female-dominated occupations. position. '3 The data on occupational pref- Adult males had patterns similar but not erences should be treated with caution. identical to those of their female counter- Participants answered the preference ques- parts; where CETA assignment altered oc- tion after they had enrolled in CETA, and cupation it tended to move both males and most had been assigned to an occupation. females to occupations dominated by the Their responses may be biased in the di- same sex. A smaller percentof males than rection of their postenrollment occupational of females shifted out of sex-typical pre-CETA jobs (16 and 24 percent, respectively). Males shifted out of sex-atypical pre-CETA jobs at almost the same rate as that of females; they 13 The occupational preference data came from questions on the CLMS that asked: "Did you want a certain shifted out of mixed occupations at some- kind of (Job/job training) when you visited the man- what higher rates. power office?" [If Yes] "What was the (job/job training) In sum, CETA changed the occupational thit you wanted?"

347 3U2 LINDA 1. E AND SUE E. BERRY ALIN

TABLE 16-7Distributionof Desired Occupation by Obtained Occupation for Male and Female CETA Jobholders(FY 1976-FY 1978) (percent) Desired Occupation Female Male Occupation of Male- Female- Male- Female- CETA Job DominatedDominated Mixed DominatedDominatedMixed Male-dominated job 41.6 6.1 9.7 84.6 31.9 50.1 Female-dominated job 40.5 77.9 43.4 4.2 43.9 7.6 Mixed job 17.9 16.0 46.8 11.3 24.2 42.3 Total 100.0 100.0 100.0 100.0 100.0 100.0 N (000) 22 140 81 253 26 104 SOURCE: Table 17, Berryman et al. (1981), p. 42. assignments. If they had no pre-enrollment or mixed jobs got them. Of the females who p. eferences, they may haveresponded to wanted and failed to get sex-atypical jobs, this question by naming their assigned CETA 69 percent ended up in female-dominated occupation. If they were assigned to an oc- jobs. Similarly, of the females who wabted cupation different from their preference, they but did not get mixed jobs, 82 percent ended may have accommodated to thediscrepancy up in female jobs. More than 75 percent of by modifying their original preference. Both the women who wanted female jobs got them; of these potential biases would produce and of those who failed to get desired female overestimates of the match between pre- jobs, almost three-quarters got mixed, not ferred and actual assigninent. As such, our male, jobs. data on the match between preferred and In sum, from FY 1976 to FY 1978 CETA actual occupational assignments represent employed or trained fewer than half of its the maximum responsiveness of CETA to female participants in male-dominated or clients' preferences. mixed occupations. The percentages in- In each fiscal year more than half of the creased across fiscal years and were higher adult female respondents indicated that they in on-the-job training than in CETA's class- had had occupational preferences at the time room training or job services. Relative to of CETA entry. 14 For those women who ex- their representation in the particular CETA pressed preferences, a small but increasing service, females in on-the-job training were proportion wanted male-dominated jobs much more likely to be assigned to female- across time (5 percent to 10 percent). An dominated occupations than were females increasing proportion (from 26 percent to 35 in classroom training. OJT's better occu- percent) wanted mixed jobs, and a declining pational desegregation record was attribut- majority (from 69 percent in FY 1976 to 55 able to the small number of female occu- percent in FY 1978) wanted jobs in female- pational slots in that activity. For women dominated occupations. whose pre-CETA job had been a male or Table 16-7 shows the CETA occupational mixed occupation, CETA employed fewer distribution of adult females relative to their than half in occupations of the same sex- preferences at CETA entry. Fewer than half compoSition type, shifting almost half of the of the females who wanted male-dominated "movers" into female occupations. For women whose pre-CETA job had been a female-dominated occupation, CETA shifted 14 The percentages were 65, 57, and 59 for FY 1976, 25 percent to a mixed or male occupation FY 1977, and FY 1978, respectively. primarily to the former. Finally, for women

