Using Self-Reports to Predict Student Performance. Research Monograph No

DOCUMENT RESUME

ED 126 116 TN 005 371

AUTHOR Baird, Leonard L. TITLE Using Self-Reports to Predict Student Performance. Research Monograph No. 7. INSTITUTION College Entrance Examination Board, New York, N.Y. SPONS AGENCY EXXON Education Foundation, New York, N.Y. REPORT NO CEEB-BM-7 PUB DATE 76 NOTE 97p. AVAILABLE FRO?; College Board Publication Orders, Eos 2815, Princeton, New Jersey 08540 ($5.00)

BDRS PRICE MF-$0.83 Plus Postage. HC Not Available from MS. DESCRIPTORS Academic Achievement; Activism; Adults; Background; *Biographical Inventories; College Attendance; College Students; Creativity; *Credibility; Educational Experience; Elementary Secondary Education; Grade Prediction; Grades (Scholastic); High Achievers; Higher Education; High School Students; *Literature Reviews; Negro Students; Occupational Choice; Predictive Measurement; *Predictive Validity; Predictor Variables; Psychometrics; Questionnaires; *Self Evaluation; Socioeconomic Status; Statistical Analysis; Student Characteristics; *Test Validity

ABSTRACT This review examines the accuracy and the . concurrent and predictive validity of brief self-report information, and evaluates the promise and problems involved in its practical use. In section one, the pc_:er of self-report information for predicting and understanding grades is reviewed. In the second section, research on the influence of background and personal information en students' progress through higher education is summarized. The third section deals with the prediction of creative or high level accomplishment. The fourth section reviews the increase in the understanding of students' educational backgrounds that biographical information provides. In the fifth section, the power of biographical information to predict vocational choices and vocational success is reviewed. The sixth section is a review of the technical and theoretical requirements of effective biographical and self-report information systems. In the last section, the general functions of biographical data are considered. Throughout the review, emphasis is placed on predictive validity of the measures in educational settings because the power of a variable to predict later performance or choices is necessary before the variable can be used for guidance or sectioning. However, wherever warranted, other aspects of the use of self-report information are also discussed. '(RC)

Documents acquired by ERIC include many informal unpublished materials notavailable from other sources. ERIC makes every effort to obtain the best copy available. Nevertheless, items of marginalreproducibility are often encountered and this affects the quality of the microfiche and hardcopy reproductions ERIC makesavailable via the ERIC Document Reproduction Service (EDRS). EDRS is not responsible for the quality of the original document.Reproductions supplied by EDRS are the best that can be made from the original. DOCUMENT RESUME

ED 126 116 TM 005 371

AUTHOR Baird, Leonard L. TITLE Using Self-Reports to Predict Student Performance. Research Monograph No. 7. INSTITUTION College Entrance Examination Board, New York, N.Y. SPONS AGENCY EXXON Education Foundation, Ner York, N.Y. REPORT NO CEEB-RM-7 PUB DATE 76 NOTE 97p. AVAILABLE PROM College Board Publication Orders, Eos 2815, Princeton, Ner Jersey 08540 ($5.00)

EDRS PRICE MP -$0.83 Plus Postage. HC Not Available from EDRS. DESCRIPTORS Academic Achievement; Activism; Adults; Background; *Biographical Inventories; College Attendance; College Students; Creativity; *Credibility; Educational Experience; Elementary Secondary Education; Grade Prediction; Grades (Scholastic); High Achievers; Higher Education; High School Students; *Literature Reviews; Negro Students; Occupational Choice; Predictive Measurement; *Predictive Validity; Predictor Variables; Psychometrics; Questionnaires; *Self Evaluation; Socioeconomic Status; Statistical Analysis; Student Characteristics; *Test Validity

ABSTRACT This review examines the accuracy and the concurrent and predictive validity of brief self-report information, and evaluates the promise and problems involved in its practical use. In section one, the pc :er of self-report information for predicting and understanding grades is reviewed. In the second section, research on the influence of background and personal information cn students' progress through higher education is summarized. The third section deals with the prediction of creative or high level accomplishment. The fourth section reviews the increase in the understanding of students' educational backgrounds that biographical information provides. In the fifth section, the power of biographical information to predict vocational choices and vocational success is reviewed. The sixth section is a review of the technical and theoretical requirements of effective biographical and self-report information systems. In the last section, the general functions of biographical data are considered. Throughout the review, emphasis is placed on predictive validity of the measures in educational settings because the power of a variable to predict later performance or choices is necessary before the variable can be used for guidance or sectioning. However, wherever warranted, other aspects of the use of self-report information are also discussed. (RC)

Documents acquired by ERIC include many informal unpublished materials not available from other sources.ERIC makes every effort to obtain the best copy available. Nevertheless, items of marginal reproducibility are oftenencountered and this affects the quality of the microfiche and hardcopy reproductions ERIC makes available via the ERIC DocumentReproduction Service (EDRS). EDRS is not responsible for the quality of the original document. Reproductions supplied byEDRS are the best that can be made from the original. Using Self-Reports to Predict Student Performance

U.S. ONPARTNIIF HT OF HEALTH. "PERMISSION TO REPROOUCE THIS EDUCATION WELFARE COPYRIGHTED MATERIAL BY, MICRO. NATIONAL INSTITUTE OF HE ONLY HA SEEOR NTEO sY EDUCATION

THIS DOCUMENT HAS BEEN REPRO- .., Arillrf TO ERIC ANO ORGANIZATIONS OPERAT OUCED EXACTLY AS RECEIVED FROM INGUNOER AGREEMENTS WITH THE NA THE PERSON OR ORGANIZATION ORIGIN- TIONAL INSTITUTE OF EOUCATION ATING IT POINTS OF VIEW OR OPINIONS FURTHER R EPROOUCTION OUTSIOE STATED DO NOT NECESSARILY REPRE- THE ERIC SYSTEM REQUIRES PERMIS SENT OFFICIAL NATIONAL INSTITUTE OF SION OF THE COPYRIGHT OWNER " EOUCATION POSIT ION OR POLICY Using Self-Reports to Predict Student Performance Research Monograph Number 7

by Leonard L. Baird Senior Research Psychologist Educational Testing Service

College Entrance Examination Board, New York, 1976

3 Editorial inquiries concerning this book should be addressed to: Editorial Office, College Entrance Exami- nation Board, 888 Seventh Avenue, New York, New York 10019. Copies of this book may be ordered from College Board Publication Orders. Box 2815. Princeton. New Jersey 08540. The price is $5.

Cover photograph by Glenn FOSS

Copyright C) 1976 by College Entrance '-'xamination 13o.tr(1. New YorIcrA11-rightleret4erved Library or Congress Catalog Number: 76.4312 1' tinted in the United States of America

4 Publication of t his monogaph has been made possible by a grant from the Exxon Education Foundation

I The theoretical requirements of efficient systems of brief-assessment information .64 Statistical methods appropriate for brief-assessment information 66 Summary 71 Section VII: Conclusions: The general functions of brief-assessment measures . 72 Broadband assessment 72 Selection and placement 73 Providing information for counseling and guidance 74 Serving as a teaching and guidance instrument for students completing the biographical questionnaire 74 Identifying students with particular characteristics or needs- . 75 Helping colleges in their planning and administrative activities 75 Providing the basis of two-way systems of information and action 75 Monitoring and understanding the transition from high school to schools, colleges and jobs 76 Conclusion 77 Reference list 78

VI Contents College students 31 High school students 32 Predicting creativity and Introduction I high level accomplishments among The validity of direct measure per se. 2 scientists and other adult groups from Section I: Predicting and understanding biographical records of accomplishment and activity 33 academic criteria 5 Some questions about direct assessment of Self-reported grades 5 biographical accomplishment scales. 35 Self-reported grades as predictors . . 9 Self- predicted academic performance and Implicaticns of these results 36 Summary 39 self-concepts of academic ability . . . . 10 Prediction of grades Section IV: Biographical information as an by general biographical information. 12 assessment of students' educational Practical implications 14 backgrounds 40 Section II: Biographical and SES 40 self-report information in the sequence Black students 44 of higher education 18 Section V: The use of College attendance 18 biographical inforination to predict The influence of biographical factors on and understand vocational choice 47 college selection 20 The validity of expressed interest as a Biographical factors in choosing measure of vocational choice 47 a residence 21 Classification systems 51 Biographical factors in choosing a major The use of biographical information and in vocational choice 22 for understanding vocational choice 53 Biographical information as a predictor of Prediction of vocational success 54 attrition in college 26 Biographical information related to Section VI: Requirements of an effective brief-assessment system . 57 student activism 28 Psychometric requirements 57 Section III: The prediction of Program and technical requirements creative and high level accomplishment with of a brief-assessment information biographical and self-report information. . 30 system 64

6 The theoretical requirements of efficient systems of brief-assessment information . 64 Statistical methods appropriate for brief-assessment information 66 Summary 71 Section VII: Conclusions: The general functions of brief-assessment measures . . 72 Broadband assessment 72 Selection and placement 73 Providing information for counseling and guidance 74 Serving as a teaching and guidance instrument for students completing the biographical questionnaire 74 Identifying students with particular characteristics or needs .. 75 Helping colleges in their planning and administrative activities 75 Providing the basis of two-way systems of information and action 75 Monitoring and understanding the transition from high school to schools, colleges and jobs 76 Conclusion 77 Reference list 78

vl Introduction tions tend to become fuzzy when they are applied to devices that yield scales based on biographical items, or when biographical items are used for This is a brief review of some of the studies that purposes other than the purpose their content have examined or demonstrated the value of brief would indicate. In general, however, they seem to assessments of students' past and current edu offer a serviceable delineation of the area. In re- cational and personal characteristics. It concen- viewing this area, emphasis is placed on the trates on measures that have been used in ed. utility of the measures. ucational settings, although some industrial appli- Tests of many kinds are extensively used by cations are also discussed. In spite of its length. admissions officers, counselors, and educators. it really is a brief review, because the studies in Intense intellectual effort and great amounts of eluded have been selected from a much greater time and money have been spent on their de- number of articles and research reports. Some of velopment and use and on technical and theoreti- the information studied is biographical in a strict cal models of tests. In contrast, relatively little sense. referring to students' previous activities, attention has been devoted to information directly family history, etc. However. most of the reviewed available from self-report instruments. Use of in- studies have used brief measures. such as self- formation of this type has also often been ne- ratings and self-expressed vocational choices. glected. This review of the uses and research on The instruments reviewed are relatively spe- biographical and other self-report information cific measures that obtain information about a does not support this lack of attention and use, person's past experiences, currcntstatus, or and. in a sense, is an attempt to study the viabili- needs. Many rely on the individual's selfexpres- ty of this information as "first-class variables." sion or self-evaluation of his characteristics, plans, Therefore, this review outlines some of the gen- motives. and goals. The scope of the review is eral functions of biographical and other self-re- necessarily broad so that it can include the wide port information, research relating to its specific variety of information that has been used in the uses, some theoretical and technical models for major studies of brief self-report information. using questionnaire information, and some of the Two aspects of the instruments should be empha- implications of self-report research for practice sized: brief self-report assessment tends to be and theory. direct rather than indirect, and itis based on In short, the purpose of this review is to ex- self-expressions rather than scales composed of amine the accuracy and the concurrent and pre- diverse, subtle items. Of course. these distinc- dictive validity of brief self-report information,

1 and to evaluate the promise and problems involved inits practical use. The issues are re- The validity of direct measure per se viewed in several arras: In the first section, the Beyond obvious and innocuous information such power of self-report information for predicting as his vocational choice or hometown size, can or and understanding grades is reviewed. Grades are will a person give an accurate account of his important in academic life, and they can have history and present status? The few studies of the critical effects on the lives of students. It is there- validity of self-reports provide a fairly consistent fore as important to understand the nature of answer to the question. As early as 1937, Walker academic: success as it is to be able to predict it. found that college students' reports of such factual Self-report information adds to that understand. information as their father's occupation and class ing. In the second section, research on the in- standing agreed very well with official records. fluence of background and personal information Harris (1946) found high validity for a question- on students' progress through higher education is naire he used in a psychiatric setting. Mosel and summarized. The third section deals with the pre- Cozan (1952) reported high validity for applica- diction of creative or high level accomplishment. tion blank work histories in industrial settings. Biographical information about individuals' past They found a high level of agreement between the accomplishments have consistently been found claims made by job applicants and the reports of to predict later high level accomplishment better past employers with respect to weekly wages, than does -any other class of information. The duration of employment, and job duties. All cor- fourth section reviews the increase in the under- relations except one were .90 or greater. Hardin standing of students' educational backgrounds and Hershey (1960) found that when workers' re- that biographical information provides. In the fifth ports of their wages on a questionnaire were section, the power of biographical information to checkedagainstcompany payrecords,the predict vocational choices and vocational success worker and company figures correlated .98 among is reviewed. The sixth section is a review of the women, and .99 among men. About eight percent technical and theoretical requirements of effec- of the sample under- or over-stated their pay by tive biographical and self-report information sys- plus or minus six percent. Interestingly, about tems. In the last section, the general functions of three times as many employees understated their biographical data are considered. Throughout the pay as overstatedit. Clausen (1968) compared review, emphasis is placed on predictive validity self-reports of voting in elections to official rec- of the measures in educational settings because ords and found an "invalidity" rate of approxi- the power of a variable to predict later perform- mately 6.9 percent. He cautions that this may be ance or choices is necessary before the variable an overestimate, for: "All errors that lead the can be used for guidance or sectioning. However, investigator astray in tracking down the record wherever warranted, other aspects of the use of of the respondent's vote, e.g., incorrectly spelled self-report information are also discussed. name, incorrect address, have the one-sided effect Before turning to the educational uses of brief of challenging the validity of the respondent's assessment, one question must be dealt with im- vote report." This is a very important point to mediately. This question is whether individuals' remember in every study of the validity of self- self-reports can be believed. To be blunt, how do reports. One should not simply assume 100 per- we know that people won't lie? cent accuracy in official records and the reports of those records. Calahan (1968) asked a number of Denver 9

2 adults questions ranging from whether they had Maxey and Ormsby (1971) studied the agreement a ,eitone in their homes to whether they contrib. between student-reported and school-reported uted to the Community Chest. The selr reports nonacademic achievement on 28 items. (They also on many factual questions were quite accurate studied the accuracy of self-reported grades, of (home ownership, 96 percent accurate; valid which more later.) Their sample included 5,775 library card, 87 percent accurate; voting in 1948 students completing the American College Test presidential elections, 86 percent accurate) while Assessment (ACT) battery. Student reports were some more socially desirable and rarer behaviors checked with school reports in 134 high schools. were reported with less accuracy (contributed to Achievement was measured in athletics, leader- Community Chest drive in 1943, 56 percent ac- ship, music, speech, drama, art, writing, and sci curate). After a variety of analyses, Calahan ence. Measurement items included, for example: noted that accuracy was higher for items con- "Edited a school paper or yearbook," and "Placed cerned with present fact, and suggested: "... first, second, or third in a regional or state science respondents generally will tend to tell the truth contest." The average level of agreement be- even when it may reflect on their prestige, pro- tween student report and school records was vided that the question of fact concerns the about 90 percent. But this did not mean that 10 respondent's present status rather than past percent of the students were exaggerating. On the events." typical item only about 6 percent of the students In samples of college students, Walsh (1967, claimed an accomplishment for which the school 1968, and 1969) has found that students generally had no record. The other four percent of the stu- provide accurate reports of their past behavior, dents were credited by the school with an achieve. even when items deal with sensitive issues, failing ment they themselves did not claim. The items on courses, for example. However, Calahan's com- which agreement between the two groups was ments seem to apply to college students, as well. greatest tended to be highly visible, easily verifi- Thus, Walsh's students seemed to have a little able items, for example: "Placed first, second, or difficulty recalling remote or insignificant events, third in a regional or state speech contest." Con- but, ". .. if an error of plus or minus .20 was per- versely, the items on which there was less agree. mitted in a student's report of his previous semes- ment tended to be behaviors about which the ter CPA, then the percentage of accuracy would school would have little information, for example: be 100 percent." Overall, Walsh found a very high "Actively campaigned to elect another student." level of accuracy. In addition, Walsh did not find No systematic differences in agreement were any difference in accuracy among interview, ques- found when the data were broken down by sex or tionnaire, or "personal-data blank." In his later family-income level. The reports of students who studies, Walsh found that the level of accuracy of made better grades agreed with school reports self-report was not changed when students were slightly less frequently than with those of students given financial or social incentives to distort their who made lower grades. The authors think this self-report. Studies of the validity of self-reported may he because of a tendency for students with grades reviewed in the next section also indicate higher grades to be more active in school social that students generally provide quite :accurate activities in ways that are unknown to school per. information about themselves. sonnel. The fact that the students' reports of As part of a comprehensive study of the accu- achievement were gathered at the same time that racy of self-reports on a questionnaire adminis. the students were completing a national assess- tered with a national college testing program, ment for college admission leads one to expect the 10

3 reports to be exaggerated. The fact that they tended not to be adds strong support to the idea that self-reports are accurate. In brief, from the limited evidence available, it appears that questionnaire responses have useful validity for the types of decisions and actions for which they will be used. Self-reports seem to be accurate when they deal with matters that are fairly recent, relevant to the person's present interests and concerns, verifiable, and, as we shall see in this review, when the questions are phrased in the best way. Although the evidence is fairly consistent with this generalization, there are several gaps in the current research literature. In particular, few studies have examined the validity of self-report among subgroups, such as ethnic mi- norities or disadvantaged students. (Freeburg, 1974, however, found correlations in the .70's between self-reports and records on topics as sensitive as the number of police contacts among disadvantaged school dropouts.) Clearly, more work is needed. In the following pages, the concurrent and predictive validity of self-report information is examined. It is often as valid as more extensive and expensive tests in similar areas. This evidence suggests that one can believe and make decisions based on self-report information in a wide. variety of areas as much as one can believe and use test information.

Ii 4

II Section I: Predicting and between self-reported and high-school-reported understanding academic criteria grades in a large sample of students who were applying to college and who knew that their re ported grades would be part of the basis for ad- This section is concerned with the utility of bio- mission. graphical and self-report information in the pre- In a more recent sample, Hanna, Bligh, and diction of grades. As criteria, grades evoke many Lenke (1970) reported a study of 1,105 eighth general questions abort the use of biographical in grade students in six states. Their figures were formation, and therefore provide a logical place to gathered by area, as well as by summing over four begin a review. The questions include, among areas. The correlations between self-reported and others: "Are biographical or sereport question- school reported grades in mathematics were .88; naire responses valid; can they be believed?" in science, .84; in English, .84; and in social "Are they related to anything; do they predict studies, .84. When grades were summed over all criteria we are interested in?" "How do they com- four areas, the correlation was .93. This study is pare with and add to information provided by particularly important because the students were tests?" and "Do they add anything to our under- eighth graders and reported their grades on simple standing of the phenomenon with which we are four-point scales. In addition, Hanna, Bligh, and concerned?" Lenke reported the means and standard devia- tions of student- and school-reported grades. The Self-reported grades largest reported difference in means was equiva- The first and perhaps most critical question is lent to a difference of .15 on a four-poiut scale. whether students' self-reports of grades and class The studies by Walsh (1967, 1968, 1969) re- rank can be believed, and whether they are valid ported earlier used criteria of "accuracy" of a as predictors. The evidence on these questions student' estimate of his CPA during his last se- begins as long ago as 1940 when Perry found cor- mester. It is possible that Walsh's estimates of relations between schooeported and self-re- accuracy are conservative, since many students ported grades of .83 for high school students with who were not on academic probation or on schol- grades above 80 percent and a correlation of .66 arship would have no reason to calciqate a pre- for those below 80 percent. Dunnette (1952) re- cise CPA, and so were required to guess. If they ported a correlation of .94 between school- and had been asked to estimate their approximate sereported grades. More recently, research by overall grade (A+, A, 8+, B, etc.) the same stu- Davidsen (1963)yieldedacorrelation of .93 dents may have been able to give a more nearly

. 12

s .

... accurateresponse.I Ins approach was used by and self-reported grades was .95. Richards and Lutz (1968) who found that meas Boldt (1973) compared the self-reported and tires of this type correlated .84 and .86 with col- school-reported grades of 4,200 students in nine lege reported CPA's among men and womenre- high schools. The self-reported grades were ob- spectively at19 four-year colleges. The same tained when the students took the Scholastic Ap- figures were .85 and .87 at 13 two-year colleges. titudeThst and completed the Student Descriptive In another sample of students at 27 two-year col- Questionnaire. The median correlation was .87. leges, Baird (1969b) carried out similar analyses lioldt concludes: "... one can see that more than and found the correlati-ms to be .83 and .86 for 98 percent of the grades reported were either men and women. In both these studies the means exactly right or only one offthere is a very small ofthecollege-reportedandstudent-reported tendency to overrate oneself but much smaller grades represented almost identical values. than the tendency to be accurate." Each student An unpublished study by the author (Baird. reported grades in mathematics courses, English n. d.) found that this level of correlation was con courses, etc. Boldt found that 79 percent of the sistent in eight separate multi-institutional sam- grades were reported exactly right. ples. This study also found that the correlations In what must be regarded as at least an ex- generally were of the same magnitude (median = haustive study, if not the definitive study, Maxey .87) in separate colleges (well over 100 separate and Ormsby (1971) studied the correlations be- colleges) as well as in the total samples; that is. tween students' self-reported grades and school- the correlation was not due to the pooling of stn. reported grades in a sample of 5,775 students in dents at varied institutions. Considering the sim- 134 schools throughout the country. As in the ple self-report scale, and the nature of college Boldt study, the grades were gathered as part of a grades, this figure (.87) is probably dote to the standardized national testing battery. The stu- limits imposed by the reliabilities of the variables. dents knew that their self-reported grades would Using a sample of the very bright National Merit be reported to colleges where they were applying scholars, Nichols and Holland (1963) found self- for admission. Overall, the correlations ranged reported and transcript-reported grades to corre- from .81for natural science grades to .86 for late .96. The largest discrepancy was one stu- grades in English. On the average, 98 percent of dent's change of a C to a B. Nichols (1966) also the students reported grades accurate within one found student- and school-reported grades to grade, 78 percentreportedexactlyaccurate correlate .83 and .82 in another sample of Merit grades, 16 percent overrepresented their grades scholars. Presumably, these students would be by one (or more) grades and, interestingly, 6 per- concerned about obtaining and retaining scholar- cent underrepresented their grades by one or ships and might have been expected to distort re- more grades. The authors believe that even this ports of their grades. high level of agreement may be an underestimate. At the graduate level, Kirk and Sereda (1969) -In the last 7 years more widespread use of asked Berkeley architecture students to report honors courses is included in high school curricu- their grades and found a high degree of accuracy; lums. This introduces a dual grading system, and 43 percent reported exactly correct grades, 78 itis possible that some confusion results when percent reported grades differing ±.1 from their students report grades earned in honors courses. transcript GPAs. and 92 percent were within ±.3 Secondly, multiple courses in the various subject of their actual grades. The largest discrepancy re- matter fields arc now offered, and consequently it ported was .7. The correlation between transcript- isdifficultfor the studentto determine what

13 -T' 6 course grade should be reported. For example. .74 between student. and transcript-reported many students report psychology as a science grades. Searching for the sources of discrepancies course, but others do not. This may lead to some between student- and transcript-reported grades, confusion when school officials report a science they conducted extensive analyses of the options grade as some may interpret it as a nonscience available to students in the SDQ. They found that course when the student has interpreted it as a the SDQ questions needed to: take account of science course. Thirdly, when students are asked multiple grades in the various academic areas to report the last grade earned in a course, they assessed: provide better options for non-conven- may interpret this as the final examination grade tional grades; provide clearer instructions to stu- or 9-week grade rather than the semester grade." dents who had not taken courses in an area; elimi- (Maxey and Ormsby. 1971, pp. 10.11) nate the use of the term "semester" because it is In addition, the authors analyzed their data by not commonly used in high schools; provide ex- various student el,aracteristies, and found no con- amples for students to use as guides: define "so- sistent differences in accuracy among students cial studies" more precisely; encourage students of different race, family income, or size of high to check their grades with authorities if they are school class. Girls were a little more nearly accu- in doubt about them; and use a common time- rate than boys, as were students with higher de- point for all students, such as "end of junior year." gree aspirations. and those who scored higher on These all seem to be useful suggestions and per- the college admissions test. These differences arc haps indicate that many students who have re- not large, however. Maxey and Ormsby make two ported grades that arc different from school-re- additional points:first, because the grades of ported grades were confused by the format of the college-bound students cover a restricted range, questions, and had responded as well as they were the correlation between self-reported and school- able to do. They probably did not intend to lie: reported grades would he even higher if the full they simply gave their best response to items that range of students were studied; and second, the seemed ambiguous to them. fact that students applying to college report accu- Armstrong and Jensen also examined a wide rate grades although they may be motivated to put variety of variables that might be related to dis- themselves in a favorable light gives strong sup- crepancies between student-reported and tran- port to the proposition that self-report information script grades. They found that students who had is valid. higher SAT scores and higher high school grades _ Armstrong and Jensen (1974) conducted an extended to report fewer discrepant grades than tensive study of the accuracy of self - reported students who stood lower on these measures. grades of some 2,775 randomly selected appli- They also found a lower incidence of discrepant cants t: colleges id the Massachusetts State Col- reporting among: out of state applicants; financial lege System. The source of information was the aid applicants; BA or MA aspirants; women; appli- Student Descriptive Questionnaire, completed by cants from college preparatory programs; appli- students when they take the SAT. Overall, they cants from large high schools; and applicants from found that 71.7 percent of all student-reported public high schools. No differences in reporting grades agreed exactly with transcript grades. and were found in breakdowns on the following vari- that 97.4 percent agreed within one grade. Al- ables: full-time/part-time/evening attendance plan: though 21 percent overrepresented their grades New England residence; marital status; housing by one or more grade, 7 percent underrepresented plan: family income: and ethnic group. them. The researchers reported a correlation of From all their analyses, the authors concluded: 14 "It seems reasonable to conclude that appli- to highly selective colleges than among students cants reported their grades accurately to the best applying to other colleges. Itis interesting that of their ability, with few exceptions. Confusion underreporting of grades seems to increase along generated by the directions in the version of the with selectivity as well as does overreporting of spQ to which applicants included in the present grades. Hanson and Lamb also examined the study responded was apparently responsible for same relations between selectivity and agreement most of the lowered level of precise accuracy between student. and school-reports of grades for noted in the present study as compared with pre. students who differ in ability. Overall, they found vious research." (p. 100) that the percentage of students whose reports of Some college administrators and admissions grades agreed with school reports increased with officers may accept the overall results of the the ability of students. The specific results were Maxey and Ormsby and Armstrong and Jensen complex, but among the students who sought ad- studies, yet retain doubts about the accuracy of mission to highly selective colleges, the level of students' grade reports when they apply to certain exact agreement in the first three quartiles of abil colleges. They may suspect that students who ity was approximately 72 percent. In the top-ability apply to highly selective colleges will distort their quartile, the level of exact agreement was nearly grades to a greater degree than will students who 87 percent. It is probable, then, that the high-abil- apply to other colleges. They may also suspect ity students who are more likely than other stu- that low-ability students might be tempted to dis- dents to be seriously considered by highly selec- tort their grades when they apply to moderately or tive colleges are also more likely to report grades highly selective colleges, moderate-ability stu- that agree with their school reports. dents when they apply to highly selective colleges, In short, research accumulated over 30 years, and so on. To study these possibilities, Hanson and using various methods, in samples of grade school Lamb (1973) compared the school- and student- students, high school students, college applicants, reported grades of 2,255 high school students who junior college students, four-year college students, applied to colleges that varied in selectivity. The and professional school students adds up to one students indicated the colleges they wished to conclusion: students' reports of their grades are apply to when they took the ACT assessment. about as usable as school-reported grades. This Their first-choice colleges were designated as low, conclusion seems particularly valid when one con- moderate, or high in selectivity, based on Astin's siders the conditions involved in some of the (1965) measure of selectivity. The percentages of sliidies. That is, even when students were faced students who reported grades in exact agreement with the pressure of maintaining scholarships, with their schools' report were 81.8 percent among applying to college, and deliberate experimental students applying to colleges low in selectivity, attempts to get them to change their responses, 83.5 percent among the students applying to col- they continued to tell the truth. leges moderate in selectivity, and 75.9 percent But what about the rare students who do not among the students applying to colleges high in give responses that are consistent with school selectivity. The corresponding figures for under- records? What can be done about them? Kirk and reporting grades were 3.7, 3.7, and 5.7, and the Sereda (1969) studied students who reported dis- figures for overreporting grades were 14.5, 12.8, crepant grades, and the authors suggest that these and 18.4. These figures show that there was a students may have particular characteristics that slightly greater discrepancy between school- and could help to identify them. They found that stu- student-reported grades among stude^ts applying dents reporting discrepant grades tended to have 15 I "It seems reasonable to conclude that appli- to highly selective colleges than among students cants reported their grades accurately to the best applying to other colleges. It is interesting that of their ability, with few exceptions. Confusion underreporting of grades seems to increase along generated by the directions in the version of the with selectivity as well as does overreporting of SDQ to which applicants included in the present grades. Hanson and Lamb also examined the study responded was apparently responsible for same relations between selectivity and agreement most of the lowered level of precise accuracy between student- and school-reports of grades for noted in the present study as compared with pre- students who differ in ability. Overall, they found vious research." (p. 100) that the percentage of students whose reports of Some college administrators and admissions grades agreed with school reports increased with officers may accept the overall results of the the ability of students. The specific results were Maxey and Ormsby and Armstrong and Jensen complex, but among the students who sought ad- studies, yet retain doubts about the accuracy of mission to highly selective colleges, the level of students' grade reports when they apply to certain exact agreement in the first three quartiles of abil- colleges. They may suspect that students who ity was approximately 72 percent. In the top-ability apply to highly selective colleges will distort their quartile, the level of exact agreement was nearly grades to a greater degree than will students who 87 percent. It is probable, then, that the highabil- apply to other colleges. They may also suspect ity students who are more likely than other stu- that low-ability students might be tempted to dis- dents to be seriously considered by highly selec- tort their grades when they apply to moderately or tive colleges are also more likely to report grades highly selective colleges, moderate-ability stu- that agree with their school reports. dents when they apply to highly selective colleges, In short, research accumulated over 30 years, and so on. To study these possibilities, Hanson and using various methods, in samples of grade school Lamb (1973) compared the school- and student- students, high school students, college applicants, reported grades of 2,255 high school students who junior college students, four-year college students, applied to colleges that varied in selectivity. The and professional school students adds up to one students indicated the colleges they wished to conclusion: students' reports of their grades are apply to when they took the ACT assessment. about as usable as school-reported grades. This Their first-choice coNges were designated as low, conclusion seems particularly valid when one con - moderate, or high in selectivity, based on Astin's side's the conditions involved in some of the (1965) measure of selectivity. The percentages of s' dies. That is, even when students were faced students who reported grades in exact agreement with the pressure of maintaining scholarships, with their schools' report were 81.8 percent among applying to college, and deliberate experimental students applying to colleges low in selectivity, attempts to get them to change their responses, 83.5 percent among the students applying to col. they continued to tell the truth. leges moderate in selectivity, and 75.9 percent But what about the rare students who do not among the students applying to colleges high in give responses that are consistent with school selectivity. The corresponding figures for under- records? What can be done about them? Kirk and reporting grades were 3.7, 3.7, and 5.7, and the Sereda (1969) studied students who reported dis- figures for overreporting grades were 14.5, 12.8, crepant grades, and the authors suggest that these and 18.4. These figures show that there was a students may have particular characteristics that slightly greater discrepancy between school- and could help to identify them. They found that stu- student-reported grades among students applying dents reporting discrepant grades tended to have 15 a parents of less than average education. were inure college grades after two years, while a college ad- academically oriented, more ambitious. and less missions test correlated only .31. The correspond- social. Ty pical discrepant reporters had parents ing figures for women were .54 and .39. Hanna, who were not well educated themselves but who Bligh, and Lenke (1970) found that self-reported placed high demands and expectations on their eighth-grade grades predicted grades in ninth- children, tending to cause the children to be grade algebra and geometry about as well as highly motivated to achieve. schoolreported grades (.55 and .57 in algebra; This description seems plausible, and leads .62 and .62 in geometry), and that they were ex- one to expect that further research would find a actly equivalentinprediction efficiency when pattern of variables on a biographical question- combined with test scores (.62 in algebra; .63 in naire that would lead to identification of students geometry). Hoyt and Munday (1968) summarized who are likely to exaggerate their grades. Such a the results of validity studies conducted by the pattern would be roughly analogous to the "lie" American College Testing Program for 437 col- scales on personality and interesttests. The leges between 1965 and 1967. A total of 273.000 schoolreported grades of students who scored students were involved. They found that student- high on such a scale could be checked. reported high school grades were better predictors of college grades, on the average (the median Self-reported grades as predictors correlation across colleges was .54), than were the Given the fact that selfreported grades seem to ACTtests (the median correlation was .48). The be accurate. are they valid as predictors? The American College Testing Program (1973) made a research answer to this question is positive, al- similar survey for the years from 1968 to 1970. In though it has not been substantiated in as many studies conducted for 419 colleges, involving settings. In a study by Hoyt (1963). the predictive 298,000 students, approximately the same results efficiency of average self-reported grades equaled were obtained. Student-reported grades were bet- that of the student's high school rank obtained ter predictors than theACT tests,on the average from his transcript. In two of the studies cited (the median correlations were .51 and .47, re- earlier(Richards,Holland,andLutz,1967; spectively). Of course, the average multiple corre- Richards andLutz,1968). self-reported high lations using both student-reported grades and school grades were better predictors of college ACTscores were higher than either alone. Cole grades than tests of academic aptitude. Moreo%er, (1969) found that self-reported high school grades these self-reported high school grades generally were superior to theACTtests in most studies had substantially larger beta weights (added more that had examined grades inspecific college to the prediction of grades) than did the test courses. Summarizing data from 19 subject areas, scores. Cole found thatstudent-reported grades pre- Similar results were obtained fur students in dicted grades better in areas representing 304 samples of both occupational and academic stu- specific courses, thatACTtests were better pre- dents in tu 0.) ear colleges by Baird (196%), in dictors in areas representing 69 courses, and that a sample of average college students (Nichols. they were equal in areas representing 44 courses. 1966). and in a number of National Merit samples Maxey (1972) compared school-reported and (Nichols and Holland. 1963. Holland and Nichols. student-reported high school grades as predictors 1964; Nichols, 1966). For example. Baird (1969b) of college academic achievement. In 26 colleges, found that among two-year college men. self- Maxey found that school-reported grades had reported high school grades correlated .44 with essentially the same predictive power as student- 16

9 reported grades. The median correlation between alone. In a second study, Brookover, et al., (1965) student-reported grades and college grades was found that when the scale was administered in .58: the median correlation between school-re- the seventh, eighth, ninth, and tenth grades, it ported grades and college grades was also .58. In predicted grades one, two, and three years later the 77 colleges that used high school rank as a with considerable efficiency, in most cases as well predictor. student-reported grades were slightly or better than IQ tests, but less well than earlier better predictors of college grades. The median grades. More recently, Binder, Jones, and Strowig correlation between college grades and student- (1970) found that the Brookover scale was a better reported high school grades was .52; the median correlate of grades in twelfth grade than an IQ correlation of college grades and high school rank test and a self-expectation scale. In four samples was .48. the correlations were .56, .51. .71, and .67. The In brief, there is considerable evidence that scale had substantially higher beta weights than self-reported grades can be as useful as school- the other measures in a multiple regression with reported grades as predictors of college grades. grades, although the multiple correlation using all measures was higher than any zero-order correla- Self-predicted academic performance and tions. self concepts of academic ability At the college level, Biggs, Roth, and Strong An additional class of variables that has recently (1970) found that the Brookover scale predicted been explored are self-made predictions of per- overall grades in college with reasonable effi- formance and self-concepts of academic ability. ciency. When compared to a multiple correlation Brookover (1962),for example, has developed based on standardized tests (including the Minne- brief scales of general self-concept of academic sota Scholastic Aptitude Test and the Academic ability (8 items), and similar scales of self-con- Achievement Scale of the Strong Vocational In- cepts of ability in the four areas of mathematics, terest Blank), the multiple correlations of the English, social studies, and science. Typical items Brookover scale and high school grades had ahout asked students were: where they thought they the same degree of accuracy for males but were ranked in their classes, whether they thought lessnearly accuratefor females.Biggs and they had the ability to be doctors, professors, etc. Tinsley (1970) used an adaptation of the Brook- The scales showed adequate reliability. For sev- over scale to predict college grades in two differ- enth grade students, the general self-concept ent samples. In both samples the self-concept of measure predicted overall grades with a correla- ability was a better predictor than standardized tion of .57 for both males and females; the correla- college admissions tests (.54 versus .45 in one tions between IQ and grades were .61 for males sample; .44 versus .36 in the other). When ability and .65 for females. The self-concept significantly was partialed out, the correlation was still sig- added to the multiple predictions of grades, how- nificant. Jones and Grieneeks (1970) found the ever. The measures of self-concept in the specific Brookover scale to be a better predictor of college areas were correlated more highly than IQ with grades than the SAT (.4.8 versus .36 for women; .49 grades in the subjects corresponding to the spe- versus .22 for men) and had the highest beta cific areas in two comparisons, were correlated as weights in multiple regression equations. Sproull highly in two comparisons, and were correlated (1969) also found that the Brookover scale admin- less in four comparisons. In every case, the multi- istered in high school predicted academic per- ple correlations using both specific area self-con- formance in college, full-time versus part-time cept and IQ were substantially higher than either status, and major area. Using a similar measure- 17

10 Peterson's motivation for grades scaleFurst ence for men. Freehill points out that questions (1966) found correlations of .17 and .53 (men and concerned with the student's perception of his women) with high school grades. higher than those relative standing in comparison with other stu- between verbalability measures and grades. dents are more useful than direct predictions of Lindsay and Alehouse (1969) found that the same grades. (The former is very much like the content motivation-for-grades scale significantly increased of the Brookover and Peterson scales, and self- the multiple correlation predicting college fresh- ratings.) man grades. after SAT verbal, SAT math, and high In another group of studies using college stu- school grades already had been included. The dents, simple one-item self-ratings of academic scale was the second best predictor, after SAT ability or scholarship have been found to predict scores for females, and high school grades for college grades better than academic ability meas- males. A summary of validity studies using the ures or previous grades in large samples of Na- motivation for grades scale in the ETS validity tional Merit Scholars (Nichols and Holland, 1963), study service (Oppenheim, 1970) found that, in college students representing a wide range of hundreds of studies. the average increase in mul- talent (Baird, 1969a), and able psychology gradu- tiple correlation was .03 after high school grades ate students (Wiggins, Blackburn, and Hackman, and SAT scores were already included. One might 1968). Holmes and Tyler (1968) found that a simple wonder how much grades and SAT scores would ranking of oneself in comparison with peers was a add to the scale if they were included in a step- better predictor of college GPA than TAT scores, wise multiple regression first. or laboratory tasks (in this study, the self-ranking Yet another approach was used by Keefer was the only significant predictor). These simple (1969) who simply asked students at the beginning self-ratings are very similar to the basic content of the college year and at the beginning of three of the Brookover and Peterson scales. other grading periods to predict their end-of- In summary, self-estimates of ability seem to he freshman-year grades. Selfpredicted grades were relatively efficient predictors of academic per- consistentlybetter predictors (the correlation formance. However, the kind of self-estimate was .51) than college entrance tests of academic must be carefully defined. Apparently the most ability (.41) awl high school CPA (.43). The ac- efficient format for self-estimates is to ask the curacy of predictions did not decline over the subjects to estimate their relative standing in a nine month period. In addition, the self-predicted group of their peers. Most students should be able grade consistently added to the multiple correla- to do this quite well after twelve years of com- tion, after aptitude and high school grades had parisons with their peers, feedback on test scores, been entered.Inearlier studies, Doleys and and the daily evidence of their performance in Renzaglia (1963) found that self-predicted grades classwork and tests. Such ex7criences should did predict grades. but with less efficiency than provide students with a conception of their own SCAT scores, and the self-predicted grades raised capacities that incorporates ability, past achieve- the multiple correlation only from .63 to .65. Free- ment, and motivation. In any ease, estimates of hill (1952) found that selfpredicted grades pre- this type seem valid, and students appear able to dicted womens' grades in science better than estimate their own ability correctly. abilitytests.Althoughself-predictedgrades Self-estimates of traits have proved to he very added to the multiple correlation in science for effective predictors in a wide variety of areas be- women, and "verbal.' courses for men, they did sides academic performance. For example, a not do so in "verbal" courses for women, and sci- number of studies at National Merit Corporation 13

