Louisiana State University LSU Digital Commons

LSU Historical Dissertations and Theses Graduate School

1972 An Analysis of Contextual Effects on Electoral Behavior. Charles E. Grenier Louisiana State University and Agricultural & Mechanical College

Follow this and additional works at: https://digitalcommons.lsu.edu/gradschool_disstheses

Recommended Citation Grenier, Charles E., "An Analysis of Contextual Effects on Electoral Behavior." (1972). LSU Historical Dissertations and Theses. 2212. https://digitalcommons.lsu.edu/gradschool_disstheses/2212

This Dissertation is brought to you for free and open access by the Graduate School at LSU Digital Commons. It has been accepted for inclusion in LSU Historical Dissertations and Theses by an authorized administrator of LSU Digital Commons. For more information, please contact [email protected]. INFORMATION TO USERS

This dissertation was produced from a microfilm copy of the original document. While the most advanced technological means to photograph and reproduce this document have been used, the quality is heavily dependent upon the quality of the original submitted.

The following explanation of techniques is provided to help you understand markings or patterns which may appear on this reproduction.

1. The sign or "target" for pages apparently lacking from the document photographed is "Missing Page(s)". If it was possible to obtain the missing page(s) or section, they are spliced into the film along with adjacent pages. This may have necessitated cutting thru an image and duplicating adjacent pages to insure you complete continuity.

2. When an image on the film is obliterated with a large round black mark, it is an indication that the photographer suspected that the copy may have moved during exposure and thus cause a blurred image. You will find a good image of the page in the adjacent frame.

3. When a map, drawing or chart, etc., was part of the material being photographed the photographer followed a definite method in "sectioning" the material. It is customary to begin photoing at the upper left hand corner of a large sheet and to continue photoing from left to right in equal sections with a small overlap. If necessary, sectioning is continued again — beginning below the first row and continuing on until complete.

4. The majority of users indicate that the textual content is of greatest value, however, a somewhat higher quality reproduction could be made from "photographs" if essential to the understanding o f the dissertation. Silver prints of "photographs" may be ordered at additional charge by writing the Order Department, giving the catalog number, title, author and specific pages you wish reproduced.

University Microfilms

300 North Zeeb Road Ann Arbor, Michigan 48106

A Xerox Education Company 72-28,346

GRENIER, Charles E., 1930- AN ANALYSIS OF CONTEXTUAL EFFECTS ON ELECTORAL BEHAVIOR.

The Louisiana State University and Agricultural and Mechanical College, Ph.D., 1972 Sociology, general

University Microfilms, A XEROX Company, Ann Arbor, Michigan AN ANALYSIS OP CONTEXTUAL EFFECTS ON ELECTORAL BEHAVIOR

A Dissertation

Submitted to the Graduate Faculty of the Louisiana State University and Agricultural and Mechanical College In partial fulfillment of the requirements for the degree of Doctor of Philosophy

in

The Department of Sociology

by Charles E. Grenier B.A. Southeastern Louisiana University, 1956 M.B.A. Louisiana State University, 1960 May, 1972 PLEASE NOTE:

Some pages may have

indistinct print.

Filmed as received.

University Microfilms, A Xerox Education Company ACKNOWLEDGEMENT

The author wishes to express his sincere appreciation and gratitude to his Major Professor, Dr. Perry H. Howard, for his

Invaluable assistance and guidance in the research and writing of this dissertation. His interest and enthusiasm made this research project an enjoyable experience.

Appreciation is also extended to Professors Vernon Parenton,

Wilfred Holton, Neil Paterson, Miles Richardson, Virgil Williams and

Quentin Jenkins for their constructive criticism.

Special gratitude is offered to my wife, Virginia, and daughter,

Cheryl. It will be impossible for me to repay them for their patience and encouragement during the past four years.

The author takes full responsibility for all the shortcomings and errors that this study might contain.