308 OCCUPATIONAL DESEGREGATION IN CETA PROGRAMS 303

who had occupational preferences at CETA TABLE 16-8Average Hourly Wage of entry, the match between preferred and ac- CETA Occupations by Sex (FY 1976-FY tual CETA occupation was much higher for 1979) (constant collars) those with preferences for female-domi- N N nated jobs than for those with preferences Occupation (000) Males(000) Females for male or mixed jobs. Professionall technical (136) 3.56 (133)3.39 Managerial/ WAGE IMPLICATIONS OF CETA administrative (39)3.78 (21) 3.56 OCCUPATIONS Sales workers (11) 3.24 (12) 2.49 Clerical (98) 3.05 (537) 2.69 The low wages of female-dominated oc- Crafts (249)3.25 (20) 2.72 cupations are ol.e of the primary reasons for Operatives (121) 3.19 (43) 2.67 trying to desegregate occupationsfor women. `Transportation From this perspective wages are the critical equipment operatives (63)3.04 (7) 2.75 basis for judging women's occupational ex- Laborers (380) 2.97 (32) 2.71 periences in CETA. We examine their wages Service (326) 2.86 (217) 2.54 during CETA and the wages paid in the Average (1,422) 3.10 (1,023) 2.76 labor force as a whole for the CETA occu- SOURCE: Table 37, Berryman and Waite (1982), p. pation in which they trained or were em- 89. ployed.

(Berryman and Waite, 1982). As the last col- In-CETA Wages umn of Table 16-9 shows, therank order of We assess sex differences in the CETA wages by CETA activity is: public service wage implications of female CETA occu- employment > on-the-job training > work pational assignments in three ways: by 1- experience >. classroom training. Wenoted digit census occupational codes, the CETA earlier that females are more apt than males service, and the sex - typicality of the occu-, are to be funneled intoTitle I. Once in Title pation. I, they are more apt than males are to be Table 16-8 shows the real average hourly funneled into classroom training and work CETA wage by sex for the 1-digit census experience activities. Thus, a much larger occupational codes. Without exception males percentage of CETA females than of CETA earn higher hourlywages' than do females males are in the two activities (work expe- in the same occupational category. '5 rience and classroom training) that receive Table 16-9 shows the real average hourly the lowest CETA wages. wage by sex and CETA activity.Both sex Table 16-10 shows that women in CETA and CETA activity affect CETA wages. If training and in CETA jobs received lower we look at wages y sex forthe same CETA wages than men did in each of thethree sex- activity, males again get systematically higher composition occupational categories (Ber- wages than do females: Theeffects of CETA ryman et al.,1981). The wage difference activity are the same for males andfe- between the sexes was greatest for the females and, as data published elsewhere male-dominated occupations, less and about show, for whites, blacks, and Hispanics the same size in the male-dominated and mixed occupations. Women in CETA job training received somewhat lower hourly wages if they trained 15 The large sample sizes make tests of significance they relatively uninformative. We examine wages for dif- in n female-dominated occupation than if ferences large enough to be signifLunt substantively. trained in either a male-dominated or mixed

309 (304 LINDA I. WAITE AND SUE E. BERRYMAN

TABLE 16-9Average Hourly Wage by CETA Activity and Sex (FY 1976-FY 1979) (constant dollars) N N N CETA Activity (000) Males (000) Females (000) Total Classroom training (210) 2.23 (297) 2.06 (507) 2.13 OJT (207) 3.22 (109) 2.70 (316) 3.04 Work experience (398) 2.51 (400) 2.38 (797) 2.45 Public service employment (794) 3.30 (476) 3.09 (1,270) 3.24 SOURCE: Table 44, Berryman and Waite (1982), p. 102. occupation. However, training in a female- "made higher wages than women did, and d aminated occupation did not reduce the for two reasons the sniallness of the differ- wages of men relative to the wages of those ence between them may be less impressive training in male-dominated and mixed oc- than it initially appears. First, CETA wages cupations. were subject to floors and ceilings, thus Women in CETA jobs received the lowest compressing the wage range for both sexes. wage rates in female-dominated occupations Second, howeverwe.categorize CETA oc- and the hig6st in mixed occupations. Al- cupations, participants in the same CETA though men in CETA jobs also received the activity or CETA occupation were probably highest wage rates in mixed occupations, more homogeneous even on unmeasured working in a female occupation did not de- characteristics that affect wages than were press their wages relative to the wages as- members of an occupation in the general sociated with male occupations. labor force. However we categorize CETA occupations - by census, code, CETA activity, or Post-CETA Wages sex-typicality - within each category worn- on's wages were on average about 90 per- We do not know the relationship between cent of the men's wages. It is not clear how the occupation of the CETA job or job train- to judge this wage record. Although the av- ing and that of participants' post-CETA jobs. erage difference between male and female However, if CETA clients train or work in wages in CETA was small, men consistently occupations whose counterparts in the labor