11 (Holland, 1961; Holland and Astin, 1962; Nichols tion to be related to grades. This research is more and Holland, 1963; Holland and Nichols, 1964) difficult to summarize since much of it has been showed that simple self-ratings on four-point on a purely pragmatic "Let's try it and see what scales on a variety of traits were among the best happens" basis, as Freeberg (1957) has pointed predictors of later college accomplishment in out. In general, earlier studies found that multiple such creative areas as art, science, leadership, predictions based on academic aptitude measures music, dramatiLs, and writing. These studies in- and preNious grades are improved slightly, but sig- dicated that such self-ratings were often better nificantly by the addition of general biographical predictors than SAT scores, personality scales, in- information (e.g.. Myers. 1952; Myers and Schultz, terest scales, and measures of student potentials. 1950; Schultz and Green, 1953; Webb, 1960; Hilton These results were based on studies of extremely and Myers, 1966). By contrast, recent studies have academically talented students. However, other had more success in the use of biographical infor- studies (e.g., Baird, 1969a) have found that self. mation, perhaps because of improvements sug- ratings continue to be very useful predictors of gested by earlier studies. For example, Denham later accomplishment among diverse groups of (1966), using a combination of biographical data students. Although the use of self-ratings may be and self-ratings, was able to account for a larger questioned where students might believe that the percentage of variance in college grades than did ratings were being used for selection purposes, SCAT scores. The combination also predicted num- students may give more honest answers than one ber of hours completed, while SCAT scores did not. would expect. Lunneborg and Lunneborg (1966a) predicted col- Self-ratings arc also highly related to vocational lege freshman grades in four areas, overall grades, and curricular choices among college students withdrawal, and change of major from biographical (Abe and Holland, 1965a, 1965b; Holland, 1968). In information, high school grades, and ability test these studies certain self-ratings distinguished scores. Multiple regression equations were devel- mong fields as well as interest test scales, and oped for each criteria. When just the first six pre- distinguished better than (1;(1 many other vari- dictors of the criteria in each equation were tabu- ables. lated, only one high school grade was selected, Finally, a number of studies by the American along with five test scores, and twenty-two bio- Council on Education have used self-ratings as graphical variables. Only biographical variables basic sources of data for a number of studies of predicted withdrawal and change of major. Nichols the characteristics of students and the influence (1966) found that specially constructed "Objective of colleges on students' educational and voca- Behavior Inventory" scales, which consisted of tional plans (American Council on Education, students' reports of their activities and prefer- 1966.1975; Astin and Patios, 1969). ences, predicted college grades almost as well as All these results lend support to the idea of self-reported grades, and, for certain groups, a using self-ratings to assess academic and non- specially developed scale from the California Psy- academic traits. chological Inventory. Starks (1967) successfully used biographical information, both alone and in Prediction of grades by combination with a reading test, to predict sub- general biographical information ject-matter learning. In addition to self-reported grades, self-concepts Nichols and Holland (1963) found a wide variety of ability and self-ratings, a variety of studies have of biographical variables predicted college grades found other, more general, biographical inform a- in a sample of extremely bright students. In gen- 19

12 moderators, has been used to identify over- and much more heavily on measures of prior academic under-achievers and students with creative ac- achievement. That is, while biographic informa- complishment in various areas. These areas will tion is not predictive of what is common to suc- be discussed at greater length later on. cess across course areas, it is essential to explain Horst (1954, 1955) has developed a system of why a student may succeed in one area and fail in differential prediction that is analogous to the another, or if he is likely to succeed in all areas, moderator system just discussed, but focuses on why achievement could be greater in one area the criteria, choosing predictors that best account than another. Where predictions are used by the for differences among criteria. For example, if individual the variability inherent in biographic grades in six areas are predicted, differential pre- data permits better choices among competing cri- diction selects the first predictor that has the terion activities. highest average correlation with the differences "For example, suppose there are some aspects between all possible pairs of criteria; that is, dif- to family life, attitudes, interests, or ways of think- ferential prediction identifies the variables that ing that are characteristic of students who do well predict the differences in grade performance in in mathematics and not characteristic of students science, humanities, social science, and so on. who do well in biology. With predictions based on The predictors subsequently selected are those this background information a student can choose which, combined with earlier selections, produce between the two fields. On the other hand, if his the highest average multiple correlation with predictions of success in the two fields are based these criteria of difference. Horst (1954) has de- on the characteristics common to those students veloped an index of differential prediction effi- who succeed in mathematics and those students ciency. The technique is discussed in Section III. who succeed in biology, then choice is impossible. Differential prediction is particularly useful for He will be informed that he will either succeed in placement decisions; for example, those made by both areas or fail in both. students and counselors who are deciding among "Differential prediction obviously demands that various majors and courses. Differential predic- there are actual differences among criteria. As- tion could help students decide which major to suming, as in this study, that such differences enter by identifying the curriculums in which exist, the relatively low correlations obtained with they would do best. Thus, differential prediction the criteria have one sourcelack of matching was not developed to identify the predictors that variability in the predictors. A search for a more predict overall absolute academic success, but complex, factorially discrete set of predictors was designed to identify the predictors of relative must necessarily accompany adoption of the dif- success. The Lunneborgs have found that bio- ferential model, for only in this way can the multi- graphical variables are highly useful as predictors ple correlations with the criteria be raised. In the of differential success in a number of studies meantime, the accuracy of the absolute model is (Lunneborg, 1966, 1968; Lunneborg and Lunne- sacrificed for the greater utility of differential pre- borg, 1966a, 1966b, 1969). Lunneborg made a con- diction to personal decision making." (p. 106) vincing case for biographical information in differential prediction in 1968: Practical implications "These findings suggest that indeed biographic Thisreview of biographical information and data are more important in making differential grades has examined research accumulated over predictions of academic achievement than in many years, using various methods, in varied sam- making absolute predictions. The latter relied ples such as grade school students, high school 21

14 oral, these variables clustered around two traits tion of grades, seem to predict high grades. that are mentioned frequently in studies of the Although general biographical information may nonintellective predictors of grades: (a) persever- predict grades in an absolute sense, and may add ance and motivation to achieve, and (b) conform. to the multiple correlation. biographical informa- ity and socialization. Roberts (1965) found similar tion may be particularly useful in two nontradi- results in another sample of very bright students. tional ways: as moderators, and as variables in He examined item correlations with college grades differential prediction. The general technique of and creative achievement in science, writing, art. moderator regression analysisisdescribed in music. speech and leadership, mid developed Section VI. Itis based on the idea that a given scales based on the most predictive items. The variable may predict a criterion better for certain researcher summarized the content of the grades subgroups than for others. For example. the best scale, "...as including obviously related aca- predictor of grades for humanities majors may be demic accomplishments (such as the self-rating the sATverbal scale, whereas for science majors of academic ability)." In addition, "... it appears the best predictor may be high school grades. In that college freshman `A' students had fewer inthis case, the "moderator" would be major field. terestg tv..:2 less active. less social and less com- Many studies have reported results of this kind. petent in a number of practical, everyday affairs Frederiksen and Melville (1954) for example, found than were `C' students. They did not have fewer that a scale of interest in engineering predicted accomplishments... but they did not have more. engineering grades for "noncompulsive" students In short, some of the popular unfavorable stereo far better than it did for "compulsive" students. types of high grade achievers are supported." In The measure of compulsiveness itself was not re- addition, the grade prediction scale had negative lated to grades or the interest scale. The general correlations with every other scale except the sci strategy of moderator analysis is to find variables ence scale. Nichols (1966) has summarized the that will identify homogeneous groups of individu content of another biographical scale for predict. als for whom predictors work in different ways. ing grades: students scoring high more often en These analyses should indicate which groups are dorsed items suggesting "...that they are re predictable, and which variables are most pre. ligious and involved in church activities, they are dictive with which sorts of people. Studies demon. interested in music and participate in musical strafing this kind of analysis include: Berdie, activities; are hard workers with well established 1961: Frederiksen and Gilbert, 1960; Ghiselli, work habits; and are shy, socially withdrawn, and 1960a. 19606. 1963; Rock, 1965: Saunders, 1956; introverted." Other variables suggested that stu Stricker,1966. However, as Klein, Rock and dents obtaining high grades were more compul- Evans (1968) pointed out, most of these studies sive and conforming. (A review of the attempts to have focused on only one variable at a time. Their describe the relation of personality to academic study illustrates a method for examining several success is beyond the scope of this monograph. possible moderators at the same time. For ex but these descriptions are consistent with many ample:the least predictable group of the law stu- studies; for example. Bendig, 1958; Bendig and dents they studied consisted of students of low Sprague, 1954: French. 1963, 1964: Gebhart and socioeconomic status, who tended to be older. Hoyt. 1958; Gough, 1953: Holland. 1959. 1961.) To They conclude, "...the value of background summarize, general biographicalvariables re- variables may lie in tneir effectiveness as modera- flecting orderliness, compulsiveness. perscver tors rather than in their utility as predictors." ance, introversion, conformity. and a high valua- This approach, using biographical information as

''2 0 13 moderators, has been used to identify over- and much more heavily on measures of prior academic under-achievers and students with creative ac- achievement. That is, while biographic informa- complishment in various areas. These areas will tion is not predictive of what is common to suc- be discussed at greater length later on. cess across course areas, it is essential to explain Horst (1954, 1955) has developed a system of why a student may succeed in one area and fail in differential prediction that is analogous to the another, or if he is likely to succeed in all areas, moderator system just discussed, but focuses on why achievement could be greater in one area the criteria, choosing predictors that best account than another. Where predictions are used by the for differences among criteria. For example, if individual the variability inherent in biographic grades in six areas are predicted, differential pre- data permits better choices among competing cri- diction selects the first predictor that has the terion activities. highest average correlation with the differences "For example, suppose there are some aspects between all possible pairs of criteria; that is, dif- to family life, attitudes, interests, or ways of think- ferential prediction identifies the variables that ing that are characteristic of students who do well predict the differences in grade performance in in mathematics and not characteristic of students science, humanities, social science, and so on. who do well in biology. With predictions based on The predictors subsequently selected are those this background information a student can choose which, combined with earlier selections, produce between the two fields. On the other hand, if his the highest average multiple correlation with predictions of success in the two fields are based these criteria of difference. Horst (1954) has de- on the characteristics common to those students veloped an index of differential prediction effi- who succeed in mathematics and those students ,ciency. The technique is discussed in Section III. who succeed in biology, then choice is impossible. Differential prediction is particularly useful for He will be informed that he will either succeed in placement decisions; for example, those made by both areas or fail in both. students and counselors who are deciding among "Differential prediction obviously demands that various majors and courses. Differential predic- there are actual differences among criteria. As- tion could help students decide which major to suming, as in this study, that such differences enter by identifying the curriculums in which exist, the relatively low correlations obtained with they would do best. Thus, differential prediction the criteria have one sourcelack of matching was not developed to identify the predictors that variability in the predictors. A search for a more predict overall absolute academic success, but complex, factorially discrete set of predictors was designed to identify the predictors of relative must necessarily accompany adoption of the dif- success. The Lunneborgs have found that bio- ferential model, for only in this way can the multi- graphical variables are highly useful us predictors ple correlations with the criteria be raised. In the of differential success in a number of studies meantime, the accuracy of the absolute model is (Lunneborg, 1966, 1968; Lunneborg and Lunne- sacrificed for the greater utility of differential pre- borg, 1966a, 1966b, 1969). Lunneborg made a con- diction to personal decision making." (p. 106) vincing case for biographical information in differential prediction in 1968: Practical implications "These findings suggest that indeed biographic This review of biographicalinformation and data are more important in making differential grades has examined research accumulated over predictions of academic achievement than in many years, using various methods, in varied sam- making absolute predictions. The latter relied ples such as grade school students, high school 21

14 students, college applicants, junior college stu- one to expect that further research could find a dents, four-year college students, and professional patternof variables on standard admissions and graduate school students. It leads one to con- forms, such as the College Board's Student De- clude that students' reports of their grades are as scriptive Questionnaire (sDQ), that would lead to reliable and valid as school-reported grades. This identification of the student who is likely to ex- conclusion seems particularly valid when one con- aggerate his grades. Such a pattern would be siders the conditions involved in some of the roughly analogous to the "lie" scales on personal. studies. That is, even when students were faced ity and interest tests. Then the college could with the pressures of maintaining scholarships check the grades of students who scored high on (the National Merit studies), applying to college the scale. However, the main point should be re- (Davidsen; Holland and Richards; Hoyt; Maxey membered: most students tell the truth. In short, and Ormsby; Boldt; Armstrong and Jensen), and it seems that self-reported grades can be legiti- deliberate experimental attempts to get them to mately used for most of the purposes for which change their responses (the Walsh studies), they school- or college-reported grades can be used, at continued to tell the truth. But what about the considerably less cost and with fewer administra- problems of actual, pragmatic usage of self-re- tive problems. For example, self-reported grades ported grades and other biographical information reported on a standardized format such as the about college applicants? There are three sorts of SDQ could be used in an information system that problems, the first concerning the technical usage would: routinely report to participating colleges of the information, the second concerning im- on the characteristics of their applicants; provide plications for policy, and the third concerning students an estimate of the grades they would ob- ethical problems. tain in specified colleges, based on recent grade The evidence concerning the technical issues prediction studies; report to colleges the charac- indicates that self-reports of grades can usually be teristics of the enrolled freshman class; and re- believed, even when the student knows that his port the information on the students' choices and self-reported grades will be used as one of the prospects to high school counselors. A system like selection criteria. Self-reported grades can be this, developed by the American College Testing gathered in routine admissions procedures, and Program, has been widely used by hundreds of used in many selection and guidance decisions. colleges, thousands of high schools, and millions But even if the overwhelming majority of students of students for more than a decade. Supporting are honest, some may not be, and it is these stu- materials provide potentially very useful guidance dents who are of concern to many colleges. What guidelines for high school counselors and stu- could be done to reduce their inaccurate grade dents. reporting? The first task would seem to be to re- The second class of problems is much more duce some of the ambiguities in the present sys- difficult and puzzling. These problems stem from tems of collecting self-reported grades. The com- the fact that certain biographical information may ments and recommendationsof Maxey and add to the prediction of grades at a statistically Ormsby and Armstrong and Jensen would be significant level, but produces dilemmas for the good starting points. Another way to decrease the user. For example, there is ample evidence that rate of distortion would be to imply that the stu- the grades of women are more predictable and dents' actual grades may be looked up, and to use tend to be higher than those of academically spot checks as a further precaution, as suggested equally able men (e.g., Abelson, 1952; Baird, earlier. The Kirk and Sereda (1969) study leads 1969b; Jacobs, 1959; Munday, 1967; Seashore, 2-

15 1962). But this result is difficult to use. Should a reached. The rest of the applicants could be ex- college deliberately admit more women because amined using the biographical data as supple- they can be more reliably expected to earn higher mentary information. grades? Should it ignore the difference, and put Another approach to the problems of using everyoneintooneregressionequation, and biographical information would be to use the re- thereby overpredict for men and underpredict sults of existing research studies as a base for for women? Would thisrepresentfavoritism further researchto determine the underlying toward men and bias against women? Should the traits that explain why students in one group do college recognize the difference and include dif- better, or are more predictable, than students in ferent admission criteria for men and women? other groups. For example, the greater predicta- Would a quota for each sex then need to be used? bility of women, compared to men, may be due to Would the introduction of different admission greater ability to concentrate, better study habits, criteria outrage academic sensibilities?What etc. These traits could be identified by research would be the reaction of various external groups and measured by new assessment procedures. and publics if one course or the other were fol- Such underlying traits could be used to select lowed? Similar problems arise if a different level students in ways that seem fairer and more legiti- of grade performance, or a different level of pre- mate than ways that rely on group differences in dictability is found for students of different ethnic predictability.' Some useful ideasforpossible re- or racial background, social class, type of high search are already available in the voluminous re- school, and so on. (Of course, the kinds of prob- search about the background, demographic, per- lems discussed here are similar to the problems sonality,and other characteristicsrelatedto connected with the use of personality or interest grades. tests.) The basic question is to what extent these A third class of problems concerns the ethical kinds of variables should be used for admissions issues involved in asking for sensitive personal decisions. The position taken here is that most information and in using that information. The variables of this type should not be used as part recentEEOCguidelines highlight the concern of the admissions decision except, rarely, in felt by many that information about their back- marginal cases. Much biographical information is grounds will be misused. This concern is war- concerned with things over which the individual ranted because, unfortunately, information about has no controland cannot change sex, race, individuals' backgrounds have been misused for social background, family education, etc. It does capricious or prejudiced reasons in a wide variety not seem reasonable to hold such things against of educational and vocational situations. Even a candidate (nor to hold them in his or her favor) when there is a rational basis for considering since they are almost accidental phenomena. group differences in achievement and predictabil- However, it may be appropriate to use some bio- ity, as in the example of men and women just dis- graphical information of certain types in marginal cussed, the ethical problems of holding something cases. The type of information might include against people who have no control over their such details as study skills, value placed on edu- past remain. Just as with the use of personality or cation, etc. These variables could be used in a interest tests for admissions purposes, each self- two-stage strategy. In the first stage, the usual report variable that is proposed for use in an as- academic measures could be used to admit stu- sessment device needs to be carefully examined dents with given chances of success until a pre- I.This idea was suggested by Robert Boldt (personal com- determined level of probability of success was munication). 23

16 for all its ethical implications. Even fairly innocuous information can be used in very unfair and objectionable ways. These considerations also suggest that each person who completes a questionnaire should be told the reasons he or she is being asked for the information requested.

24 17 irvvvr,,, I

Section II: Biographical and patterns for males and females (Educational Test- self-report information in ing Service, 1957; McGuire, 1967; Sewell and the sequence of higher education Shah. 1967, 1968; Prediger, 1969). A number of other studies have found that parental encouragement was positively related to educational achieve. Biographical information and other brief self- ment above and beyond SES and intelligence report measures have been related to all aspects (Bordua, 1960; Rehberg and Westby, 1967; Sewell of higher education, including college attendance, and Shah, 1968; Simpson, 1962). Of course, these choosing and attending a college, choosing a living variables are interconnected with the parents' group, choosing a major, general adjustment to ability and willingness to contribute to financial college, and withdrawal from college. This sec- support for college costs. tion reviews studies bearing on these areas. The studies just mentioned need to be studied carefully, since many investigations indicate that College attendance general "socioeconomic status" is an abstraction College attendance is related to a great variety of that reflects many different, if connected, vari- things (e.g., Beezer and Hjelm, 1961; Berdie and ables. These include heredity, the physical sur- Hood, 1965; Wilson, 1966). However, most of the roundings and facilities available, diet, health research on college attendance has been con- care, language patterns, the neighborhood and cerned with two determinants: intelligence and school setting, the family's provision of educa- social class. The higher the student stands on tional experiences, the family's ability to finance either of these variables, the more likely is he or education, parental aspirations, and child rearing she to attend college. Studies using both variables practices. Each of these areas has been the sub- have produced different, but complementary re- ject of study and research. For example, the ef- sults. Some studies have found that intelligence is fects of child rearing practices on achievement a more important influence than socioeconomic behavior have been examined by Goode (1964), status (Schoenfeldt, 1966,1968; Nam and Coward, Rosen (1964), Morrow and Wilson (1961), Douvan 1962). Other studies have found that socioeco- and Adelson (1966), Elder (1963), and Rehberg, nomic status (SES) is a better predictor than in- Sinclair, and Schaefer (1970). These studies gen- telligence (Kahl, 1953; Medsker and-Trent, 1965; erally support the idea that child rearing practices Rogoff, 1963; White, 1952). Some other studies are related to achievement behavior, but asReh have found SES and intelligence to be approxi- berg, et al., conclude, the size of the "... degree mately equal in influence, or to have different of association have led us to believe that there is 25

18 more interfamilial variance in achievement be- through which this effect works. They also found havior unaccounted for than accounted for." This that the higher the ability composition of the conclusion probably applies to each of the com- school the more depressed were the students' ponents of "socioeconomic status." Although the educational plans, and the higher the social status relative importance of these variables isstill a composition of the school, the higher the students' matter of professional controversy among sociolo- plans. They found that the depressive effect of gists (see the continuing discussion of the Jencks some schools on their students' plans were mainly et al., book Inequality) it is clear that they are im- the result of the lower class rank that students in portant determinants of students' aspirations and a high ability school had, and the enhancing effect achievements. The importance of these variables of social class composition resulted from the like- leads to several pragmatic possibilities. The in- lihood that the student would enroll in a college fluence of SES on educational aspirations and preparatory curriculum and associate with college achievement suggests that special recruitment of oriented peers in the high status schools. bright students from low SES backgrounds may One of the largest data files accumulated in the be necessary. The importance of parental en- history of educational research, Project TALENT, couragement may entail counseling or educa- has been analyzed by Folger, Astin, and Bayer tional programs involving parents as well as stu- (1970). Using a subsample of this vast source of dents. And, of course, financial aid policies are information, the authors studied 8,746 males and related to parental willingness to contribute. 6,794 females who had completed the TALENT Recent evidence suggests that the context of battery of tests and questionnaires as high school the local high school also influences student as- seniors, and who had responded to a five-year pirations and achievement (Wilson, 1959; Mi- followup survey (Flanagan and Cooley, 1966). The chael, 1961; Turner, 1964; Campbell and Alex- researchers used several criteria that will be ex ander, 1965; Boyle, 19 , we nd Armer, 1966; amined because they are germane to the topics McDill, Rigsby, a Id Myers, 1969; Meyer, 1970; reviewed here. The first criterion was college at- Hauser, 1970). As Bain and Anderson's recent tendance or nonattendance. The researchers (1974) review of these studies indicates, the high presented zero-order correlations of 38 variables school context is probably less powerful than with the criterion, partial correlations of each other variables related to the family. However, its variable with the criterion with the effect of the influence suggests that schools with many low remaining 37 variables partialed out, partial corre- SES students may need to make special efforts to lations of each variable with the criterion with all -1...._-11 encourage able students to attend college. These the other variables not in the same domain (such studies have shown that students who are enrolled as ability scores, temperament measures, etc.), in high schools with a high proportion of students multiple correlations of all the variables in a from high socioeconomic backgrounds are more specified domain with the criterion, and multiple- likely to attend college than equally able students partial correlation of all variables in the domain in schools with a low proportion. However, Meyer with the criterion, partialing out the effect of all (1970) has pointed out that this positive effect other variables. Obviously this variety of analyses masks a negative one. Since high status schools provides a number of ways to study the phenom- often have more able students, the able students' ena of college attendance. Whichever system was aspirations are lowered, presumably because of used, however, seniors' plans to attend college the 'ligher level of competition. Alexander and and mathematical aptitude test scores predicted Eckland (1974) tried to analyze the mechanisms college attendance best. The three domains most 26

19 strongly related to college attendance were, in decreasing order. college commitment variables The influence of biographical factors (including college plans. encouragement of col- on college selection lege plans by the student's parents and peers). Many factors are involved in the choice of a col- ability variables (vocabulary, general information. lege.students'characteristics, their families' creativity, abstract reasoning, and mathematical characteristics, their peer group relations, the aptitude scores), and socioeconomic variables social and economic context, the information (family income. father's occupation, father's edu- available to them, and the characteristics of the cation, mother's education, and number of books college. For example, there is evidence that stu- in the home). Domains related to a lesser extent dent, of higher SES are more likely to emphasize were (in descending order): interest variables, the prestige, intellectuality, and social character variables describing the high school the student of the college in their choice than do low SES stu- attended, the student's marital and parental sta- dents (Baird, 1969a; Rossi and Coleman. 1964). tus, the student's family condition, residence The most striking result in studies of college variables, temperament variables, and ethnic and choice is the lack of information that character. religious status. These results indicate that rela- izes most student choices (Holland. 1959; Rossi tively simple direct measures of students' back- and Coleman, 1964; Trent, 1965). As Holland

grounds and experiences clearly contribute to the (1959) puts it, students generally ".. . make prediction of college attendance. choices in the same way that consumers often, if As the Folger, Astin, and Bayer study indicated, not usually, buy household goods; they select a variety of Fiographical variables influence col- colleges by means of vague notions which they lege attendance such as the urban or rural setting seldom can document meaningfully.I of the home (Sewell, 1964). high school size (Cof- Regardless of student ignorance, a vast litera- felt and Hobbs, 1964; Simmons, 1963; Stroup and ture indicates that, consciously or not, by their Andrew, 1959). size of family (Educational Testing own selection or the colleges', students with dif- Service, 1957; Medsker and Trent, 1965; Sim- ferent backgrounds and traits enter different mons, 1963) and religion (Bordua, 1960: Medsker kinds of colleges (e.g., Astin, 19656; Baird and and Trent. 1965; Trent, 1965; Trent and Medsker, Holland, 1969; Astin, Patios, and Creager, 1967; 1968). Race, of course, is also related to college Hood and Swanson, 1965; Medsker and Trent, attendance. Each of these differences suggests 1965; Schoenfeldt, 1966, 1968; Rossi and Cole- new recruitment strategies, counseling practices, man, 1964). Many of these studies indicate that and educational programs. Also, taken together. students from high SES backgrounds are more these variables suggest that perhaps researchers likely to enter prestigious and selective institu- should focus their energies not only on under- tions, and less likely to enter such less prestigious standing and measuring the impact of the vari- institutions as junior colleges. ables on aspirations and achievement, but should A study by Hoyt (1968) indicated that a number also attempt to develop techniques to modify the of "biographical'. items were related to the char- sometimes deleterious influence of these vari- acteristics of the 169 colleges chosen by students. ables. 1. Item is an area where "biographical- instruments could be of great use by assessing the degree of knowledge that sup dents have about colleges. Those students found to have little realistic knowledge t mild be directed to college guides. general description.: and to litho- materials that would help them to make a more informed choice. 27

20 The biographical items included a checklist of edu- SAT of students, and .60 with the intellectuality of cational goals (Baird, 1967h), level of educational the students. Folger, Astin and Bayer (1970) also aspiration. parttime work expectations, college reported a wide variety of background correlates attractions, influence of parents and peers, and of attendance at nonselective or selective senior nonacademic (creative) accomplishments. The colleges. percentage or means of entering freshmen classes These results suggest that the students "seon these variables were correlated with the Astin lect" colleges with characteristics roughly con- (1965a) scores of college environments. In 18 of 19 gruent with their own; it also suggests that col- cases a multiple R of student characteristics leges "select" students with particular charac- significantly correlated with college characteris- teristics. Perhaps a better description of this tics. In a similar study, Astin (1968a) also found process would be the "flow and distribution of that average student characteristics were corre- students to institutions" which is not a random lated with the environment scores of 246 colleges. process. There are several significant implica- Excluding two physical factorsspread of cam- tions. First, it may be possible to make the choice pus. and friendliness of the dorm counselor meas- of college more rational by providing information ures of student characteristics from a biographi- about the choice process, and about colleges that cal questionnaire had correlations of .50 or higher the student might not otherwise consider. Second, with 27 of 33 remaining environmental dimen- colleges might seek to enroll a greater diversity of sions. The median highest correlation was .63, the students. This might be done through a student median second highest correlation was .58. Eight locater service, for example, the College Board's of the correlations were above .72. In an earlier Student Search Service. Thirst, these results add study, Astin (1965a) developed other environ- support to the tentative conclusions of earlier re- mental measures, and related derived measures searchers (e.g., Darley, 1956) that "the students of student "input" characteristics to them. These make the college." That is, a great many of the correlations were also approximately in the range significant aspects of the college environment are just cited. determined by the characteristics of the students Pace (1969) has also reported correlations be. attending them. While the evidence for this con tween student characteristics and the dimensions elusion is too complex and voluminous to review of environment in his College and University En- here, its ramifications can be outlined. A college vironment Scales. For example, the "Practicality" could profitably examine the characteristics of its of the colleges correlated .74 with the mean SAT incoming classes. By including a wide variety of scores of entering students, .52 with majoring in biographical data (and by referring to appropriate the humanities, and .63 with self-ratings of un- research) colleges could obtain useful descriptions conventionality. "Community" correlated .52 of the needs of their students, and gain some with the pragmatism of freshmen (Astin's input idea of the environments they will help create. If score), and .62 with having no extracurricular ac- colleges wish to change their environment, prob- tivities. "Awareness" correlated .53 with status ably the simplest and most effective way to do so of freshmen, .47 with majoring in the humanities, is to recruit and select new students with differ- and .42 with lack of a religious preference. "Pro. ent characteristics. priety" correlated .57 with the proportion of males in the college, .48 with planning a career in educa- Biographical factors in choosing a residence tion, and .55 with attending church regularly. Students who choose various kinds of residences "Scholarship" was correlated .60 with the mean clearlydiffer insocioeconomic status (Baird, 28 21 1969b; Dollar, 1966; Jackson and Winkler, 1964; where the members can relate frequently and in- Kaludis and Zatkin, 1966; Levine and Sussman, timately with one another. These differences 1960; Scott, 1965). Students entering fraternities should not be exaggerated, however; most differ- or sororities generally come from higher SES ences are small, and there are no differences on homes than do students who enter other kinds of many variables. residences or who live at home. In addition, those These results are interesting and plausible. who enter fraternities and sororities generally par- However, college administrators and others who ticipated in more extracurricular activities in high work with students may well wonder how they school (Baird, 1969b) and describe themselves as could use such information. It is particularly im- being oriented toward social activity (e.g., Scott, portant to know how information of this type 1965; Stark, 1965; Baird, 1969b). could be used to make residence groups more ef- Johnson (1970) and Miller (1973) compared fra- fective agencies of educational and psychological ternity members and independents on the scales growth. Biographical-Wpm-la( ion -cou-Idhel p-ad--- of the College Student Questionnaire (Peterson, ministrators and student personnel workers un- 1965). Both investigators found that fraternity derstand the backgrounds, aspirations, and ex- members had lower scores on measures of in- pectations of students in various kinds of resi- dependence of peers and of liberalness. Miller dences. But a potentially more important use of found that fraternity members scored lower on biographical information is as a basis for experi- the scales of cultural sophistication and higher on mental manipulation of the composition and family social status. Both investigators found the group dynamics of residence groups. A handful of groups were equal on motivation for grades, in- studies have attempted such manipulation. Brown dependence from family, and social conscience. (1968) found experimental effects on students' (Reviews of the many studies of residence groups choice of major, friendship patterns, and satisfac- are provided by Feldman and Newcomb. 1969, tion. Morishima (1966) found experimental resi- and Longino and Kart. 1973, to which the reader dence effects on attitudes and "scholarly orienta- is referred.) tion." Beal and Williams (1968) produced effects Research using biographical and other brief on student satisfaction by changing the pattern of self-report information indicates that fraternity residential grouping. These studies need to be and sorority members, in comparison with those replicated and extended, of course. But they are in other residence groupings, tend to come from suggestive of the kinds of possible actions that higher economic backgrounds, to be more so- could be taken to increase the educational and cially oriented, to plan to be active in campus adjustment outcomes of students. affairs, and to be more socially assertive. These differences are largely the result of self-selection Biographical factors in choosing and selection by the groups (Michigan Student a major and in vocational choice Survey, 1967a, 1967b). Feldman and Newcomb's, The choice of a major field may be an even more and Longino and Kart's reviews suggest that col- complex process (but perhaps more understand- lege residence also influences withdrawal rates, able) than the choice of college (Crites, 1970; Hol- and changes in attitudes and values. Most re- land, 1973; Osipow, 1973; Super & Crites, 1962). searchers agree that the peer-group influence is (The biographical correlates of choosing particu- probably the most important factor in the educa- lar career fields are reviewed in section V. Much tional growth of students (Newcomb and Wilson, of the evidence in that section applies to the 1966), especially when the groups are small, and choice of major.) Again, SES is related to the 29

22 1969b; Dollar, 1966; Jackson and Winkler, 1964; where the members can relate frequently and in- Kaludis and Zatkiti, 1966; Levine and Sussman, timately with one another. These differences 1960; Scott, 1965). Students entering fraternities should not be exaggerated, however; most differ- or sororities generally come from higher SES ences are small, and there are no differences on homes than do students who enter other kinds of many variables. residences or who live at home. In addition, those These results are interesting and plausible. who enter fraternities and sororities generally par- However, college administrators and others who ticipated in more extracurricular activities in high work with students may well wonder how they school (Baird, 1969b) and describe themselves as could use such information. It is particularly im- being oriented toward social activity (e.g., Scott, portant to know how information of this type 1965; Stark, 1965; Baird, 1969b). could be used to make residence groups more ef- Johnson (1970) and Miller (1973) compared fra- fective agencies of educational and psychological ternity members and independents on the scales growth. Biographical-inforinat ion-could--hel p-ad-- of the College Student Questionnaire (Peterson, in i nist ra t or s and student personnel workers un- 1965). Both investigators found that fraternity derstand the backgrounds, aspirations, and ex- members had lower scores on measures of in- pectations of students in various kinds of resi- dependence of peers and of liberalness. Miller dences. But a potentially more important use of found that fraternity members scored lower on biographical information is as a basis for experi- the scales of cultural sophistication and higher on mental manipulationof the composition and family social status. Both investigators found the group dynamics of residence groups. A handful of groups were equal on motivation for grades, in- studies have attempted such manipulation. Brown dependence from family, and social conscience. (1968) found experimental effects on students' (Reviews of the many studies of residence groups choice of major, friendship patterns, and satisfac- are provided by Feldman and Newcomb. 1969. tion. Morishima (1966) found experimental resi- and Longino and Kart, 1973, to which the reader dence effects on attitudes and "scholarly orienta- is referred.) tion." Beal and Williams (1968) produced effects Research using biographical and other brief on student satisfaction by changing the pattern of self-report information indicates that fraternity residential grouping. These studies need to be and sorority members, in comparison with those replicated and extended, of course. But they are in other residence groupings, tend to come from suggestive of the kinds of possible actions that higher economic backgrounds, to be more so- could be taken to increase the educational and cially oriented, to plan to be active in campus adjustment outcomes of students. affairs, and to be more socially assertive. These differences are largely the result of selfselection Biographical factors in choosing and selection by the groups (Michigan Student a major and in vocational choice Survey, 1967a, 1967b). Feldman and Newcomb's. The choice of a major field may be an even more and Longino and Kart's reviews suggest that col- complex process (but perhaps more understand- lege residence also influences withdrawal rates, able) than the choice of college (Crites, 1970; Hob and changes in attitudes and values. Most re- land, 1973; Osipow, 1973; Super & Crites, 1962). searchers agree that the peer-group influence is (The biographical correlates of choosing particu- probably the most important factor in the educa- lar career fields are reviewed in section V. Much tional growth of students (Newcomb and Wilson, of the evidence in that section applies to the 1966). especially when the groups are small, and choice of major.) Again, SES is related to the 29

22 choice of majors. High SES students tend to "over- original and creative." Students choosing busi- choose" arts and humanities, social science, pre- ness had lower academic performances. Catholics law and political science, and premedicine. (Stud- are relatively more attracted to business, and ies of law students (Warkov and Zelan, 1965; blacks less attracted. Interest in people, SES, and Baird, 1974) and medical students [Rogoff, 1957; hometown were not strongly related to the choice Baird, 1974) indicate that these professions draw of business. The choice of physical science was heavily from the upper and middle classes.) Low negatively related to interest in working with peo- SES students tend to "overchoose" education, ple. The choice of the physical sciences was as- engineering, and certain sciences (Davis, 19651); sociated with the occupational value of originality Baird, Clark, and Hartnett, 1973). among men and with "making money" among Students in different fields also vary on many women. Students choosing the physical sciences other "biographical" variablesself-ratings (Abe were more often high academic performers. SES and Holland, 1965; Holland, 1968), life goals (Abe was unrelated to freshman choice of physical sci- and Holland, 1965; Holland, 1968), and _values ence, but low status students, particularly those sought from jobs (Davis, 1965; Slocum, 1974; with appropriate values, were more likely to re- Thistlethwaite, 1963). The largest studies of this main in science and to shift into it during college. type are Davis' Undergraduate Career Decisions Davis (pp. 75-76) summarizes the major pattern (19651) and Great Aspirations (1965a), Astin and of findings for the correlates of change: Panos' Educational and Vocational Development In general, the items that discriminate between of College Students (1969), and Baird's Careers defectors and those remaining in a field are the and Curricula (1974). Davis' study was concerned same as those distinguishing between recruits with the changes in students' career choices dur- and nonrecruits. Thus, the same sorts of charac- ing their undergraduate years. The biographical teristics are related to remaining in a given field variables associated with original choice, reten and shifting intoit from a different freshman tion of choice, and recruitment to choice in nine choice. general occupational groups were studied. The In general, the items related to choice during groups included: education, business, humani- college are the same items that associated with ties, other professions, law, physical science, freshman choice. Thus, differentiation during col- medicine, and engineering. The associations in lege tends to confinue directions of selection al- these analyses were generally moderate but typ- ready begun at entry into college. ically are as high as those reported in similar The social sciences constitute an exception to analyses conducted with personality and other these generalizations, the suggestion being that test scales. Davis summarized his results for each college experience leads to a shift in the sort of field. In some cases the summaries required a student attracted to these fields. number of qualifications, as when a general trend Davis' conclusions suggest that undermanned did not hold for a particular subgroup. As ex- fields could recruit students from groups of stu- amples of his results, his summaries for business dents with particular characteristics. and the physical sciences can be outlined. Busi- In contrast to Davis and to Baird, Astin and ness is a masculine field among freshmen and Panos (1969) had actual data on students' char- tends to recruit proportionally more men during acteristics when they were freshmen rather than the course of college. Students choosing business retrospective accounts. Astin and Panos found placed more emphasis on the values of "making a that freshman career choice was generally the lot of money," and less on "opportunities to be best predictor of final career choice when stu- 30 23 dents were followed up w hen they were seniors. Analyses indicated that the sample included pro Major field added to the pred4 lion in most cases. portionately slightly fewer minority students than Other variables. such as sex, highest degree did the nonrespondents. but the sample did not sought. parents' income, and high school grades. seem to be biased in any other way, and included also added to the multiple regression equations. an extremely wide variety of students. (Astin and Panos found that income w as related The College Senior Survey included a great to career or final major field, whereas parental deal of self-reported biographical, personal, at- education was not.) They concluded: titudinal and educational information about stu "By far the best predictors of the student's dents. Reports of theirGRE, LSATandMCAT final major field and career choice were his initial scores were also obtained. choices as reported four years earlier at the time The followup questionnaire ascertained stu- of matriculation. The characteristic having the dents' educational and vocational activities. The next greatest predictive value over the four-year criteria used were: working full-time; pursuing interval was the student's sex. The trend toward graduate study in the arts or humanities; pursuing greater differentiation of sex roles during the un- graduate study in the social sciences; pursuing dergraduate years was pronounced, with men graduate study in the biological and physical sci- gravitating toward the more `masculine' careers ences, attending medical school; and attending (i.e.. those that were initially preferred more by law school. men than by women) and women gravitating to- Baird studied the multiple regression results ward the more 'feminine' careers. The same was when pustcollege activities were used as criteria true for major field. and the senior survey data were used as predic- . .In general, the student's career choice tors. These analyses were made twice, first with. tended to move into closer conformity with the out including senior plans for the activity, and inure popular choices among his fellow students. then including them. This was done because This effect was particularly evident in the case of plans correlated so highly with activity. Baird's the fields of engineering, teaching, law, and busi- results for each activity are summarized below. ness." (Astin and Panos, 1969, p. 132) Working full time. The results for working In contrast to Davis and to Astin and Panos, (R = .56 with plans, .40 without) indicate the var- Baird (1974) examined the relation between bio- ious reasons for working: poor grades, poor test graphical and other simple self-report information performance,lateconsiderationof advanced about students w hen they were seniors and their study, parents' education and influence, being actual educational activities a year later. The re- married, and lowself-conceptions of ability. search was specifically designed to gain detailed Wont( n were more likely to be working rather information about the factors that determine who than attending graduate or professional school, goes where to graduate and professional school. although they had better grades than men, on the The data for this study came from a follow up of average. The variables suggested a life history of a national survey of a sample of college seniors lack of parental and peer encouragement of edu- who replied to a questionnaire, The College Sen- cational aspirations either because of the stu- ior Survey, in the spring of 1971 (Baird, Clark. dent's sex, peers, or background, associated with and Hartnett, 1973). Some 7.734 people who had poorer academic performance. been seniors in 94 colleges across the country Graduate work in the arts or humanities. The replied to a followup questionnaire designed to zero-order correlates of graduate work in the arts determine their activities a year after college. or humanities present a familiar picture of stu-