11 TABLE OF CONTENTS

PAGE

ACKNOWLEDGEMENTS...... 11

LIST OF TABLES...... vl

LIST OF FIGURES ...... viii

ABSTRACT...... *

CHAPTER

I INTRODUCTION: THE STATEMENT OF THE PROBLEM, AND THE THEORETICAL AND METHODOLOGICAL FRAMEWORK...... 1

Concepts and Strategies...... 4

General Problem: Two Stages of Investigation...... 7

Political Ecological Analysis...... 10

Political Ecological Variables and Analytical Procedure...... 10

Factor Analysis of Latent Political Dispositions ...... 11

Cluster Procedure...... 12

Contextual Analysis...... 15

Survey Level Variables ...... 20

Dependent Variables...... 20

Independent Variables...... 21

II. LATENT POLITICAL DISPOSITION PROFILE ANALYSIS...... 22

Factor Analysis Procedure...... 23

Factor Analysis Results...... 25

Analysis of Factor Dimensions...... 30

Urban-Republicanism...... 30

Catholic-Moderation...... 32

Raclal-Radlcalism...... 35

H i CHAPTER PAGE

Validity Teat...... 38

Factor Structure Compariaon...... 39

III. DELINEATION OF POLITICAL ECOLOGICAL SUBREGIONS ...... 42

Parish Cluster Procedure ...... 42

Eight Political Ecological Subregions...... 44

Urban French ...... 46

Agrarian French...... 48

Piney Woods-Hllls...... 49

Black B e l t ...... 50

Urban North...... 51

Urban South...... 52

Emergent-Transitional...... 53

Orleans...... 54

Howard's Voter Types Revisited ...... 55

IV. CONTEXTUAL EFFECTS ...... 60

Variables...... 61

Contextual Effects Model ...... 63

Hypotheses...... 65

F i n d i n g s ...... 67

Voter Profile Analysis ...... 76

Causal Model of Environmental Effects...... 83

General Hypotheses...... 84

Nixon Model...... 87

Wallace Model...... 93

Humphrey Model ...... 95

V. GENERAL SUMMARY ...... 99

iv PAGE

APPENDIX I. PER CENT TABULATION OF CANDIDATE SUPPORT FOR THREE CANDIDATES IN THE 1968 PRESIDENTIAL ELECTION, CATHOLIC POPULATION, BLACK POPULATION, AND URBAN POPULATION, BY PARISH...... 108

APPENDIX II. CONTOUR MAPS OF SELECTED AGGREGATE LEVEL VARIABLES EMPLOYED IN THE LATENT DISPOSITION ANALYSIS ...... Ill

APPENDIX III. INTERCORRELATION MATRIX OF SELECTED PARISH LEVEL POLITICAL ECOLOGICAL VARIABLES ...... 143

BIBLIOGRAPHY...... 154

VITA...... 161

v LIST OF TABLES

TABLE PAGE

I Percentage of Total State Vote ...... 2

II A Rotated Factor Matrix of Selected Soclo-Cultural and Political Indicators Representing the 1960-70 Period.... 26

III Dimension I - Urban-Republicanism: Factor Loadings...... 27

IV Standardized: Factor Scores for the Rotated Factor Solution...... 28

V Dimension II - Catholic-Moderation: Factor Loadings .... 33

VI Dimension III - Racial-Radicalism: Factor Loadings...... 36

VII Rotated Factor Matrix of 36 Selected Gubernatorial, Presidential, and Socio-Cultural Variables ...... 40

VIII Urban French - Cluster I: Factor Scores ...... 47

IX Agrarian-French - Cluster II: Factor Scores ...... 49

X Piney Woods-Hllls - Cluster III: Factor Scores...... 50

XI Black Belt - Cluster IV: Factors Scores ...... 51

XII Urban-North - Cluster V: Factor Scores...... 52

XIII Urban-South - Cluster VI: Factor Scores ...... 53

XIV Emergent-Transitional - Cluster VII: Factor Scores...... 54

XV Orleans - Cluster VIII: Factor Scores ...... 55

XVI Least-Squares Analysis of Variance for Selected Individual Level and Aggregate Level Attitudes and Their First-Order Interactions ...... 69

XVII Candidate Choice Means for the Three Candidates for Black and White Voters ...... 70