TABLE 16-10Average Hourly Wage for CETA Trainees and CETA Jobholders by Sex Typicality of Occupation (FY 1976-FY 1978) (constant dollars) Female CETA Activity/Sex Male Typicality of Occupation Total White Black Hispanic Total White Black Hispanic Trainees Male-dominated 3.63 3.66 3.61 3.46 3.00. 3.04 2.96 2.82 Female- dominated 3.52 3.64 3.39 3.46 2.79 2.77 2.78 2.93 Mixed 3.32 3.37 3.19 3.25 2.89 2.88 2.94 2.94 N (000) 103 70 16 13 56 37 13 5 Jo' Holders Male-dominated 3.34 3.43 3.11 3.25 3.12 3.21 2.95 3.18 Female-dominated 3.34 3.41 3.25 3.10 2.90 2.94 2.84 2.79 Mixed 3.53 3.60 3.36 3.42 3.35 3.39 3.18 3.45 N (000) 609 396 145 48 407 251 112 29 SOURCE: Tables 20 and 21, Berryman et al. (1981), pp. 50-51.

310 OCCUPATIONAL DESEGREGATION IN CETA PROGRAMS 305

TABLE 16-11FY 1976-FY 1979 CETA Occupational Distribution by Sex and CE1,1 Activity and the 1979 Unemployment Rates and Median Wages of Occupations in the Unsubsidized Sector Males Females Unsubsidized Sector 1979 Median Weekly Earnings' 4 CETA CETA CETA CETA 1979 (Full-time Wage and Ciccupational Training Job Training Job Unemploy- Salary Workers) Category' (percent) (percent) (percent) (percent) ment Ratesb (dollars) Piofessional/ technical 5.6 10.8 6.4, 14.4 2.4 316 Managerial/ administrative 2.0 2.8 1.0 2.2 2.1 349 Sales workers 2.5 0.3 2.1 0.8 3.9 254 Clerical 6.1 7.2 53.3 53.5 4.6 195 Crafts 35.5 14.4 4.3 1.4 4.5 303 Operatives 28.7 4.0 11.9 1.4 8.4 211 Transportation equipment operatives 4.0 4.4 0.3 G.7 5.4 272 Laborers 7.7 30.1 1.0 3.5 10.8 206 Service 7.9 25.9 19.6 22.1 7.3 164 Total or average 100.0 100.0 100.0 100.0 5.8 244 N 375,246 1,189,725 332,945 872,307 ° These are the 1-digit census occupational categories. They exclude three categories that do not occur in the CETA occupational structure: Farmers and Farm Managers, Farm Laborers and Supervisors, and Private House- hold Workers. b From Table A-23, Department of Labor (1980), p. 257. From Table 704, Bureau of the Census (1980), p. 424.