O 24 dents who rate themselves high in relevant abili- earlier studies of students' career plans (Baird, ties (creativity, artistic and writing ability) and 1973; Astin and Palms. 1969; and Davis, 1965) sug- low in science-related abilities. who have been gesting the importance of early consideration and rewarded for their work by faculty encourage- interest in a field, self-confidence in one's rele- ment or grades, and who began college with a vant abilities, high test scores, and academic suc- career in arts or humanities in mind. The multiple cess. The results have a number of implications regression results for graduate work in the hu- for theories of vocational choice and for practical manities (R = .SO with plans, .38 without) indicate policies, such as encouraging certain groups to the relevance of high verbal ability, grades. and a consider advanced study, encouraging early con rejection of income as a basis for vocational sideration of a variety of careers, and counsel- choice. ing intervention to encourage proper curricular Graduate work in biological or physical science. choices by students. The correlational results (R = .68 with plans, .53 These studies indicate that many students' without) presented a picture of the future science career choices change during college, but most graduate student as a student high in mathe- students who do change move to similar fields, matical ability who wishes to contribute to knowl- such as from physics to chemistry. Simplified, edge. who is not particularly interested in people, and generalized, and ignoring many complica- and who gained the attention, encouragement and tions, this process suggests that whether a stu- employment of the science faculty. dent remains in a field or changes to a new one The correlates of graduate study in the social usually seems to depend on the congruency of the sciences (R = .52 with plans. .33 without) indi- demands and characteristics of the field with his cated that the future social science graduate stu- needs and characteristics. If the student's present dents planned an advanced degree, and were con- field is the more congruent, the student will re- fidentof their academic abilities. They also main in it; if another field is more congruent, the tended to have planned a career in social science student will move to it. as college freshmen. Students' views of their own talents seem to be The correlates of studying law (R = .81 with highly related to the decision process (Holland, plans, .49 without) indicated that prospective law 1973, has summarized some of the evidence). In students tended to be bright males who came each of these areas, students seem to seek the from families who encouraged advanced educa- careers most in line with their characteristics. tion. They considered advanced training rela- All of this makes the process of choosing a tively early, entered college with law in mind, but career field sound incredibly rational. In fact, were not encouraged to seek further training by choices are often made subconsciously, or as a their colleges' faculty. They tended to have con- consequence of chance encounters, course sched- fidence in their ability and tended to voice an in- ule changes, etc. The results just described are terest in status, prestige, and income as reasons general ones, applying to large groups of students. for their choice of profession. Many people, quite happy in their current ca- The results (R = .84 with plans, .65 without) reers, would have difficulty giving a coherent, not suggested that the most salient correlates of going to mention rational, account of how they chose it. to medical school were an early initial choice of However, even if a particular individual's choices medicine, greater concern for service than money, may be hard to explain (Holland, 1973), there is a and high test scores. great deal of rationality in the process of changes The results of this study are consistent with of groups of students. Obviously, the chances 32

25 . that a student would choose an appropriate career A second kind of biographical variable related would be increased if the choice were made ra- to attrition in college is the students' desire for tionally. Unfortunately, such rationality requires the degree and interest in college work. For ex- so much knowledge about people, careers, their ample, Trent and Medsker (1968) found that the interaction and the future, that only the broadest single variable most predictive of dropping out of sort of guidance can be provided at this time. college (to judge from the size of the X2 figures) was a simple question about the students' feelings Biographical information as a predictor about the importance of college. of attrition in college Bayer (1968), using a sample of 8,567 partici- The problem of school dropouts has been a con- pants in Project TALENT, studied the relation of cern to educators for nearly a century. Since thirty-eight psychological and demographic vari- Sputnik, of course, there has been great concern ables to educational outcomes. For men, dropping about the "loss of talent." Many people believe out versus completing college was most strongly that one of society's goals should be to provide associated with ability variables, and then by opportunities for everyone to pursue his or her family of procreation variables (e.g., age at which educational goals to the fullest. The magnitude the student planned to be married); for females, of this concern has led to the development of a the strongest associations were family of pro- voluminous literature. One review of research lit- creation variables, and then ability variables. For erature on dropouts found some 800 references both sexes, socioeconomic status had slightly published before June 1956 (Blough, 1956). A lower associations with dropping out. A similar more recent review of studies of school dropouts pattern of results held for the criterion of delayed (Varner, 1967) referred to 149 studies or sum- completion of college versus completion in five maries. Most of these were dated in the 1960's. years, except that college commitment (plans for 70 The literature concentrating on college students college, college plans encouraged by parents and is not quite as vast, but itis still large (see the peers) was the best predictor for women, and the reviews by Eckland, 1964; Iffert, 1957; Iffert and second best predictor for men. Clarke, 1965; Spady, 1970; Summerskill, 1962; The Folger, Astin and Bayer study (1970), de- Tinto, 1975). scribed earlier in the section on college atten- In general, attrition during college is related to dance, also examined the variables associated both previous academic success and academic with dropping out or completing senior college. ability as well as socioeconomic status, and also to Ability variables were most strongly associated other variables. However, the relative importance with completion of college, followed by the stu- of these variables in various studies is difficult to dents' marital and parental status, interest vari- assess because of problems of sampling, equating, ables, temperament variables, parents' socioeco- and the range of student input on the variables. nomic status, college, commitment variables, and Research is also complicated by varying defini- ethnic and religious background variables. tions of dropping out or attrition (Eckland, 1964) Among the very bright students followed by the and by the multiplicity of factors involved in leav- National Merit Scholarship Corporation, Astin ing college (Sexton, 1965). In the studies that have (1964) found that Merit finalists who drop out of included both ability and SES, there have been college tended to come from lower SES back- mixed results. However, it is clear that SES con- grounds, to have lower ranks in high school, to tinues to influence students' educational accom- plan initially to get lower college degrees, and to plishment. apply for fewer scholarships. (Personality test dif-

3 3 " social, college-environmental. and personal vari- his results. Astin recommends a variety of admin- ables, as well as things that are unique to each istrative actions to reduce dropout rates. These person w ho withdraws fromco..ege.11 However. the include. offering work-study programs and avoid- most important personal variables seem to be aca- ing loans, encouraging employment, especially by demic achievement, family background. motiva- offering on-campus jobs, and encouraging resi- tional importance of the degree. aspirations. and dency in dormitories. Astin also found that stu- self-concept. Self-report questionnaires seem to dent-bodyhomogeneity enhances persistence; be a useful device for assessing many of these i.e., students are less likely to drop out if they at- variables. tend a college whose other students are similar in Biographical information may hdp to identify religion, race, and size of their hometown. Ob- the potential dropout, but can it aid in attempts to viously, this last finding is more difficult to use. keep the potential dropout from becoming an ac- Astin makes additional recommendations con- tual dropout? One contribution that biographical cerning counseling, tuition, and facilities con- information can make is to diagnose the particular struction. As Astin's study indicates, there are reason for possibly dropping out. For example, ways to use research findings to influence student different approaches might be taken depending attrition in college. on whether the primary reason for possibly drop. ping out was financial, poor study skills, family Biographical information related to conflicts, lack of peer group support, choice of an student activism inappropriate major, or choice of an inappropriate In the recent flood of materials about student ac- college (Knoell, 1966). By its provision of broad- tivism, there have been a few large scale, com- band information, biographical data can help petently performed empirical studies (see the counselors and administrators plan the best ap- bibliography by Altbach and Kelley, 1973). For proaches to help students stay in college. These example, Jansen, Winborn, and Martinson (1968) approaches might go beyond traditional counsel- found that background variables, simple self- ing (see Island, 1969; Korn, 1969) or group coun- descriptions and self-reports of behavior distin- seling (Anderson, 1969), to include experimental guished between different types of student lead- residence groupings (e.g., Brown, 1968), special ers at Indiana University. The groups included courses, or change of college. Perhaps the provi- political action groups, residence hall leaders, sion of reliable information about the realities of fraternity leaders, and activities leaders. Astin colleges could help students know what they are (196814 found that background information pre- getting into, and encourage them to adapt to col- dicted with some efficiency whether students lege requirements with greater skill. would participate in protests against the Vietnam Astin (1975) has recently attempted to carry out War, racialdiscrimination, or the policies of sonic of these tasks. He presented the usual find- the college's administration. Furthermore, Astin ings, based on 41.000 freshmen at 358 represen- found that demographic characteristics of the tative colleges, which showed that dropping out freshman class predicted the extent of protest on was predicted by low academic ability, poor aca- campuses with considerable efficiency, when the demic and family background, poor study habits, institution was used as the unit of analysis. Kahn and low aspirations. However. Astin developed and Bowers (1970) found that socioeconomic worksheets, based on these results, that adminis- status was related to activism, but that this was trators can use to compute dropout proneness for the case only in highly selective institutions. The individual students or groups of students. From same pattern also held for the number of hours

28 ferences were also found.) Watley (1968) found Spady has noted that studies of the relation of that, for his sample of Merit Scholars, initial edu- attitudes to attrition need to be examined care- cational aspirations assessed in 1957 were gen- fully. erally related to educational attainment in 1964 "...when socioeconomic and attitudinal fac- e.g., among the men who .eventually obtained a tors are considered simultaneously, the advan- doctorate or professional degree, 100 percent had tages thought to accrue to individuals with par- indicated that they planned such a degree seven ticular kinds of attitudes do not exist indepen- years earlier. However, the accuracy was not high dently of their family background. It is background 42 percent of men without any degree had also experiences, then, that both lead to and account said they planned a doctorate. Fairly sizable dif- for the attitudinal differences often associated ferences between the students who had obtained with attrition." (Spady, 1970, pp. 69.70) a degree and those who had not were found in He then notes the interaction of the background family background variables, and in students' and dispositions that students bring to college self-ratings on drive to achieve, perseverance, with the characteristics of the college and with 1P'i concern with scholarship, expressiveness, emo- specific college experiences. tionalstability, and self-understanding. Differ- "The major inference to be drawn from this en- ences on the California Psychological Inventory tire set of findings would appear to be that sur- were also found. Recent studies by Chase (1968) vival in college is dependent largely on a clear and as well as by Rossman and Kirk (1970) ',lave also realistic set of goals and having interests that are shown withdrawal to be related to a wide variety compatible with the influences and expectations of biographical variables. of departmental faculty and curricula. Men in In the largest national study of attrition ever particular, however, appear to maintain high ex- conducted, Astin (1972) examined various criteria pectations despite the academic realities of col- of attrition in college. Studying a sample of 45,432 lege life.... students enrolled in 194 four-year colleges, Astin "In the main, then, the findings discussed here used information about the students gathered suggest that interpersonal relationships facilitate when they were freshmen to predict attrition over greater integration of the student into the social four years. Using the criteria of either "obtained system of the college. To the extent that peer the bachelor's degree or still enrolled," Astin group norms either emphasize or denigrate aca- found that self-reported high school grade point demic endeavor, they may also influence achieve- average was the single best predictor of per- ment within the academic system, but this in- sistence followed by academic test scores, sex fluence is more often implied in theory than veri- (male), students' assignment of low probability to fied empirically." (Spady, 1970, p. 72; p. 77) being married in college, being a nonsmoker, and Spady then goes on to develop a theoretical not being employed during college. Of course, the model of dropping out, which he tested in a later reverse of these variables predicted dropping out. study (Spady, 1971). This kind of theoretical work Astin found similar results in analyses of attrition seems a necessary step in furthering our knowl- in two-year colleges; lie also found similar results edge about dropping out, and the role played by when he used other criteria of attrition. background variables. Further empirical work When reviewing these studies, it is well to re- may add to our stack of materials on dropping out, member that some theoretical explanations are but some conceptual guides through the maze of needed to account for dropout, as Spady (1970) results is needed. and Tinto (1975) have emphasized. For example, In summary, attrition in college is the result of

3 4.

27 social, college environmental. and personal vari- his results. Astin recommends a variety of admin- ables, as well as things that are unique to each istrative actions to reduce dropout rates. These person who withdraws from college. llowe'ei. the include: offering work-study programs and avoid- most important personal variables seem to be aca- ing loans, encouraging employment, especially by demic achievement. family background. motiva- offering on-campus jobs; and encouraging resi- tional importance of the degree. aspirations, and dency in dormitories. Astin also found that stu- selconcept. Self-report questionnaires seem to dent-bodyhomogeneity enhances persistence; be a useful device for assessing many of these i.e., students are less likely to drop out if they at- variables. tend a college whose other students are similar in Biographical information may hdp to identify religion, race, and size of their hometown. Ob- the potential dropout. but can it aid in attempts to viously, this last finding is more difficult to use. keep the potential dropout from becoming an ac- Astin makes additional recommendations con- tual dropout? One contribution that biographical cerning counseling, tuition, and facilities con- information can make is to diagnose the particular struction. As Astin's study indicates, there are reason for possibly dropping out. For example, ways to use research findings to influence student different approaches might be taken depending attrition in college. on whether the primary reason for possibly drop. ping out was financial, poor study skills, family Biographical information related to conflicts, lack of peer group support, choice of an student activism inappropriate major, or choice of an inappropriate In the recent flood of materials about student ac- college (Knoell, 1966). By its provision of broad- tivism, there have been a few large scale, com- band information, biographical data can help petently performed empirical studies (see the counselors and administrators plan the best ap- bibliography by Altbach and Kelley, 1973). For proaches to help students stay in college. These example, Jansen, Winborn, and Martinson (1968) approaches might go beyond traditional counsel- found that background variables, simple self- ing (see Island, 1969; Korn, 1969) or group coun- descriptions and self-reports of behavior distin- seling (Anderson, 1969), to include experimental guished between different types of student lead- residence groupings (e.g., Brown, 1968), special ers at Indiana University. The groups included courses, or change of college. Perhaps the provi- political action groups, residence hall leaders, sion of reliable information about the realities of fraternity leaders, and activities leaders. Astin colleges could help students know what they are (1968b) found that background information pre- getting into, and encourage them to adapt to col- dicted with some efficiency whether students lege requirements with greater skill. would participate in protests against the Vietnam Astin (1975) has recently attempted to carry out War, racialdiscrimination, or the policies of some of these tasks. He presented the usual find- the college's administration. Furthermore, Astin ings, based on 41,000 freshmen at 358 represen- found that demographic characteristics of the tative colleges, which showed that dropping out. freshman class predicted the extent of protest on was predicted by low academic ability, pour aca- campuses with considerable efficiency, when the demic and family background, poor study habits, institution was used as the unit of analysis. Kahn and low aspirations. However, Astin developed and Bowers (1970) found that socioeconomic worksheets, based on these results, that adminis- status was related to activism, but that this was trators can use to compute dropout proneness for the case only in highly selective institutions. The individual students or groups of students. From same pattern also held for the number of hours

28 spent studying, grades, and majoring in social and tend to be employed in occupations for which science or humanities. Being "intellectually ori- advanced educational attainment is a primary req. ented" was related to activism in colleges at all tiisite. levels of selectivity. The typical activist's family is quite secular; Baird (1970) used many of the types of bio- however, a significant minority of activists come graphical information reviewed inthis mono- from families with a strong religious orientation. graph. Using a scale of degree of activism, Baird The typical activist's parents are politically lib- found that students exhibiting many behaviors eral; the proportion of activists who have "con- one would classify as "activism" were charac- verted" from a background of conservatism is terized by: self-ratings describing themselves as quite small, as is the proportion who have parents socially ascendent and capable, socially sensitive who are themselves left-wing or liberal activists. and gregarious, aesthetically talented and inde- The religious secularism and political liberal- pendent; life goals concerned with holding power ism characteristic of activist families are expres- and directing others, helping and community serv- sions of an underlying cluster of values articulated ice, and intellectual activity; high school non- by parents and shared by their activist offspring. academic achievements, in every area, especially Activists tend to come from homes in which leadership; and interest, potential, competency, a relatively democratic and equalitarian child- and personality scales suggesting social and lead- rearing ideology was emphasized; there is little ership abilities, writing and artistic talent, and evidence, however, for the popular view that stu- self-assertiveness. dent protest is related to parental over-permis- Flacks (1970) reviewed the empirical evidence siveness or indulgence. concerning the social bases of activism, and came Although the prototypic background of student to several conclusions: activists is the "educated humanist" family, fac- Movement participants tend to be recruited tors other than family background are also im- from the most selective universities and colleges; portant in determining recruitment to the move- the highest incidence of offcampus and oncampus ment, and forms of participation in it. protest activity has occurred at major state and Although Flacks was not concerned with the private universities and prestigious liberal arts issue of biographical information, it is apparent colleges. (This may be because such institutions that many of his conclusions are based on bio- attract or select "protest-prone" students.) graphical data. It may be that it is not possible to Student protesters are rarely recruited from understand the roots of activism without bio- among those with below average grades; there is graphical information. some tendency for those with high grades to be It is difficult to suggest possible uses of such disproportionately represented in protest activity. analyses. It might be possible to use this informa- In terms of aptitudes and interests, activitists tion to identify possible activists, but the idea of tend to be more "academic" than nonactivists. barring students of this type would be abhorent to (Perhaps a better way of putting this is that ac- many college officials. Furthermore, such stu- tivists are more intellectually oriented. Activists dents would be inclined to attend selective col- do not always hold traditional academic values.) leges, be intellectually oriented, come from lib- Activists are disproportionately recruited from eral, educated, and humanist families, and have a particular social background: they are the sons unusual potentials for leadership; and these are and daughters of high-income families, in which just the sorts of students many colleges look for both parents have at least four years of college and that society needs. 36

29 16

Section III: The prediction of here is concerned with biographical and other creative and high level accomplishment simple predictors of creativity and high level per- with biographical formance. Studies of the creative process (Golann, and self-report information 1963) or of the personality of persons who achieve at high levels (recently reviewed by Dellas and Gaier, 1970) are, of course, very valuable in our People have been concerned with the prediction search for the bases of high level accomplish- of creativity and high level accomplishment for ment, but this review concentrates on biograph- many years. Thousands of years ago, the Chinese ical or background information about previous ac- developed the imperial examination system in an tivities and accomplishments. Without attempting effort to find people who would be outstanding of- to citeall the comparative success of studies ficials and ministers of the state. More recently, using different classes of variables to predict high researchers have undertaken a wide variety of level accomplishment, we simply assert that, from studies to determine the antecedents of high level the current evidence, the conclusion of Taylor and accomplishment in science, writing, creative arts, Holland (1964) still stands: biographical informa- and leadership. This review summarizes some of tion is consistently the best predictor of creative this research, and shows that there is evidence and high level performance. Further, information that creativity and high level accomplishment can about the biographical precursors and the devel- be predicted with some success, even if we do not opment of accomplishment could lead to changes have a complete understanding of the processes in personal, educational, or organizational prac- of creativity and achievement. Much of this re- tices that would foster greater accomplishment. search has been conducted with samples of col- This review also concentrates on studies of real- lege students and professionals working in in- lifecriteria of accomplishment, following the dustry. guidelines of MacKinnon (1962). These include: Consequently, since our major interest is in re- originality, uniqueness, or statistical rarity; ad- gard to what the research findings suggest might aptation to reality, aiding in the achievement of be done to improve educational admissions, our some real-life goal, such as a scientific or aes- argument will have to be somewhat indirect. How- thetic problem; and sustained activity leading to ever, many of the questions that apply to admis- the development, evaluation, and elaboration of sions have been addressed in research in other an original idea. Studies based on such criteria as areas. Their results are more relevant than they having a "creative" profile on a personality test, might at first appear to be. The material reviewed or other arbitrary classifications devised by a re- 37

30 searcher receive less attention. accomplishment in college was accomplishment Biographical inventories of earlier activity and in the same area in high school, as measured by accomplishment have been related to creativity simple check lists of nonacademic achievements. and high level accomplishment in several popula- (Similar results have been found in a large sample tions: college students, high school students, sci- study of more typical students, Baird, 1969a.) entists, and professional people (professors, doc- Other National Merit studies by Roberts (1965) tors, lawyers, and so on). These studies will be and Nichols (1966) studied the item correlates of reviewed in the sections that follow. They are high level accomplishment. Roberts developed presented here as evidence fortile power of scales for six areas of accomplishment: science, measures of accomplishment at one level to pre- art, writing, music, leadership, and speech (as dict accomplishment at another. Their relevance defined by the same sort of items described for college admissions decision-making will be earlier). In general, more achievers in each area discussed in the final section. endorsed the items expressing interest, activity, or competence in each area than did the non- College students achievers. These items tended to be directly re Creativity and high level accomplishment among lated to the kinds of accomplishments later ex- students has been examined in many studies. As hibited in college. As Roberts states: "Many of a natural outgrowth of their concern for identify- the items in each scale were directly content- ing talent, the National Merit Scholarship Cor- related to the area of criterion achievement, and poration reported a series of studies concerned a fair number were related to other specific areas with high level accomplishment. These studies of activity and achievement." For example, in the found many significant relationships between bio- science scale, more than half the positive pre- graphical information and achievement in college. dictors were "direct indicators of scientific activ- "Achievement" measures consisted of such state- ity or interest and several others may be `techno- ments as: "Had a scientific or scholarly paper pub- logical'innature (for example, photography, liLled (or in press) in a scientific or professional nature collections)." Nichols' correlations also journal," "Received an award for acting, playwrit- indicated that previous behaviors were generally ing or other phase of drama," "Was editor or fea- the best predictors of high level accomplishment ture writer for collegiate paper, annual, magazine, in both a Merit sample and a sample representing or anthology, etc.," "Composed music which has a broad range of talent. (Biographical information been given at least one public performance," about previous accomplishments was a better pre- "Won a prize or award in an art competition, dictor than the personality, interest, or ability painting, sculpture, ceramics, etc.," "Organized scales that Nichols also used in his study.) Similar a college political group or campaign." Scales National Merit studies reached similar conclu- were developed for six areas: science, art, music, sions (Holland, 1961; Holland and Astin, 1962). leadership, drama, and writing. In studies by Hol- Other studies, using large samples of average land and Nichols (1964), and Nichols and Holland students, have shown that scales measuring high (1963), nearly every major test that has been sug- school nonacademic accomplishment are the best gested for the prediction of accomplishment was predictors of later accomplishment in college and used in the predictor battery, including person- have sufficiently high correlations to be of prac- ality scales of all sorts, interest measures, as tical use (Richards, Holland, and Lutz, 1967; sessments of cognitive styles, "creativity tests," Holland and Richards, 1965, 1967; Richards and and high level ability tests. The best predictor of Lutz, 1968; Baird, 1969a). Ability, personality, and 3;

31 interest measures were generally poor predictors previous activity and accomplishments, to high in these studies. school students participating id a National Sci- The need for measures of out-of class accom- ence Foundation summer program. Using ratings plishment in addition to measures of academic ac- of the creativity of the students' research per- complishment is evidenced by the fact that, in all formance as the criteria, the biographical ques- these studies, there was little relation between tionnaire proved to be the best overall predictor. grades, academic ability as measured bytest Schaefer and Anastasi (1968), and Anastasi and scores, and later accomplishment. The need is Schaefer (1969), developed biographical inventory further emphasized by the work of Wallach ante keys against criteria of creative accomplishment Wing (1969) who replicated these studies in their among high school boys and girls. Separate keys study of Duke University students in which little were developed for science and for art and crea- relationship between academic and nonacademic tive writing. They were cross-validated in second achievement was found using methods other than criterion groups. Cross-validated validity coeffi- correlations.Baird (1968)similarly compared cients among the boys were .35 and .64 for the bright and average students and found little av- science and art- writing scales, respectively. For erage difference intheir nonacademic accom- girls, art and writing were predicted in a cross- plishments. Elton and Shevel (1969) further clari- validation with correlations of .34 and .55, re- fied the issue by examining individual items on spectively. Using a similar biographical inventory six scales of accomplishment and finding that and the same sample, Schaefer (1969) was able some out-of-class accomplishments were related to predict creative performance in art for boys to measures of academic talent but about an equal (.65), writing for girls (.55), and, in combination number showed a negative or no relationship. The with personality scales, science for boys (.48), and same lack of relation was found by Holland and art for girls (.55). In their discussion of the con- Richards, 1965, 1967; Richards. Holland, and tentsof these scales, Anastasi and Schaefer Lutz,1967:Richards and Lutz, 1968; Baird, (1969) pointed to the common characteristics of 1969a. b; and the American College Testing Pro- high performing adolescents (with some support gram, 1973. from other studies). These were: continuity and pervasiveness of interest in the students' chosen High school students field; prevalence of unusual, novel, and diverse The studies of high school students provide some experiences; and the educational superiority of what indirect evidence of the power of measures the students' family background. The first point of previous accomplishment to predict later ac- deserves some reemphasis. Two recent studies complishment. They are reviewed here because (Baird, 1968,1969b) indicate that accomplishment they show that previous activity and experiences uften begins in adolescence or before in explora- are also related to later accomplishment, as well tory activity, often resulting in recognized achieve- as earlier accomplishment. Long term activity ment. Baird and Richards (1968) and Baird (1969h) and interest in an :reit may not result in publicly found that such accomplishment seldom begins in recognizable accomplishments, but they do show college; there are few "late bloomers." The great that behaviors consistent with later accomplish- majority of students who show accomplishments ment are important; accomplishment does nut ap- in college showed similar activities in high school. pear overnight. Taylor, Cooley, and Nielson (1963) Anastasi and Schaefer (1969) point out: applied a biographical questionnaire, developed "Typically, the highly creative adolescent girl on NASA scientists and which concentrated on in this study had manifested an absorbing interest

3 9 32 in her field since childhood and her creative ac- speetively. The content of the items suggests high tivities had received recognition through exhibi- self-confidence and high selfconception. "This tions, publication, prizes, or awards. Her initial interpretation is reinforced by the frequency with interest was thus reinforced and reinforced early which the high criterion groups say that they: in life by persons in authority, such as parents and have more readily taken advantage of opporto li- elementary school- teachers. The continuity of lies presented them: consider their achievements creative achievement over time is corroborated by thus far to be greater than those of others with the findings of other investigations, notably Hel- the same education: work more quickly than son's (1966. 1967) research with college women, others; and prefer to have many things 'on the the surveys of National Merit Scholarship finalists fire' simultaneously." It might be noted That this (Holland and Astin, 1962; Nichols and Holland. description was based primarily on answers to 1963). and our own earlier study of creative high factual questions about the scientists' accom- school boys (Schaefer and Anastasi. 1968)." plishments. Chambers (1964) used both biograph- Similar results are reported in studies of indus- ical and personality test variables to study creativ- trial, scientific. and professional samples, as re- ityin chemists and psychologists. Three per- viewed below. sonality scales and 16 biographical items were significantly related to the criterion of creativity. Predicting creativity Compared to their less creative colleagues, the and high level accomplishments more creative scientists more often had fathers among scientists and other adult groups who were professional men, graduated from high from biographical records school earlier, spent more hours per week (more of accomplishment and activity than 50) in study and research in graduate schell, Biographical variables dealing with past accom- published more articles then, and more often met plishments, past a:ti ;t'l. and interest similar to their graduate school expenses by scholarships those just described have been found to charac- and fellowships than by part-time work. terize scientists who have demonstrated a high McDermid (1965) found that biographical vari- level of accomplishment. For example. Roe (1952) ables were the best predictors of ^upervisory and found many unusual biographical characteristics peer ratings of high level (in this ease, creative) of scientists m her sample. Kulberg and Owens performance. McDermid also used personality (1960) and Morrisson, Owens, Glennon, and Al- tests (the California Psychological Inventory and bright \1962) found that biographical information the Adjective Check List), an interest test (The correlated v,;th the creativity, professional inter- Vocational Preference Inventory), a high level in- est, and research competence of engineers and telligence test (Concept Mastery Test), the Social scientists. Albright and Glennon (1961) found that Insight Test. and Welsh Figure Preference Test. biographical variables distinguished between su- All these tests had been used in other studies of pervisory and research oriented scientists at all creativity, but were not useful in McDermid's levels of a laboratory organization. Smith, Al- sample of engineering personnel. McDermid con- bright, and Glennon (1961) also found that bio- cludes: "The correlations obtained in this study graphical information predicted rated scientific between paper and pencil tests and the criteria competence, rated creativity, and number of pat- of creativity were so low as to be virtually useless ents within a group of research scientists. These for predictive purposes; biographical data, on the three criteria were predicted in a cross-validation other hand, proved to be significant as predictor3 sample with correlations of .61, .52, and .52. re- of both supervisory and peer ratings of creativity.

4 0

I I This finding, of course, is quite consistent with for men and .00 for women, and the median corre- the practical dictum that the best predictor of lation between ACT composite scores and adult future performance is past performance...." accomplishments was .06 for men and JO for Taylor and Ellison (1967) summarized eight women. The median correlation between college years of work on the identification of biographical grades and adult accomplishments was .09 for predictors of scientific performance. In the last all students combined. This study is important NASA scientist samples the cross-validated corre- because it shows the long range validity of the lations with ratings of creativity were .41, with biographical accomplishment scales, even after the number of publications .62, and with GS level the intervention of college and work, and illus- .72. The factors in the Taylor and Ellison study trates, again, their superiority over other meas- were consistent both with other studies of ac- ures. complishment in science and the studies of stu- In summary, the studies we have reviewed sup- dents just summarized. For the highly performing port the conclusions reached by Baird (1969a): scientists, as for the students, the tendency was "There is some consensus, then, that students to have a conception of themselves as capable of who later achieve... (in creative activity, as well high level professional performance, to be in- as academic activity) have engaged in activities dependent of others' opinions, be greatly dedi- and developed skills related to that area, have cated to their work, to work very hard, to have conscious goals and desires to achieve in that clear ideas of their goals (which they set at a high area, and describe themselves as having ability level), and to be intellectually orienteda trait in that area. that developed early in adolescence. ". .. The achiever...has a history of activities Finally, NIunday and Davis (1974) have shown and achievements related to his present achieve. thatbiographical accomplishment scalesad- ment. He is motivated to achieve in this area and ministered in high school predicted adult ac- accurately assesseshis own talents. Perhaps complishment six years later. The adult accom- rather than attempting to develop new scales to plishmentmeasures included such items as: describe some universal creative mind, psycholo- "was author or coauthor of scholarly or scientific gists should concentrate on the development of article accepted for publication in a popular or more accurate and reliable measures of past ac- professional journal or presented as a public lec- tivities, goals, and self-description." ture," "received an award for acting or some These results and those of the student samples other phase of drama," "sold one or more works suggest that measures of accomplishment could of art to collectors, museums, or the general pub- be used for the early identification of students lic," "won a literary award or prize for creative with the potential for high level accomplishment, writing," "composed or arranged music which and as one of the bases for selecting students for was publicly performed," and "been a candidate special programs. In most of these studies, bio- for election to school board, city, county, or state graphical information about past accomplishment office." was the best predictor of later performance, bet- The median correlation between the high school ter than ability, interest or personality tests, sug- accomplishment scales and the corresponding gesting the power of these variables for particular adult scales was .25 for men and .26 for women purposes. when graduates and nongraduates are combined. In contrast, the median correlation between high school grades and adult accomplishments was .03 41

34 This finding, of course, is quite consistent with for men and .00 for women, and the median corre the practical dictum that the best predictor of lation between ACT composite scores and adult future performance is past performance...." accomplishments was .06 for men and .10 for Taylor and Ellison (1967) summarized eight women. The median correlation between college years of work on the identification of biographical grades and adult accomplishments was .09 for predictors of scientific performance. In the last all students combined. This study is important NASA scientist samples the cross-validated corre- because it shows the long range validity of the lations with ratings of creativity were .41, with biographical accomplishment scales, even after the number of publications .62, and with GS level the intervention of college and work, and illus- .72. The factors in the Taylor and Ellison study trates, again, their superiority over other meas- were consistent both with other studies of ac- ures. complishment in science and the studies of stu- In summary, the studies we have reviewed sup- dents just summarized. For the highly performing port the conclusions reached by Baird (1969a): scientists, as for the students, the tendency was "There is some consensus, then, that students to have a conception of themselves as capable of who later achieve... (in creative activity, as well high level professional performance, to be in- as academic activity) have engaged in activities dependent of others' opinions, be greatly dedi- and developed skills related to that area, have cated to their work, to work very hard, to have conscious goals and desires to achieve in that clear ideas of their goals (which they set at a high area, and describe themselves as having ability level), and to be intellectually orienteda trait in that area. that developed early in adolescence. "... The achiever... has a history of activities Finally, NIunday and Davis (1974) have shown and achievements related to his present achieve- thatbiographicalaccomplishment scalesad- ment. He is motivated to achieve in this area and ministered in high school predicted adult ac- accurately assesses his own talents. Perhaps complishment six years later. The adult accom- rather than attempting to develop new scales to plishmentmeasures included such items as: describe some universal creative mind, psycholo- "was author or coauthor of scholarly or scientific gists should concentrate on the development of article accepted for publication in a popular or more accurate and reliable measures of past ac- professional journal or presented as a public lee- tivities, goals, and selfdescription." tine," "received an award for acting or some These results and those of the student samples other phase of drama," "sold one or more works suggest that measures of accomplishment could of art to collectors, museums, or the general pub- be used for the early identification of students lic," "won a literary award or prize for creative with the potential for high level accomplishment, writing," "composed or arranged music which and as one of the bases for selecting students for was publicly performed," and "been a candidate special programs. In most of these studies, bio- for election to school board, city, county, or state graphical information about past accomplishment office." was the best predictor of later performance, bet- The median correlation between the high school ter than ability, interest or personality tests, sug- accomplishment scales and the corresponding gesting the power of these variables for particular adult scales was .25 for men and .26 for women purposes. when graduates and nongraduates are combined. In contrast, the median correlation between high school grades and adult accomplishments was .03 41

34 the brevity of the scales used in these studies. Some questions about direct assessment of In short, it appears that biographical accomplish- biographical accomplishment scales ment scales can be constructed that meet stan- We have just seen the power of biographical ac- dard psychometric requirements. complishment information to predict subsequent Can biographical accomplishment scales be high level accomplishment. The studies just re- used in practice?Biographical accomplishment viewed indicate that this information is consider- measures have seldom been used inreal-life ably more useful than most other kinds of infor- studies of the selection of college or graduate mation. However, before one considers using students, and hence there are few guidelines for these kinds of data in practice we need to answer the person who would like to make use of these three questions about it: Can we believe students' variables. Several industrial studies provide some reports? Can measures be constructed that meet stimulating suggestions. but they are few and far standard psychometric criteria? How could such between. Certainly, very few, if any, colleges or measures be used in real-life selection, place- graduate departments have made past extra- ment. or guidance situations? academic accomplishment the most important Can we believe students' reports?As the earlier basis of their selection procedures. However, a review of the validity of self-reported accomplish- study by Baird and Richards (1968) simulates ments (Maxey and Ormsby, 1971) and the review what would happenifvariousselectionpro- of the accuracy of selfreported grades indicated, cedures were followed for admission to college. students' reports of their high level accomplish- This study suggests some of the practical prob- ments can be believed. The validity of questions lems of using accomplishment data in selection aboutpastaccomplishmentsappearsuseful decisions. The authors compared the results if: enough for the decisions and actions that they only academic criteria were used to admit stu- could be used for. dents to college; only criteria based on previous Can biographical measures of accomplishment creative accomplishment in each of six areas be made psychometrically adequate?The studies were used; and both academic and creative ac- of the scales developed by the National Merit complishment were used. A close examination of Scholarship Program (Nichols and Holland, 1963), the results shows that an educational institution the research on more typical college students cannot have everything. For example, if an insti- (Richards, Holland. and Lutz, 1967; R;chards and tution selected students only for high level ac- Lutz, 1968). and the operational work of the complishment rather than for grades, it would in- American College Testing Program (ACT Tech- crease its dropout rate. However,an institution nical Report, 1973) show that biographical ac- could still make use of nonacademic predictors complishment scales can be constructed with of creative accomplishment. For example, as adequate reliability. Occasional skewness in the Baird and Richards suggest, ". .. a college could scales does not present a serious limitation (Hol- decide which areas of achievement it wished to land and Richards, 1967). The validity of the emphasize; that is, whether it preferred more or scales does not seem to be affected by restrictions fewer students with potentials for achievement in of range on academic talent (Holland and Rich- leadership or science, art or writing, speech and ards, 1967; Baird, 1969a). The validity of the drama or music." scales, discussed herein earlier, also indicates A college or graduate school's choice of a par- the psychometric adequacy of the scales. All of ticular selection strategy is a function of the out- the results may be underestimates because of comes it values most. Institutions must choose the

. ; 42

35 the brevity of the scales used in these studies. Some questions about direct assessment of In short, it appears that biographical accomplish- biographical accomplishment scales ment scales can be constructed that meet stan We have just seen the power of biographical ac- dard psychometric requirements. complishment information to predict subsequent Can biographical accomplishment scales be high level accomplishment. The studies just re- used in practice?Biographical accomplishment viewed indicate that this information is consider- measures have seldom been used inreal-life ably inure useful than most other kinds of infor- studies of the selection of college or graduate mation. However, before one considers using students, and hence there are few guidelines for these kinds of data in practice we need to answer the person who would like to make use of these three questions about it: Can we believe students' variables. Several industrial studies provide some reports? Can measures be constructed that meet stimulating suggestions, but they are few and far standard psychometric criteria? How could such between. Certainly, very few, if any, colleges or measures be used in real-life selection, place- graduate departments have made past extra- ment, or guidance situations? academic accomplishment the most important Can we believe students' reports?As the earlier basis of their selection procedures. However, a review of the validity of self-reported accomplish- study by Baird and Richards (1968) simulates ments (Maxey and Ormsby, 1971) and the review what would happenif various selection proof the accuracy of sereported grades indicated, cedures were followed for admission to college. students' reports of their high level accomplish- This study suggests some of the practical prob- ments can be believed. The validity of questions lems of using accomplishment data in selection about pastaccomplishmentsappearsuseful decisions. The authors compared the results if: enough for the decisions and actions that they only academic criteria were used to admit stu- could be used for. dents to college; only criteria based on previous Can biographical measures of accomplishment creative accomplishment in each of six areas be made psychometrically adequate?The studies were used; and both academic and creative ac- of the scales developed by the National Merit complishment were used. A close examination of Scholarship Program (Nichols and Holland, 1963), the results shows that an educational institution the research on more typical college students cannot have everything. For example, if an insti- (Richards, Holland, and Lutz, 1967; Echards and tution selected students only for high level ac- Lutz,1968), and the operational work of the complishment rather than for grades, it would in- American College Testing Program (ACT Tech- crease its dropout rate. However, an institution nical Report, 1973) show that biographical ac- could still make use of nonacademic predictors complishment scales can be constructed with of creative accomplishment. For example, as adequate reliability. Occasional skewness in the Baird and Richards suggest, ". .. a college could scales does not present a serious limitation (Hol- decide which areas of achievement it wished to land and Richards, 1967). The validity of the emphasize; that is, whether it preferred more or scales does nut seem to be affected by restrictions fewer students with potentials for achievement in of range on academic talent (Holland and Rich- leadership or science, art or writing, speech and ards, 1967; Baird, 1969a). The validity of the drama or music." scales, discussed herein earlier, also indicates A college or graduate school's choice of a par- the psychometric adequacy of the scales. All of ticular selection strategy is a function of the out- the results may be underestimates because of comes it values most. Institutions must choose the 42

35 relative value of obtaining: a group of students students. They wish to have some way of sole( ting who will attain high grades: a group of students students, who will be outstanding students and who will achieve in the at complishment areas it who will eventuallyc ontribute most tosonlet). is interested in: a group of students who will not As the review of research indicates, the most el. drop out. or some other group of students that it ficient information for predicting future atconn values. While an institution can obtain a student plishments is data on previous accomplishments. body that will show various proportions of these The studies reviewed show that the best predie outcomes, it would be hard pressed to find an in- tors of future high level. real-life accomplishment coming group of students thatis desirable in in writing. science, art, music, and leadership are every way. On the other hand, a college or gradu- similar accomplishments, albeit at a lower level. ate department can obtain a group of students in previous years. In fact. as with all other be who will be aligned inure (loosely with its purposes havioral and scientific prediction, which is based and goals. on the consistency of the same or similar phe nomena over time, the studies indicate that up to Implicationsofthese results now the most effective predictor of high level More and more students delay entry to college accomplishment isclearlypast high level be after high school, leave college for a few years, haviors of the same or similar types. People who and seek unusual work or other experiences dur- have been outstanding in a wide variety of areas ing their breaks from their studies. In addition. in science, literature, creative arts, and public more students work during college. and more affairs have been shown to have had accomplish. older students are entering colleges. Many of meats in those areas in their college years. The these students have had educationally valuable institution that wishes to have graduates who will experiences outside the classroom, for which be outstanding in their fields in the future might they receive no credit. Many colleges have also well consider the previous accomplishments of begun a wide variety of off-campus programs of their applicants. To date, information about past independent study, work experience, and public accomplishments has proved to be a far better service, and so on. In this way. many students predictorof highlevel accomplishment than have opportunities to develop and demonstrate measures of ability, interests. or personality. As talents they would not otherwise have had. the review ids., indicates. scales of real-life ac- Clearly. tests and traditional undergraduate complishment, can be constructed that are reli- transcripts do not provide adequate means of able. usable, and seldom faked. They can be used recognition for these kinds of learning and at- in bele( don decisions in a variety of ways. They complishment. For these reasons, it is important SCUM particularly useful when there is a need to to find ways to assess the at omplishmentsul assess talents somewhat removed front at ademit students. ability. such as artistic capacity, musical skill. A second purpose in assessing students' a( - ability to write expressively and forcefully. dra- complishments is to se'« I students w Ito are likely matic power. and the intuition needed to devise to be productive. to be creative, to provide leader- a si ientific experiment. As these examples sug- ship. and to make a contribution to their fields. rreqthe assessment of talent is more difficult in Many admissions committees, faced with large mune areas than in others and. consequently. the numbers of applicants and dwindling funds, feel predi( five power of the variables will vary front the need for some way to assess the high level area to area. In any case. these measures cannot nominstitution-sponsoredat ( omplishmentsof replace measures of academic talent: they simply

43 36 written questionnaire items that are often in- live years after graduation. The discrepancy in e bided as an afterthought. It is therefore surpris- these results probably results from the smaller ing that the biographical items do as well as most number of categories used by Sharp and Kt as- of the tests. The second point is that apparently negor. Rather than the individual majors used by simple information can tap t he strut turc of fib) e bo- the National Sciene c Foundation, they used nine logical domains about as well as tests, as Norris groups of majorse.g.. humanities and arts. and Katz found. lithe strut t Ore old psye hologie al health, natural science, et(.Thus, the postgrad- domain is not coherent and well integrated. it is uate careers of college graduates seem to be fairly unlikely that either simple self-estimates or tests stable. These figures agree reasonably well with will find it. If a characteristic is not an important those developed by Slocum (1974)for professional. part of a domain, it probably will have few cor- technical, and managerial workers. (However. relates with other variables. If it is important. it jobs based on less education generally are less should have numerous correlates, and, it seems stable.) probable. should be assessable by relatively sim- Studies byFlanagan. etal. (1966) indicated ple self-reports or simple estimates. that vocational choice gradually seemed to be- In sum, the evidence lead; one to conclude, ( time more stable as students progressed through with Dolliver (1969), Whitney (19691. and Holland high school. (Theories of vocational development and Lutz (1968). that there seems no justification such as Super's, and Rosenberg's would predict forthelong-standing biasagainstexpressed such a result. of course). The average consistency choice in favor of tests. The practical implications for 30 listed occupations in the twelfth grade of this «me lusion will be discussed later in this when «onpared to those listed one year later was section. only 28 percent for men and 40 percent for Apart from their comparative validity. how pre- women. However. as reported earlier. Cooley's dictive are expressed ( hoi«s? Whitney's recent study of the same data divided these occupations review (1969) of this question. previously cited. into six e ategories and found a 54 percent stabil- brings together the relevant evidence from the ity over the four years of high school. large scale longitudinal studies IN > 3.000) that Using another research strategy at the college bear on expressed ( {mice (the Strong study and level. Astin and Palms (1969) studied the voca- the two Cooley studies have already been men- tional choices of 3.821 subjects over four years of tioned). A large study of college graduates (on- college. They grouped 30 career choices into 12 ducted by the National S( um«. Foundation (1963) ( ategories. and developed multiple correlations to found that. for 32 o« upations. die median per- predict finale areer choice from a wide variety of ( entage of graduates employed (o tea( hing in the bioge ((phi( al information. The correlations ranged area of their major field two years later was 1.0 from .23 to .53. Freshman expressed choke was percent for men and 29 pert ent for women. The the best predi( for for seven categories and second majors varied greatly in ;lability hum 91 pert ent best for the remaining five. (Major field choke in pharmac y for men and 91 per ent in edit( ation was the best predictor for three of those cate- for women to 4 pert ent in philosophy for men and gories, degree goal for two. and sex for one.) The two pert mt in religion for women. In «intrast. results of these varied studies suggest that self- Sharp and Krasnegor (1966) followed another expressed choke becomes a better and more large sample of college graduates, and found that stable medic-tor of vocational behavior as people 74 pert mt of the nn -u and 83 pert mt of the bet time older. The results also suggest that the women were employed in their major field area predictive power of expressed choice varies with

...- ,. 0 i 50 relative value of obtaining: a group of students students. They w ish to have sonic way of sele( ting who will attain high grades; a group of students students who will be outstanding students and who will achieve in the a( t omplishment areas it who will eventually( ontribute most to society. is interested in: a group of students who w ill not As the review of research indicates, the most ef- drop out, or some other group of students that it ficient information fin predicting future a( cum- values. While an institution can obtain a student plishments is data on previous accomplishments. body that will show various proportions of these The studies reviewed show that the best predic- outcomes, it would be hard pressed to find in- tors of future high Iceel, real-life accomplishment coming group of students thatis desirable in in writing. science, art. music. and leadership are every way. On the other hand. a college or gradu- similar accomplishments. albeit at a lower level, ate department can obtain a group of students in previous years. hi fact, as with all other be- who will be aligned more (dose!), witlt its purposes havioral and scientific prediction, which is based and goals. on the consistency of the sante or similar pile- »omena over time, the studies indicate that up to implicationsofthese results now the most effective predictor of high level More and more students delay entry to college accomplishmentisclearly past high level be- after high s( hold. leave college fur a few years. haviors of the same or similar types. People who and seek unusual work or other experiences (lur- have been outstanding in a wide variety of areas ing their breaks from their studies. In addition. in science, literature, creative arts, and public more students work during college, and more affairs have been shown to have had accomplish- older students are entering colleges. Many of ments in those areas in their college years. The these students have had educationally valuable institution that wishes to have graduates who will experiences outside the classroom. for which be outstanding in their fields in the future might they receive no credit. Many colleges have also well consider the previous accomplishments of begun a wide variety of off-campus programs of their applicants. To date. information about past independent study. work experience. and public accomplishments has proved to be a far better service. and so on. In this way. many students predictor of highlevel accomplishment than have opportunities to develop and demonstrate measures of ability, interests. or personality. As talents they would not otherwise have had. the review also indicates. scales of real-life ac- Clearly. tests and traditional undergraduate complishments can be construe ted that arc reli- transcripts do not provide adequate means of able, usable. and seldom faked. They can be used recognition for these kinds of learning and ac- in selection decisions in a variety of ways. They complishment. For these reasons. it is important seem particularly useful when there is a need to to find ways to assess the a« omplishments of assess talents somewhat removed from academic students. ability. such as artistic capacity. musical skill. A second purpose in assessing students' ,Ic- ability to write expressively and forcefully. dra- uoi»pliSliniciits is to seder t students w Ito are likely mathc power. and the intuition needed to devise to be productive. to be creative. to provide leader- a s( ientific experiment. As these examples sug- ship. and to make a contribution to their fields. gest, the assessment of talent is more diffi( tilt in Many admissions ( ontinittees. faced with large some areas than in others and. consequently. the numbers of applicants and dwindling funds. feel predic tive power of the variables will vary from the need for some way to assess the high level area to area. In any case, these measures cannot noninstitution-sponsored accomplishmentsof replace measures of academic talent: they simply

4 3 36 provide indications of capacity in and out of physical science, etc. Thus, different accomplish- class in other areas that are useful for specific ments need to be assessed and summarized in purposes. In this way, the range of talents that different areas. Another assessment decision is institutions consider in their applicants could be about the level of behavior or involvement to he greatly expanded. Thus, an institution could not assessed. Should only high level accomplish- only select students who will get good grades, but ments thatreceive public recognition be as- students who will be good organizers of research, sessed? Perhaps the best answer to this question leaders in political and nonpolitical organizations, is to point omit that high level accomplishments good writers, and inventive experimenters. are usually preceded by a great deal of similar Another important reason for developing meas- effort that is not or cannot he given public recogni- ures of in- and out-of-class activity is that the tion. It is also important to assess these accom- student applying for study has a right to be able plishments. However, the emphasis should be to present his or her skills, talents, and achieve- placed on accomplishments that show long term ments to selection committees. As recommended persistence and originality and require complex by the College Board's Commission on Tests, skills for their attainment (Holland and Richards, students should have some choices in the picture 1965). Given these criteria, a range of accomplish- of themselves that selection officers see. (There ments should be assessed, from fairly common is another positive outcome of the inclusion of but private ones, to rare and public ones. The this sort of information. The students who com- goal would be to assess bands of competency, so plete a form that asks fur their personal accom- that students could be given scores indicating plishments feel that they are being taken more different levels of accomplishment. seriously, and that they have had a chance to Of course, each college needs to validate the present their best side.) information about earlier accomplishments in Assuming the accomplishments of informal its own environment, just as colleges need to vali- education are important, what accomplishments date admissions tests and high school records in should we assess. how should we assess them, their own institutions. Moreover, if many colleges and can we assess them (i.e.,isitpractically actually began to select students on the basis of feasible to assess students' accomplishments in their earlier accomplishment itis possible that a regular, programmatic way)? students would begin to exaggerate their achieve.- What to assess? The studies examined in the ments. This aspect will be discussed in the next review suggest that questions on accomplish- section. ments are better predictors, the more closely they How to assess? Once we have decided what to resemble the real life criteria they are designed assess, we need to decide how to go about doing to predict, the more direct they arc, and the more it. The method chosen should be as simple, clear, recent are the behaviors to which they refer. It and direct as possible, so that students will have would seem that the kinds of school accomplish- little difficulty responding accurately, and so that ments that are educationally and socially rele- the results can be interpreted readily. However, vant, and similar kinds of previous accomplish- the approach should also deal with the fact that ments can be identified. The desired behaviors students may exaggerate their achievements. will vary from field to field. of course. The ac- Essentially, there i6 concern that some students complishments important in biological science will fake their responses to any self-reports of will not necessarily be important in art, those in accomplishments unless there is some mecha- thehumanitiesnotnecessarily importantin nism to check their claims. Although research 4

37 indicates that faking in these areas is very rare, phasized here is that itis much more important some mechanism for verifying student assertions to consider creative accomplishments of students needs to be developed in order that the claims in the admissions decision than it is to avoid con- of students may be believed by admissions com- sidering the occasional student who exaggerates. mittees. Clearly, the claims of every student can- There arc many honest students whose accom- not be checked. Students' interests and accomplishments are several. And, according to em- plishments are too diverse and widespread to pirical research, those who exaggerate are not allow any systematic and thorough documenta- numerous. In any case, one or a combination of tion. What could be done in lieu of such checks? the suggested systems could be used to identify There are several possibilities. One is simply to those students who are possibly exaggerating; identify the student who claims too many achieve. and, thus, to keep their number to a minimum. ments in too many areas. For example, the stu- What is feasible? Assuming answers to the dent who claims to have served as class president. questions of what and how to assess, we need to written a novel accepted for publication, won a consider what would be feasible for a university state music contest, published a scientific article or for an organization like the College Board. The as well as having sold several paintings is suspect questions used should be appropriate for mechan- (Holland and Richards, 1967). Cutting scores can ical data processing. Today, in relation to the be devised for identifying such exaggerators, and current volume of candidates, the available tech- (to allow for the possibility of a modern Leonardo nology, and the need for compatability with op- da Vinci) some methods devised to ask the stu- erational systems now in use, the most efficient dent for proof of the accomplishments, including system would be one in which the questions are the time, place, and exact character of the ac- presented in a multiple response format that complishment. Another strategy would be to fol- could be edited and recorded by optical scanning low the example of Kirk and Sereda (1969) and scoring machines. This requirement makes write- construct a "profile" of the person who is likely in responses, for example those used in the Inde- to exaggerate. (Kirk and Sereda found that stu- pendent Activities Questionnaire developed by dents who exaggerated reports of their past Klein when he was at Educational Testing Service, grades were ambitious, worried about status, and more difficult to use. However, it isunlikely that had parents who pressured them to achieve.) any single list of accomplishments could cover A third possibility would be to audit a small the vast diversity of human activity and accom- percentage of the students' forms on accomplish- plishment. Strong consideration must be given ments that can be verified easily. (The forms to write-in forms, free-response formats and such would probably include a number of accomplish- techniques as microfiche photographs of stu- ments that would be extremely important but dents' productions so that this diversity can be difficult to verify. If only those easily verifiable assessed with thoroughness and fairness to the were included, the content would be too narrow student. Thus, a variety of formats should be con- to be predictive.) Students might be either re- sidered. Each technique would require a develop- quired or requested to identify someone,,s ho ment project to construct standard assessment could verify each major area of activity. procedures, evaluation criteria, and reporting A fourth possibility is to include fake items in systems. Whatever the format, the questions the listof accomplishments, such as "Placed should have face validity to candidates, and they first, second or third in the area Sears Science should be briefly stated and relatively simple. Program Contest." The point that should be em- They should not be complicated, tedious, and tax- 4 5

I ing to the student's memory of exact details class of variables proved nearly so useful. The (charges that were leveled against the Indepen- information seems appropriate to the assessee, dent Activities Questionnaire). In addition, the it can be made psychometrically adequate, and questions should reflect important cultural and it can be used in various selection strategies. A background differences among applicants. It may strong case can be made for the utility and value be especially important to consider the lack of of biographical accomplishment information. opportunitiesforachievement among disadvantaged students. Science equipment, art supplies, musical instruments and so on are beyond the budget of many disadvantaged families. To be fair,it may be necessary to consider disadvantaged groups separately.

Summary The implications can be stated fairly briefly. (I) Since the consensus of the studies indicates that information about past accomplishments is the best predictor of later accomplishments, ad. missions committees who wish to select students with the greatest potential for future accomplish. ment should look for evidence of students' past accomplishments. (2) Satisfactory measures of past accomplishments have been constructed at several levels, so it seems plausible to believe that such measures can be constructed for many admissions decisions at other levels7(.3)-The mess ures that have been constructed appear to have adequate reliability, accuracy, and validity, so it seems plausible-to-think-that- measures deemed adequate in these ways can be developed for most admissions decisions. (4) The measures seem in- dependentof academic aptitude.sosimilar measures for admissions would probably add a good deal of information of a new kind to the admissions situation, and (5) studies simulating the use of these measures show that different selec. tion strategies produce different results. Schools, therefore, shouldnot see these measures as panaceas, but as a new kind of information. We have found biographical accomplishment information to provide useful prediction of later high level accomplishment in a wide variety of samples and settings. In these studies, no other , 4 o

39 Section IV: Biographical information to examine the items and differences found in a as an assessment of study of social class and the urban school (Herriott students' educational backgrounds and St. John, 1966). These authors present details on broken homes, welfare payments, parents' employment and unemployment, transfer rate Students bring their homes, peers, and conununi- from other schools, adequacy of clothing, condi- ties with them when they enter college. The gen- tion of teeth, diet, quality of medical care, paren- eral degree of compatibility of these background tal attendance at school events, parental interest factors with the college experience can have a inchildren's academic or extracurricular ac- powerful effect on students' satisfaction as well complishments, supervision of students, delin- as on their educational growth and development. quency, disturbances in class, being held back a year. student effort on assignments, and degree of SES congruity with teachers. Students from working or lower-middle class It is unlikely that many students entering col- homes, or from rural backgrounds, generally have lege would have ali the problems that this list had less stimulating intellectual and cultural ex- suggests, but some low SES students probably periences. As summarized by Feldman and New- have come from homes and schools that ham- comb (1969). students of lower SES. compared to pered rather than aided their academic growth, their higher SES counterparts: are less culturally and these experiences affect their approach to sophisticated (Peterson, 1965, 1968; Hartnett and college. These educational differences have been Peterson, 1968: McLaughlin. 1965); have had found to be related to adjustment in college in a more restricted range of experiences (Baird, several studies. Lane (1960), for example, found 1967; Matteson, 1955); and are more likely to be that female students from lower SES backgrounds, oriented to college in terms of vocational or pro- when compared to students from higher SES back- fessional training and less likely to be oriented in grounds, had a more difficult time adjusting to terms of intellectual growth (Baird, 1967; Gott- certain academic and social demands at Stanford, lieb, 1962; Lane. 1960). (Some of the voluminous as measured by self-report and peer ratings. studies of behavioral differences among social Similar results were found among Vassar fresh- classes are summarized in Centers, 1949; and men (F'reedman, 1967), and Yale men (Davie, Berelson and Stiener, 1964. They are reviewed in 1958). The results are not unequivocal, for dis- a popular book by Packard, 1961.) To illustrate sertations by Whyte (1963) and McLaughlin (1965) what differences in SES can represent, it is useful indicated that lower SES students at Cornell and

4 7 Harvard were as satisfied with theirs ollege. if not c ludo*andhistory. and anegative relation more so. as were higher 5f.s stulents. But overall between changes in liberalism and part-tim em- measures of satisfaction mayau be the most ploy ment. Changes in the scales of social con- sensitive measures of adaptation to college. It science wine related to enrollment in psychology seems probable that stud.. ins from varied has k- courses, attendance atconcerts and lectures. grounds will hav e diff rent adjustment problems and participation in intramural competition. This in college. study does not show how specific college ex- Jones and Finnell (1972) directly studied the periences might have different impacts on stu- relationships between college experiences and at- dents from different backgrounds. titudes of students from economically deprived Baird (1969) attempted to study the relative in- backgrounds. Using the College Student Ques. fluence of social class background and other vari- tionnaire (Peterson. 19651. they found that: ables on students' aspirations.In a randomly ... students from economically deprived back. selected sample of 21,110 college applicants. grounds (141 experience attitude changes. On one Baird compared students with different degree scale family independence the change was plans and different family income levels. He was significant. Obviously. the influence of the low particularly interested in students from low insocioeconomic bac kground did not neutralize the come families who aspired to Ph.D. or profes- Witten« provided by the college environment. sional degrees and students from high income attitude c {inges in students from economically families who aspired onlyto complete junior deprived backgrounds. however, resembled the college. Students were compared on their re- changes that occurredincollege students in sponses to the Student Profile Section of the ACT general.... The only significant difference in the assessment. a self-report instrument concerned attitude changes in economically deprived stu- with the students' plans, educational values, and dents and college students in general w as in the academic and nonacademic achievements. Baird area of family independence. This means that concluded: students from low socioeconomic backgrounds "The most obvious general trend in these re- became more autonomous in their attitude toward sults is that students with discrepant family intheir parents and families:Ile significant change comes and degree goals were more like other inattitude may be attributedto a situation students with the same degree goals than they whereby their two years in college had introduced were like students from families with similar these students to an env min went which afforded incomes. In a few comparisons the differences more physical comforts and ether attractions than between degree goals were accentuated for the could be provided by their families, hence an at- students with discrepant family income and de- titude of less psychological dependence insofar gree goals. Thus, although there was a positive as their families are concerned. (p. 317) relation between family income and degree plans. Jones and Finnell also correlated the extent of income .alone did not seem to he as powerful a attitude change with spec ific college experiences. determinant of degree plans as other characteris- Thus, students who enrolled in speech and dra- tics considered singly. matic arts courses changed significantly more "Within this sample, there was a slight positive than other students on the scale of family inde- association between family income and the ACT pendence. There was also a positive relation Composite Score. This suggests that the students between changes in scores on the cultural sophis- from less wealthy families who aspired to high tication scale and enrollment in courses in psy. level degrees were over-achievers, while students

4 8

41 from wealthy families who aspired to low level during freshman registration or orientation peri- degrees were under-achievers. These results also ods, and related their responses to a followup suggestthatthese studentsvaryingeneral questionnaire administered four years later. Holm- "drive" and motivation to achieve. This idea strom concluded: gains support from the scores of these groups on "The implications of the present study are the nonacademic achievement scales. The level clear. Aside from the expected dissimilarities in of involvement in nonacademic achievement also their demographic and background characteris- suggests that the students with higher degree tics(e.g., father's occupation), the low-income goals had enthusiasm as well as a high activity undergraduate does not differ dramatically from level. Thus a number of variables suggest that his more affluent classmates. He shares with them students of low family income who plan various the same life goals, degree aspirations, activities, degrees select appropriate goals and see them- and interests. He may be more likely to dropout selves as potential holders of those degrees. but only temporarily; he may shy away from stu- "Thestudentsfromwealthyhomes who dent demonstrations; he may get slightly lower planned only a junior college degree had low ACT grades: but, overall, the likelihood of his attaining Composite Scores and generally low scores on his degree in four years is reasonably close to that the nonacademic achievement scales, except in of his more privileged classmates. art. When they chose a college, they placed less "Although the source of financial support is an importance on high scholastic standards or finan- important factor in determining the academic suc- cial aid offers (they were probably unlikely to cess of both groups of students, it should be em- have received such offers). phasized that the low-income student is particu- "Finally, students with discrepant family in- larly dependent on certain sources of support if come and degree plans generally seem to have he is to make satisfactory progress. chosen appropriate degree goals, considering "Finally,low-incomestudentswhoattend their other characteristics. That is, students with highly selective institutions do better than their similar degree plans were more like one another counterparts who attend less selective institu- than they were like students from families with tions, and this holds true even when differences similar incomes, suggesting that there is a pool of in ability and in the availability of financial aid talented students who, at the beginning of college, are controlled for. That is, given two students plan to obtain advanced degrees and enter the who are equally able and who receive the same professions regardless of family income. The ex- kind and amount of financial support one stu- tent to which they attain these goals contributes dent at a very selective institution, the other at greatly to social and economic mobility within a relatively unselective one the former student American society." (pp. 19.20) will be more likely to complete the baccalaureate In what is probably the largest scale investiga- in four years. Although the reasons underlying tion of the effects of low family income on the this relationship are not immediately obvious, careers of students yet conducted. Holmstrom the present results suggest that, whenever pos- (1973) studied a sample of 34.346 students in 252 sible, low-income students should be advised to colleges. He compared a subsample of low-income apply to more selective institutions. It is to be students with a subsample of other students on a hoped that future research will clarify the special wide variety of variables collected on the Ameri- impact of institutional selectivity on low-income can Council on Education's Student Information students." (pp. 19.20) Form, a self-administered questionnaire filled out There is ample evidence that social background 43

42 influences students' vocational choices as well portant, the way of lifeit presents essentially as their educational decisions. Ginsberg, et al., alien; the rewards at the end of the line not worth- (1951) found that boys from high- and low-income while or not worth the effort, and the goals not families went through the same stages of voca- achievable. The educational system is seen as tional development, but their choices were quite hostile and, at the least, offers the minimum of different. High-income boys, even at an early age. rewards." Itis no surprise, then, that low SES tended to think they would go to college, and students show up as lacking in achievement mo- were oriented toward the professions. Low-in- tivation and certain middle class values, for ex- come boys tended to think of skilled jobs that ample, deferment of gratification for later re- would pay more than their father's. "They believe wards. (Such characteristics are often associated there is little that they can do (about vocational with upward mobility.) The low SES student must planning) beyond selecting a high school where be helped to understand that a rise in society by theycanpursueanappropriatevocational one's own efforts is possible, and must be shown course." how education is (or can be) related to this rise. Werts (1966) foundthatsocialclass was Teachers, adults in the community, and guidance strongly related toinitial career choices. In a officers may serve as role models that will en- subsequent study. Werts (1967) found that when gender educational aspirations. There is some evi- freshmen college students were grouped by aca- dence (Baird,1970; Spady, 1971) that extra- demic performance in high school, their career curricular successes can lead students to raise choices were still related to their social class. their level of aspiration. If lower-class students The professions and social science were above could be encouraged to engage in a variety of average in SES; education and engineering were academic and nonacademic activities, their self- below average. Werts also found that career concepts and aspirations might he changed. Voca- choices depended upon ability at each level of tional counseling can also encourage students to social class, that controls for ability barely af- avoid premature decisions, and to explore some fected the influence of social class, and that con- of the educational and vocational options open to trols for social class barely affected the influence them. of ability. SES and ability seemed to be about Feldman and Newcomb (1969) have suggested equally important as influences on career choice an "incongruity" hypothesis about the overall re- among college students. lation of background to college impact which also Once we accept the fact that SES has many leads to a number of possible actions. They hy- effects on education and career choice, what can pothesize that ".. .the college will have the be done about it? Bei lin and Bergin (1956) sug- greatest impact on students whose orientations gested several ways to increase the utilization of are incongruent with the dominant orientation of talent in lower SES groups. Scholarship programs the college. By the same token, change will be should be directed more intensively to low SES greater among students whose previous environ- students. In addition to funds, this entails distrib- ment is discontinuous with that of the college uting information about the programs, encourag- than among those whose environment is con- ing students to apply, and requires the willingness tinuous." After examining a variety of studies of selection boards to choose truly needy stu- with different designs, purposes, measures, and dents. Bei lin points out that scholarships and oc- results, they concluded that the influence of back. cupational information are not enough; low SES ground is complex. youth often feel that "... education is not im- "Our best guess at the moment is that a college 50

43 is most likely to have the largest impact on stu- purposes; the accuracy of students' reports of dents who experience aoatinuing series of not- their families' status. Very few studies have ex- too.threat ening discontinuities. Too great a diver- amined this question, and only one has examined gence between student and college, especially the accuracy of students of the age to apply for initially, may result in the marshalling of resis- college. Kerckhoff, Mason, and Puss (1973) ex- tances. Too little might mean no impetus for amined the degree of agreement between stu- change. From this point of view, a college's ob. dents' reports and their parents' reports of the jectives might include a tolerance, or even a de- parents' education. and the fathers' occupation. sire, for those discrepancies that can stimulate The students filled out questionnaires, and the change and growth." (p. 332) parents were interviewed. The correlations be- Feldman and Newcomb's working hypothesis, tweenparents' and students'reports of the along with other results on cultural sophistication, mothers' education was .84 for white students suggest a number of possible courses of action and .83 for black students. For fathers' education, for colleges. Colleges might seek a certain num- the correlations were .89 and .81 respectively, ber of students moderately incongruent with the and for fathers' occupation, the correlations were college's environment. Knowledge of the intellec- .93 and .74 respectively. tual and cultural backgrounds of students could help instructors choose teaching strategies that Black students will result in the maximal growth of their students. Knowledge of the backgrounds of students may With the same knowledge, colleges could organize he especially important for minority group stu- extracurricular programs that would offer a vari- dents, notably, black students. Studies of the ety of educational experiences to aid the develop- educational backgrounds and aspirations of black ment of students. Counseling programs could use students have accumulated rapidly (e.g., An- knowledge of student's backgrounds as one basis tonovsky and Lerner, 1959; Bayer and Boruch, for understanding students' problems in adjusting 1969;Dregef andMiller,1968; Epps, 1969; to college. Fichter, 1967; Hartnett, 1970; Hull and Davies, These possibilities may be especially important 1973; Lott and Lott, 1963; Rosen, 1959). Perhaps for students who are the first in their families to the most comprehensive of large scale studies enter higher education. For example, as Cross's among these are those by Bayer (1972), Bayer and (1968) review suggests. two-year colleges must Boruch (1969). and Fichter (1967). Bayer and face the problem of dealing with great numbers of Burnell sampled black and white students in 358 students with no guidelines about what to expec t colleges, and weighted the replies so that they from higher education. Colleges beginning open- would be representative of black and white stu- door policies will very probably face similar prob- dents in all colleges. Their results suggest the lems. breadth of useful information that can be obtained Methods for assessing social doss. There are a from biographical questionnaires. Their findings great variety of methods for assessing SES. These show that greater proportions of black than of can be broken down into the subjective, the repu- white students were women; were older;re- tational, and the objective approaches. They arc ported lower high school grades; participated in beyond the s( opt of this monograph, and the in- extracurricular activities and had more non- terestedreader should consulttextbooks on academic achievements; had higher degree aspi- sociological measurement for details. However. rations (hutless often planned a professional one question is particularly relevant to the present degree); chose majors in the social sciences or

44 business (but less often ( hose majors in the physical of all black students. but college administra- cal sciences or engineeing), applied to more than tors ds well as others might profit from sonic re- one college; were influencedintheir college views of the life experiences of blacks as provided choice by individuals (and were relatively less by the reviews of Dreger and Miller (1968), Riess- influenced by characteristics of the institution): man (1962), Pettigrew (1964), Parsons and Clark grew up in a large city; came from families with (1965). and Deutch. Katz, and Jensen (1968). less education, occupational status, and income: However, as greater numbers of blacks enter relied on loans, scholarships, or grants for finan- college, more colleges will have to plan for stu- cial support (and less often relied on family. dents who may have some of the problems of dis- savings, or employment), and felt unsure whether advantaged students. Some of these include: they would have sufficient funds to complete nonacademic deviant values and behavioral ex- their education; assigned greater personal im- pectations;relativelackof languagefacility; portance to such longterm goals as being an au- relative deprivation of familial suppo, and press thority in one's field. obtaining recognition from for achievement, relative lack of support for aca- one's peers, being well off financially, helping demic achievement from peers and lack of adult others in difficulty, and becoming a community examples of academic or other success; and rela- leader, expected that they would participate in tively low level of self-esteem. Warden reviewed a demonstration, be elected to a student office, the research in these areas, discussed the aca- and obtain academic honors (and were less likely demic and social consequences of cdch area, out- to expect that they would change their major or lined various programs and actions that have career choice, transfer to another college, or been used to influence these factors, and made marry during college); had somewhat different suggestions about other possible approaches. A study habits: had engaged in different activities. detailed discussionofthis research (the problems and had somewhat more restrictive and utilitarian and programs) is beyond the scope of this mono- attitudes toward higher education. These results graph. However, it is important to point out that are quite variedin content and meaning. but biographical information can assess the standing they suggest a number of possible practical ac- of students in the five areas listed above, and can tions and policies, which will be discussed briefly help colleges prepare for the educational and later. But itis clear that a college with the kind psychological needs and requirements of black of information Bayer and Boruch provide would students.' Of course, studies will need to be done be better able to plan for groups of black students. to show the predictive validity of biographical Bayer (1972) replicated the earlier Bayer and data for blacks, and evaluations of the success of Boruch study by comparing 12,927 blacks with intervention programs conducted. 158,111 nonblaek freshmen enrolled in 324 col- From the first part of this section, itis clear leges, in 1971. Bayer found few differences from that similar considerations apply to low-income. the earlier report. However. there was a slight low SES, white students. Many face some of the shift in degree aspirations among blacks, greater same kinds of problems many black students face. proportions aspired to either less than the bacca- The specific content of the problems may differ, laureate or to a professional degree. However. but the general aspects are similar in many re- the basic picture remained the same in the an- spects. For example. the peer groups of blacks swers to the great majority of questions in their 1. The July 1973 issue of theJournal of College Student Per. survey. sonnelsuggests sonic approaches for working with minority Black college students are undouhtably at ypi- students. 5 2

45 and lowincoe whites may be different, but the central problem is the same: the lack of peer sup. port for academic achievement. The five areas listed above seem to apply to many white students as well as to many black students. And biographi cal information can be useful for analyzing these factors for whites aswell as for blacks.

46 Section V: The use of Since the SVIB :s the most extensively studied biographical information to predict and and carefully validated interest inventory (for understand vocational choice example, Strong, 1935, 1943, 1955; Darley and Hagenah, 1955; Campbell, 1971) the comparison of its results with those of expressed choice are Biographical idormation has been used in the probably the most powerful test of the relative study of vocationdl choice in three main ways: validity of expressed choice versus inventories as a substitute for vocational interest tests: as tha; can be made. After carefully reviewing the part of vocational classification syste. .s; and as probitals.interminology, correspondence be- a way to understand the factors in vocatknal de- tween the svm and expressed interests, reliabil- velopment. ity,validity, and experimental design, Dolliver (1969) concluded: The validity of expressed interest as "There is only a moderate degree of overlap be- a measure of vocational choice tween the results of the SVIB and the results of an Although interest inventories are the traditional expressed - interest method. tool of vocational counselors, a number of bio- The reliability of the SVIB exceeds that of ex- graphical questionnaire items have been pro- pressed interests. The reliability of expressed in- posed as adjuncts or substitutes for interest tests. terests is moderately low. Of these, simple expressed vocational choice ap- The predictive validity of expressed interests is pears to be the most useful. The small amber of at least as great as the predictive valid.it; of the studies in this area center around two questions: svm. In no study where direct comparison was how w does expressed choice do in comparison made (Dyer, 1939; Enright and Pinneau, 1955; with interest inventories, and how well can we pre- McArthur and Stevens, 1955) was the SVIB as ac- dict from expressed vocational choice alone? curate as the expressed interests in predicting oc- Dolliver (1969) has conducted an extensive re- cupation engaged in. view of studies comparing the relative merits of There is an apparent discrepancy, since ex- the Strong Vocational Interest Blank (SVIB) in pressed interests do not seem highly reliable but comparison with direct expressions of vocational yet seem highly valid. This result may be due to choice. (Throughout this discussion, expressed the observation made by several authors that ex- choice refers to responses to such questions as pressed interests which develop early are highly "What job do you plan to be in when you graduate predictive. Thus, there appears to be a closer link from college?" or "What do you want to do ?") between the reliability and validity for expressed 0. 47

1 interests than for the SVIIt. Vernotelindzey Study of Values was. 64 percent. There isIII/ e( 111(111 e 14/ show that the sv 1B is I lolland and Lutz (1968) found that self - expressed superior to expressed interests. (:nitislore, and interests had considerably higher predictive va- others have apparently supposed a body of ex- lidity than Holland's own Vocational Preference perimental evidence w hich dues nut exist. Ex- Inventory, in a large sample of college freshmen pressed interests have apparentlybeen given followed up a year later. In a study oldie Project little credence bet ause of studies of "self-esti- TALE:\ r sample. Cooley(1967) found thatex- mated- interests. studies of tovcrlap between in- pressed ehoites in ninth grade predicted grade 12 ventoried and expressed interests, and ttudies expressed choices as well or better than a 17 scale which measured the reliability of expressedin- interest inventor y that was used (54 vs. S3 percent Ierests. The evidence from those studies was correct predictions). Richards (1970). also using taken to be sufficient basis for ruling out the pos- Projet t TALENT data, found that simple expressed sibility that expressed interests might show sub- choice (a "one-item test-) predicted nearly as stantial predictive validity..\s the review indi- well as the best combination of variables taken cates. there were very few studies which directly from 138 variables. compared the sv IB with expressed interests be- Cade and Suliah (1975) compared the predit tive t Jose of the (1,SSMDIA11/1) of the superiority of the accuracy of self-expressed career preferences. collected as part of the American College Testing hi addition. as 1C hit ney 11969) points out, a Progt ain't, routinely administered Strident Profile careful reading of Strong 0953) show s that Strong Se( lion, with an interest test. the Vocational Pref- obtained a c etrielation of .69 between freshman etem c inventory. Both instruments w ere com- him e and m t upation engaged in 19 years Wet. pleted before tire students entered college. Tire The median t orrelation between S' 1K interest pat- t riteria w ere. the majors in which students grad- terns over the saint)(Tirl ws..s .72. Stang him- uated from ci itege. and the initial full-time em- self (1955) states. "1 inrates by the author of the ploy meat of the students as recorded by their uni-

. .. validity of freshmen chide es of or ( upation.s versity's placement center. All variables were are much higher than those previously reported. lassified into Holland's system (1966). The sub- they indicate that sin int Imbrues are related in the jects were 131 male graduates. Expressed elude es majority of c ases to the work the students will do were significantly more nearly accurate than the in later life even w hen they fail to enter the spe- interest test information in predic ting final major cific occupation named in the freshman year:* (67 percent were in the same category using ex- In addition to the studies comparing expressed pressed choice 'Versus 50 percent using the tests). choices and the svIB. four other large-sample as w ell as the initial career fields the students en- studies have e ompared other measures w ith ex- tered (62 per( cut versus 49 pert ent). Expressed pressed Noe atonal t mule e. Cooley (1963) studied t hole es were also superior when the definition of the development of .-cientdieinterest in boys. at curacy was expanded to include e;tegories of Eleventh grade boys predie ted what field they fields that w eret ongruent with the cate5ories would be in. M hen these sell.predit tions were "freshmen c bolt e- or "test scores.- at cording to I militated with the field the student was in after Holland's theory. Fut graduating major. the per- three y ears, the -Int- rate was 50 percent. A c OM- centages were 81 percent for expressed choice bination of interest scalefrom the Knelt r and 71 percent fur the interest test:. for initial ',e- erence Retard also had a 50 per«bnt hit rate; the t imation, the respet tive percentages were 81 and rate for a combination of scales from Allimrt 76 percent. The authors coneluded:

3 48 "Although the authors do not recommend the Norris and Katz also factor analyzed their abandonment of the use of interest tests, the data measures and found: do suggest that college counselors can eliminate "The factor structure of interests remained re the need to require every client who has a voca- markably consistent across different measures, tional or educational concern to take an interest acrossdifferentgroups, and acrossdifferent test. Furthermore, the results suggest that coun- times. Thus, the structure of interests was very selors can have confidence that for many clients similar whether AIM or students own ratings of in- expressed choices have as much validity, if not terests were used: whether the group was com- more, for use in long range college and career posed of male or female enrollees in two-year or planning as inventorized interests have. Inven- four-year colleges, or high school graduates who toried interests may continue to be valuable tools were not enrolled in any educational institution; for counselor use with clients who are undecided whether the interests were measured in grade 11 about vocational plans or with uninformed clients or grade 13." (p. 157) who can learn from the experience of test taking Thus, the basic structure of the student's rat- about the range of occupations in the world of ings of interests was quite similar to the structure work.. . .Yet, it does not seem wise for coun- of the interest tests, suggesting that they assess selors to rely only upon the administration of in- the same psychological domain. terest inventories for every client who considers Norris and Katz also found that student-rated a long-range vocational decision. (p. 120) interests were just as valid as Alm in predicting Finally, in aarefully conducted study, Norris grade 12 or 13 grades, although neither added and Katz (1970) gave a battery of tests to nearly very much to multiple correlations that included 20.000 eleventh grade students in a national sam- the PSAT scores. They conclude: ple. These students were followed up as twelfth "In short, this extension to occupational in graders and again a year after high school gradua- terests of the network of relationships among AIM tion. resulting in a final sample of approximately scores obtained in grade 11 and ratings of inter- 6,000 students. The investigators were thus able ests in grades 11, 12 and 13 suggested that inter- to study the prediction of grade 12 and grade 13 ests occupy a rather well integrated and coherent marks and Interests, using both standard predic- territory in individuals' self-concepts. It is perhaps tion and differential prediction. The basic instru- this very integrity and coherence of interests that ments were a biographical questionnaire, the Pre- has made simple ratings of academic interests liminary Scholastic Aptitude Test (PSAT) and a just as valid predictors as the full AIM scales of carefully constructed interestinstrument. the every criterion used in the study. For use in pre- Academic Interest Measures (Aim). Both twelfth diction, anything AIM can do. si (self-ratings of in- and thirteenth grade interests were predicted terests) can do faster. For use as a criterion meas- about as well by simple eleventh grade (or twelfth ure as recommended in Part I, AIM retains a num- grade) ratings of interest in various subjects as by ber of advantages.'' (pp. 158-159) AIM. The simple ratings of interests also provided This study illustrates several important points somewhat higher differentialvalidities in both about biographical variables. As in many other follownp periods. Another simple itemratings studies, ranch greater attention was given to the of occupational valuesdiscriminated among in- development of the scaled instrument than to the tended major field groups that were not well dif- simple self-estimates. In many studies, this rela ferentiated in a discriminant analysis by PSAT and tive emphasis has tended to pit carefully eon AIM. structed scale scores against rather haphazardly 5,6 written questionnaire items thatarc oftenif). five years after graduation. The discrepancy in chided as an ally' thought. It is therefore surpris these results probably results from the smaller ing that the biographical items do as well a, most number of categories used by Sharp and Kras- of the tests. The second point is tItat apparently negol. Rather than the individual majors used by simple information can tap the strut ture of psy t lio- the National Science Foundation, they used Mne logical domains about as well as tests. as Norris groups of majorse.g., humanities and arts. and Katz found. If the strut lure of a psyt hologit al health, natural science. etc. Thus, the postgrad- domain is not coherent and well integrated. it is uate careers of college graduates seen] to be fairly unlikely that either simple selfestimates or tests stable. These figures agree reasonably well with will find it. If a characteristic is not an important those dev eloped by Sloe um (1974) for professional, part of a domain, it probably will have fewt or. technical, and managerial workers. (However, relates with other variables. If it is important. it jobs based on less education generally are less should have numerous correlates. and, it seems stable.) probable, should be assessable Ity relatively sint Studies byFlanagan, etal. (1966) indicated pie selfreports or simple estimates. that vocational choice gradually seemed to be. In SUM. the evidence leads one to conclude, mne more stable as students progressed through with Dolliver (1969), Whitney (19691. and Holland high school. (Theories of vocational development and Lutz (1968). that there seems no justification such as Super's, and Rosenberg's would predict forthelong-standing biasagainstexpressed such a result, of course). The average consistency choice in favor of tests. The practical implications for 30 listed occupations in the twelfth grade of this «int lusion will be discussed later in this w hen compared to those listed one year later was only 28 percentfor men and 40 Percent for Apart from their comparative validity. how pre- women. However, as reported earlier, Cooley's dictive are expressed choir es? Whitney's ret ent study of the same data divided these occupations review (1969) of this question. previously into six t ategories and found a 54 percent stabil- brings together the relevant evidence from the ity over the four years of high school. large scale longitudinal studies IN > 3,000) that Using another research strategy at the college hear on expressed t hoit e (the Strong study and level, Astin and Patios (1969) studied the voca- the two Cooley studies have already been men- tional choices of 3,821 subjects over four years of tioned). A large study of «dleg graduates con- tallege. They grouped 40 career choices into 12 ducted by the National St km e Foundation (19631 ategories, and developed multiple correlations to found that, for 32 oft upations, the median per predict final t areer chid( e from a wide variety of entage of graduates employed of teat fling it) the biographical information. The correlations ranged area of their major field two years later was 10 from .23 to .53. Freshman expressed choice was pert ent for men and 29 pert rill for women. The the best predu for fin seven categories and second majors varied greatly in stability from 91 pert ent best for the remaining five. (Major field choir e in pharmacy for men and 91 pert mit in (*du( ation was the best predit for for three of those cate- for women to 4 pert blit in philosophy fin men and gories. degree goal for two, and sex for one.) The two pert ent in religion for women. In «intrast. results of these varied studies suggest that self' Sharp and Krasnegor (1966) followed another expressedt hoicc becomes a better and more large sample of college graduates, and found that stable predit for of vocational behavior as people 71 per ent of the men and 83 percent of the become older. The results also suggest that the women were employed in their major field area predit tive pow er of expressed choice varies with