market as a whole have high unemployment Femalesin CETA job training had about rates, they should have less chance of cap- the same occupational distribution as that of italizing on then CETA occupational expe- females in CETA jobs. About 75 percent of rience. If the labor market counterparts of the women in each of these activities fell their CETA -,ccupations have low wages and into two occupations: clerical and service, CETA clients obtain a post-CETA Job in the both with low wages rates in the labor mar- same occupation as their CETA occupation, ket as a whole. The service occupation also their wages will be low. had relatively higher unemployment rates. Table 16-11 shows how CETA males and Males in CETA job training had different females distributed across the 1-digit census occupational distributions than those of males occupational codes by CETA service (train- in CETA jobs. Of those in CETA jobs, more ing and jobs). It also shows the 1979 un- than CO percent fell into two occupations: employment rates and median weekly earn- laborer and service, both with low wage rates ings for these occupations in the labor market and relatively high unemployment rates. For as a whole. The occupations with the highest males in CETA job training, almost two- 1979 unemployment rates were the opera- thirds fell into two different occupations: crafts tive, laborer, and service occupations; those and operatives. The former had a moderate with the lowest median weekly wage rates unemployment rate and relatively high wage were the clerical, operative, laborer, and rate; the latter, a relatively high unemploy- service occupations. ment rate and low wage rate. 306 LINDAJ. WAITE AND SUE E. BERRYMAN

Thus, from FY 1976 to FY 1979 CETA Title I for FY 1978. When we controlled on employed most women in occupations with variables that should affect title and activity low wages in the labor market as a whole. assignments, we found women overrepre- CETA did not use training to alter the pro, sented in training activities, especially in portion of women in occupationsoffering classroom training relative to OJT, and in relatively little economic security. CETA income-transfer jobs relative to jobs de- employed most men in two of the least eco- signed to lead to permanent, unsubsidized nomically secure occupations with relatively employment. low median wages and high annual employ- For all three years, female CETA partic- ment rates. However, CETA used training ipants were concentrated in female-domi- to reduce the percentage of men in the four nated occupations, although the concentra- low-wage occupations from two-thirds to one- tion was less among CETA job trainees than half. among CETA jobholders. CETA placed only about 40 percent of the women whose last pre-CETA occupation had been male-dom- Conclusion inated or who had, expressed a preference Training or working in male-dominated or at CETA entry for. a male-dominated oc- mixed occupations gave womenligher CETA cupation. In both cases, for those not placed wages than those from training orworking in a male-dominated occupation, about two- in female-dominated occupations. How- thirds were placed in sex-typical occupa- ever, CETA wages were consistentlylower tions. for women than for men in the same census However we categorized the CETA oc- occupation, in the same CETA service, or cupation, within each category women's in the same sex-.:omposition category. wages were about 90 percent of men's wages. Of those in CETA jobs, CETA employed Although the wage difference between men 80 percent of the women and 67 percent of and women was not large, it was consistent, the men in the four occupations whose un- and for reasons discussed earlier in this pa- subsidized counterparts had the lowest wages per, the smallness of the difference may be and/or high unemployment rates. For those less impressive than it appears. In terms of in CETA training, CETA did not alter the their post-CETA prospects, about three- percentage of women in lower wage occu- fourths of the women in CETA jobs and in pations, but reduced the percentage of men CETA job training were employed or trained in these occupations from 67 to 50 percent. in occupations that paid low wages in the however we judge CETA's occupational general labor market: service and clerical desegregation record, the bottom line of that occupations. record for women their CETA wages and The policy implications of these data are post-CETA economic prospects is not not clear for three major reasons: impressive. 1. In connection with CETA's reauthorization, CETA prime sponsors were di- SUMMARY AND POLICY IMPLICATIONS rected to reduce occupational sex segre- The analyses reported in this paper are gation in CETA. Our analyses provide abase - useful as baseline information about women line for assessing CETA's response to that and CETA for the three years prior to directive but not its current occupational CETA's 1978 reauthorization. We found that, status. relative to then eligibility for CETA, women 2. One of the reasons for desegregating were underrepresented inCETA's Titles I, CETA occupations was to improve women's 11, and VI for all three fiscal years except in post-CETA wages. At this juncture we lack