, " so the number and appropriateness of the categories variety of studies at National Merit Corporation, used to classify choices. Holland's system has also been applied to samples of typical college students (Holland and Lutz, Classification systems 1968: Holland and Whitney, 1968). These studies Various classifications were used in the studies showed the system to be reasonably efficient, at just reviewed with various rationales. Usually, least over the period from application to college they were "eyeball" classifications based on as- to the end of the first year of college. Holland and sumed similarity. In contrast to these classifica Whitney (1968) found that 69 percent of the men's bons. Roe (Roe, et al., 1966) and Holland (1966) choices were in the same major class over that have developed classification systems based on period, with another 10 percent in a theory-de- psychological analyses of vocational behavior. In fined "closely related" subgroup ina related these systems, when individuals change their vo- major class. Comparable figures for women were cational choices, they are predicted to change to 84 percent and nine percent. However, as Whit- psychologically related classes. Thus, changes ney (1969) notes, although this "proportion of cor- are evaluated according to their conformity to rect predictions is promising, ...the followup theoretical expectations. In both systems, rela- period was short and the criterion employed was tivelydirect and simple explanations are pro- expressed choice, and not actual employment." vided for the vocational behavior under consid- Holland (1973) has clearly explained the devel- eration. Roe et al. (1966), using Roe's eight cate- opment and rationale of his system, and has pre- gory scheme to study the patterns of movement sented extensive evidence for its validity in more when men change jobs, found that 68 percent than a hundred studies. One of these is particu- moved to jobs in the same major group. Other larly important. moves were most often to related categories. Sub- "Holland, Sorensen, Clark, Nafziger, and Blum : equent studies by Hutchinson and Roe (1968). by (1973) applied the intermediate form of the classi- Osipow (1968), and by Roe and Hutchinson (1969), fication to a national sample of retrospective work have indicated that the Roe System reflects the histories (N = 973) to test the predictive efficiency nonrandom nature of occupational behavior, and of the classification and related hypotheses from that the probability of movement from one group Holland's theory of careers. Analyses were per- to another varies directly with the psychological formed by organizing and reorganizing the work nearness of one group to another. (Meir. 1970, has histories according to the classification. recently suggested an alternate to the Roe System "The classification appears to order lower-level that he believes will improve its predictions). occupational histories in an efficient way, well Holland's system explains changes a similar beyond chance. Over five- and ten-year intervals, way: movementistopsychologicallyrelated 6 x 6 tables show that the percentages of people classes (Holland. 1966. 1973). Holland's theory is remaining in the same main categories are 77.3 based on a six-fold typology of vocational chokes: and 74.2. Other analyses imply that all three let- Realistic type or technical, skilled trades, and ters in an occupational code have predictive i- some engineering fields; Intellectual or scientific lidity. Still other hypotheses from the theory re- and some technical fields; Artistic or artistic, veal that the "consistency" of a man's initial musical and literary fields; Social or educational occupational code forecasts both his level of and social welfare fields; Enterprising or mana- achievement and his tendency to change occupa- gerial and sales fields; and Conventional or office tional categories. Last, a man's level of education, and clerical fields. Originally tested through a income, and prestige is predictable from his ini-

5 (3

51 tial occupational code in accordance with the hy- occupational choices and career occupations. The potheses about education and vli( ational hiev e- classification systems correctly placed many more inem. All such analyses were statisticallysig- individuals than would be expected by chance. As nificant and usually substantial in size.- (p. 80) the individuals become more and more involved Holland and Gottfredson (1974) st udied the psy in the actual world of work, their choices and chologieal meaning and predictive value of an in- activities become more and more stable. For dividual's vocational aspirations by applying Hol- example, the high school senior occupational land's typology to the vocational aspirations of choices were in exactly the same Holland cate- four samples: high school juniors. college juniors. gory as were the actual occupations pursued one employed adults, and a sample of college fresh- year later in 57.8 percent of the cases, the actual men followed up a year later. The basic data con. occupation five years later in 44.7 percent of the sisted of the vocational daydreams reported in cases, and actual occupation eleven years later in Holland's Self Directed Search (Holland, 1971). 38.6 percent of the cases. (All of the percentages Answers to three questions were sought; Do a reported here are taken from the Holland system person's current and retrospective vocational as- because it was the most effective system in every pirations possess psychological coherence or arc comparison.) Individuals in one occupation in they unrelated to each other? Does the degree of 1961 were in an occupation in the same category coherence among a pcison's vocational aspira- four years later in 55.9 percent of the cases, and tions have any psych( logical meaning? Can a per- ten years later in 46.4 percent of the cases. Fi- son's vocational aspirations be coded and St/ in- nally, individuals in an occupation in 1965 were marized to yield an efficient index of future voca- in an occupation in the same category six years tional aspirations? It was found that vocational later in 63.3 percent of the cases. The analyses of aspirations. categorized according to the theory, the systems provided additional information about were at least as efficient as interest scales and the choice process. Most of the changes in occu were often considerably more efficient in the pre- pat ional categories were to adjacent categories in diction of later aspirations or choices. They also the systems; few changes were to categories far found that the more closely related were an removed from the original ones. For example. it dividual's aspirations fur several vocations. the %V as much more likely that a draftsman would be- more predictive they were of subsequent voca- come an electronics technician than a life in- tional aspirations, and the better the individual surance agent. The authors also found a general scored on a decision-making task. This study tendency for a movement away from intellectual shows that vocational aspirations ( an add to the careers to careers in business and sales. This understanding of voc ational choir es when they study. when considered in conjunction with Rich- are organized into theory-based categories. ards' (1970) results, also based on Project TALENT Probably the most onvincing demonstration of data. indicates that when the choices are grouped the predictive power of categorical systems is into theory-based categories of vocations, self- Mt Laughlin and Ticdeman's (1971) ( omparison of reports of vocational c }ikes arc more predictive the hit-rates of the Holland. Roe, and Flanagan of later choices and careers than is test informa- occupational classification systems in predicting tion. career areas one, five and 11 years after high The results of these studies suggest a number school. Following up tb 1960 Project TALENT of possible practical act ions. First, students might sr. -n plc in 1961. 1965. and 1971, the authors ex- be sent the results of their questionnaires and test amilid the change and stability in the sample's results with guidelines for examining the con.

J 0 52 gruency between their own interests. abilities. and psychological resources and the character. The use of biographical information for istics of people currently in their field of choice. understanding vocational choice as well as with the requirements of their field of There have been many studies of the correlates training or vocation. Perhaps general "congru. and determinates of vocational choice (reviewed ency scores" could aid this effort. Students could in Crite's voluminous and comprehensive Voca' also be guided to smallpackage "libraries" of tional Psychology). These studies have led sev vocational and educational information that would eral psychologists to develop theories of career describe the characteristics of people in the field development (five theories are summarized and as well as the requirements for the field. Self- evaluated by Osipow, 1968). Several of these the. classification schemes of the sort reviewed in Sec- ories particularly emphasize biographical vari tion VII, could be used. Students would group ables in the development of vocational interests. their characteristics and interests into classes and Roe (1956, 1964), for example, has particularly investigate the occupations within the occupa emphasized the influence of child rearing prac tional classes congruent with them. In addition. tires and early experience on vocational selection. by using such results in combination with classi Grigg (1959) and Hagen (1960) concluded that fications, a counselor could use students' tenta- their studies of students did not support Roe's tive choices and their histories of choices to guide particular hypothesis that a warmth-cold dimen them to relevant materials. and to diagnose oc- sion in childhood exerts an influence on the later cupational conflicts. Undecided students or stu- choice of a vocation. Utton (1962) also did not find dents unable to make up their minds, could also evidence to support Roe's idea. Switzer, Grigg. roughly classify their own interests and charac- Miller, and Young (1962) did find relations be- teristics and study the occupations in the class tween students' vocational choices and childhood congruent with them. An undecided student could experiences. but in the opposite direction from also be encouraged to study the entire classifica- Roe's prediction. Roe and Siegelman (1964). in tion. The possibility of using biographical infor contrast, did find evidence that the amount of mat ion to locate people for particular fields has early social experience was related to the amount already been suggested. A national center for of later interest in people. They also point out that storage and dissemination of information would other background factors, such as socioeconomic be needed to do it effectively. background and education, as well as abilities, The research reviewed here provides a good play a role in the choice of a specific vocation. deal of information; but dearly more information. Holland (1961, 1962, 1963) has also studied research, and integration of research is needed, parental influences on vocational choices. In early as well as dissemination of that information and studies, Holland (1962, 1963) found that some its incorporation inpractice. In general. these parental attitudes were related to some of his classification schemes. Niiih arc simple to apply groupsofvocationalchoices.For example, and inexpensive. seem to offer a promise of im- mothers of students with Conventional choices proving prediction and understanding of voca- had more authoritarian attitudes. Fathers of boys tional choice something that simple expressed in the Intellectual category valued their childrens' choice alone and most interest tests do not do. curiosity. whereas fathers of boys in the Con- ventional category hoped their sons would he "happy and well adjusted." These results are generally plausible and tend to suggest that the

60 53 origin of interestsis related more to parental and stabilization substage; and adValliCIIICIII sub- values and interest, than to spec ifie child-rearing stage. Super (1963) also outlines five vocational practices. although many parental values will developmental tasks: crystallizing a vocational probably influence parent-child relationships in preference; specifying a vocational preference: subtle and complex ways. implementing a vocational preference: stabilizing Holland (1963. 1964. 1968) has also found that in a()cation: and consolidating status and ad- a variety of self-ratings. life goals, bobbies, extra- vancing in a vocation. Super, Kowalski. and Got- curricularactivities, and creative accomplish- kin (1970) have presented evidence that such ments had logical relations with the types of variables as social status. grades. and participa- choice. Holland (1968) found that, in general, the tion in school and community activities predict same attributes that distinguished between the later vocational choice. Of course, Super'sear six types also distinguished between students phasis On the development of self-concept and within the main types who had different second- itsrelationship to vocational decision and his ary interests. In this study, Holland also found emphasis on the development of vocational ma- that when both the first and second initial voca- turity are closely aligned to developmental as- tional choices of students were in the same cate- ',cuts of vocational choice. Many of the studies gory, their chokes were inure stable over a year Super reviews are related to biographical in- of college than were those of students whose formation, although Super does not present them choices were in different categories. In related in this light. Norton (1971) reviewed the research research, Holland and Nichols (1964) found that on vocational development up to the time of his remaining in a field is associated with the pos- writing, and presents a number of additional session of attributes similar to those of the typical studies that are related to biographical assess- studentin the field. Holland (1973) has sum- ment. marized other research on the development of In summary, there is a considerable body of vocational dispositions. literature that demonstrates the importance of klthough research based on the approach of biographical variables in vocational development. Ginsburg and his associates (1951) has not used many of which can be assessed by relatively biographical informationextensively, the gen- simple self-reports. eral ideas of their approach could be used to cate- gorize some biographical variables. Ginsburg as- Prediction of vocational success sumes three general periods: the fantasy period, 13iographical and self-descriptive information has comprising the period when children have no been related to success in a variety of occupa- basis in facts for their ideas. and imagine many tions. (The research on creativity in occupational careers: the tentative period. when children ex- settings was reviewed in Section III.) The gen- plore their MN n interests, abilities. and enjoyment eral approach taken in most studies has been of various kinds of activities. and the realistic pe- essentially empirical. Typically, as described by riod, iv hen students try outv arious possibilities Baehr and Williams (1964 the general procedure and finally crystallize their choices in a specific is to administer a biographical questionnaire to a vocational pattern. group of applicants or job holders. Then the dis- Super and his colleagues (1957. 1963) have de- rimination pow( r of each item in predicting veloped a similar model, consisting of use sub- sonic criterion of success in the occupation is de- stages of NO( ational bellaN1111. tentative( hole e termined. The items that discriminate arc finally substage. transition substage; trial substage, trial ( ombined into a score, which is used to cross-

6 i $4 validate the results in another similar sample. In (1962) estimated that they could reduce turnover general, this approach or similar approaches with among office workers from 53 percent to 38 per- greater or lesser sophistication. has been used to cent by use of biographical variables. Fleishman predict two kinds of criteria: attrition, turnover, and Berniger (1960) were able to predict tenure and tenure; and ratings of effectiveness or suc- among secretarial and clerical women workers in cess and criteria of performance. It is not the pur- a cross-validation sample with correlations of .77 pose here to review all the work in the area but and .57, respectively. The key misclassified only some representative studies may be mentioned. 22 percent of the short tenure group. Shott, Al- Attrition, turnover, and tenure.Employers arc bright, and Glennon (1963) found similar results naturally concerned with reducing attrition and in samples of male and female workers. Buel turnover, and increasing tenure. If they could (1964), was able to predict turnover with a corre- eliminate people who are likely to leave their lation of .33. More recent studies obtained similar companies theycould reduce personnel costs results. as reviewed by Schuh (1967) and England considerably.It may be useful for colleges to (1971). However, one serious need is to revalidate study this area, since it has many parallels with the weightings every few years because the nature the problems of reducing the dropout rate in col- of the criteria, the situation, and the applicant leges. group can change. Schwab and Oliver (1974), for Schuh (1967) reviewed hundreds of studies of example, presented the results of four studies employee tenure and found 21 that related bio- that had originally used weighted biographical graphical data to tenure. In general, biographical data to predict tenure with original correlations information predicted tenure well. England (1971) running from .37 to .66. In a cross-validation four has also reported good results in the studies of years later, however, no correlation was higher tenure as related to bi,graphical information col- than .10. This suggests the importance of cross- lected from weighted application blanks. Some validation and revalidation. of these studies may he mentioned as examples. Proficiency ratings and performance criteria. In one study, Mosel and Wade (1951) were able to In many studies, biographical information about predict job tenure among department store clerks individual's personal histories predicted ratings with a biographical inventory. In a cross-valida- of proficiency and success. For example, Cureton tion sample. the scale correlated .41 with job ten- (see Henry, 1965) commented that the single item ure. and the authors suggested tentative cutting "Did you ever build a model airplane that flew?" scores on the scale that might be used to increase predicted rated success in flight training during tenure. Dnnnette, Kirchner. Erikson. and Balms World War II almost as well as the entire Air 119601 were able to predict tenure (using tetra- Force battery of tests. Asher (1972) has reviewed clume ( orrelations) among female office workers the relative predictive power of biographical or with correlations of .74. .61. and .38 over one. personal data compared to intelligence, ability, two, and three years. Whether or not this decline aptitude, and personality tests. He summarized in correlation is a result of sample attrition and the results of the many studies analyzed earlier increased homogeneity, it suggests that the op- by Ghiselli (1966) and Ghiselli and Barthol (1953). timal weighting of biographical responses should Biographical or personal data were clearly better be reviewed every few years to be sure that the predictors of job proficiency in a wide variety of best prediction is made. The authors also found a settings. Asher explains the results as due to: the correlation of .73for keypunch operators. In fact that biographical data includes systematic, another industrial setting. American Oil Company comprehensive factual information about an in- 62

SS dividual;the items selectedforbiographical correlations with each criteria were as follows: forms are more relevant to the ( riteriun than other with paired-comparison performance ratings .42; variables; and the items refer to behaviors that with mean sales volume .50; maximum sales have a point-to-point correspondence with the volume rank .30; route difficulty .27; and tenure criterion, as, for example, high school grades have as a salesman .30. These correlations sometimes to college grades. (In some cases, the point-to- included some unexpected results. For example, point correspondence may be artifactual, since tenure was negatively related to vocational satis- the same measures of content are used. Thus, it faction. The authors explain this as due to the may be unclear whether there is continuity of be- older salesmen who were resigned to the occupa- havior or simply consistency in response.) The tion, rather than satisfied with it. superiority of biographical data is particularly im- The usefulness and validity of biographical in- pressive when some of the complexities of criteria formation in industrial and organizational deci- and diversity of groups are considered. Ratings of sions seems to be solidly documented. Biograph- success have many psychometric difficulties, but ical information seems to predict criteria that are are often the most practical criteria for many roles complex and multifaceted with equal or better which are multifaceted, or have few observable success than more time-consuming tests. criteria of success. For example, Kirkpatrick (1960) used ratings of the overall job performance of Chamber of Commerce executives, and found that items in a biographical questionnaire dealing with education, extracurricular activities, employment records. etc., were related to this criterion. In a cross validation an extraordinarily high correlation of .81 was obtained. Lockwood and Parsons (1960) were able to predict ratings of supervisory effectiveness in a cross validation sample with a correlation of .53. In a sample of airmen, Harding and Bottenberg (1961) correlated biographical and attitude variables with absences, ratings, and rankings of performance, with correlations of .38, .43, and .41 respectively. Real life criteria of success arc difficult to construct in many fields, either because "success" is so complicated, or because success is based on many intangibles. Perhaps these difficulties explain the factthat the majority of industrial studies have been of sales occupations, where the criteria is often measurable in dollars. The general complications in this area are indicated by a recent study by Baehr and Williams (1968). They used scores on a biographical inventory based on an earlier factor analysis. These factored scores were correlated with five criteria. The multiple 63

56 Section VI: Requirements of be applied. (This discussion uses the general an effective brief-assessment system categories of Robinson, Rush. and Rush, 1968.) Item construction criteria: proper sampling of content. Brief-assessment measures should meet This section is concerned with the various tech- thiscriterion, both inoverall content and in nical, statistical, operational, and theoretical re- specific items. For example, a questionnaire de- quirements of effective systems of assessment signed to be used in college admissions work based on brief assessments. should include items reflecting the major concerns of students, schools, colleges, and other Psychometric requirements parties concerned with college admissions. In The psychometric requirements of information addition, each item should accurately obtain the provided by biographical questionnaires vary with information necessary for the decisions for which the uses to which the information is put. The re- the item will be used. For example, an item con- quirements are minimal when questionnaires are cerned with a student's requests for assistance used as brief screening devices, for example in from his college that did not include "assistance locating students who say they will need further in finding part-time employment'. would be in- help in particular subject areas. At more complex complete. Of course, the items selected should levels. the traditional psychometric standards are be chosen after a careful review of earlier studies, difficult to apply to biographical questionnaires. and they should be pilot tested, and retained only For example, as discussed in Section V. a stu- if they were as varied and useful as assumed. dent's self-expressed vocational choice may not (Factor analyses of biographical information, such be highly reliable in itself, but, as part of a typ- as those conducted by Freeberg and Rock, 1969, ology and theory of vocational choices, it can be may be useful for this purpose by suggesting a good predictor of later vocational behavior. important variables that should be contained in Thus, an evaluation of self-expressed choice on the instrument in order to achieve scale indepen- psychometric grounds must include an evaluation dence and coherence.) of theories and ty pologies as well as single re- Simplicity and clarity of item wording.This sponses. It is necessary to extend and revise tradi- criterion is especially critic al in brief-assessment tional standards to evaluate these kinds of inno- measures because in many instances, individual vative approaches. Given these restrictions in the items must stand by themselves as pieces of in- construction of self report instruments, however, formation. In contrast, a lengthy scale can in- many traditional measurement ( riteria can still clude many items with marginal relations to the

57 content of the scale, but which. w hen taken as the options in numbers, write the options to in- a whole. add to the. total «overage of the scale. In- e lode all alternatives, or, if this is not possible, dividual itemsinbiographical and self-report provide an "escape" option; give the items a questionnaires must usually be directed exae tly neutral or pleasant connotation. Owens also sug- to the point. any subtlety of wording or hidden gested that items should not ask for information purpose usually will add to misinterpretation and beyond the memory of the respondent, that state- increase the error variam e. This is not to say that ments be positively worded, and that responses the items should not be constructed with care and should not be defined in qualitative terms, such attention to the subtle nuances of words; just the as "seldom." "occasionally," and "frequently." opposite.'l'o use a classical example, Centers (Asher also discusses other aspects of writing 949) was skeptical about the large percentage biographical items including a number of sug- of people in a survey who said they were "middle gestions for future research.) class" when asked to place themselves in the These principles appear obvious, but are often lower, middle, or upper class. Ile simply added difficultto implement. Clarity, directness, and theadditionalcategory"working class" and simplicity are difficult to obtain, but should be found that the majority of those who had previ- sought as goals. As Kuder (1970) concludes: ously identified themselves as "middle class" now "A person w ho follows these principles should identified themselves as "working class." These not be surprised if he produces a deceptively people had not wished to call themselves "lower simple questionnaire which looks like something class" bee ause theyfeltthat designation had a junior high school student might have thrown moral and e tiltural a onnotations in addition to au together on a rainy Saturday afternoon. That's the economic. caning. In addition, the questions way it should look!" (p. 225) should not be. offensive, unpleasant, or threaten- Item analysis.When items are grouped into ing to any of the groups using the instrument. This scales, they can be subject to examination by guideline should be followed e autiously. sine e it traditional item analysis techniques; item inter- is probably impossible to develop an item that no correlations, item discrimination analyses, and one would find offensive. At a minimum, the items especially the correlations of the item. -with ex- should be reviewed by repiesentatis es Of the ternal criteria (item validities). Individual items, groups that would be using it. For example, a however, are more di& tilt to evaluate. The evalu- broad-band questionnaire for t allege admissions ation tends to be based on the extent to whit h the should be reviewed by students of different ability item discriminates among criterion groups, its levels, interests, bat kgrounds. anti ethnic groups. correlates with criterion variables, and its rela- Consideration should also be given to different es tions to other kinds of information, including in age and set. NV, hen the instrument is adminis- tether biographical items or other tests designed tered. the groups should be told the purpose of to measure the same. variables. the instrument and how its information would be Scale conAtruetion. Most of the considerations used. in constructing biographical and self - report meas- ,1sher 11972) has ret end) summarized a good ures arc. similar to those in the construction of deal of earlier work in an article entitled. "The achievement and attitude scales (e.g., Lindquist, Biographie al Item: CanIt Be Improved?" lie 1951. Adkins, 1971). However, two related prob- notes the four rules developed by Owens. Glen- lems are particularly important in biographical 11011, and Albright (1966) for writing items that self-report measures. scale keying and scaling. w ill be stable. be brief. wherever possible express Biographical self-report sc ales may cover an un-

6' 58 usually wide area of content and are consequently misuse of the information. difficult to key. For example, a composite scale of Control of response set: acquiescence and social the educational stimulation of the home might in- desirability.The control of acquiescence can clude items on the' existence of art supplies, sometimes be a problem in brief-assessment in- globes, microscopes, or tape-recording equipment struments,althoughprobablylessfrequently in the home: the number of books in the library than in personality tests or opinion surveys. As in and the number of magazines and newspapers to those measures, the constructor of brief-assess- which the family subscribed: the attitudes toward ment devices must he concerned with the context education held by the parents; the neighborhood of individual items, and the stems in the items. context, etc. Itis difficult to make any a priori The constructor of a self-report instrument can judgments about the relatis e importance of these lessen "social desirability" by trying, where it is specificsforgeneraleducationalstimulation. appropriate to do so, to equate item stems for Even when item statistics might indicate that social desirability, and by the use of forced-choice such items form a statistically coherent scale. item formats. Another approach is to develop a care needs to be taken that conceptually and em- score based on the choice of socially desirable pirically distinct concepts are not merged into a alternatives (e.g., following Smith's, 1967, pro- scale that would be too heterogenous to be useful. cedure). The technical problems in scaling biographical Since many biographical and self-report ques- self-report scales are fairly straightforward. How- tionnaires are concerned with accomplishments ever, there are several conceptual problems. For and achievements, another control should be example. a scale for the prediction of accomplish- made for any tendencies for exaggeration. One ment may include private, unverifiable, and fairly way is to develop a scale of unusual accomplish- lowlevel accomplishmentsfor example, writing ments that are uncorrelated or negatively corre- a short story on one's own, public, verifiable. but lated, so that a person who claims many of these fairly low-level accomplishments such as writing rare uncorrelated accomplishments is very proba- a short story for a writing class, private, difficult bly exaggerating his accomplishments. (Holland to verify accomplishments of a fairly high level and Richards, 1967, developed a scale of this for example, writing a novel that was not accepted kind to control for exaggeration of non-academic by a publisher: public, and verifiable accomplish- achievements.) The fakability of the items is a ments, of varying levels for example, winning a particularly important problem in biographical school or college prize fur writing: and public. and self-report questionnaires as previously dis- verifiable, high level accomplishments (having a cussed in Section III. Since the items are de- story published by a national magazine) and out- signed to be simple, direct, and to the point, they standingpublicaccomplishments (winninga are eminently fakable, particularly when their National Book Award priz). Although a scale of purpose is often quite transparent. Given these thistype,. based on these varied accomplish- conditions, there seem to be few ways to reduce ments, would assess a continuum of accomplish- the fakability of the items. But there may be ment, it is still possible that one applicant would several ways to reduce the actual incidence of have three low-level accomplishments in contrast faking or distortion. And here, ironically, the to another applicant who would have one high- very transparency of the items may be useful. level accomplishment. Thus, scaling the levels of Since the items often cover factual matters that accomplishment needs to be done very carefully. can be checked, the respondent may feel that his and interpretive guides written to minimize the responses will be checked with realitr if he dis- 6 6

59 torts them. The instrut lions to the questionnaire to apply to individual item measures. The excep- could be vi tine!' in sin lea way as 1011111)1) that un- tion is test-retest reliability. and even here the usual responses will be checked correlation formulas are not always applicable. The sample.It seems a truism to note that bio- since many items do not form "scales" in any graphical and self-report questionnaires should sense. Rather, the basic pointis whether re- be developed from a Sample that is representa- spondents make the same choices in an item tive of the population in which the questionnaire w hen the item is presented a second time (i.e.. is to be used. However, brief-assessment instru- stability). However. even stability may be a prob- ments List% in prat tit e, been used in populations lem. For example, family income or educational considerably different from the samples used to plans may actually change over a period. Various develop them. In addition. because of the seeming statistics (mild he used to estimate the extent of simplicity of biographical and self-report ques- this consistency, but the extent to which these tionnaire items. questionnaires are sometimes statistics produce an overall estimate of reliabil- developed on inadequate samples. and are in- ity" as accurate or powerful as the more typical validly generalized to a population of interest to reliability estimates is questionable. Where scales the developer. Thus. biases are sometimes built are used, item-total scale relationships are in- intothe items selected. Questionnaires to be dicative of reliability (internal consistency). used with very diverse groups. such as high Homogeneity vs. heterogeneity.Since biographi- st Imof students, or college applicants should be cal and self-report questionnaires are generally drawn from developmental efforts based on a designed to serve a number of purposes. some sample representing that diversity. The nature heterogeneity of content should be desired as well of this sample will influence the adequacy of the as expected. Individual sc ales should have in- items asked and the way they are interpreted. Fin ternal homogeneity, however. and scales of bio- example. in a question about family income. the graphical questionaaires should show factorial incomen ategtnies usefulina sample of Is invariance over time (Freeberg and Rock. 1969). League students might not be appropriate for use Validity. Theitems and scales of biographical with most college students. and self-report measures should be shown to dis- .Vormatite information.Detailed information t riminate among important groups and correlate about the groups used for 1101-111t, is as important with relevant criteria. They should also maintain for biographic al and self-report questionnaires as their !omen of discrimination and validity in cross - itis for norms for test scales. If items are to be validation samples. Special consideration should used as basin information, then distributions of be given to validation of biographical and self- pert entages n housing eat h alternative should be report questionnaires in situations similar to those prepared for samples of each group that would in which they will be used in practice. For ex- be a major user of the instrument, and for each ample. questionnaires developed for use in col- group thatrepresents ariteriongroup. The lege admissions should be validated with a sample mean, standard deviation, and distribution of of applit ants, in the kind of selet lion situation st ores for these groups on any sn ales developed for which the questionnaire will be used, rather in a !Hier-assessment instrument should also be than in a sample of admitted students. reported. The review of literature in the previous sections Rellabiho. Traditional concepts of reliability illustrated the multiple uses of biographical and applytost ales used in biographic al and self- self-report questionnaires. The biographical and report questionnaires. but are generally difficult self-reportinformation usuallydiscriminated

67 60 among criterion groups as well as test informa- outline four basic problems: the variability or tion. In addition, the information discriminated reliability of actual performance; the reliability among groups that have many characteristics in oftheobservanceofhumanperformance; common; for example, between more and less whether performance involves a single dimension creative Ph.D. chemists, matched for age and ex- or several dimensions, and, if several dimensions perience. Thus. the power of biographical and exist, how one may deal with the dimensions in self-report informationtodistinguishrelevant practice; and how performance on the same task groups based on numerous performance charac- may be affected by being required in different teristics seems well established. However, an ef- situations the problem of generalizability. fort should be made to insure the relevance and On the firstpoint. the variability of perfor- applicability of the criterion groups chosen, with mance. Fiske and Rice (1955) outlined three types special care taken to avoid generalizing from ex- of behavioral variability: spontaneous or random treme groups. For example, information that dis- variability; systematic variability; and variability tinguishes between English majors and shop due to task differences. Other studies suggest metalworking majors at a two-) earcollegeshould that "people are performing to arrive at a rather not he used its a general guide for distinguishing limited outcome, but they get there using different betweentransferand occupationalstudents. performances or responses." (Ronan and Priers, Similarly, information distinguishing between a p. 89). An argument can be made that biographi- single unselective lot al two-year college and a cal information about past accomplishments and highly selective nations! university (for example. experiences provides information about the typi- comparing San Jose City College and Stanford cal performance in the past, rather than the maxi- University) is an unreliable guide for guidance of mal perfelance assessed by tests, and, in this most students in choosing a college. sense, assesses the systematic variability of past The criterion.The criterion is the most impor- performance. Biographical information can also tant variable in all validation. The importance of provide data about the differences in the tasks the criterion to self-report measures is suggested and task situations in which the performance. by a book edited by W. W. Ronan and Erich Prien. occurred. Perspectives on the Measurement of Human Per- Related to the second point on leliability of ob. formance(1971). The book consists of reviews servation. Kipnis (1960) discussed the problems and readings concerning the criterion problem. when ratings are used in situations where sonic

The authors attempt to show that "...the prob- extraneous value (salary or promotion) is depen lem is actually that of shifting research emphasis dent on the rating. including propinquity of the from individual differences (as in psychological rater and ratee, social setting, expressions of tests) to understanding human performance in the criticism. and specific behaviors affecting ratings. Ronan and l'rien (1971) suggest sonic things that 'real world ... theriterion, as an evaluative index of performance behavior, isreally a sub- might be done to improve the reliability of ob- problem in the field of understanding human servation when these drawbacks are considered: performance . . . (arid) the study of human per- recognize that almost every task has more than formance, per se. has been a sadly neglected area or.e dimension; seek facts that can be collected of research, and. in consequence. future progress about a person, other than ratings; recognize that must include the shiftin research mentioned some may be assessable only by ratings. above if any significant progress in performance Biographical and self-report information about prediction is to Inc accomplished." (p. v). They past performance obviously provides data about 68

I I the variety of rapier behavior that may be rele- Il':1',11111on olganizational behavior has been vant to ( uirent ,11111/ilil,, as suggested in Se( lion evident. However,larch and Simon I1958) point I t'. It also provides Wally other fele% ant fa( is that out that die writings about organizations are s( at- mayinfluenceindividuals'pet human«. and tired and diverse, and that the literature dis- ratings of their performam e. It is often the only closes large discrepant ies betw ern hypotheses type of information analogous to the inhumation and ev idene t. The literat ore c ontains many asser (1)11ected by ratings. Thus, bioglaphical and el bons, often with little data to back them up. Re- report information has (Ionsiderable potential for sear( II on this toph has traditionally been carried matching oust .vedperformance, and.( nose out through laboratory investigations, field experi (wend), has potential as a sour( r of predh toi intuits. and the innaorganizational approach. Al- variables. though little laboratory research has been di. Dunnette (1963). in a dim ussion relevant to the restedtowardtheinvestigationof industrial Ronan and 1'6 n's third issue, the dimensionality organizations per se. much laboratory research of performance, has emphasized that perfoun- which has been conducted on small groups may ante behavior needs to be thought of as a c omplex beconsidered to have relevance to the process event that requires study before any predictive and perhaps particularly the unprogranied activi- tudie. are ted. A varietyof studies have ties of groups that occur in formal organizations... demonstiated the(4)niplexityof job perform- Most of the current literature of research on ant e even in (ostensibly simple positions. t hiefly ical-life organizations has been provided by the by1,1( tor analysis. Often, the dimensions identi- intrainganizationalapproach.Suchresearch fiedin job performan«e are negatively related, typically involves the investigation of one or a Cnfortunately, most studies of performan«.air 1,111,111 number of firms. The possibility of general- -one shot- affairs that do not build on previous izing from such studies has been necessarily research. and that are not repli( ated in subse- ( urtailed because there has been no sampling of quent researt h. Consequently, itisdifficult to organizations, of time periods. or ( ontrol over the outline the important dimensions in 1.1Cff111" Mall( e relevant organizational variables which would ex- ill many areas. sin( e there is no c ()herein pattern plain the circumstances under which relation. in the available research. ships do or do not occur:. (p. 300) Biographical and sell-report information seems Dunteman found that organizational attributes to provide. a better ham. of information about the and individual behavior in 234 industrial (organi- variety and( omplexity of ( arlier performan«. zations were highly idependent, and concludes and expriiem es than do most test batteries. The that as more organizational variables are c onsid- inhumation ( oriel led can be -tailored- to refle( t crud, orig7md relationships among variables be- the important dimensions of the perfin man«. ( come altered and take on new significant e. Simi- teriaind the (lin Ide !o. of the lea! riteria larly. Bo( k, Baird. and Linn 0972) found that ((l- ( an be matt lied byjoint [nation about similarly iege c haiat tristi( s have all impa( t on the amount complex and rule vain cadier performan«.s and students lean]. even when the students' initial experiene es. characteristics are taken into account. Thus, the Dioneman 11966) has ( ()lulu( led a plovoc alive organizational ( context would seem to have a per study that bears on 1Z4111,111 and Prier's last issue. %asiv e effcl t on performance and c riteria. the influent e of the situation on pellormain c. Biographic al and sell-report informati in would Duntemnan pointed out that. seem to have considerable potential for obtaining -In re( cot theory and consequent information about the organizational context of 6.)

62 earlier performance. It (-mild provide important olated more directly to future behavior. Forex- data for moderator analysis and for understanding ample. biographical data may be especially useful earlier performance. Foeexample. Hackman. in assessing students' educational experiences Wiggins, and Bass (1970) have found that informa- and suggesting reasonable bases to explain their tion about the college a student attendedwas academic performance. In addition, since bio- more useful than GU scores and most grades for graphical and self-report information assess char- the prediction of graduate school performance acteristics that are factorially complex. it is likely six years after enrollment. thatitwill provide predictors of similarly fae- In short, information about earlier performance torially complex criteria. Thus, for all these rea- and contexts would seem to reflect the diversity sons, biographical and self-report information and complexity of real-life criteria, since it is itself seems to have many properties that meet Cron - a record of real-life events. bach's prescription for validation as explanation. Cronbach (1970) has emphasized that vali- However, some of the advantages just claimed dation examines the soundness of all the inter- must be viewed with caution, and this represents pretations of a test descriptive and explanatory the other side of the argument. The very "inter- interpretations as well as situation-bound predic- pretability.' and "common sense- quality of bio- tions. That is, tests, or any standardized informa- graphical and self-report data may lead to misin- tion, can be used to produce a description ofa terpretation of the predictive validities obtained person. This description has many implications in research. When "common sense- is incorrect, that will bear on future decisions. "To validatea it leads to error. For example, most people think description is to answer the question: When per- they know why rich and poor people differ, and sons are described in this way on the basis of how to interpret differences due to age but their these data, how much confidence( an be placed ideas may be quite erroneous. The problem is due in each of the implications of the description'?" to the frequent error or exaggeration of these This approach to validity eats two ways in itsap- common sense interpretations. Thus, the task for plication to biographical variables. On one side, researchers using biographical and self-report it(-an be said any assessment period is an arti- variables often is not just to establish a valid in- ficial occasion for observation, used to predict terpretation, but to correct the unwarranted inter- subsequent observations in situations arising in pretations that tend to arise. This frequently re- the natural course of the person's work or school- quires a great deal of substantive research into ing. In contrast to tests, biographical information the nature of the particular variable, whether it measures obtain broad-based data about a per- be social class, sex, race, college attended, or son's behavior insimilar "natural situations whatever. Beyond the empirical research into from the past, and thus can provide relevantpre- what is true, theoretical constructs to aid under- dictor information. Biographical information us- standing are required. All of this suggests that we ually refers to typical performance. whereas tests need much more substantive research on the often assess maximal performance. Furthermore, meaning of many biographical and self-report in- in contrast to tests that arc samplings of behaviors formation variables before we can say they are or attitudes usually in a few narrow areas, bio- highly valid in Cronbach's sense. However, many graphical data are likely to assess large chunks other biographical and selfreport variables are of relevant past behavior and to include many reasonably valid in Cronbach's sense, and several diverse areas. Thus, biographical variables are important ones show equal or greater validity often readily interpretable and ran be extrap- than comparable tests. 70

63 system could consist of additional information. as Program and technical requirements in the College Locator System or a career.sug- of (1 briefassessment information system gestor system, or guides to further information, \lany of the technical requirements of brief-as- for example. references touseful information sessment systems have already been met by the about colleges and occupations. Another backup CollegeEntrance Examination Board, Educa- and self-report system could consist of well tional Testing Service. and the 1merican College validated tests that provide information for any Testing Program in their respective student in- other decisions that would remain, for example, a formation forms, but itis still useful to disc US!, valid test of clerical aptitude that could be used those requirements. As with all other aspects, the for students reporting some interest and talent in requirements are determined by the uses to which clerical occupations. This system would provide the biographical and self-report data are put. A students with a place to go. and materials to aid multipurpose system has, of course, multiple re- their thinking after they have received their score quirements. The first requirement is simplicity reports and prediction equations. All of these and clarity of format, so that the items can be various requirements seem feasible, although the answered easily and directly. A second require- work and funds necessary may be very great. ment is rapid turnaround time, so that the student, college, or school ran make use of the re- The theoretical requirements of efficient systems sults in their decisions. These two requirements dictate the kind of data processing and scoring of brief-assessment information systems used, and bear on the next requirement Itis striking that virtually all studies using bio- easily understood reporting services. The for- graphical and self-report information are strictly mat of these report: will be different for different empirical. and that the number of articles users, and each depends on certain backup sys- reviewed in the previous pages have led to very tems. For example, in order to tell each user what few theoretical statements. To some degree this the chances are thata patticular student will lack of theory is due to the practical nature of the earn at least a C average at a certain college, a problems investigated. However, the most plausi- previous prediction study must have been made. ble reason for the lack of theory is the general low The data from this study should he fed into a cen- regard in which biographical and self-report vari- tral information system, where it can be the basis ables are held by many theorists and researchers. for processing student scores and information. Such variables appear to be simple. common- and the results fed bat k to the various users. In sensit al, even banal. Since they are "obvious" addition to these kinds of pro( essing programs. they may well be overlooked as basic data for reporting programs are also neededfor example. theoretical systems. Of course. biographical and assemblies of rosters and reports for counselors self-report differences are indirectly used to test or agent ies. summarization programs lot t °lieges theories. for example, when students with differ- or other agent its that want profiles of groups of ent majors are expected to show certain traits, students, and so on. Then a system of interpretive or when students reporting high grades are taken materials must be written and distributed so that as a group of "high achievers" in a theory of the reports can be understood and used most ef- at hievement motivation.Ingeneral, however, fectively. In addition, an ideal system would have there have been few systematic efforts to use bio, several backup systems of student and institu- graphical and self-report information as the basis tional interpretative materials, For example. one of t heories.

64 71. Of course. as the many useful applications of fluency of the peer group on achievement motiva brief data reviewed in this monograph tom) or arranging success experiences (since have shown, theories are not required for the theory suggests that self' concept and self.eonfi practical use of biographical and selfreport in. deuce have a strong influence on acievement). formation. In this sense, brief measures are simi In addition, biographical and selreport informa lar to most tests, which are also directed toward lion can help shape theory. Such data can provide practical questions, and are often unrelated to the raw materials for a variety of theoretical theory. The major uses of both tests and bio approaches. and can test the validity of theory graphical data. such as predicting grade:. vora based propositions. tional choices, or supervisors' ratings, can be Another way of suggesting the importance of treated simply as empiriezd problems, and with theory for use with biographical and selfreport considerable success. However, there are several data is to examine the role of theory in serving the good reasons to hope for more theoretical systems general functions of brief-assessment information that are based on or can be applied to. biographic outlined in Section VII: (I) Theory can increase cal and selfreport information. The most impor- the chances that the broad-band information pro tant reason for hope is that a brief assessment vided by brifassessent measures includes the information system can be more efficient and ef. appropriate variables to insure that there will he fetive if based on testable theoretical ideas, as no -gaps- in the spectrum of relevant information the review of the prediction of vocational choice provided by the measures. (2) By insuring compre- indicate(:.With previouslyverifiedtheoretical hensive and relevant coverage, theory can help constructs, it is possible to evaluate the results of make selection and placement decisions more a particular :turfy or measure with greater suc nearly rational. Theory can suggest additional ess. For exampi, a theory of industrial creativity predictor variables and criteria that can help could help explain why a particular predictor did make the decisions more precise and relevant. not work, and suggest other predictors that would In addition, theory can suggest variables that can be more efficient. A theory could also suggest be used for differential piedictiom or as modera variables or information that should be included tors. (3) Theory applied to biographical and self- in a system. A biographical and selfreport in report information can be a powerful source of formation system can also take advantage of hypotheses for counseling and guidance and can theoretical relations. For example. by postulating suggest a wider range of appropriate techniques theoretical relations among types of vocational for counseling the client. GO For the same reasons. choices. a system could refer students with a students or others who work through a question tentative vo( ational choice to information about naire, using it as a learning instrument. can be related choices. This example bears on another referred to the clevant materials or to reasonable use of theory in biographical and self-report sys options to enhance their self knowledge. (5)Theo terns suggesting appropriate actions for the re ries applied to students %,ith special needs ollid cipientsof the information generated by the suggest appropriate actions. (6) Helping colleges system. For example. by using theory-based in. or other agencies in their planning and adminis- formation, a counselor could select the most ape trative activities conceivably could be aided by propriate counseling technique to use with an theory. but the application is yet to be shown. underachieving bright studeni who collies for (7) As the examples discussed in this section sug. counseling. The counselor mil at choose between gest, theory could greatly aid a two-way system of group counseling (since theory indicates the in information and action. by defining appropriate 72

65 assessment variables, developing relevant bac k- distributions. Although there have been a number up systems and materials. and guiding the student of recent developments in statistical techniques or client to the most appropriate materials. 18/ usingcategoricalinformation(forexample, Finally. theory can help us understand the transi- Cocka, 1973: Rock. 1974: Weds and Linn, 1970, tion from high school to colleges and jobs. by sug- traditional (and the most powerful) methods de- gesting the personal. social. and economic factors pend on assumptions of normal, continuous vari- involved in ambition. educational ac hie% einem. ables. While some of these techniques are so ro- and real-life accomplishment. In sum, it seems bust that their assumptions need not be met to that theory could help biographical and he. limpid the last detail. many may be used only with great information systems fulfill many of their potential aution. The standard tetimiques of analyzing functions. data involving proportions, categorical informa- Of course. the potential contribution of theory tion, measures of association. and su un. are out- to systems, as suggested in the last paragraph, lined in numerous sources. depends on the actual construction of theories Rather than repeat those descriptions, this using brief-assessment data. Theories can be con- section will be devoted to some topics that seem structed from such data. as some of the examples particularly relevantto biographical and self- in this monograph indicate. At a minimum. bio- report data. the use of factor analysis. multiple graphical and self-report data can be used in the regression, canonical correlation, differential pre- first stage of inquiry for some theories to test out diction and multiple absolute prediction. and the ideas. More precise testing of the ideas could be use of biographical variables as moderators. left to later well-validated instruments or experi Brief assessment information and factor anal- mnts. Another use of biographical data is in the ysis.Factor analysis is a method that has attracted opposite direction. That is. ideas developed in the both passionate adherents and passionate de- laboratory or with measures using small samples tractors. Among the adherents there arc continu- can be tested in broad scale samples using bio- ing controversies concerning the proper rotation graphical and self-report questionnaires. There techniques,estimatesof communalities.etc. are at least three areas where the use of bio- These controversies are beyond the scope of this graphical self-report data seems to be leading to monograph, but they have consequences for the theoretical conceptions:creative achievement. interpretation of factors and the ways the data educational aspirations and attainment. and as- can be arranged and rearranged. With some limi- sessing the stimulus qualities of the environment. tations. itis possible to obtain results much in If biographical andself-reportinformationis line with those the factor analyst expects. One viewed by researchers as better than "sec ond- analyst may come to solutions quite different class data,'' more theoretical ideas may be ex- from those of another using the same data, and pected. the factor solution is directly dependent on the number and variety of variables that constitute Statistical methods appropriate for the input data. For these reasons, factor analysis brief-assessment information must be used with considerable caution. The problem of choosing the most appropriate Factor analysis should be used with special statistical technique for the use of biographical t are with biographical and self-report data, par- and self-report data is adiffit ult one. be ause tic ularly when each item is supposed to serve a most data of this type consists of responses to different purpose. as in broad-band assessments. categorical questions. which may have unusual In this case. many of the factors may have load-

7 66 ings on only a few items, and the common factors one must be cautious. because of the ways in may account fur relatively little of the variance. which the statistical technique may be used. Mul- The specific factors may be the must valuable in tiple regression can both discover and obscure im- the factorial makeup of the biographical question- portantrelations.Interaction effects are espe- naire. As Cronbaeh (1970) has in ritten: cially tricky with multiple regression techniques. -Too energetic a wielding of the scrub brush Biographical and other self-report information of parsimony scrapes away significant informa- can be useful for multiple prediction for several tion. A baby leopard with his spots scrubbed off reasons. First, the variables often refer to vari- is no doubt a kitten. but saying that leopards have ables that arc similar to the criterion variables, for a high loading on the eat factor does not imply example, past grades predicting future grades. that one should discard the concept of leopard past sales volume predicting future sales volume, and take the kitten home for a pet." and so on. A practical advantage is that they can In spite of the caveats outlined above, factor provide broadband assessment without using an analysis can still be a useful beginning fur devis- elaborate testing battery. The multiple regression ing stable biographical and self-report scales, par- procedure can then select the combination from ticularly when the instrument is designed for a the diverse biographical variables that may lead singlepurpose. such aspredicting academic to the most efficient prediction. The diversity of achievement or sales success (e.g., Baehr and content may increase the correlation. Here, how- Williams, 1967). In such analyses, a great number ever, is one drawback of studies using biograph- of items may be reduced to a much smaller num- ical and self-report datathey have generally ber of the most coherent and important dimen- been extremely empirical, and the results predict- sions that can, in turn, suggest scales or new items ing an outcome in one setting can often not be to assess those dimensions. The results may also generalized to other settings. To some degree suggest interpretations of the factors that can lead these failures in replication are due to differences to a rational as well as an empirical approach to in measures, samples, conditions, and so on, but describing the content area (Baehr and Williams. much of the difficulty lies in the emphasis on em- 1968). A rational interpretation may be especially piricism itself. If there were more interpretation critical when a longitudinal study is conducted, so and attempts to state hypothesized reasons for a that the biographical and self-report factors that result, the findings of one study could be com- are relatively invariant may be assessed, the rela- pared with those of another with some profit, and tive strengths of the factors at each time may be the similarities and differences in results could be determined: and changes in the factor structure understood and evaluated better. may be described. Of course, the measures devel- When an overall assessment also includes test oped from the results of the analyses must be data, biographical and self-report information can evaluated against external criteria. Later sections be a source of suppressor variables. because this suggest some ways to do that. type of information assesses things that are lac- Multiple regression and canonical correlation. torially complex. That is, some biographical and Multiple regression is discussed briefly in many self-reportvariables may involve variance not basic statisticaltexts.Multiple regression ap- found in the criterion that is found in other pre- pliedto biographical and self-report data has dictors. In sum, biographical information may be been widely used in industrial settings. and in a quite useful in predicting a wide variety of out- number of educational settings (e.g., Astin and comes with multiple regression. since brief-as- 1'a nos. 1969: Flanagan, et al., 1966). Here again. sessment data can include a great deal of non-test

74 67 information. not only assessing ogniti% e arid111111- !cello! mane . on multiple c riteria. using multiple (411411ilil 1/s)( hologic al halts. but also assessing li/lS.lit multiple differential lurch( thin the social rind background %ariablcs. solution sought is one that will find the perform- A few studies have used a related ter liniqu. anc area in which each person will clo best. In canonical correlation. This method begins with multiple absolute predic Hon. the solution sought multiple predictors and multiple c riteria (or any would seek a maximal performance for the entire other two sets of variables) and attempts to find sample on all c litria. Hurst (1966 des( laces Hu- the linear combinations in each that will result in t it techniques as follows: the maximum c orrclation between the lt%1) In multiple differential predic tion the problem reltli11/11 is simil.1t to isIII use .1 set of multiple predit tors to !credit factor analysis in this way. after one set of linear eat II 4.1 number of different criteria. Here we combinations is wino% ed from the first-order c or- ish to provide differential prediction so that. relation matrix. other sets of linear c ombiliations having estimated eath one of a number of differ- an be removed from the residual c orrelation ma- ent c ritcrion measure:, for each person in the sam- trix. This femme of the ter Imique shows the vari- ple on the basis ,,f-pretlic tot measures. we ( an de- ous ways in which the sets of variables are re- termine for w hick of the critefirm measures the lated. Canonical correlation. therefore. seems a person might be best suited. The problem is to get highly useful way to examine the overall relations predic thins of sut c est, w hich will give maximum between theoretic allyrelated types of variables. differentiation for the individuals with reference such as interests and other personal traits. (For to the estimated criterion measures. details of the method. see (:ooley and I.ohnes. The solution of the predictor selection prob 1971.) This method has been used in .t few studies leni assume:, that we have a relatively large Ill1111 involving hiographit al and self-report data. e.g.. her of predictor measures, and that we wish to de- Baird (1970). liutif other methods must be in- termine which of these measures should be se- terptctetl with caution. c atomic al c orrelation war- lected so that when they are used with the rant special caution. The weightings in the vet appropriate prediction formulas, we will get tors may be dire to .pet ificN.11-14111 «'S or c oninion mum differentiation among the estimates of the ariam Sometimes. are St/ (1/111- criterionmeasures. These estimated criterion 1)1(A the 111.1% face information I% measures will provide a basis then for assigning 1141(1 .11111 diflic critics in inter pretation. flic c arioni- eat It person. ur 11/I advising each person about the c al proc edirre sometimes suppresses stone vari- activity for which he may be best suited... ables while maximizing others. and seems lit be \lultiple absolute prediction of multiple t ri highly dependent carthe variables unc hided in teria is slightly different from the model we 11,1% each data set.inally. the researcher must be just discussed... sore that both sets «insist of variables that c an be The problem then is to select from a larger represented meaningfully by %cc tors c imposed of pool of potential predittoi variables those whit I' additive c ornbinations of separate elements. (See will give the highest sum of multiple t orrclations (;iillahorn's 1967 disc lIS,11/11 of her results.) with the multiple criterion measures. regardless iol predit Ilan am! nailtopleabsolute of the t orrelations among the prctlic red c lit erion pretlirt ion. The first of these l411 IerlIllifille de- variables... veloped by Horst (1931. 19331. has already been "It should be noted that in general the formulas mentioned ill Set tutu I. In bush 111110(1s (Il prob- for the Koh( for selec non in the multiple tliffer- lem is to make the best (let isi1111S for maximizing ntial approach willnotselect the same pre-

7 5 68 dictors as in the case of the multiple absolute drawback of scales developed using these meth- prediction approach. The latter approach will ods is that they have generally not been inter tend to select predictors which measure variane e. preted. but simply used for empirical studies. common to all the criterion variables. and w ill Thus. when the utility of the scale in a new sam- therefore cause correlation among the predicted ple is less. there are no guidelines for understand- criteria. On the other hand. for differential predie ing why. or for attempts to improve the scale. Of Lion the selection procedure tends to select those course this need not be the ease and the investi- predictors which do not predict that variance gator could at least try to characterize the content which is common to the criterion measures." (pp. of the scale. Examining the correlation of the 361-362) scale with other measures or factor analyzing the Lunneborg's fine statement of the ease for the scale w ould also probably be useful. (2) A similar use of biographical information in multiple differ- technique in which item responses are correlated ential prediction was quoted in Section 1. Much with the difference between actual and predicted the same reasoning applies to multiple absolute criterion scores has been used by England (1971). predictors: that is.itis likely that biographical The scales developed by this method are then and selfassessment information will include vari- used as predictors in a multiple regression equa- ables related to the multiple criteria. Of course. tion. Again. these scales tend to be atheoretical. both of these methods also require careful inter- (3) Berdie 0961. 1969a. 196910. in his studies of ',relation and replication of results. intra-individual variability and predictability, has The use of biographical and self-report vari- suggested that persons whose performances on ables as single and multiple moderators.Modera- tests are variable will also be less predictable than tor research has attempted to find subgroups of others. Variability itself seems to Berdie to be a individuals who are relatively homogeneous in generalized trait, a construct in its own right. Ap- terms of their predictability, and/or for whom a parentlyitmay be measured in many ways. predictor or set of regression weights are espe- Berdie has shown consistent but non-significant cially appropriate. Among the major approaches differences in R's between high variability and to the identification of moderator variables are the how variability students. However. his approach following:(I) Ghiselh's method which selects has also been basically empirical. If the construct items that correlate most highly with the absolute of variability is to be supported, it needs much difference between standardized predictor and more evidence to validate it, and to show its rela- criterion scores (Ghiselli. 1960a. 1960b. 1963). tions to other traits. but Berdie's results are highly Using this approach. a scale can then be (level- suggestive. (-1) Finally. the search for scales or at- oped to predict the degree of correspondence be- tributes, either single or multiple. that charac- tween predictor and criterion variables expected terize more or less predictable people have been for any person. Jorgenson 11970) has extended this the main emphasis of the most active researchers logic to identify items that discriminated among in moderator research. Since this research is ex- groups whose criterion scores were predicted by tensive, it is dealt with at greater length here. one predictor better than by another. Jorgenson The single moderator research has already also identified items that discriminated between been mentioned in Section I. While this research a group whose criterion scores were predicted has led to many suggestive results, a more fruitful well by a multiple regression equation. and a approach seems to be the use of multiple modera- group whose scores were predicted poorly. The tors (Rock. Barone. and Linn. 1967). From pre- later scale added slightly to the multiple R. One liminary analyses. the researcher may hypothe- 76

69 Size Several variables that he thinks will have a tions of biographit al and self-report information moderating effect on the relationship between to the area. For pure selection purposes, it would predictors and a tritelion. The data on multiple he highly useful to know which applicants were moderators form an n-dimensional spat c. Using predictable, and which were not. The decision- a procedure developed by Ward (1963) the method maker t e,uld bele( t more predictable persons in clusters the cases into groups based on their simi- groups and simply eliminate the others. if there larity to one another on the dimensions. Ward's were a large applicant pool. Ile could also apply technique begins with many groups, with only a different selection ratios to people of different few cases in eat h. and successively c ollapses the predictability, SI/ that the probability of success groups into fewer groups with more cases in each, would remain the same, simply raising the cutting and each group successively combined with the stores as predictabilitydeclines. Another ap- ones most similar to it until the researcher none es proach would be a sequential strategy, accepting a large increment in tae within-group sum of those w ho were predictable and who had high squares. The multiple regression predicting the predictor scores, and asking for more information criteria are calculated and compared in order to from further assessments for those who were less see if, in fat t. the groupings based on the modera- predictable. Another potential advantage (more in tors led to different levels of prediction. As de- principle than in practice) is the possibility of .4c ribed by Flaugher and Rock (1969) the success finding predit tors that apply more to one group of the grouping t an be evaluated according to the than to another. Understanding the differences in degree it serves at least three purposes. whether the pattern of predictors could, potentially, be of it increases us erall !medic tability across groups; great substantive interest for example. if men whether it identifies groups that are most differ- use abilities different from those of women in entin predictability. and whetheritidentifies showing successful performance in executive po- groups that are most different in predicted and sitions. If moderators can isolate groups with dif- obtained scores on the t riterion (the method has ferent predictor patterns, a greater variety of pre- been used to study over- and under-prediction). di( tors may be considered in initial assessment The computer program that t arries out the t alt u- batteries. That is. a predictor may not be highly lations for these methods t an be altered to reach useful across an entire sample, but (mild still be solutions that will meet eat h of the three functions useful for a particular subgroup identified by mod- just listed. Flaugher and Rock (1969) have used erator analysis. variants of this technique to study under- and Again, the moderator analyses may make the over-at hievement, Klein, Rut k, and Evans (1968) basis for decision more apparent by identifying to study at ademit at hievement, and Roc k, Baird. variables thatinfluent e predictability. For ex- and Linn (1972) to study the relation betweentul- amples, if a scale of psychopathology moderated leget harm teristii s and mean at hievement on the predit tion of success in sales, one might wish to Graduate Record Examination Area Tests. The re- reconsider his selection policies. The same kind sults of these various studies are t omplex, but of result might make one reconsider his criterion suggest that the method c an profitably be applied measure. Another stimulus for reconsidering a to a wide variety of problems. criterion would be when it was found to be highly This brief dist ussion ofno approat hes predit ted for a small NUL et of people, but poorly does not do juitie.,s to the area, but two issues predicted 14 the majority of cases. should still be dist missed. the prat tit al benefits of Moderator techniques can also be applied use- defining moderators and the panic ular t ontribu- fully to a variety of substantiv. problems. stn h as

I 70 under. and overachievenient (Flatighet and Rock. find that high social class was associated with 1969) or under- and overpredi( tion tllobart and over-prediction. but that middle, rather than low. Runnette, 19671. In analyses of tins ty pe. itis im- social class was associated with under-prediction. portantto remember the cautions of Zedeek (1971) and NIeNcinar (1969). That is. a moderator Summary may simply be correlated with differences on the This section has been devoted to a brief review of criterion measure that stem from correct and some techniques that can be used with brief as- over- or under-prediction. The analyses should sessment data. While a number of cautions were control for that possibility. One of the most in- suggested for these techniques, some seem to be triguing kinds of result is when the variables that quite applicable to biogrE phical and self-report indicate under prediction or achievement are dif- data, and several would seem to take particular ferent from those that indicate over-prediction or advantage of the potential of biographical and achievement, that is, when the moderators do not self.report information. operate in a linear manner. Another area of potential substantive value lies in the attempts to study variability per se, considering it as a psychological construct. The value of biographical and self-report information, again, lies in its potential for broadband assessment. That is, it is probably more likely that a moderator that is uncorrelated with the criterion, but which influences the predictability of the criterion, will be found in a variety of biographical and self-report data. rather than in a test battery. In addition, biographical and self-report variables are generally easily interpretable. so that one may reach a clearer understanding of the reasons for their moderating effect.Biographical variables also often have social value, in the same way a :t widely agreed-upon criterionhasas social value, so one may evaluate the entire selection procedure in terms of who is selected and who is not with better understanding of why the result occurs. Biographicalandself-reportinformationalso seems to provide unusual opportunities for learning inure about criteria. For example, in the section on grades, the negative personal character isties associated with grade-getting skill was described. A final advantage of biographical and self-report variables is that they can often have nonlinear relations as moderators, thus being especially useful in the under. and over-prediction design. For example, hypothetically one might 78

71 Section VII: Conclusions: accurately measures what it is supposed to meas- The general functions ure. The dilemma occurs because, in general, of brief-assessment measures any increase in bandwidth reduces fidelity; greater fidelity reduces bandwidth. The dilemma has led test constructors to favor the approach described The values of brief-assessment information in- by Cronbach (1970): , hide the breadth of its coverage. its utility in a "The classical psychometric ideal is the instru- variety of det ision-making situations, its potential ment with high fidelity and low bandwidth. A col- use as a learning experience, its ability to screen lege aptitude test tries to answer just one question large numbers Of people inexpensis el} and quit kly, with great accuracy. It concentrates its content in and the ease with which it is obtained. a narrow range using homogeneous items to increase reliability. Its parts are highly correlated, Broadband assessment hence part scores give little information for choos- Perhaps the most important t ontribution of brief- ing between majors or diagnosing weaknesses. assessment information is that it provides broad- (p. 181) band assessment of many variables. as demon- In contrast, the approach used in most brief- strated in this review. Cionbach and Gleser (1957) assessment instruments isto cover many vari- defined "broadband- assessment as assessment ables, that is, to seek wide bandwidth. (The ex- that attempts to give a rough answer to several amination of the question of fidelity in the earlier questions rather than a pre ise answer to one or sections of this review showed that there is con- two questions. The greater breadth of broadband siderable evidence that brief item information measures, how ever, has other Costs. That is, the t an have verycry good fidelity and also be part of an user of broadband assessment must a onfront the instrument with wide bandwidth.) "bandwidth-fidelity derived from Shan- The broad band of information provided by non's (1919) information theory. Bandwidth refers brief-assessment may be particularly useful in to the amount of complexity of information that a several ways. OffitIlltlit atm- tries to t 'ive) in a given space or First, biographical questionnaires can easily in- time. The analogue for this in testing is the num- t hide a variety of information directed toward a ber of independent se or s one tries to obtain from number of decisions of varying importance and a given test or testing period. "Fidelity" refers to complexity. If only a simple decision isto be the tut tam y and pre( ision of a communication. made, there is no need for such broadband assess- its analogue in testing is the extent to which a test ment. For example, colleges interested only in the

7 J72 academic performam r of admitted students may vide information about areas fur w hi( I1 no present not be coin ernd with theit values. personal Ill high-fidelity instrument exists. but in which it is social adjustment, or probable careers after col- ne( essary to make decisions. Even if the informa- lege. In contrast. many colleges ate cunt el tied tion provided by these instruments in these areas with multiple out( ol»es and must make decisions is or relatively poor quality,it has sonic value in many meas. They need validated broadband in vs hen thereis no better basis for choice. Of formation fur such purposes as seC t toning, &min ( (ose. the choices must be evaluated by later fol- soling, choosing those who need remedial work. lownp and feedback. selecting anumg financial aid candidates. and for help in educational diagnosis. In addition, many Selection and placement educational decisions ate multifaceted and r The greatest contribution brief-assessment meas- quire information on many aspects of the student. ures make to selection and placement is based and this can be provided by brief assessment in- on their( apacity to provide broadband assess- struments more economically than by tests. Of ment, as just described. This broadband assess- course. each decision needstobe validated ment can provide information about the multiple against subsequent outcomes. characteristics needed for success in complex Se( ond.briefassessment instruments( dlitter toles. By providing additional information. bio- bload band 411114o illation and ( an pout/ ide rough- graphical and self-report questionnaires can make to-good data for decisions that can be confirmed the original selection-rejection decision a better or ( hanged at a lard time, thereby in( rasing the del ision. (The contribution this information makes efliciem y of the assc,smnt pro( rdure. For ex- to the prediction of college grades was reviewed ample. a student vs ho checks that he would like uu Section I.) The contribution of broadband ques- help with his leading skills on a questionnaire tionnaire information may be especially valuable should not be sent automatically to a remedial where the performance or decision is evaluated reading class. but should be telerred to the ( ()tin- by multiple criteria. or course, when multiple cri- seling centet for testing of his reading level. Simi- teria are uncorrelated, or negatively correlated. it larly, a student who checks that he would like to is difficult to develop a single selection strategy he considered for advanced placement in mathe- that will maxit»ize a total payoff function. as il- matics classes Wright be asked to take an exami- lustrated by the study by Baird and Richards nation prepared by the mathematics department. (19681. In these examples, itis clear that the question. Another advantage of brief-assessment proce- name procedure is more efficient and less costly dures for selection purposes is their flexibility and than giving all tests needed lily all decisions to diversity. They can be "tailored'' for various deci- ever) entering student. Furthermore, the btoad sion situations more easily than a an test batteries. hamd m10'1)1,164,1( provided by the questionnaire The Ilse of a "tailored questionnaire must be con- served as the first stage in a two -stage decision firmed by further evidence. of course. process. In the Se( 011d stage, the Initial possible Biographic al and self-report questionnaires may decision leads to a final de( ision based on high- be partic ularly useful for the decisions faced by fidelity measures. This role in sequential deci- two groups of ( ()lieges at opposite ends of the se- sions may be among the most useful of the fun( Iec dimension. Highly selective colleges and tion,orbrief-assessment instruments. universities face the problem of choosing from a Finally. the broadband information provided by large pool of equally well-qualified applicants. Bio- briefassessment nisi:mm1as ( an sometimes pro- graphical and self-report questionnaires can pro- 8 0

73 vide infornwtion about these St11(10111S'110ilata dmic abilities. pa,t surecs,es, and educational Sering as a teaching and experiences. This iniorniation can be used as one guidance instrument for students completing of the bases for selection. Opendoor colleges, in the biographical questionnaire «mtrast, fa«. the !emblem of dealing with large 13iogiaphi(al and ....elf-report questionnaires t an, number, of students w lio are at many levels of potentially, serve as teat ping and guidan«. de- a( admit preparation. For these t idleg,vb, brif- vi«.s. By asking students to work through them assessment instruments ( an be useful diagnostit cons( ientiously students tan be trained to 110th e devit es, they may (...d.t) be one of ow rev, ways to the same kinds of ( onsistent it's anduufii( is that meet the di& ult adininistrativ e problems of as- a «aniselor would find important. This use of sessing large numbers of students. For both kinds questionnaires presupposes some bat kup mate- of ( olleges, differential or modeiated piedi( non rials that help the student or client to ask appro- may he useful, as suggested in Sections I and VI. priate questions about his own responses. Halm (1969) has developed materials of this type for use Prori(ling information for with adults who wish to evaluate their own career counseling will guidance objectives. These devices guide the adult to a Brief assessments ( an provide ( eninseling ()flit es critical examination of his own plans and t hard( w it ;nth( atoms of students' problems, and tan be istics. Holland (1971) has developed a system used to dist 41%er thus( students who mustneed as- that helps students organize their occupational ststant t'. cr. instruments of this type taut daydreams, activities,( onipetencies, interests, be used Lyt ounselors to se( life kit kgruund in- and self-estimates into a pattern, w hi( It can then formation before meeting with a student and t an 1w compared with patterns of people in various help in (housing any additional tests the student groups of occupations. Computerized systems might ',nuke( tiv el)take. "Tyler (1961) suggests such as Katz's System fur Interactive Guidance several ways that "( ase ret (an he used by and Information are based on the same ideas. ounselors as aids espet iallyin redo( ing the These systems illustrate some of the possible amount of interview time that must be devoted to ways iriograplii( al and self-report questions can collecting factual information. Im used to help students examine their choices By providing simultaneous infoi illation about and ((insider new possibilities to explore. students' personal. so( tad, and a( adeinit back- set ind way brief-assessment instruments. ( an grounds, edut ational and vo( atiunal plans, titian- be used as teat fling and self-guidance devil es is ( taI needs, past a( e omplisliments, attitudes to- by using items or groups of items that will inform ward their abilities, and so on, biograplii( al and students about options and requirements. For ex- (..t.11-irport instruments t an be lit h sour( es of hy- ample, questions about w hat the student wants potheses or guidant e ()flit ers work- from his college ( (mid help the student re( (insider ing AA reh student. Tile) enable t his reasons for ( housing t ()lieges, dad tuuld help spot tfitt ollSistent it's or Int ollsiStent ICS him( Insider inure appropriate( ()lieges. Simi- andt until( Isun students' bat kgrotlinis. larly. a t het klist of at adeinic skills that includes and1)1,111S. items as, for example, "1 can underline a textbook so thatI can review the most important points late.- could help students locate their academic weaknesses, and suggestspecific-skillsthey should develop.

8I.

74 Summaries of such information can also help in Identifying students with planning extracurricularactivitiesaswellas particular characteristics orneeds health services (Lutz. 19681. The College Board's The previous uses suggest how brief - assessment Summary Reporting System and the American information could also be used to identify stu- College Testing Plogram's Student Profile Service dents with spec ial ( haracteristit s or needs. Thus, already provide much of this information. Break- students from minority groups w ho wish to avail dow IIS of summary data for spec ial groups of stu- themselves or programs designed especially fur dents c an help c ()lieges prepare and plan for the minority groups could indicate theii interest by needs of such groups as commuting students, men the relatively innocuous process of filling out an and w omen, older students. minority students and information blank. Simdarly. students w ho wished working students (Hoyt and Monday, 1968). By assistance in numerous areas could alert their col- summariLing the characteristics and needs of leges to their needs. Students who simply desire these students. colleges can also avoid mistakes information about part ic ular college programs based on lack of information in working with could also use the questionnaire as a way of com- these groups. municating their interests to several colleges. In addition. colleges receiving a description of the Providing the basis of two-way systems educational needs of their incoming classes could of information and action prepare for them by, for example, creating a Byprovidinginformationaboutstudentsto course in reading skills if a soffit ient number of schools, colleges, and agencies, biographical and students indicated tin) needed alit-. Che( klists self-report information can form the basis for two for students w ho wish assistance on information way systems of information. The first direction of from their college are part of both the College flow of information is from students to institutions Board's Student Des( mitive Questionnaire and and agencies. Properly utilized, this information the American College Testing Prop am's Student could help colleges locate students with charac- Profile Section. teristics the college values, as the College Board's Student Search Service does, and the colleges Helping colleges in their planning could send in forn:ation about themselves to the and administrative activities students so identified. Of course. students have to A systematic ( (illection of biographic al and plan- agree to allow their names to be sent to interested ning information c an be of great assistance in the colleges. Similar procedures could be used by day-to-day work of c colleges. These data ( an Form programs and agencies interested in the recruit- the basis of a student record system by providing ment of talent, such as national and state scholar- common information for all students on a wide va- ship agent its. One example of this is the com- riety of variables. By assessing the extent of stu- bined PsAT-National Merit examinations which dents' needs for financ ial aid, the distribution of have been taken by millions of students. These probable majors, and so on, summaries of ques- groups could then turn the flow of information tionnaire data( an help in budgetary planning. back to the student, providing him with useful ma- Summaries of' students' needs and hope's about terials that ( ould inc rease his options and widen college ( an help t «lieges plan their orientation at Ins understanding of his choices. A central na- tivitics. Knowledge of students' plans for housing. tional agent y could also be the source of general commuting, and so forth( an help ( ()lieges pre- information that could aid the student and his pare Cut those aspec is (cf. student personnel work. counselors. As an information clearinghouse, the 82

I national agent vt ould pr44v ice 4 areet and o« upa- ineparation, work experim e. geogiaphit (night. Clonal infot illation about soec di( fields designatt d edit atto11,11bat kgi ound. and mina( alit:Atiltat by the student. or44(1141 guide hits to sour( es of omplishments..1national program. gathering information.If the student inda ate('thatla data about studentbaracteristi( s. and obtaining wanted help in (lending about a vin ation 40 a- data about the distrdoition of students to varunls rcer. he ((dd also be guided to appropriate ma- progiams. so hoots, )1411.t. and fields would Ill 40,1de terials. Similar sery if es t ould be available to stu- the basis for des( tiptively answering questions dents who arehousing a college. Information similar to these. "How do students finam e their about designated ( alleges. guidain e to appropri- will( ations?-What are the effects of scholar- ate materials. information about how tohoo.te a ships and loans'?-What are the obstacles- edu college. and so on. In addition. w ttli enough infor- ( ational, physical, and social- to further dufa- mation about theharacteristit s of students and don?" "What kinds of students enter various their piclerences fur( hard( teristics of 4 ()liege:, fields?- -Why r ''What kinds of students enter they might attend. a(Aural agent y could operate various kinds of ollegesr ''Why ?" 'What kinds a ()liege lot ator servi( e. .111 ideal set vo e would of students may the small college hope to ob- proyide names and desc t ipuons of()lieges that fit tain?" -What kinds Ala) the large college hope the ( hard( teristit s that students Indic ate are im- to obtain ?" ''Hovs many minority students enter portant to them. and that fit theharm_ teristit s of various programs, colleges, and jobs?" -Where the student. For example. information about Yale and N)11) do they enter?- -What are the problems would be of little value to most students with SAT fat ed by the disadvantaged and the poor?- -What scores of 100. Theentral agent( ould also be a is the plight of the middle classes' ?" "What hap- dearitighousefor informationdire( ted t4Mdrd pens to the undermotivated. and the 'overmoti- specific echo ational needs, and ( uuld send mate- vated.?- few4/1* these descriptive ques- rials to the student or guide him to materials un tions (44(11d he answered by test data alone. The how to stud) in4 °liege. how to improve reading answers will probably change as society and will- skills. guides to reviewing :lie( Ili( sulk)et Is, bat I.- ( ation ( hange.,It is. therefore. important to moni- ground reading for college. and living away from tor the Changes continuously. home. Biographical and selfreport information may be particularly valuable in examining the flow of ,llonitoring and understanding talent because it can broaden our conceptions of the transition frorn high school to talent and talent loss. Although the terms "talent" schools, colleges, and jobs and "talent loss" ran be variously defined, edu. NIonitoring the flow of students from high st hool ( ators and social st intistsustomarily have de- to ,1(144(js :4( /Wok. ( ()lieges. and jobs is aa input- fined "talent" in terms a a single dimension - Lint task bit 111,111) reason.. It provides data that a( adeinit aptitude-usually masared by school are essential to students, ages( les, and institu- grades or at adentic aptitude test scores. Those tions I OM erlIell With till' flow 441 t,1leul and the ob- students s( oring above aertain level are -tal- slat les to that flow. Studies of manpower allot a- ented,- while the others. by implit ation. are "un- non, rr( ruitment of talent. st holarship pull( talented." ,tny "talent loss- is the percentage of and polit ies for minority poops all depend 4411 in- -talented- students who failto attend college. formation 4/ithis ty pr. Biographic al information This kind of definition appears eminently prat ti- c an c (attribute to this monitoling supplying t al at first glam e. but the prat tical advantages of data about .444 141(1 4411411114ha( kground, al adt'llItt definitions based on a( adtiM measures do not

8 3 76 justifythene orientpopularity.1s we have partit Air. this means looking for eidein e that shown. at admit measures ale not effn ient fore- eat Il st ore added to the profile (and eat h differ- casters of a great range of talented performant e. em e bet wren st ore:, in the profile) is reasonably Before we can as.t.e...ss the degree of "talent loss- generalizableand then going on to ask for evi- in even an approximate way. we need to know dence that sut Ilrefilled information relates to many things. sonic of which t an be prof ided non-test variables." This review has indicated biographical information. First. we need to de- that dirt t information from questionnaire items scribe the sot iall)relevant lath oinc:,that we is generalizable. useful for spec ifit decisions. and hope "talent" will attain. NNe then 'Hard to know related to non-test and non-questionnaire vari- what kinds of human abilities are essential for the ables. The potential of information of this type attainment of these outcomes. as well as the ap- Ina) not yet have been tapped. Biographical and propriate environmental and sot idl t nnditiuns. n- self-report questionnaire rcasearch has notre- nally, we need to know whit Il kinds of tiaining ceived anything like the intellet foal and financial programs best develop and train people with the support given to traditional tests. And, as this re- required abilities so that they will attain the out- view has suggested, more systematic work needs comes we value. Clearly, we must know much to be done to examine and improve the utility and more and make 111<111) value judgment_ before we validity of biographit al and self-report informa- t an speak act orate!)of "talent loss.- Better tion. It seems that such work would be of value to judgments may be made when biographit al and thet rectors and users of test information. Fi- self-report information is available. nally. in considering the importance of biographical information for a complete description of all Conclusion individual, it is instructive to note, as Dailey (1960) Thissection has suggested some of the potential has, When a noted psyt hologist dies, the com- uses of brief-assessment information. Biographi memorative article describes his life and work cal and self-report information seems to have there is no c ase on record in whit h a journal pub- great possibilities, espec sally when part of a sys- fishtail the test scores of a de«aased scientist as an tematic national information system. The previ- adequate des( ription of him as a person... ous sections examined the evidence for the validity and utility of brieassessment variables for a number of purposes. the technical and statistical treatment of biographical and selfreport vari ables, the psychometric and other requirements of useful biographical and selfrport information systems. and the issues involved in the use of in formation of this type in various kinds of deci. sinus. Cronbach (1970) suggests that any attempt to add to traditional measures elicits a healthy con servative response. The conservative response led Cronbach to hope "... for benefits from each in novative proposal. since that is the only way prog ress in testing can conic, while insisting that each proposal to increase bandwidth prove its merit. In 84 77 Reference list Slizlients with diirlent vos Jtion.31 elhoit 4(.7' Research Report No. 1. 1965b. \ lickon. It P. Si N.diffrein ,n prethetahilit) of college, InIro(luctiolt giddy. 1-://ice ational and !'sir hologu a! ,tleasuretnent. 1952 Ca lahan. I'. of re.pondent tirtiee in the I) yr 12.638.611. 1.1110 e ciudi. Piddle ()pinion Qtsarterh, 1968. 32.607 6.:1. \ merit an (.tile ge Testing Program. lasescum stiolente on the ( latro.n. \. Audit), lore report, Puhlo ()pinion it a) to college\ mina. Iowa City. 10% a 1973 Qmirterh. 1968. 32. 388.606 \ mem an Council on Elie atom. National norm, for mitering F reeberg. I'm-tom! 111111111114 4111011. 1971, t ollege fre-bournI- -toil aninialk um the 4(.F Research Re 1 on Edo Hardin. F.., and I ler lieeI, I t mar) 01 employee repent - ports -env-. 11a-hington. I). (:,.\ mem all CO lllll nn in ;me Journo/ of ipp/iet/ Pc)cho/og). 1960. 11, cation. 1966.1975. 269.273. \ rin.trong. It. and Jen,n. J. Report 1111 1I.IN,J1 Mon,. State Ilarre.. IIJ.I he Cornell Selectee hide. - it-11,1'111 College mud!, mudent reported grades Draft Re rie. elitegmeasEnna. . Veit Aadetn) al St sent r, 1916. port. 1974. 16.591.603. \ min. A. Ir'ho goes where to college% Chicago: Science Be- Maxey. E. Jand ()mud)). 1.IThe elecurm i01 -elf.report cearell Associates, 1965.

information t tillectril 011 the Nt 1 lem Batter). High Biel \ still. A. W., and Pang, 13. J The educational and vocational grades szitl item, of mimic ,i(10111( einent.I(.7 Re. development of college students Washington. D. ( '..Aineri cear'h Report Va. 1.3. I4.v. a Ciiv, I et% a: 1 uth. or. 1971 rail Council on F:dueation, 1969 Nhe.el. J. N.. and (au/an. I.. A I he a« ti rat N of apple( a:011 Baird. l.. L. The predie Min of accomplishment in college.111 Wank work 111.4 Journal of A prized holog). 1931. exploration of the process of achievement. foam& Coun 36. ;163.369 \cling l's)rholog), 1969a, /6 246.254. V, Aker,. h. I'. F.xamining personal meridian item. of Baird. I.I.. Predicting grades in 0« upational and at admen( lion limn. :allele. Journal of haat at tonal Rec,ar.h, 1937, 31, tarn ula in t 'NO Near s. Journal of Educational Alias 281.282. arensent, 1969b 6. 247.254. Aalsla, R. 13\ alieliitd Ireport Journal of C calm; Baird, 1..I.. A review of studies 4,1 selreported grades Ps), halog).1967. /1, 18-23. published manuscript,

ah.h. A 13 1alithie of 44411 report notht r look Journal 01 Belielig. \. V Comparative validity of ernpirn al temperament Counseling I'm holog, 1 168. 13. 180.186 test kers in predit wig student achieNeinent in psychology Walsh. A. B. Selfrport ler mu tinde suable and di-. Journal ("Educational Research, 1958..51, 34-38. tomtit 4 fon rrz/ of Lotinseling Pc )c holo/n, 1969. Bi neut.\.'3'and Sprague. J.I.. The Guilford-Zimmerman !6.367 171, Temperament Survey a. a predtetor of achievement level and achievement fluctuation in introductory psychology. SCC/i0/3 Journal of Applied Psicholog), 1954.38, 409.413. Aloe, I. and Rolland. J. I .I 'Anion 4414 "Ile gt In 'lout n. !krill( .11 F. Inuit Huh. Anal variability and predit t htl