312 10. OCCUPATIONAL DESEGREGATION IN CETA PROGRAMS 307 analyses that show that being trained or em- that job programs for women should care- ployed in male-dominated CETA occupa- fully assess their goals and the ways in which tions positively affects women's post-CETA the sex composition of the occupation affects economic outcomes. If the effects aie neg- the chances of achieving those goals. ative for reasons substantially beyond the control of CETA, trade-offs between these two policy objectives reduced occupational sex segregation and economic self-suf- REFERENCES'' ficiency have to be made. Barrett, Nancy S. 3. The male-dominated occupations for 1979 "Women in the Job Market: Occupations, which most female CETA participants will Earnings, and Career Opportunities." P. 39 in be eligible are blue-collar occupations. Ralph E. Smith, ed., The Subtle Revolution. Structural changes in the American econ- Washington, D.C.: The Urban Institute Press. Berryman, Sue E., and Linda J. Waite omy and the occupational consequences 1982 Hispanics and CETA: Isaias of Access, Dis- of these changes imply that we need to tribution, and Eqt:ity. Unpublished working re-examine what occupations 4ex-typical paper, Rand Corporation, Santa Monica, Calif. or atypical best equip CETA participants Berryman, Sue E., Winston i{. Chow, and Robert M. ,Bell for economic self-sufficiency. 1981 CETA: Is It. Equitable for WartinnP N-1683- NCEP.. Santa Monica, Calif.; Rand Corpora- The avowed purpose of job programs, in- tion. Bureau of the Census cluding CETA, is to improve the prospects 1980 Statistical Abstract of the United States: 1980 of those who lack the skills to obtain ac- (101st ed.) Washington, D.C.: U.S. Govern- ceptable enployment on their own. This ment Printing Office. ° decent wages. 1981 Current Population Reports, Series P-20, No. means getting people jobs at 363.PopulationProfile ;he United States, Moving women into jobs currently filled 1980. Washington, D.C.: U.S. Department of predominately by men is desirable to the Commerce. Available from the U.S. Super- intendent of Documents, U.S. Government extent that it serves this purpose. But a' Printing Office. number of factors may decrease the utility Chiswick, Barry, J. Fackler, June O'Neill, and Solomon of occupational desegregation as a means to Polacheck the ends espoused by job programs. First, 1974 "The Effect of Occupation on Race and Sex Differences in Hourly Earning." Pp. 219-228 most women eligible to participate in job in Proceedings of the American Statistical As- programs could enter white-collar occupa- ) sociation. Washington, D.C.: American Sta- tions only through stereotypically female jobs tistical Association. such as clerical work. The male-dominated Department of Labor 1980 Employment and Training Report of the Pres- jobs potentially available to then tend to be ident. Washington, D.C. blue collar, primarily st,!rvice, operative and, Mirengoff, William, and Lester Rindler perhaps, crafts jobs. Many of these occu- 1978 CETA: Manpower Programs Under Local Control. Washington, D.C.: National Acad- pations show high rates of unemployment emy of Sciences. currently, and women seeking to enter them Sawhill, Isabel V. would face competition from large numbers 1973 "The Economics of Discrimination Against of men. Second, the structural changes now `Women: Some New Findings." Journal of Hu- taking place in the economy make unskilled man Resources 8: 383-396. 1976 "Discrimination and Pot,erty Among Women and semiskilled blue-collar jobs especially Who Held Families.* Pp. 201-211 in Martha susceptible to technical obsolescence. Third, Blaxall and Barbara Reagan, eds.. Women and little empirical evidence exists on the suc- thi Workplace. Chicago: University of Chicago Press: cess of occupational integration as a mech- Wolf, Wendy C., and Rachel Rosenfa anism for improving the employment pros- 1978 "Sex Structure of Occupations and Job Mobil- pects of women. For these reasons, we argue ity." Social Forces 56(3):823-844.

313 r.

Commentary

WENDY C.WOLF

The paper by Waite and Berrynian deals to the time that this "new" legislation took with the Comprehensive EMployment and effect. The relevant question to be answered Training Act (CETA) system and its record is, therefore, how was CETA doing before in providing access to nontraditional jobs. this additional emphasis was put on services . CETA represented a potential area,of fed- to women. eral intervention but also a Potential force Another critical point to consider is that that may have perpetuated sex-segregated Berryman and Waite often mention public job choice. It also represented a system in service employment (PSE). PSE has been which there had been considerable effort out of vogue for a while. For this reason, during the last decade to improve services the focus of my comments is on classroom to women and, in fact, to improve nontra- training, on-the-job training, and adult work ditional optiods for women. When inspect- experience.