1Nindt tn.: .1111 illert 111 4111411 111.901 lit IdI( 1 R4 ,f art h ',cantina' and l'stchological Meacurenient, 1961, 2!. 663 Boron No, 3, 1963a 676

.3, . 11011.1101, J.I. st 111.11111 01 ill, Ir. Ann, 11. Meg., I)\ .11.411..1I)and Strong. S. II If-mad at admen(

78 pro4l11 t oat. and at .014 nodpi H0:811.1114 t. Sleomo, in, to I or.., Edm ntionol ond hologo al 11m:sorrow/a. 1964. Evaluation to Grodam e. 19:0. 3. 81 8.1 21.321329. Higg'. 04 A4. and To1, 14.5 1). J 11..1-111111.1414 .14 .1411-11114pr4 Furt. E. J. \ alidit% of , obn 4 115 1.114 of mmisaiion 144 1111110W, Journal of Adoentionol R.seon h.197(1. 61, 193 10 1114 1111gat. Jilt .14 Illes v1111.111.Et/tit ottonol and l's$ 197. cholopeal SI ensurer/lent. 1966.26, 927.933. Ilinder. I). .. J... and :41.44)4.ig. R. . Nomsmollot (.4 bliart. (.. (....11111 1144)t. I), P. Pt r.onalit)114 ed, of Ire 6,4..4-11.report sartable, a prodo tor. a .14 ho -st - .11111 11, t r .14 Int %mg fre.lin14 u. Jourool of Ipplo,1 l'sit hul mont. The Journol uJ Edumitionol Re%eurch. 1970. 63. 361. op.1958. 12. 123.128. 366. (4111.41b. E. E. The predict:on of predirtabilit. k.'iliiristforso/ 1144141t. li. F. At. tlt.11-5 01 grade !mint ,tatao . t.11.-. 11 41 ..1111 and hologo nl 11m:surer/sent. 1960.1. the (.4411ege Entrant r Examination 1144.u414.4111.414-.4 riptom G111.4111, E. E. Diflerentiali 4444 of 11,1, in It 1111:4 of the .1% t 11t.l1

blank. N14. and . Prim 114111. \ J.. Edo, at .tl .5111155 1114 II 1111 predit t for a 0. en individual Edistationni Sr Moe. 1973. and l'.ehologo of Measurement. I9601>.20. 675.684. ilrookm or. ,II. ...)(4).4uncept uJ nhrlrt 1 tool st hoot h see e F: E. N loch-rating rife( I. and differential rellabilit) ment I. Looperatm. lir,elltt IIl'roirt 1No. 843. Nlit higan andability. Journal of 1pplied P.,holog). 1963. 17.81 -86 Nate I. [liverit). (Mitt ol Itt..4-art la and Publo alone, Dt (,ough. II. The 4 011,11111 11011 0f a 0.011.114) t.t .11e to medic t comber. 1962. ..4.11441.1),to.aelairvontent. Journal of Applied l'scholoi.... 14044km-cr. W. 11.et,d. Se/fq oneept Of abilit I mid school 1953..17.261.266. achleirment II. COOper,1111 l'tor( 1\o. 1(436. Hanna. G S . llhgh. II. V.. and I rake, J. \I. Student reported Mit lugait Nato 1 imer.ity. Olio o of lir.earch and Public a- grade a. predic 1or.. i'ersonn/ / (:nalorne Journal. t; ( )4.6410.r. 1965. 1910. 8.465.468. lade. N. I/dime:mai s,thdus in the \lI tt. I( 7 Krum', h la 1144.44n. G. Ii.. and 1..im11.11.11. Les el of rollego 40.14.4-11) it y and Report No. 30. lo% (at\ . 144)(..1- rho Ndin.1 ic an 1.4411111 the act mat yof (4.4.1f-report ml high ,( 1144441 grade.. Vorth- ing Program. 1969. ::::: g re.eareli report eitt41 in American (.ollrge Davoen, 0 \I. ;mint.% 44I IA1,1141111\ 441 stiolont 4.1f.reported 1.1.41gf.1111, 1 5 51.'01 ng student% un the 11411 to College, Vol I

high ,c104441 grade.. NInnoo. 144% .144%,t. The tilor14.111 10%.1 Cit ION.!. Author. 1973. ( :.urge '11.4.ting Program. 1963, 11111,111. T. and NI)cr44.. A, I'Growth Study II. Per.onal Donham. C. 1111 predimion of collegt. %MI Ino 1).14 kground experiem r. and ,4-1144441 achievement: in- graplural data and 44-11rating...l npub11 .114.41 D( toad so.tigati :::: of the 4 0111rib11lioti of (pm.tionnairo data to ;ma .ortat . l lin er.tts of \ rkan.a.. 1966. deismprediction. ReAcurch Donau: Rll 66.5. Princeton. E. J.. and It on/aglia. G \.\(-(11r.14 5 01.111411.111 Ifft1111 N.) .K41111 .111011.11 Totaing Servi( e. February. 1966 Unit of follegt. gradv.. Person/04 ond Gaol e Joarnol. 114411.iiid. J. I,. T114. redo tom ofd ollege grade. from the Cali 1963. I 328.530. Imola INoliologaeal Inventor) and the Seliola.tie Aptitude Dunnette. \II). wirary 01 -44'1414qm; report 4.41 honor point Te.t. Journalof Eilmdtemled 1939. 50. 133 I 12 ,% sage.. Journal of Applied 11% 1 r holog. 1932. 16. 20.22 Holland. JI..Croat:\ 4. and .1t adoinit performance .1111011g rodorikm.n. \.. and Gilbert. k. (.. F. It eplo anon 01 .14141 of 1.111111ra .1tIolt 54. rill,. Journal of Educational differential medic-1.'1411u S. Edm otzunol find 0sseholmeleal 191,1..57. 136.117. tlemurenzent. 1960. 211. 739 767. Holland. J. I.. Eploration. of a thoory 441 voc ational choice: \ I. F roderikon. . and Nick Mr. I)Differential prodit \ longitudinal .W41) u.ung .1 .implc of ty ant al t ,1!!111 111 in the 0.1. nI ifst 1 ore.. Edit 01/4/111// Mid !'Si, hulogo al 411.111., Juonul of Applied l',3\rhola,to ,Slonogroph%, 1968. tlnorrement. 1934. / /, 617 636 \ 0, 1 F rerborg. N. E 1114Inograplin al inlorination blank .4prt - I 14,11.11111. J I. . andqui. A. W. The prodo t111 of a( ademb do tor 441 (atident .141111 s. 1111 111.\ 1.t sic 55Pt s/ h0/44g84441 lie arti.114 ..c It illiflt and 111 1.11 311111 st Hit-Ill Of 111141ergrAtiat.

)torts, 1967, 30.911 921. 441 ,uperior 44.411441,14d11aptitude. Journal of Educational Frrelidl. \I, F. Student 44.olf.euntiato, Wilda:tee fit .elec 11.411 l'%holop. 1962..i3. 1.12.143 Of I ollr.e. (.44//rge ond 1 nit er.slt5. 1932. :7. 233.212. Holland. J. I... and N1( hob, Ii(. Prediction of academic and .rut 11. J. V,I.nnyrar,Wsr protIll 11.111 of t "114 go ma)oreld 1 Xltat Wilt (liar .11 1111 t'1114111illI lollge. Joarmil of 1...dor is grade. lo 111111.11 lot.11111111dr, 11111114411:111 ;114,011.1bl% :maul l'%ehology. 1964. 55.3.1-65.

111e.Ittlifet,Edit/ otio/1/// tint/ Pl5tho logo of11mi:oat-meta. Holland. J.I,and Ito bards. JNI.. Jr. Al adonin ;1144111011 l')63, 23.7(>7.771 .11%111111110 .111.01111411411111111:Corrolated or micorrelated? Frrmih.3, V .I, parati.t. predo I 441 high...4114ml grade, Journal of Elite noon:! l'srholop. 1965. 56. 163 171 by puredartor aptitude. information and per.onalitlora.. Holm-4. I) S . and Tvlor. J. D. fbre4 t ser.u. prmoctive 11:11.1% 86 79 Int 111 .1 Ito %edit in 111.1 I% .1 11.11 ./0////11d or/ (..0102//1/1)( :-141'11 all195,11'101g o LP\ lu1ub11.hsd I(.11nnu!!'svhul.. gs. 1908. /2, 712 717. mailm niot. loss a (.05. loss a. l'he 1merit.1111 "liege Teting 114.1-.4.I'.1 techimitse lot the deselopmedt ohllerential Program. 1972. pro oho (loll haw is.1'18, hoo/oe:to oil1/ /...ofith1951. 48 \0119 s. I., J.. and Orm14%. 1. I. Flo ao > ol elf re jooI t III \n. 9, %holt. number .180 formation led on the \tI 'fest 11.8tter. High Imol 11.1.1. P. 1 ithumid lor 1111- t Lynn III ol a mishap'. Ai.. pad.. and itt-ali 11.11 41, .8.15 moo ao 55 1111 ad. 7 /it lot, preoli. 11.1111.111. rs,l'Slo /1010/.1.10 (1/1/ ,/(1,////s,191;. oo no Is pall \a. t.;. 104..4(.4ts. lossa.'sada. 1971. 10, \ 0, 1s11010 11' 390 \I la% I 1. Predicting grade. irmg \tI data. Ed/I/Inn ,,,,,, llost I).l'.flirt 1. sear. .1 rt- aro II In Ilio\ merit an I .11. mid PA I. ha/PA:U.11 lietIvuretnem, 1967. 27.101 l06. Is 'Wig I'.11't 3rt I'%, Itologo \ oo \ Is. ( lloograplait 1.4o tor, and at ado mai ai hat st In( 111 \I 1963. \ 11 1.N perlinental 1110.1 Ig.1111111. Erlaaa I and l'%)ehalag.

!lost. D. P.. and \I 14%. I..\)4411, .0!14.)[e eAlt,1141144, 41 Ion! %least, retnent. 1912. /2.411 126 (AI)loossa, Mr 11110 xi. au ( .dl' go 'R -.tang Program, 1908. \ It.( mid :.so holt/. I). Pa collo ling .11 4111111111.11 lin s 11

Jo, oh-. J. \ \ 144184dt mid al hit N., ant Hi1111 .1.11 1'1 IIIpro 111111t N1111.1 0tH\ mishit Interet Q,111111ionnaari Ediso 41st tang high 110.1 .014 t 1. rsunnr1 41,4,1 Gil 14111 III t. Joni rolonli/ Ono/ Pst( hotIngoo mil Sled: %III 4.111041. 1950. 10. (1.741-003. mil. 19:49. 32 131 \ it ho. Ii.(. \ lomintelleetivo. predictors of :whirs 111

_how., J. mid (.lit IL-, I,. \ It .1,111 *o1 I. 11 5 01011 41, sollm go . i.tha an 1 ma l'shalaga ,Ileauire mem. 1966. preolietor .(134,14t114 .11.13J4,4Is11tit. The JOU/ Mil 011.:11a a 20, 899 913. maga! firwaa I,. 1970. 01. 201-20.1. Nit hook. 11.( and !holland. J.I. Predio non oof the lirt )ear

1.11 I. r. is. F.so pr.1, 1 1* .01 .1111 ,. 110 s5 111111 11% 01 .ollego perlormano t of high .1141144de tisolood. 14 g. 11014 III-Jo////10101 E//11«Thomal h. 1969. 01. oil 11 graphs. 1903, 77. \,. 7. is look 11111111,111" -.1 0141111111111111. I) Summary of tmlie using the \ loon% Mims for

Fork. It1.. alio! It 11,1.I \ urao s ol 4. If ro porit 41 1.115 go l.rado .114, Program Report., P tom N. .1.. Edut grade ,o.o.rago. .111111 11,11%11 1141.15( ol non411111 de., repaid tional Servs( e. 1970. poll( r.,ho/10 /1 11 ,,,,,, 0/1/1 Ps 1/ hood.);. oil 11, at. Po irs. I. I). .of high hood 4155 rap soinpisio ol 196'1.19. 117 Ir ,,,,, tudent. equitates of their high ellood grades. So' haul S I'. I(.u.k. I)1 .401,1 Esau,. F. R. r1111 11e of multiple 1 Surust ). 19,10. 52. 03 01. iii.olo rotor. 111 .11 .1411 1111,In t 10.11 foo/1//0/1 0/ kJ/0 111001/1/ 1'414 a44,1. I(E. 74., host to/ ,,,,,,, 1tollege Student Quenon,' Ilrnsrrmr,lt. 19(41.5.1'51 160 inure% Prm«.14111. F:1111( .111.0,11 It.ting Servo( t. 1965. ( . \ 11 L011111.11 .111 % aliditool Ili, Rio hank J. \I . Jr.. Ilalaud. JI.. and lad/. S Pro olio non tong \alional lut< rI Blank\, ado 11111 ko 111( s, 1111 111 .ol studs nia, I 0811111,1iiiieilt m ( 0111 g5. Journal Of EdlII S1 ale and this >mob sit (,)11e11.1111.11ro. \ Iota% atom for noun! l'ischnIngt .1967, 58, 343 .151. I.raolt:so all6du. al 1 and Psis h./Flo a/ I/ens/Jr/ men:. Rd hard. J.\I.. Ji.41111 l.ulr. S. 11. l'rs di< 1111g -41011111 al 1969. 2'). 189.19 3 compliIntient .liege Irons the \I 1ment Journal

ladisd-loorg. 1 F.. \re.o. aro Isr8,% oolilit11,iliangton Pio .11 Kiln, totonnal %len int e went. 1908, 17.29.

ollt golo ling Program Journos/ 441 1,1114 ,11144/3411114444 stir. otolot a .1 Pro oho 11011 1.1 ,.115 goII( doorman( of .4a110 nor tnent. 1906, I. 117 166 tsioleast VI1S(. Reset:11h Report 1. 5, 1905.

laisiad !nog. 1 V. apho %al ialoloan ohlft r, II, 11., D. 1 Immo% mg Ilitpro do Isl.,: of ao ads mit .o 1148 loco pa t old lion Jou/ to/I id ataaail In' at. 1908. 15 111115 1.1111,111.1 1110,1( f.1101"&, l moublilit d dm !oral Ili, 207 210, 11 rim . Pstrolut l in% crot). 1965.

1 . 4 1 1109 Inorg., I K. ,1101 1 11111n boo g, I' 11 Tlit psi olio loon ol dal 5.1111 de 1.,I I. \ loodo.r.itorarialolc.s ui meth( non. Edo( es lo so rat* u,, 1.1Liss 11001 8, rim main lad. ntson(11 tuna! anti l's )5 la nIn);u mil Ileamtrernent1')50, /6. 209 222 I'll, /III/Ill:I, nl th.lisu rem, tit, 1966. 20, 91:0 1 ...-alabitz.I)(... and I.reen. II.F.. Jr. Neill( ling 411.111(.011(

1,ffiltio 1144rg, 1 F.. .4s1.1 Lunen o loorg. I'\ \. Ir, Imotallt 11.0.1 .11 111*.1 1111 111 551111 .1 In ss .111111011 11111 rt I ttswiloomiaire

jos flotilla/11 1 j11-1 4111 It 4111,011 11111 111 5 115 .110111.11111 k.oho nunnul ,111,1 PA 1 alIlemsurement.1913, 1,

11.1110.11 s. 1110,1,11 I.,. Jaarnal 115111:rd l'whang 19694 ;1,61 ; i, 209-211 Se.111.1-0. III.\ \'1105 n are more predit table than men Jour I, -loorg. I'\\ and 1 0110, borg. ( 1181doff, no 111,.11 pis nu/ ,,/ (.5m115e11 rig !'s>, lining). 1902. 9.261 271).

111111.11 ono gograolt Iran blow .ipluo solos :proull. isII1 litlo 1.414o: lot I, el 11 111g11 1 114.101 If 1 1 I/1

Ed/soots, lllll /mu/ /'sto hoo/oRoos/I/ea 41114 Mtn!. NW,. 20. 11 .1313 11111.1131111)411111 1/111,1/11 111.11 .1411 11111.11 11101.4 nil 111 917 923. 111 the 1 0111111 lllll 1)1'0111.ge. npillshlied doctoral 411erta \1,9915, \ompasioit toadomt repotted and Ilool I \h, higast State l 111% end). 1909.

8 ;80 Stark... I). I), The istilstation ol loop aphmal intormamos an the rance.n 1973. prediction of arade a. 111., mem I sipubliIsed dor aural \ \. \\and P.M.. R. J. The .'.fur it rid 10/011011(1/ di--ertati .... Purdue 1 no es -it% . 1967 tle".1/""ent o/ en! /.r '11"1"1". \ \a.lnn bm. I). Co. \ "(en( St ticker, 1..1, Comiptillss% it Irrator .araabl., \ rplat I an C....azt 11 on 1:,411,/ .11 ion, 1969. t and warn-mos. Journal of Ipplleil IN,: hobo:a. 196(y. \\\ It. J.. and ( J. \ \ ational ,,,,, HO. 50. 331.333. for enterang r "liege Irt,Insten - tall. 1966, Reward: Re. \\ alh, IS.\ alidit )nl ellryport. Journal of Counsel tog port., I to!. \o, 1. \\ alongton. D. \ merit C 11 on l'sat liologa,1967. I I. 18.23. Edus Mimi. 1967 \\ a1,11. \\IS. \ alolst% of ellurport. \ not her I. Joarnal Of Ram. J klidel011. J. G Imof r ontet and peer tutlu I'iahologa 1968, 1 i, 180.186 cnci on edit, al it mai of .1.101e rnl. /(et ton ojEtiffir fl- 1; alIs. ItSell- report under Inallc sustle.srable and dr-. trona/ Rewart h, 1974. //'9,9 146. tort fatale. Jour nal of (.01111Ct fang l'sar hologa. 19(0). Band. I.I.. Fansal% m. .o.. 1 the r harm ten-a ir ..1 t bound qtaltatlIC; N .. Report So. 17, low.: (.11). ebb. (. I he l mums% t. sah.hts .1 tss., loographst al in 10%4. .\ 1113'3 It ,tong Program. 19673. senior) Journal o/ 10101 l'sar hologa1960. 11. 177 Baird. I,. I.. Thv etlus at tonal g...11of sollegt. I I !mulls. 0 I 183. Resent': h Report Vo, 19, lama lova: Amerman College \\ iggiss. \.. Ills. Limns. \1and 11a, k man, IL Prs.,liellost of "11(ting Program, 1967b lam s ear graduate so m pN. Isolog% Pro: ;loth Baird. L. I,Pastern...1 cdso atsonal aoration-I( 7' Research ;Inntral 1.ot:tent:on.1, mem anI \ woo,. Report 12 1,...,1 Cus. Iowa\ niers( Collt-ge '11,ting 1'X18, Program. 1%9a, Baird. 1.. L.iu elret t,.It allege restlens e group on -au. SectionII deal.' eoneeps.,, ;mak,. and at hieventenl, i'rtanne/ and 11011.1:41. J. I \ siphon ..1 r allege leeInnen. and (afahaa efoornal, 1969b. 17. 1013.1021 I. Student. s ith 41111.1 mit rhos. e of major held. / Baird. \\ Iw prolelz,A .lady of Andens al &SA.. In J seoreh Report \r,1 lom a I at% , Iowa \ menca is College Fodor and I), Long. (EdJ. Protest! Stu:lent at ttrism to 11-4ing Program. 196.1 Anterwa. Nev. loll: Alorrow and Co., 1970. \ Irander. IN_ and F 1.1,11111,13. K. Lonte.nial vitro m the Natl. I.I.. Careers and nrrtcalaI report on the at It; flies high chool edsational all, 111 prove... Report \o, 178. awl urn s of graduates a sear after leaf tog eollege. Prince. Center for Seem/ Organ:valor: of ;)ehools. Jnlm. Hop. ton, \ J.. Education,a'he.tlugSersn r, 1971 kin. 1 em eret %, Baltimore. \ 1971. I \l. J.. and Hartnett. I(.I'. The gnu/tunes. \ !that-Is, P G.. and kelleN. D. II. -Itnert: fret student. .1 %e'er tett Prin. on. N J.F:dut ,mina/ Serve. 1973 InhltifrgetiplIN an %Indent al MI an a in! related lOpa.s Baird. I. and Holland. J. I., Th,. flov, of high hoof xing109 Itrok 1973 411.113 10 college., .111t1 )01e. \ re.e.ansination .11 \ Istler.on, \11. Groul. ( otilrslIng. Re: le" ol b.1," at imzul k'' ostse old .pst..tn. hs the 11o. of mills spit rods/ no of 3.11013 march. 1969. 10. 209.226 rather than In a Ingle ar adrunt Journal of llama A \\\ ,Personal and ton sronmental factor- a01 sated Resources, 1969, I. 22 37. ollege dropout.: Jimmy: high aptitude indent. Jour Baird. I,I... and 111. hard..1\I.. Jr The effects of rl(belltIg full o/ Kilo. rztronal I's:. 61.1g., 1961. ; 219-227. t "liege tudent I))C.11-1011. 1.11141 of laugh chood arlsiew \ 111 \. \\11 ho goes Is here to f allege' CM( stun e 11 mem. Iii Research Report ,Vo 2.1. 1o% .1 ( :11). Iowa, \ men. ..'arch 1963a ('olley:1, Tring Program. 1968. A -ism A. \ \, ( °liege welt-sem of s ors .lad. M.,. Col legs 11.1\ er. \ Tht. ollege dropout. Vac tor Mfrs ting CIllor 011egt. andful eraaa, 19631r. 10. 282 297. .8111)1e1O.11. SOC10/0}:1 of Edo: anon. 1968, 41.305.316. \.1n.\\\ .The r o//ege entt ron men t .\\ alungton. I) ( lia er. A E.. and liontrls. 11F. The Islas k audent ses A merit an .\ merit ,111( mom 114.11 Et1114.,11.41t. 1968.1 l allege.0,t; Re-Sears h Report, Vol 1. No. 2, 1969 \ Ann, ,\R. Per.ostal t ul :mom tional delerilisihm1 of lit Beal. P, F., and \A slhan. I)A. Ass e penmen! a.1111 alted dent ao us tomI).'imir,rnen I and halal:Ilan in Grath:fa,. houang aigussient at the 1 usverst% of Oregon 19686, 1.I 19. '.;.Z tiros //ossulag. Research. February. 1968. 1.2.

A \ .College dropout,\ national profile.(( I/6. Bet /sq. R. II.. and I ',elm, II.F:, Factor. t elatud s.1 r allege' al

searrh Report:.o/. s o. I ahusgiost. I)(. \ merit an r S. Departnirest of I leall It, Eau/ atom, and \\ el C don Kdoration, 1972. fare. Cooperative Heeart It\ lonogra ph No. 8 0)F:310231 \ \\/'relenting :Indents limn droppIng out.1 /pup \kaIlington 1". S. Government Printing Mee. 1961 multaInstitational4111111of dr"1,""4."'" herd,,', R. F',. and Hood. \It/Jet-scions for to:nor:on plan: of 88 (....1,.. The kmai Eaglet" /11/;11$1 foul snorers rs 101 idler grasIntolitiii -I m I (jilt. N., J.. Prt.sitis eliall. critof \1' l l's 1965. 1961 T B\ s ran if .i,.11% so- of s I, , tell 11 1.111 Ie his rot sirs Homo s. H. \I. Eshis at tonal sdrotilis .ition in the I sided St.its-. son the problem of soli Idrool t s.I npulildid dos turd) Sat laita:11-111 I nun on,1970. silt 102.109. dissertation. t nuersot ) of Pitt-lough. P)16. Holland. J.I.,Dui tor:sunsuitofs "liege e huire. (..ollego. and Bonilla. I). J. Educotssonal aspiration. and parental sire. on I nil erns% . 1959..i.i. 11.28. s °liege. SO( col Forces. I 9(A). (8 ,'262 269. I Isollonsl. J. I.. E.ploration of a thesor% of S111.11101111 t-lisoice: Bodes. R Ills sdles 1 of II° high .s food on iiislent ait .1 \ I.\ longitudinal 111114 using .1 .ample of hpn.tl a oilg an Journal of Sot rule/ 1, 1'166.:1. 628 619, i11111 Jsitornoll of ,lppbeol hologx Monographs. 19141. Woo, n, H. D. \I.tuipuLuion of the en, irosinientol prs. in .1 col. ;:!. No. I. I. gsre isle 111 c hall. i'erao lllll / and Lueduno t Journal. 1968. Holland. J. I..11(441111;I al ea tonal him es. A Mewl of 1r,reerc. 16..555.560 ( :ar.. \. J.: Presitit-s.11.111. Inc.. 1973. Campbell. ond \I. %ander. C. \, Sims torol a Ifoe t and I load. \ E. 11..11)1w of e °liege for is.11.1t interps rtoonal relationship..losers( an Journal ol :no osiltigx. t% jos of .tulle lit! I. S. Deportint lit of Health. Eden moon. onsi 1965. 71. 284.289 Welfare, Cooperative Research Project No. 2182. Alinsit.sp I-Woo .( I Ills non los rvosting 1111141 rsal)fr.( Iiition Journal oli. Nosh sit( sonost hog Bureau. I ns% t rot ) of \I :solo. (..ol lege .`u toilent l'erson 1968.9, I 65 I 70. 1965.

offelt. JJ.. and Ihololo. S In and out of allege. Oklal .1 1). I'. 1)1,1 n1101111411111 1/11 lino 1If therf.ny lllllfotir Oklahoma Sian Be ill sit. for lisgliri 1.:sisis at lllll . 1931 se sirts ollegt Alen so mown/ and Et ulnas Ion In Loulan,e. Crites, J. 0. 1 0: it:zonal psx( hologx ,Nst.%. 'souk: \ loGraw 1968. I. 16.26. 1970. I fferc R. E. Retention and withdrawal of College students. Dade). J l'romise and performano e4 st utl obi! a, nail . Ds !hutment 1,1 Ile.tlth, Folucat and W chart% ()fru c to hie, einem lir: higher eilin anonBs rkle%. Cold . (.eider of Etitit ation Bulletin 1958. No. 1. \\ 4Iiington. D. (... l . S. for the ".anal of higher 1:11111 atom. I necer,nt of California. Cos eminent Printing Office. 1937. 1956 I (fell. Eand Clarke. B. S. College applicant:, entrant:, \ Grew aspires:son( his ago\ Wine Piso.s, 19634. slropotits. Department of !kolas, Esitss ation. and W ri 1).11111. J \I no le rgrailtiezt at err tie: o coons, Chit ago. \ !slim fartOflis tof Edm alum 131111ctin 1963. NO. 29. WstIiingtsm. 19655. D. (:.. IS. Gtovs.rinnnt Printing Office. 1965. I)I14r. R J S'111111 sits harm Is reostis and 1 111111of hotiing. Is D. I). Cotin.cling student. with sloes sal problem:. Res Juurnnl of Leil lege Student Personnel. 1966. 117 130. teen of Eoltit &lona/ Researt h, 1969..10, 239 230. Dols% an. E Ws Ison. J7'h e adolen roil e xperien«. \ et. Jos kon. stsisl isikls-r. IC (.. \ e omparisom sol pletigt. and York: W des. 1966 insleptmlent.. i'ercon an/ and Col:lance Journal, 1964..1.7. E:ckland, B. K. A moiree of error int "'legs. attrition Ando., 379.382. Soo Hilton dui ninni. 1961. (8. 60.72. Josown, I). C.. W inborn. B. B. and \ I:truss...co. W. D. (.horse ter - Eshis allonal *fs ting it s Bile kgsoosinol 1.1110r, (11,11111g 1.10 aam 1.111'11 with I osmosis !...4 1.111,1111ill .11 tom it oder hod agx. 1968. 1.5. 512.:)(i2. Iolls gsplaii .ltd1 soils g, t nrulluu lit among ptiblishigli ship. Journal of(.011115(.14ng oo 11.01 student, Prisi«.6m. \ J \ minor. 1957 .Inn -k.. . Smith.Al.. Ackland. II.. Bone. NI, J.. Cohen. I).. Eider. (,Parental pious r legitimation and 11.1 Vire, I1,41III, (owl,. 11..Ile sits. B.. and NIn Nelson. S. /negno/sr), /1 re. .1.,1,,,.111.Sorzonnern1963. 26. 50-65. 0101 anent of the effect of familx oriel schooling on America. Feldman, K. \.. also Newt ostilo. T. \I. The unpin t of college. on New York. Bash Book., 1972, students '4.111 F1-.11111%1 II.10./11 t Bos,, 1969 Johmon. K. Fraternit% member. I ompored unit isidepentls.nts Hat k.. W Ii'prate.st The so, isil hoses. tinstudent on he bis 'graph's al se ale of the I cl Journal of College .Sin. movement In J. Foster and I), Long (Eds.,. ProtestStudent dent l'ersorniel. 1970. I I , 284.285. n(nt Lan In1 :nem a New York Niorrow and ( , 1970. pp. Kahl. .1. A. Ethis at al and 1U 1111/.111011.11 sipirsitiose. of 131.157 man. boss, I/arrow:I Educational Ref (etc, 1953. 2.1, Flosiagan. to and ( :oode). W. W ,i'rojecr Miens One., ear 186203, foil lino snioliesFinal report to the IS. Ofli«. of Eau, a- Kahn. 11.. and Bower.. W . J. I he ...socialt sintet of the rank Ilion. Cooperative iirpwors 11 Pron« I\u, 2333. Pitt shurgli. ansfilso -lush sitae II% 1st. A test of four h)pothess.. Suetol I. ins s.rsit Pitt...burgh. 1966. ogl, of Education. 1970. 4 1.38.55. oolger, JK.,\ slosh II and 114.er. \I' Human resourt es (... and %Atkin. \ nanomy of a pledgee hos:, imirnal and higher eiltir ainin Nev. YorkII Issell Sage 11.undat0ml. of College Si tulent l'ersonnel. 1966. 7. 282 284. 1970 K IL I)\1A critical review of research on college dropout.

8 82 In I..N. Persil', I.I:!tell, and NtU,ah inple The \e1,, b. \ and A ikon. K. tE.ds )college peer 0//re dropout I th' ,,,n ot tdefir fart etoll. groups IIns ago Udine. 1966. \, J. Princeton I mser.iis 1960. it S 81. 0,ipors. SII, 771...///1, career de, elninnent t2nd ed.) \ et. K sssss .11,N. Higher education program.: and ardent des clop. \ tipletois (.rotors Croft,. 1973.

went. Her wit of Ednennontil Re.,enr. h. 1969. 1;1.1:2 l'al e, I , Ciiiige andnl, resits 1;n11,0ninent :wale, re, it I,es we. allot :Nti-wan. I.,\. Totn,dla,- anti nit uI numnal 12nd .41 elan, \ J.Vorits. atontal Test in fraternal, pledging inter,. on Journal 0/ ioelolosn. NM). iiig : she. 1960 I.."). 391,099. Peterson. ItI. 7./ al ,mural. College Strulent (lnesrson

1.onguso. (.. I... a...1 Katt. ( I he ...liege Is Aetna . nacres. elan. \ J.. Ellin 1961 lisors /ma/mil i(/ Prediger. I) I Ro.gi,npl11t al data Ilifiieuliating ...liege at Per 1 I. 118.12; tender, and noattender, atsdral.11., abilas les el,. Paper \I, Dill. F. Itigslos. I.( and les el 1).. Jr Educational read atmen( lllll and (oasilaili S11111 tattoo. 1.1, Innate, of high ,clitool,: Their elle{ t- and sources. 'liner: - %egas. 196'). ran Journal 01 Sol 1010g1. 1969.; I. 5()7.58() Rehberg. R.\., lam. b. and :NI:barer. A. F.\ doleseelit 11. of le -1 tire- and at hies e men thelms tor,f. Is awlionts -I met ti re and pa- In...graph'. al data for the de-, rapt and prediel 11211i rental 01 1.III/ailoll lora. to e. lineman Journal 0/ Sin la grade educational ''tile I npublislied do( 'oral di. -crow 19:0. 7.3. 1012.1011. do n. l 110,4-I-sill, 101 Illuun -. 196; Rehberg. It\ . and \\e -tits, I)I.Parental en, Iniragmnt.

\leilsker.I..11,.. awl Irene. J , blie wlluew r of dalerent 111 t 1111.10ofi, edu, ,111011, 1.11111IN Nrtifai trot or wile 0) puldo higher in,titut ion, on4 441.:4attenolam pendent dell nnun,unl. of adole,i eat tutu anoint' lie. la 11..1.1 s.trsilI Joe...110111k.11141 .11011.1 14Nels. I S. Dep..: t tom.! Stunt Forces. 1%7. Ii, 362 -.17 went of Health. Edin anon. and \\'Ilan tooperatise T. \ The de. r-ion 1111.4111111. In K \lertoit. Prot. 1. \ 138 lierkeles. (Ad. I nner.uts of I alp ( (.. Ready:. and I'.I.. Kendall. The student plt% sit MU forma. 1965 \ aril I ins t.r.th 1957. les el% J. .I Intl' elle. 1, on ...liege intention,'linen. It I. \ Lo, al ,oi ial Ann tart and edit( ational -elrt lion In an Journal ol Sin lologl. 1970. 76.19 70. \II11.11,r," 1. Howl. and \, Nwlerson dnen

Michael. J. t high Iwo!Innate,. and 1)1.111,1..1 entering i of Nun, r,01101111 ,andMOlel1 Irl'Order 171 the sot :idol:,of4,Itt lege. l'oblie (Iin num Queure,II.1961. 185 191 anon \ York: Free Pr. ,. 1963, Michigan Student Studs. P)(17.1.It n Iraiermin, at 111,I 111Ncr .11111,,,, lilt Ili Nadroitt and it ..mono growl' "its \ Prelimmars report 1 r the 01 Braid. .S.0 :0/ For, es. 196 MI 351 g.1111/.1111011 Z11141). \ dew Ilea as 1 or :Not tat Re-can Ito,,I. F. II., and (ulrman. J Fa.) 1)erertnindrirs 1,,n1 I on 111Nt.r.10 of Nlielugan, sequn es 0/ .011egy I Innt e, ago: \ Atonal Opinion Ite Ingan ..-111414.1,1 Muds. 196;6Ian ,..ren nit-- I111N or -cart h ( entr. t of ago. 1961 -its of Ingass Prelim:1.ns report loo alit Tnulent Orgatia J. E.. and Kirk. li F.6 ton, related to per,i,teno

/mom. \tinNrkor'n.411.111- for 1.11 Is. .1.141 ssltlidr,assal among innasty -its Journa/ o1( unit

1 \Ii lugan. seling l'nehulogs. 1970. 17.56.62 \Idler. 1,. I). Di.tin, tate11.11.1.1er-1,in of Ir,ninuts T. hortil, Idt, I.F. Po-t high edin anionIn I Rana

Journal 0/ College :)rudenr !Yr....mei, 1973. I I. 126.129. gan and .%N. ( ) Prop( tI \I 1:.% IOne year fol Nlors,lunia, J. K E.liet n, on ..1.1111111 .14 1111.N. 1111111 of r. ,114 ',iv. Department of Education and hall grouping, 1).1,...1 on a, adeintniwor, In(II. !{ogles A -lime. (ooprr.WSr 11e,eari h Prme, l\. 2113. Pitt, (EII I, /{ales, it on to irdeent 11111111 Pam.,(11111:,1)1 the Sr nth burgh. 1)(6. I ASI/I lllll :al'iron, 0/ the for In %Wilt WWI ReN,(111 si Imenli Ida,I. E, F.11111.1114.11 alts r high 111.01 ';0,1010g1 7I \ 1.: stale I ins er,its of \ es. loll, at ( ortlanol. Edut anon. 1968. I i. 350 369 1966 ',eon. \,I aloes and or);nnzlitioni 40.11 of 011ie/ Nlorrov.. and NN &on. K, (. Fassuls relation, of laiglit high and StIrOl lilt'sCIIII .1114.I{.111111 \11( \ 1961 aelnes nig and under .1. hies mg high -t 11001 hos, Child Dr %CIL %\ II,I in11111111111 of re,ull ill i .tadI °liege plan, elopmnt. PHA . 501510 !men, an Sill 10101:1, 01 Re, rein, 1%1, 29, 21..18. \ am, 1 and Cow ling. J. IIE a, tor. related toollega Sewell. NN,h1,, and \niter, lb \I ighborliood tonrext and pendant e of farm and :ion lain high 11..01 gra.litaI ,. 1960. allege /nen:, nn Sur lo/ogi, al He; re. 1966, IIII') IS. Department of I ounmerce. Bureau of the (.en,11,. I II,. I()8. \n 12 I) I Department of,,mace, f So,. ell. %N. II.. and Shall. I'So. Mei °noun. stain,. noel Depart linmt of Ngrieulture. 1962 by rut 4...11111 111. .111.11111114111411 Inglir ,dot anon So, 10/ogs

4) 0

83 of Er/m.1/1ms, I')67.III. 123. Aatle, I). J( ,ler iirogre of \ lent I V0110001 Moil II. \\IIand ',halt. I'so. sal natenial eu. ttttt .ihaiorship 1,01100101001 war( It /{e /lulls 1. \11I. 1968. agt mem. .111.1 1-41111.111011.11 .1.1111.111011. 1/01/11/1/1 ./0111/11// \ 11111 .11. /Air, in prop.r ,0011 ul the nainrol gran ih St000 101:). 1968. 71. 55').572 per.onalit% Net, leak: Ilolt.1952 tati. oithilotalisig 1,0.1111111.111III 10111W 1..111 R 01.1111,\ Itt olt utat gin ;Idiom 41 ot sal 11.1,1...11111 41 eiti faitar of eart laJonr/sai .PI General V11.110111 of 11101 .1100111/0)....Ittieri.011:10i 0501.:0 01 16.10.11. hologl. 1965. 72. 101.326 1939.2 5, 8.16 813 Sististion. N. (,. Junior 44114.ge going plan and at *nal 444114 g4 11111 1111 rifig -twit Iii.lttrilainand age sit of

altemlam 4-441 a. ads ally 01r110: 11101 I 1 .1.11110. 111 hang.In \I. Nt 444.4iti6 and E K. A 114411

1111 tat of Georgia for the 1')61.62 ,./ 11 I filosh peer groups.III( ago. \ Mine. 1966 doetoral 411----ertation. l tot emitof Georgia. 1963 Sisiip,on.It.I.Parental influence. antieurator .01 lab/atom. Section III and 44tial bacis. an .S444 10101:0 ul Bel Joe, 1962. J7. 116rig61 F... and l.leauion, JIl.Peronal 1411or) torn latf 117-522. of 1,10o al lotIt' i arren aviration. Journal of lpidred Slocum. 0, r 01, liana!areel, 1211 t41.1. ( hit ago haltpg1.1961. 15. 281.281. 197 I. ;nm College rting Piogram. :11441/ng ./m/ents on the Si6o1%, 1/romitil (rout flight r4.41114 atom. \n n a I /04 allege. 1ul6ot. Iowa (at)Iowa. 1973 inlinal) n to tt asol 1111114 /tura hatsp. 19711, /. 61.8.; 11t.e.t.e.i. and St hat ler. (,I.. filupraphit al c maim(

Smith .(.Statti.,44 hu to mem. 5041 1111111%.1111111 111 1111 arti(it and lilt ran 414 ,tti%II) 111adult '01'111 Joss/naiuJ \ /urn( an high S1 /100! /6-i 1971.79.:s79, toL .4/9//tet/Nleho/ogl. 1969. ;1. 267.273. *41.1rk. \ICostistolt4 r and Ii 1111114 t11.111 11111111i. 1 oilman-41 Baird. I,. I..-rin lin and af rap duds sit nel and (pus dam e Ian rnal.1965. 11. 277 281. Lelni animal /'v, holagit al 111.0.1a renienl. 1968. 28, 891 Stamp. F , and \ I) 1. liarrier totallege attnd:meg.: 89') Ltfl port 441 a --link 441- Lm tor 1'11.11111 61 111111.1111111.11 di I.- 1 111 lire di, tun of in I ',Hug,

4/11111111.11111 1/1 1111.111 Imo! gradmat 1 s 1)4 part1111111 411 t xidoration of tin true of at h utt /111111. Joarnal Cinin 111.41111. FAN( .111011..11111 \1.11.111. l 444414eranor Iteart It Pro) celing holags, 1969a. /6.216.251. t I/1111111MM 1.441 at c ompliIlinent re 11108 Magnolia. \ nt .dh;IA 1970. Baird.I..I.F.14 tor, in the Suisamerkill. J. 1/romost. Irmo 4 ollrgtlit N. Sanford iFal 110111 high .11101/11111 01111:1 1lea.s0renient and Eialliation us The 11101(( 00 4 0111./.91:1 //III hologsral and weial I illerpre Gaiplan«.,19691). 2. 5.18. last an al the higher learning. \ lurk A 114. . 1962. Baird. I. and lilt hard.J.\I.. Jr The effm of 'ells Ung Sillier. I).F.,. and ( .rule. J, ()tlppnrcling to,atvonal 4114 '.111141'. k11111. of 111811 19.4 .o 1111-te fro,, ed,1 Nev, %or'. Harper and 11,4%.. 1962. 111/.111 !CT 1C(.eur, h Report Xi, 23, Iowa ( .1 ) ,ots a, Amer;

Itiqletli%atte. I).I,lit, rustilll 111 .1111111'1111111111 01 L111.111101 Ian ( reI mg Program. 1968. 101111:1 11/111111, F mil report 411at Whit lit nog do re (.liandu n.J.\. lit !mingItt rtotalil) and Islograli11i4 .11 la( lon nurture itt and training 441 mu 114 4 111.111 1.110111111.tulle ist to 'ado 1a ativib, i'st4/144/49:n 110alographl. 1961. higher mItit ational plogram 1)epartitient of Health. 87.7. Falin anon. and At Ilan.. (.00lo rail% tIii eart It Pam, I No, \I.. and Gait r. E. I.. Ideotifit .1111/111111 reato.11)Tin in 1)198 \JAR anderIttlt Ino. erit 1963 /10/116151 l'sv ltnGtu n/ Ballet in. 19-0. 71. 55.7.5, I onto. 1/romint Iron' lin:1141 414o atom \ Iht ors 111.11 %%11110 Khan. (.. F.. and She %el. I.. R. \\ Ito c. 151111111 d! lit anal) i of of1140111 re4oft hfitI left01 kt111( 05101105 14,1'01( II. 11ov meat. 117 Rewarp h Report Vu. 1/. lots,' (a),144%a. 1971. / i, 89 126 \ merman College Prograsn. 1969 Irent. J. .Per-4itai la, 64n In 444114 gi1 Imo tPang r read it (..taunt, S. F.1.)4 hollow( .11 141.1\ ( redo. ft)PsIr hologo al .111;111.11 meeting"! the ( .,Ile ;e I.:sin-amt. Examination lioaid, Bulletin. 1963. 60. 518.565 1965. I le1-011. It 1'ertomtit) of women %all imaginahve and anti...w- Item. I. .. and \14414.4 r,I !felon,/ high 4111po01 I imp inter, i.Tintiltof ;me-41111111h. Iffigigtalit), and other 1,(..r, 1968 haraieritie in their ereatix it). Journal of l'ervonalitl. 'huller, 11.IIthe 10 101'011letl 01 011110510115.111 F1'4111'110 1966, 11. 1.25 ( handler. 1961 114.1.4411. R. Sex difference,. in eaalive t)14-. Journal of Per. \ antler. S. F..Senor(dropout,l {rcear,h caniniarl. /967 S/. tuna/to. 1967. 15. 211-233. Aalutigtosa. I)( F.11111.111011\ ..114 1.1114m. 1967 Holland. I 1,.Creall%1 .11111at adnni performato tamong toil Ian.ILon ills ill 150 1110Allig ( ago talt nit 41,'1144!1.14111- Journal of dii«aitazal /'P,1( !dine. 1965. 1961. 52. 136.117

84 91 J. I. .1111! N. Ins soredit too. of at admin.. iandelitt.. \animal Vent S

artititscientilitand 1.11at Imo. einem of undorgradm Report. 1965, 1, 5. ates of superior scholastn awn tido Journal of Edinational Roe. 1The making of a scientist. \ev. lark: Dodd. \lead. 1962.5.). 132.113. 1952. Holland. J, I... and \ 11 1106. . Pretht lion of at adenn, and S. 11(10fer. C. F. Tile prod.. lion of 1 reatise ae !dew...cut from extracurri. ular 111111111411 III1 11110gr. J0014101 of Ethic a- J I....graph.. al 'mentor)Edit stional and l's)chological tional l's$cholog$ .1961. 53.35.65 lieu curem oat, 1969. 29. .131.437. llollantl. J. I... and Ru hards. J. NI,. Jr. At athunit and 11011.11.1 St haeft.r. (.. E., and 111(1,1.1(.1. 1.1 biographit al in entory fur 11111111 .1(;e1/1111/11(.111111.111. Correlated or um urrelated? Jour 'druid') mgrail 111 J111111'10 1 nt 110) Journal of mil of Educational l's$eholog.). 1965..56. 165.174, l's$eholog). 1968.52,42.48. Holland. J. L.. and Richards. J. \I.. Jr. 11 atleinit and non al a Smith, ANJ.. .klbright, I.. E.. and Clemion. J. 11 The pretht denim acoompli.1 111 111 .1relIrl.0111.11 1'...11111/11. 01 -ICI t 11f retart It (1.1up0tent 0 and ( rya tivit) from perottal (hulls taking the Amens an ("Alegi. Test. Co//ege out/ em lotory. Journal of.4pplied Ps icholtig). 1961. 4.5, 59-62 iersit). 1967. 13. 60.71, 'Ev..10r. C. W Cooly:, G. \I.. and Nielsen, E C. W(4011014; kirk. B. A.. and :loreda. I. A11 ma( y of ...11-report( (I gradt high hoot students, v.1111 1 harm teristit a needed in rcear. orage and ( haraoteritoof . and dist repant report er ss ark. Univer.ot, of Clap. 1963 (mimeo.). Edneational and l's$t hological Ileasureineni. 1969.29, 117 Taylor. (:. AN.. and Ellison. R. L. Biographi( al predictor, of sr( 155. (-nuke performance. Science, 1967. 1.5.5, 1075.1080. Milberg. (,. F,.. and (hool. AN., :Nrille life Miler) antecril. Taylor. C. AN.. and Holland. JI.l'redicturs of 1 reative per- (-lit. 01 engineering interests. Journal of Edit( ational P.)i formant e. In (:. W. Taylor (Ed.). Creatiuty progress and holog$, 1960.51. 2631. potential. New York: \leGraw.1111.1961.

Nlacknimm. D. A. The nature and nurtort. oft reative talent. \Nana( 1..N1. A.. and W ing, C. W., Jr. The talented atudent 4 inenian P.ssi hologist, 1962. 17, 184.195. ;ululation of the ereatiitI-iiiteIligence distinction. Now Nlaxey. E. M.. and Orillbv. 1J. The at curacy of selreport York. I lolt. Rinehart and Winston, 1969. information millet toil on the t Tet Battery: High chool grades and items of nonaadt. . ement. icr Re. Section IV search Report \o. 15.1.0,, a City , Iowa. Author. 1971. it lllll Lerner. NI. J. Oct lipational apiratimis of NleDermid, (.. D. Some correlate. of cream It:, m engineering boner 1 lass negroes and %hue )outh. Sue Jai Problems, 1959. personnel. Journal of pplied lehologl. 1965.49, 14.19, 7. 132.138. Nlorrison, 11F.. f)uon... W ., (dennolt. J. I).. and ,1Ibright. Baird. I..I.. F.111111) Ill( .111Ie .11111I Ile ( 11.1r.11 toritii s of college- L. E. Factored hie Illtors ante. tolent of inclutrial recart bound tudent. T Research Report Vu 17 10% .1 City. perform:in( e. Journal of Applied holog$, 1962. 16. 281 lov..t. American (:ollegon.sting Program. 1967 2111. Baird. I..I.. Pattern of educational aspiration. 7C7 Research \I.el. J. N.. and oral) of Report .Vu. 32. Iowa City. Iowa. Ameria an College Te.t lug (-11/111. \\"le blank work Journal of Applied l's)cholop."Pli""1" 1932. Program. 1969. $6. 365.369. Baird. I..I.Overapirers and miderapirers Student. with \holiday. I,. A.. and Davis. J. (.. lanetie. of achievement alter (10o repaid ability and degree plan. 1'ntowed at 1970 Na- college: Perz,pectio on the meaning of at adonin talent. tional C1111111 ilOil Nleasuremnt m 1.:(1111..1Ii0/11111.01illg, ACT Research Report Vo. .57 low,. (AI), Iowa: 'Author. 1971 Nichols. R. (.. Nonantellect IYO predictor. 01 ai Ineyement Baser, i.1. K. The him k ( (Alegi. freshman Charm teristi( s and college. A du«itional and l'xIchological Ileasurenient. 1966, ref nut trendIli' Research Reports, Vol 7. Vo 3 A'aIling- 26.899-915 ton. D. C.: Atnerwan Council on 1.:clut alien. 1972. Nichol, R. (... and Holland. J. I.. Predn non 01 the firsty ear Bayer. A. E.. and Bonn h. R. F. The hilt, Is student in American college perlormam f of high .11111111110 vtadinis. PS% h0/0.(f. «Alegi.. I(.E Research Reports. li,l.1. Vo. 2 Washington. .rat 3lonogruphs. 1963, 77, No. 7. whole number .171) I) C.. mer.( an Council oil Education. 1969. 111( harder, IAl Jr., Holland. -I. L.. and Lot/. S. ANPredit uou Beihn. IL, and Bergin. The ...Mal mobility of a Hunted 01 student .111 01111111111111.111III tollego. Journal of Educa- urban group and some Until...won. for counseling l'erson- tional l's)rholog). 1967, 58.343.355. nel and Gm dalleCJOIlrna I, 1956. 14.:41.552, lbchartk, J, NI Jr.. and Lutz. S. ANPretlit Ung tutlent a.. ton, Borelon. B.. and Steiner. (:. A. Jima n behavior intneei- pi...fitment in college from the 11I A st.Ineni. Journal of tar) of scientific findings. New York. Harcourt. Brace :Ind Educational lieu curem en! 1968. i, 17.29, World. 1964 Robert, 11J. Prediction of collego performain a of superior Comers. 11 The p.m holog) of social classes Priut 01011. 92 85

I I' ton I or,:t% Pre... 1919 Nita:n.141 1 ins el -its . 1960 ( K /4, putl, , 4/, A:,stnel. net I de... tuition 1 "IL \J.. and I sat. It l/2,,, out/ se lute +meth. lot 1.. in... 1963. l'1M4 rt,nt, \ J I41114 .1114411.41I. A nig ,-1).11 c, 1968 Ilolt, 1100 14.111 .iiid

1)a% J mid iii. , It \I \ 1.111. 011. \\ erc1elm as tot'., \\ 11%1 IPA,' hos," la/ Pro/dein, \, t .1Its an at Its litt k 14.1, fonutn/ o/ (.onnst /ing i's 11.o.t.n. Conn. Lilt. ( siitr11). 1958 I/to/on:a. 19ii. 2. 1i

\Ihat/ I. and Jiist n,\ 11 a Fd, t1m sal ,loss. \ ItI II. Idclitis and itei sonalio.;\ aid 4.1 . 0/14/01I 0/ 11,, Ih//tIll rttS. rk 1111. VA:al 10.141.0rd 414 01111III 1011ege 111414111,I11101141111111 11.11 hart and \\ 1968 (Mal 11.11 drat ). er.its. 1961. Dregs-1-, R\Iand \Idle 1, K ( at( ns, holoAlt .al Pat kat tl. \The status .seehet.. lork , Poo kr( Book.. 1961. ofnt gn.a and .slut, ,111 tin1 mu d slain 1919 Parson,. 'I'. and ( lark. KII.( F:11..1 The 1egrut anrn. nn 1965. Ps$, hodogt, ta/ /1/1//etni ilnnu lapis Men/. 19(01, II"-ton. 1 longlinon Mifflin. 1961. 70. I. lg. Pt.lci son. It. E. Tel hull al manual. Collette Strident ene5t inn 1; n late of .t..1.14 11114al 1111 a 110'111 .00001,11101114 nen el1'1014 1 1011. \. J . Palo. .11 ttttt Srrs n r. 1965.

Urn .11141 .MI1111111 11110.11110 ;4144 .111414 Ills f 11111 Ha Is Is 1, Isou. IIE.I lit student lilt 111 \ mem an lintlato csItit at ion NW), .1170, /hredtdres. 1968. 97 tiII. 29.1..317 ldniala. K\and \ Ist 011110. \I The wino, vile/;. nn l's it Ign 1..I prn/de /,/ the. ttpree1 met if WI \11, 1011, student.. San F ran. 1st 1969. 4111 \ si.t rand ( 1961. Them Winne!! depIlled t held. \1 a lark. 1 lai pea r. 10,4 1eh II6',0/1011, int olonionanth ne Ato .1111ges 111 I. Heald. ,due 1371 \\ and Ito,. 1962. (.0s(into ut Printing 01h. 4. 1967. 11w-en. IL 1 II.,. t., villain its, and 1111..11 11111e1111111 11.1rom14. .'we whip. net le11P/39, .!1. 17.0, 1 11.1 4101,01, \I /11, I , , et/ler-WM r sal( F f.1114 1.0 0.III- I merit ...e5 Mt.-. 1967 '-pads.\\ l,, Stain,. at 1111.111111.111..11111 111011V.1111111 111 the

1111/111:::.I I'm.(11-g. . \s, Ind. ..11141 1111111.1. J I \ 110'0 an lush hold St Iwo! Retie:1.1971,79..179 II) I. (1. pot .soulAll, In digit 11111 I, 10 11 xt /If 10/ theta 1\1 11 . I'. and D5 sorilk. I rh, hrw t 011 4. mkt war. I t flu larklOlt ink.'ins.sitPI/ 19;1 anon p11idosut of Ole 01olools student, fuurned (dellege ( oatlitb, I)Sot aal lass. hat s eon lit, and the .111(lent l'enonnel. 197.1. / 1. 2')3.291. c1"9 09" S. hurl Re, ,e"1962. :IL 273-286 \1, F: Sot oal t Ids, .11111 111111.11t dryer 5 liooct oft °liege II.')ot 1/./1.141 4h(Itneirtnn. 19(16, P). 71.83. I I artn. 0. 11 T 111111 rt sit 4 in ch 4 14 II41111111dt .11014 0111 In .1

111.111,411 11411,4 I 11 111411I. 111414 10.4. 11.4411114.11a11% \\ ( List in s hoit t. 'mitt \b11tis and ticgro and uadtlao11alis 5. line lost notion. So. Adel- S", 0//0/:$ Ethit sawn. 1%7. 1(1,.118 .138, e at tun, I')70. I I. 119 116 \\ Its te. I)II. `sot AA alientat Ion among t Auden( s. al tin II. iiI'. Pt 11I 5,11. II. F.IIItgool, int It it in..14. dos towd di...yr' atom. (111111(.11I 111%ei111. 19(1.1. facto: in at minimal and hat kgromial eau es among kg,' I elinien 111, udogi Ill Edo, (.Nun. 1968,I, 22' 237 Section V I It t I . It.. and "IIs din lass toad the, ntlnin 111( 114 .111 011 l11.I'rt 4114 Luny; t 1014140ft 111111.0 el' 11011110 r4411 lord,: John \\ des, 1966 114.1 data \uthor. 1962 its I I,oast non. II oa 111. 000 .dud, Ill.110 111. 5 11111.1 110111 110 \.h, 1,II, 111114144:41,114141, al 111 01. ( stn 11 ininms sl! I al" lergradua tr.:Ii /.1/. war, h Reinnis, 1,11 8. n tttttt iv/ i'st, htdogt, 1972. 2 27,1-269 \ an,\\, and P,111,,,, 11, J. The 11115/ /wow,/ \1.thin,41,.n. \ .111 140110 11 011 1%11111.1111411. 1973. 111111. \ F ..11141 1/.1% \I\\1111 hL0 k nth iit w liagla. d /"1"11, net 41.1ung141. I). l..\till 1'1- Ill atom\ sc. and bibliogi Aldo. formut/ o/ (.0/it .Student 1 ,111 l "1111(11,M E(1111.111.,11, 16% ttttt tel. 1971. / I. SW) 112 Baehr, \IF:. and \\ ii. Prediction of -.alt.. st 11ou1 las itoriall% th t1 noun d 1111114 11,1011 of p1 noual hat k Lao J I) ..11141 J 4,0011\\ Kt L14.11,1'11. lo 1,, t al 111111 ;II ;;round 41.11,1 Journal ofI ',plied l'ite hologl. 1968. ;2. 98 1 510 1 .1.111 .4 ^111111 WI' Iron' t 11111.1.111 ally .1,

I111%1.11 t. k iOnu,1. /ourtra/ (..//evc .Student /'ci (tome/ 1(1.1 t onciir- 1972. / 1 1 118 Burt, \\I)\ ohnitan. female s Ica it al turno K1 is klioil,\ I \ I a.1,, \\\Imid 1,0, nia It.,,t, r. tat and lone 410 10. sa111111 4.111 a %..110111.11 a1111,111.111111.111401k (arm liddre 1. pans lamas ;),), nil foternal al Inttl zed Ps5e hohnt.. 146.1. 18. 180 1112. ogt Wiwi, 1973. Pr. 219 217 ( amphrli. I). P //ant/hot,/, ,/or the ,Strong for altnaurl haereli Stanford I ins cr.th Prc,. 1971. 1 an,\\ 1 h/ loo t Ia.. gad oil. \ lid% ..1..1.111 IllunA Stanford. ( \ \. (.utee, des chnont ,t/ /,/it lent, %/1 ( ooni fad 111 11114.111 4,0110 II 1 141.411410.11. 1111.,10/1411 1111 ridipm. moil, s. \\

86 lieseart h Plop.. t No 116,1 ( )11. ..1 atom. 196 t Jou of Loa ag I's It hologs, 1961. 8. 82-87

1 \\\\Immo. tom. .1 ii1011:: 4,n. n'I 1.011 11.1 1 a du/ t s..e atimial re« plan., Jim, n fil ofIpp11ed I's It holog Iformvuo/15. 1101 , I OM ,11111 Iwo -11 ar 1 11110 Ps,/ 1967. 1. No. 1. part 2. I16. logi, al Ilionographs. 11062. ...t,12(r1

Darle5. J .and I lagenali.I 1o, olio/nil Interest measure 11,41.trid. .1 I N.1111. 1)1'"."1""' "I .1 111'" "I "" meat. Theors nn,I prilent \I 41ml,,l no, er.it% ..1 \Iw and 4111411 51 nu III. Itoill cat lot, di* 11..11 11141% 11e...1,1 19-11 i' itedoeti Ill 1(17,1flfs. 1963. /2.111 301

1)//111%41/. R.II.Nti,.1.1,411.0ational ISI,u1k \ 11,41.11111. I hi/do/aril/115 ol a 1/11.14/ t 11/10/ uhonul a hot eI pre..ed 50t aUun.tl int« e.t. le% low Pst,hofotm I one "'"' P" oh. ".""1".11 \i""`"' N. 1" Ir ("'ill lents. NW/. ;2, 03 107 .1111 I. Prole...mai :N..1-5114% 1061 I) 1 tt. \I. H.. F in1111/.1, F.IN, F.n, km. I. ii., and liana-. 11.11and. J I \ psi 114,14P .111.1`'1" 31' `' l4, mr 11"` "" \. Pledn hug turn.% ..1 teinalt Mkt 1101.105, 110. aiiil Immo fields Joanna/ of (,inn5e/tog hithutl, spurn,' Idmintstration. 1060. 2 {, 11 10 1066. / {.278.288

1)%er. I).I. 161 teltion bett en 50...tional mit let 4,i no J 1 11101 .11011, 01 a du ors 111 s111 ,W1111.11 (11111( ...ne.. and their tililpit.iit liedot it .1..1 I elk 1 1 .11. Jour/1.t/ 1011011411 ml 11141,11111/: .1 0111,14 ol I%iu,tl011"P' n/ 1////11/ti P1 1/ hob/g1. 1934,! 28(1 288. Journo/ 11/Ipplietl l'5%, holog% ,1068. 52. No. I. England, ( l)ei el 0/Intent eind use e,/ .righted 11/1/111f falfffi part 2. 1..17

flf141Ire%cold. \limit...polIndu.lnal R. !atome I elm 1. 11.11..1.1.I I The sell dire, tell watt h. Palo \ ( (o11 int, et sit% .11 1971 ulUn P.5eliologi.t Pre... 1971 Enright. IIt.. and Pumeam I(Predit to, t %Aim of skiing 1 11.111nd, II..1/tilang 110 wound t boo esI thew% ofate en l 111/411..14 01 111/.11 1011 .11111 the ..,N1rong. \ auonal Ititcretd Engles.. (Ill. N. J.. l're10144. 11.'11.1:14. 1073 Blank 1.5er 1111/.104, %ear. Itnet o an l's%, holoost, 1035. ID, 114411mol. I. 1,...11111 (,otilrt dson. (,. I).\ ppl% mg a 15molog% to 121.125. 50cati0ial Report N... 176, Cenil lor ( tr. lamtgall. J. 1 . rt . Prom II I one lea/ /01110 -n1. stud ganitation of :N. John itopt,u t 1051 isit% . 1071. les. Palo \I to. ( Pimp 1I \ I F I Mit \1014,, .111 In Holland. J. I... and .,\\I lie tirtdo t15 e 5.thie .411 Ille 1.01. 1966 t 600 1 ol %motion Per mon,/ and Gordan, r j(1111 F: \and Bern:gel. J. Ont. 55 as to u mho e ofb, e P168. /6, 128 131 turn.% er, Personnel, 1960. 17. 611.71) Holland. I.I... .1111 N.11110Ik, Ii(.., Faplor,Uion. ttiilleor5 NI.. and ;;Nolialt. I)\o. atomal Pr, lerem e Ins entor% 5.. atlonal III\ longitudinal -mil% flange in high point 4 oi1e, pre.rd meth. lure 01 maim h11d ..1 l'et sonnel and Gnt,1un, e Journal. 1961. college mapa and rarter entr% JuDrhol Lounieling l'%1 11. 231242 ch01nt:1, 1075, 27. 117.121 Holland. J. I.. :Noreti-en. . IC. Clark. J. I'.. Nal/.11:0r, I). II.. (.1..el6. F.F,. The 1,111,1111 of f noniron& aptitude tests. \rss and Blum, I) ppl)ing an 0(1 lipational clas.ifu alio., I. a ork , des, 1966 repre,e 11 MO 1 I' .1111 1Ple of work Journal of litp/tet/ C.111.elli. F.E.. and !Lulli.!. I. P 5alidil5 per.onalit% P.s51 hohti.:5. 1073. 58, 31-11 1115entorie. 1 he 1.141 1 1011 01 1.1111/11/1 re, The Jou rn,r1 o/ JI. , and Whittle). I)8. Change in the vorational phrd l'sii hologs . 1953. . 18.20 plan. Mirth' students: Orde-iv or Z.111110111? .1(3' Research

(41tubcrg. E.. GizeMurg. ;N.\\ . \ StIrml. and Hernia. J1.. Report, Ao. 26, loss a , loss .r\ mem an College .n-ting ()«-opationeil ehou e: In appt 4,w h to a general t hear% \es. Program. 1068. York: Columbia I no, t.r.it% 1)51 I Intelm1.011, 'F S and Rm., I. A. :Ntudis 01 oct 'mammal his \. F.(.1111,1110..1 e t and parental altitude. tors. Part 11, .11 1 1..1( 1/11 11/., 441 :44cupational groolea on the test .1 Rm. -; 1151)ml...se. Journal of (.onopel Ing holog% Roe .5 stem infirm,/ of (:ounceling NI( hol tun, 1068. /5. 1959, 6. 153-156. 107 110. Hagen. I). (..11-4.0r and honk .01,140.1111 re. \ lest of Roe's Kirkpatru k, JJ.11,1( kground Cm tors that lead to earl titi%.

orv. Joann,/ of Counseling l'ischologs, 1060. '. 23156. .411 1 I men can Ps1( hologist, 1060, 1.5, 177. I larding,. F. 11. and Bottenberg. R. A.Efiet 1 01'1 I...nal c hai Lot II(: , and Partons. S. () Itelation.lup 01 personal aoteristies on r41.1131131,11114 1411V1111i 41111111111, 11111 per history inlornialmn to tin1.1 rformant prodlit tom sups r forman. e. Journal of Ipplsed Ps1( &dog% , 1061. 15, 128.130 5t 1Irs Engineering I tolu ct l'sle &dogs. 1960. 2, 20 26 Ieitry, F..It. Research f renee on the use ol au tobtograph- \I. \ rtliiir. ( and :Ntt 5, 11,, 1 R. Tin ,,,lohnon of e pre..d teal data as ply hologn al ',refill 1011 Greensboro. \ C I 4111111.111.11%%lilt in% entorit t! intereAt." four The 161,,,kon 19n-,. ternt, ear 1.11.55011 Jont no! ol 'Ipphetl l'sst &dog% ,19 ii, lion,uoi. J.I.. ' explor, lipational mks 0, 181 189 9' 87

I N14.1,atigblin.I).II. and '114.41t.nian. I)\Ides en.% t..ir caret.' dine. 1971.

...431,0111% and ',how., 1.114, II 41in 1 \1 I NI dala trong. F:I' .1i l'rtIII 1150 5a11411 441 %4It atiotial nitcrt tLtouh t6, F 13.4.44,114, 11.,11.1401.111/1 It 141 01 1.11 It .4. J440,00/ !Alf 441101.41 1).11( 1111119 :". 1933. .131.3N, ation .at J440/110/ o/ 1,04/1/0041/ ill hal 1411, 1971, *), Nroiig. F.. IN., Jr. //tett/mm/1 interest. /1/ men mit/ le ettilell. >law. 177.1%, lord: .tailltord 1 nistr.it) 1913. \ F.I F:nipirit Alt of Ii40 .14111 Wit tit 141 / F h,. Jr. \,1141114. 141 01 t 111/.1111011.11 1 111111 1'Ed1/I 1111,111411 Joorntil 0/ Con nsehng h01 on,/ I's% holopt IleaAaremnt, 1953. /.1. 110.121. ous. 1970. /7. 11 18 14411g. . Jr. 140-orunuil interests /8 Icor. ofter , allege, N144.cl. J, \ \,1111 . \ %%. avid,/ atom blank 1,41 41,011 m%.1-41% \ lintit-o4ta 1955. irtlin tom of t inn.% tt 411 II pawn. lit -44411 It ,k. Poi >Nor, I)F.. (I j U.. 11111111111.1. R. (. . II. ( her. n'ii I's), hologl, 1931 I, I77 181 0.41. P. I,and \Nartiatli. F'I worm/oil derelopment 1 National SI If F10111111.111011 Tic I/41 a, Ott, theal. goilt /fano um lot leSt (111 Is. Nt s% lot Is. Burt an 441 Pubis. atom,. grey, \ SI(LI 26, \\ a.diington. D. inked Nate. 1,445,441 'lean 4411egt. (.oltonbia 1 iii%t.r.ot%. 1957. anent Printing, 011at t.. 196'1 tipt.1, I), It and Goitlan, E. II. Houndeting Norm. I,. and Km/. \I Ho int int 1,l of at a11y4(410 out r Mid Mal ape, high A) hoof \ cislollsIt tit her. (1.1It gt 0,1, Part II: ht prellirti%0 %alitlitit. 401\ adennt Intertoa Pte.... 1970 \Ica-airy, F lit ...cart ItBuilt-tin 7067. Pron 11,441. 'oilier. I). E . \ latlin. N.. and Jortlaan. P, Edit ational e. 1970 ( diet r des tdopment. :N.11-4 ow t.141 them). Reicati llono- Norton. I I rug [call of re ',wart 11 on I,I alwi.d dreltiit gtanh\u.I, lark(,ollege Entratirt \ animation melt 186,,199 in II.I Peter., .1m1 J. ( 11a11-.1.4(. 1E111. 1344,041.1963. ational :::no/anroil (Jr11 dr. /11(111711.111 i.2.4141. d I\l %%itit.r. I)h . Grigg.\. \Idler, J. and Amillg. 1.41k\ %Ilan. 1971 Earls rwrielat I., and tort iiptitiolial eliont t\ of Roe'. 0.11,4%., IIthetilit %4l, wee, del flop:new NI,. Nod,\p la% potlit,i, Journal of Lonnwlsng hologl. 1962. 9. Va.

Orton ( entin s 1 roll., 1968 (8 \I 1a \\ %%hal inajoi in 1 olleg4 ' Papt t 11,11. (. Re( .1111.1111.11(411.c htId relation. a.. detrnimant 1 prt tot 41ai 1970 \ own, all liologit al taboo 04.1114,11.11 t 111411 ti. Journal al (*.ouneling holog(, 1962, 9,49.3:1. Hot .\ The ps hobo; I .1 atnull/ion% %. 1 111 Is 1111111 111 S. I),II, Vieth, king from t. S1,( .111011.11 c 110111. 1910 \rt.% tt.5%. Pr sonnel and (Aolanee Journal, 1969, 18, 279. Hubbard. \\ Hutt bur..on,I'.F... and Bateman, '1". "Indio, of III 11114,11 al lit.tot%. Part I. pplo I hang... and the 1.1..,1111 .1114111 01 to 1 1111411011, Joornal of lmoneling l'i( S8. Vl hologl. 1966, / I, 387 WI. \11144(.. I) . lest , otnhtl, Hon. lIeseloiamva andtnielprela-

Rot \and nut, 61u.wt. I of 1111 111.111011.11 ion of 441 lae (wield teis 1211 Ulu,,. (.. to,.PartIIIThe 011 upau,n,tl gomp, 01 1114' Merrill. 1971

1;, of (.ouns l'S 44 /Mb/141. 1969. 14. \ r. .I..1 I III blaquatolii( alIt m. ( aa. 44 hl inipro%t 41' l'er- 190 .395. Aonnel hologl, 197'2. 23. 251 269 Rot \...41111 It t l 4414.414. \lFlit on w 0I out 0 t/./' 1/14 \ \. \\ an41 1'.11141II.1.The cit., h,,/and 1 Of 11111111(11 1. .1-100g1"0. \1,u II .01 '1/ lel °P°1"/ 0/°II ei:P °Ha.. "4.1"0°°. I)I \ Wen" Hell mid (01/1q1,111( t V....01( 1.111,011. 1901 ( an Gount d on F,duc ation. 19(19. "I huh. \ I 11,1 da 1.111141. \ (II Bat hr, \Iand \\ G. 13.1, 1010r1%ing per. 01 theIII. I P 11(11(1.!4. 1967.20, I 13 152 .4411,11 14.4, kgrointtl data and then relation...1mi to ot 1 111)a-

110,414I)I'and 01154r. R I Pit Ill, ling It mil/ 1.1,11 11144 111.11.11I 1.1IIII .111011. Journal of !palled /'s. 1')67 itr.1141tilral \ titltracto. i'si ;/. 11(3') hoteqa, 1971. 27. 125 128 11,1011r. \IE.. and \\ dham., (.. II. Prediction of -.ale-. -.net '41,1,1i, I\I. and INra..114 got 11 ei t(eat. alter the olleg Iron, at 1.40,111% drier lllll ted dinien.losa, pt.ronal hat k d.ptel !I mph, I went 111411,10 1,I SII1 1.11 SI 14 111 tIII ground data. Jonrool of ',pulled /'% holop, 1908. 52, 98.

-earth, \\ a,linntton. I) 1 1966 10.1.

G.I . \Ili, . I E and (.11 4141041, J tin); Baud. I,I,. 'Hie illation. 01 504 alinal triterr.t. to Itgoal.. iiirii.,%t IIII.444 .11110111.111 .1 olliationPr rsount Il's1 II rating. til alnlit% and pi roinalit% trait., and pottoitia hologl, 1%1. /O, 213 219 1.)r.all !nese:tatult. Joittno/ of Etlut (atonal I/ensure/new, 144e44(41. \\ I n,,unwound art:12nd 0414( lin ago\I 1970, 7, 23.1.2.19.

88 94) Ben lie, R. F. Intramidis idual s.11 .11 ht. and pi i 1..t1 1101.111, and Ihnlin II, .\I. I). I), .4101 lllll lit1,1 mod. t atm- azzonal and I's t. 1010g0 al 1/ 0511, no Ill.1961. 21. 661 at I.., 1111,1111, 16, pr du tom, iminagi :1.111 Ile, ils 676. 111'. Pont/Z(110 1/1p1/01PAChttlt/g%, 1967. 31, 10.61. Bin. R. F. Infra indi5illtialI, ilipmal .411(4110145 and pi. I lolland. J. and !In hard,. J \, ad, inn and 11011.14

11. tabilits .Echi (aroma 111,/ rst ho/op i/ 1/east, n/. 41, 11114 414 4 01111011111114 AltII 41 111111 1 111,1i1S1t1111.11ul '411 11)69a. 29, 233.237. dent, taking the\1111.I11.111 l .II,- g,-(..o//ege um/ 1 zu, Berdie. R. E. coly.s.tem s and generalt/a61111% 01 ultra in. :er.stil. 1967. 13, 60.71

11.110.11 sanabda.. Journal al 1/,/,/,/ t. 19691,. Ilorl I'.\ I,, 10annn6.1 161th s, 101/110 ill 111 .1 differ, 1111.11 .3.4. 33. II. l'sleltolaga alIlonographs. 1934. 68. Center... R. Thy psl ho/ogl ul so, ta/lassesl't involon. \ J. /.. 9. %%hole lllll 111 r 380.

et011 I. erth 1919 I \ for the des elopinctil ill .1 multiple 1.0o1e5 \\ and 1,0111ty... 1', 14Ilalizzarzare pro. lares Int hill 1,11 tilt 11.51140.1iil n. PA It holorgl, nlIlonograplis. 1911 the /what 'oral si fences.. 12111 ell 1 \ lork: \\ Ilys . 1971 4,9 \ 0. 5, ss hole Inindwr .590. 1 sonliaeli ,I.. J. f.mentin/s oI psi, /w/oizt a/ sesnu;,: 4.141 I'd I P.%%1 /10/1,,CII al nit (tsar, men: and ',refit( it,, lIchnzazt. lark: 1140 pl.: and Rnss. 1970. 41$1 . allswill. 1966

Dinnicur. \II).\ uloddu d n101111101 t d sandatlou and jorgt D.(1I'1, ,ii, 11111 ad pit th, 11111!11%.1r1 tion 11. Journal 0111,1,11111 I's te htzlzan. 1963. 17. .11I 41,1. Journal oilKtlaz trizzatalIlea suretneta, 1970. *7. 317,332. 171 181 Ihinteman. I.II. Organi,rat tonal t 01111111010. .11111 Lt 11.,. or III I) s debt uuuaul. 01 upt -441 m. l'erAw, 231 mama nal manula luring 01 gam/allow. Journal 01I1, nel holop. 1960. 13..373 391 plied l'st ho/ogl, 1966. ilk .100 301. I'., 1{0( L. I).\.. and KAMP... TI,, m.o of multiple England. "i ".4:1""1"PPI",111."1 1110111 l',1101' III at 4011'1110 pit tilt 11011. Journal 0f Et/W(1110/1d hitt/ILA ices, 4,11\111111, apoli,11111wd ial ( 1/ea smenient. 1968..3. 131 161) ins er.11% 1 \Iimtr.0la. 197 i louder, dome print iple. of mime.! inea.urement. Edn F 1ke, I).\\.iml I \!tura ilidmdual i i Ilonal and hultnaz ul 1/ea./nen/en/. 1970. 8). 203 226 311613. I'.0 holzazz«11 Halle:tn. 1933, 32, 217 210. I indtpi. F:./An ationn/ measurement \\ 1). Flanagan. I. 1. . et al. l'iojel tI \I\ I one seat 10110%,up \ men, an C0'11)(11011 Ellin at wn. 1931

ie 1%00 \11, I AIL, \i 1 \I Olin. \m. in an In Mali 11,.1.(.., and :Nmon. II.\Orgazaz:tazons 41"rk .fatal,'. tor Itc,earYli. 1966. Ilvs, 1918 Fl,u, hers 11.1,, and k. I)\.\ multiply mod, :atm. Jp- \I( \t mar, Q. \1..(14,1..160n ,1 4111110411.r.litir tet 11111(1111Journal proadi It, illi .11 o,1- 411d underal Ines 01 I'sIzologl. 1969. 33. 69-72 Juurrtul tti Elift atIonal lie,nnremenl, 1969, r,. 223228. (1%. \\\(dnnoii. J. IL. and \ILnght, I. E\ merican Freeberg. \ and Rut k. I)\Fat tonal 111%41141M a 04 s)1101;:n \ iallon tentili \11,i11,('ommittee. gtaplin al 1.14-6,rIlnlizzatzate I3rhu, iota) Researz h, 1969, 1)1% Yaon 11.I1(111110g 01 I rle hisita 1maz( ( 1, 191210 \ (:.: Emintlanon, 1966. 1,111e111.F. F..I he predn Iwo 01 prydn tabilit%P.du, ailtznal I'liti. R.. and Rit11. Iv ItIlezz5nt es of polztz and 1'51( holtzglz al Ileasurenzeta, 1960a. 20..4 8 t 1104-r.11,. oI Michigan :,01,,., Ite...ar, 11 ("en- (.111-4.11t. 1)11L.01111.16.,11 1 te.t in term, 01 Ilse 411(111.11 111.0616, for ."00 1.11111',..11411. 1969. V. all %SIM prl-fill11 .1is ru 11111s11111,11. /*AU ,'ism 111 Rw I.. I).\\ ppropriate method for tin ipta re, anal% and l'slz holog al Ilea sttrenzetzi. 19606, 20. 673 681 11ategort al Bala l's s bolugu 1971. 8 /. 1012 13 (..111.4111. \ Indeut mg elf t. and dilierntial 44 liallilit 14 k, 1)\ , Baud. I I and I inn. I.1 litera lion Ito %%eel! and salillirs Journal 01 Ipplol'slz haloes. 196.1. 17. 81 86 alh gzOft 4101 inzerzratz 1...thz«z- 1 on 1..a.F..F. lit.greaon anal% .1. 01 proloo twitalell data mutat Re eat, 11.10nrnal. 1972, 9. 119.161 PsIchtzlorza al lIztlleita, 1973. 79. 21.27 140 k. I)\Barone. J. 1... and Linn. 141..\ 6.1-0..111 computer J, F\ lulu% ariaw approa in sines data ply pngrani for 41 Milli( r,111'11 11'1,,A11 prctli Ed(' 141:1011 04 1,11 for, k11,1 (mtlinan ,orals ( 'atonal and l'slr ha/0gal 1/eauirerrin1, 1967.27. 709 711. and 1 multiple iegrt ..Mil .111411.111tollit al or:, 1,nm Ronan, \\aml 11. E l'erz,pe :arc on the measure me11101. Ileitis meta, II hat total R so art hIlanoprtiplos, wens hatzuzzi perfartrizazz r\, ss 101i (.1111111" \o, 67.1. 1967. I .rttf 1971

Markman. R.I..\\ \ and \R. l'n dit non 01 'mink I ) .( n ling 6.1. It re,110111 111 limy; term .1.,4 6.1.41 ssals it 1,.s 1101,g. K ti .11111ade ,tn h l'uhl ()atrium Quartz-HI )toned and 1,010/z al Iletzszaenzetzi. 1970. tit, 363.171 1967, 31.87.91, 96

89 Ward. J. H.. Jr. Hierarchical grouping to optimize an objective function.Journal of the American Statistical Association, 1963.58, 236.244. Wens. C. E.. and Linn. R. L. Causal assumptions in various procedures for the least squares analysis of categorical data. Psychological Bulletin.1971.75, 430-431. Zedeck, S. Problems with the use of "moderator" variables. Psychological Bulletin. 1971.76, 295-310. Section VII Baird, L. L., and Richards, .1, M., Jr. The effects of selecting college students by various kinds of high school achievement.ACT Research Report No. 23.Iowa City. Iowa: Ameri. can College Testing Program. 1968. Cronbach, L. J.Essentials of psychological testing(3rd ed.). New York: Harper and Row. 1970. Cronbach. L. J., and Cleser. F.C. Psychological tests and personnel decisions.Urbana. III.: University of Illinois Press. 1957. Dailey. C. A. The life history as a criterion of assessment. Journal of Counseling Psychology.1960. 7, 20-23. Hahn. M. E.Planning ahead after forty: The process of psycho- evaluation with selfstudy projects.Beverly Hills, Calif.: Western Psychological Services. 1969. Holland. J. L.The self - directed search.Palo Alto. Calif.: Con. suiting Psychologists Press. 1971. Hoyt, D. P, and Munday, L. A.Your college freshmen.Iowa City. Iowa: American College Testing Program. 1968. Lutz, S. W. Do they do what they say they will do? ACTRe- search Report No. 24.Iowa City. Iowa: American College Testing Program, 1968. Shannon. C. E. A mathematical theory of communication. Bell System Technical Journal, 1949.27.379-423. Tyler, L. T.The work of the counselor(2nd ed.). New York: Appleton-Century-Crofts, 1961.