ing this system, one could look at the rec- . Even prior to the 1978 amendmpts, which ordthat is, have the kinds of services were designed to encourage the expansion improved (i.e., offering nontraditional' op- of nontraditional opportunities for women, tions)? What is the impact of this system there. were slightly increasing proportions itself on women and men? of women moving into nontraditional, Before commenting on this paper, it is within CETA and into nontraditional train- critical to consider a little history. In 1978 ing,riespite the fact that, within CETA, the there were considerable changes in the CETA overall proportion of women being exposed legislation to make it more responsive to the to nontraditional options was not high. But needs of women, both in terms of serving the CETA systeip was not keeping pace with them in numbers relative to their propor- women's changing aspirations. An insAc- tion in the eligible population and in trying tion of the aspirations oewomen for nontra- to increase their nontraditional options. Un- ditional work reveals they were rising at a fortunately, the data used by Berryman and faster rate than was the opening up of non- Waite came from the pre-1978 period. So traditional career options within CETA. It they are looking at the CETA system prior is interesting to note that the aspirations

.308 314 $

COMMENTARY 309 among women for nontraditional work were I live one minor caution about Berryman growing (which I think is very divergent from and Waite's analysis. They talk about as- the evidence Marini and Brinton present in signment either to on-the-job training or Chapter 11 about high school girls) but that classroom training and how that increases the whole system wasn't changing fast enough or decreases one's likelihood, of being in or to accommodate these changes. getting into a sex-typed job. One has to be Berryman" and Waite show the relation- careful about assuming directionality be- ship between aspiration and the type of tween the two. The fact is that clerical train- service received. Of the women who had ing occurs in the classroom.° So if a CETA traditionally female, aspirations, 77 percent participant is going into clerical training, he were placed in female-typed jobs; of those or she is assigned to classroom training. So with 'male aspirations, 41.5 percent were I don't really think that you can treat one placed in ma' e-typed jobs, and 40 percent as exogenous and one as endogenous. They into female-typed jobs; of those with "mixed" are jointly determined. . aspirations, 46.8 percent were placed in This paper shows some potential for change mixed occupations, and 43 percent into fe- in the CETA system, especially since it was male- typed occupations. This suggests that done in the preamendment days. It also. if one enters the CETA system with non- shows that the CETA system has helped traditional aspirations, one is likely to get perpetuate the status quo in terms of oc- funneled into female-typed jobs anywaj,. cupational segregation. The Berryman and Waite paper makes a It is important to realize that Berryman big deal about wagesI am less likely to and Waite describe CETA before the 1978 be so excited about this issue. The National amendments. From 1978 to 1983, specific Research Council's Committee on Evalua- language was added to the law to encourage tion of Employment and Training Programs sex equity and the movementiof women into once produced a table showing male-female nontraditional jobs. It should be noted that differences in wages in :CETA. It showed in the new Job Training Partnership Act, that women earned 90 percent of what men very little proscriptive language is included earned in CETA. This is a bit misleading, to help legislate fair and equitable treatment lowever, in part due to the faCt that within for women. CETA there was a floor and a ceiling on wages, thus little variation.

315 t

Concluding Remarks

FRANCINE D. BLAU

Employment' segregation by sex is one of ,sensitive to the level of aggregation of the the most persistent and pervasive charac- data. Even detailed (three-digit) census oc- teristics of the labqr market. The papers in cupational categories group some typically this volume help us to understand the di- male and some typically female Jobs into mensions of such segregation, its funda- apparently integrated categories. To .the ex- mental causes, and its consequences for teLt that 'men and women workers in the women's economic status. They also shed same occupational categories are segregated some light on the effectiveness of policy in- by firm, aggregation 'acrosefirms will result terventions. in an miderestimate of the magnitude of sex In Chapter 2, Andrea Beller gives `us the segregation in employment. Using estab- heartening news that, on the basis of anal- lishment-level data and defining segregation ysis of Current Population Survey data, the in terms of the employer's own job cate- tendency of men and women to be segre- gories, Bielby and Baron find a striking pat- gated by occupation declined noticeably over tern of complete sex segregation by occu- the 1970s; particularly. among younger pation at the establishment level for a majority women. Moreover, this decline was at a rate of firms (including some single-sex enter- nearly three times that of the extremely small prises) and an extremely high level of seg- decline that occurred during the 1960s. We regation for the others. Their work raises must remember, however, that the mag- the question of whether estimates of trends nitude of segregation remains high. Fur-. in the magnitude of occupational segrega- thermole, although women increased their tion based on such microdata might differ share of many traditionally male managerial from Beller's estimates on the basis of ag- and professional occupations during the gregate data. Less than a quarter of the firms 1970s, this was not the case for heavily male in Bielby and Bardn's longitudinal sample craft and operative jobs. experienced any declines in the degree of In Chapter 3 Bielby and Baron demon- occupational sex segregation during the late strate that estimates of the magnitude of sex 4960s and early 1970s. segregation in employment are extremely These are the contours of occupational

310 316 '.

CONCLUDING REMARKS 311

segregation by sex in the U.S. labor market. mitigate our concern over the issue of oc- In assessing the significance of occupational cupational sex segregation. First, the mag- segregation, we must know more-abou its nitude of movement is fairly small that is, causes and consequences. From a policy most women workers did not change the sex perspective, an understanding of the con- label of their jobs over the periods analyzed. sequences is crucial for assessing how im- Moreover, it is dithcult to judge the mag- portant a problem it is, while an analysis of nitude of ,the probability of such moves in its causes helps us determine the most ef- the abstract. We need to know more about 'fective tools for attacking it. the desire for. such quiv" es on the part of The studies I reviewed in Chapter 7 sug- workers and how easy it is for women rel- gest that occupational segregation does re- ative to men to move out of (generally lower- duce the earnings of women, although there paying) female jobs into (generally higher- are considerable problems inprecisely es- paying) male jobs. It is instructive in this timating the magnitude of thig effect. A par- regard that both studies find the probability ticular problem is that overcrowding in the of a man moving from a predominantly fe- female sector may adversely affect women's male to a predominantly male job to be con- wages in the male sector as well. This im- siderably higher Than the probability of a parts a downward bias to, estimates of the woman making such a move. Furthermore, wage consequences of segregation based on women whose previous jobs were predom- comparisons of wommis wages in female and inantly male were much more likely to change male occupations. It is also important to point the sex label of their jobs than women whose out that the focus on earnings does not take previous jobs were predominantly female. into account the possibly negative nonpe- Second, Bielby and Baron's findings raise cuniary consequences' of sex segregation in the questidn of how much of what appear employment for women. Intuitively one feels to shifts in the sex labels of jobs as mea- that occupational segregation reinforces cul- sured by aggregate data actually entail's tural notions that men and women differ changes in the sex label of the individual's fundamentally in capabilities, preferences, job at the establishment level. 9. 'land social and economic roles. The causes of occupationa1 segregation are Our appraisal of the seriousness of the often classified in terms of supply- versus problem of occupational segregation may demand-side factors. The major supply-side depend in part on the ease with which women.explanations considered in this volume are are able to move between the male andfe- sex-role socialization and the human capital male sectors. The studies in Chapters 4 and model. In Chapter 11, Marini and Brinton 10 do find some movement of women from describe how the socialization process influ- st predominantly female to predominantly male ences the occupational orientation of men jobs and vice versa. In Chapter 4, Rosenfeld and women as well as the role they see mar- found that such changes in the sex labels of ket work playing in their lives. It is the latter occupations were experienced by about one- difference between men and women that is third of women job changers over a one- emphasized by the human capital explana- year period, while in Chapter 10 Corcoran, tion critiqued by Corcoran et al. According Duncan, and Ponza found that about one- to this view, because women anticipate third dell employed women made such shifts shorter and less continuous work lives than over a five-year period. Thesefindings sug- men, it will be in their economic self-inter- gest that the male and female sectors ought est to choose female occupations, which re- not tole viewed as watertight compart- quire smaller human capital investments and ments, but they do not in my opinion greatly have lower wage penaltitc for time spent

317 312 FRANCINE D. BLAU out of the labor force. Their ownresearch lustrated by a discussion of the impact of a and that of others summarized by Corcoran few government policies. On one hand, as ,et al. does not tend to support the human Roos and Reskin point out, governmental capital ,model: In particular, women with income tax and social security policy tend discontinuous work careers appear to be no to discourage female labor force participa- more likely to work atfemale jobs than tion. To the extent that the human capital women with more continuouswork -expe- explanation. has merit, the resulting de- rience. Nor did their findings suggest that crease in women's work lives would increase the selection of female jobs is consistent with the likelihood of their entering traditionally a strategy of maximizinglifetime' income, female jobs. Furthermore, while govern- given shorter, more disrupted work lives. ment training programs provide an oppor- On the demand side, I point out in Chap- tunity for intervention to reduce segregation ter 7 that employers 'may be motivated to by training individuals for sex-atypical jobs, exclude women from particular jobs because Waite and Berryznan's research suggests in of the belief that they would be less stable Chapter .16 that occupational training under or productive workers than men. Evenif the Comprehensive Employment and not initially correct, such views canbecome -.Training Act (CETA) is sex segregated to a self-fulfilling prophesies if women are then great extent. given fewer incentives than men to become On the other hand, of course, the gov- stable, productive workers. Roos and Reskin ernment administers an impressive array of in Chapter 13 emphasize institutional fac- antidiscrimination legislation and regula- tors in their review of a variety ofbarriers tion's. There is.some debate over the effec- to female employment in traditionallymale tiveness of these activities, and it is un- jobs at four points in the job allocation pro- doubtedly difficult to measure governmental cess: pre-employment training, accessand impact in this area. But it is significant that assignment to jobs., mobility, and retention. O'Farrell and Harlan report in Chapter 15 The operation of these barriers does not rely that governmental pressure was an impor- on conscious, overtdiscriminatory acts on tant factor promoting change within some the part of employers. Rather, the everyday firms. They also find that the v....nployinent operation of the system works against female growth of firms provides opportunities for employment in traditionally male jobs. A integration and thus also facilitates change. clearer understanding of the functions ,these The government's macroeconomic policies institutional mechanisms serve is of long- and their impact on overall business con- run importance indevising effective strat- ditions are therefore another way in which egies to remove'these obstacles to women's government may have an effect on the em- advancement. ployment opportunities of womenthe While the papers in this volume do not quality of jobs obtained as well as the prob- endeavor to fully evaluate the impact of fed- ability of finding a job at all. eral government intervention in this area. While all movements toward occupational an interesting view emergesof what such integration should be welcome, it is impor- an evaluation wouldentail. A variety of ways tant to realize that the movement of women in which the government potentially influ- into male jobs does not always bring women ences women's economic status areidenti. significantly closer to economic parity with fled. Some of these effects are positive and men. For one thing,occupational sex seg- some are negative. Anoverall assessment of regation may be replaced by female en- the government's impact would necessitate claves at the lower levels of male job lad- identifying the net effect of all its many pol- dersa process O'Farrell and Harlan term icies and programs. This point may beil- resegregation. For another, O'Farrell and

318 CONCLUDING REMARKS 313

Harlan note that woinen may gain access to also a major factor. Such discrimination is male jobs' ust as they are becoming tech- deeply entrenched: Within the workplace, nologically obsolete. Indeed, Strober argues a myriad of institutional mechanisms work in Chapter 8 that, in general, women are to perpetuate segregation of work along sex restricted to the leftovers, the jobs that men lines without requiring overt, conscious acts do not want. of discrimination on the part of employers. Sex segregation in employment remains, While some government policies work to a pervasive feature of the labor market and reduce sex segregation in employment, others a major cause of women's lower earnings. actually help perpetuate it. Thus, change While sex differe'nces in socialization and will not be easyyet such change is essen- the voluntary choices that women make in tial if we are to move substantially closer to their selection of jobs may play a role in the goal of economic parity between women producing sex segregation in employment, and men in the labor market. labor market discrimination is undoubtedly

.16

319