XVIII Candidate Choice Means of the Three Presidential Candidates for Protestants and Catholic Voters...... 71

vi TABLE PAGE

XIX Candidate Support Means for Categories of Interaction Between Race and Variations in Minority Group Population Density (Brace) ...... 71

XX Mean Nixon Support by Race...... 72

XXI Mean Humphrey Support (Erace) ...... 73

XXII Candidate Support Means for Categories of Interaction Between Religion and Variations in Catholic Population Density (Erel)...... 75

XXIII Means for Interaction Effects of Individual Level Cate­ gories of Religion and R a c e ...... 76

XXIV Candidate Support Means for Categories of Interaction Between Race and Region ...... 80

XXV Candidate Support Means for Categories of Interaction Between Race and Residence...... 80

XXVI Mean Candidate Support for White Residents in the Eight Political-Ecological Areas...... 82

XXVII Intercorrelations of Variables for a Subsample of White Louisiana Voters in the 1968 Presidential Election...... 88

XXVIII Intercorrelations of Variables for a Subsample of White Catholic Voters in the 1968 Presidential Election ...... 89

XXIX Intercorrelations of Variables for a Subsample of White Protestant Voters in the 1968 Presidential Election .... 90

XXX Per Cent Candidate Support in 1966 Presidential Election and Selected Ecological Variables ...... 109

XXXI An Intercorrelation Matrix of Selected Aggregate Level Variables for the 64 Louisiana P a r i s h e s...... 144

vii LIST OF FIGURES

FIGURE PACE

1 Popular Support of National Democratic Party Candidates 2 as a Percentage of the Total Vote from 1932 to 1968 . . .

2 Analysis Flowchart...... 9

3 Latent Disposition Profile I: Urban Republicanism. . . . 31

4 Latent Disposition Profile II: Catholic Moderation . . . 34

5 Latent Disposition Profile III:Racial Radical ..... 37

6 Grenier's Political Ecological Sub-Regions...... 45

7 Louisiana Voter Type Areas...... 56

8 Hypothesized Relationships for a Generalized Conservative Candidate Choice Model...... 85

9 Nixon Candidate Choice Model...... 91

10 Wallace Candidate Choice Model...... 94

11 Humphrey Candidate Choice M o d e l ...... 96

12 Per Cent Candidate Support for Humphrey 1968...... 112

13 Per Cent Candidate Support for Nixon 1968 ...... 113

14 Per Cent Candidate Support for Wallace 1968 ...... 114

15 Voter Participation in the 1968 National Election .... 115

16 Per Cent Blacks Registered as a Ration to Total Registration 1968 ...... 116

17 Per Cent Candidate Support Edwards 1972 ...... 117

18 Per Cent Candidate Support for Goldwater 1964 ...... 118

19 Per Cent Candidate Support for Johnson 1964 ...... 119

20 Per Cent Candidate Support for Kennedy 1960 ...... 120

vill IGURE PAGE

21 Per Cent Candidate Support for Nixon 1960 ...... 121

22 Per Cent Support for States Rights Candidate 1960 .... 122

23 Proportion Catholic Population, 1971...... 123

24 Proportion Black Population 1960...... 124

25 Racial Political Equality Index: Based on the Relation­ ship of Eligible Registered Blacks to Whites In 1960. . . 125

26 Per Capita Income 1968...... 126

27 Change in Per Capita Income 1960-1968 ...... 127

28 Proportion of Families with Disposable Income Less than $3000 ...... 128

29 Proportion of Families with Disposable Income Greater than $10,000...... 129

30 Per Cent Unemployment 1968...... 130

31 Per Capita Public Assistance 1967 ...... 131

32 Education: Median School Year Completed by Persons 25 Years Old and Over, 1960 ...... 132

33 Per Cent Urban Population 1970...... 133

34 Proportion of the Population Classified Rural Farm 1960 . 134

35 Proportion of the Total Workforce in Manufacturing 1960 . 135

36 Proportion of the Total Workforce Classified as Laborer 1960...... 136

37 Proportion of the Total Workforce in Government 1960. . . 137

38 Population Change 1960-1970 ...... 138

39 Proportion of Population Under 15 Years ...... 139

40 Proportion of Population 15 to 65 Years Old ...... 140

41 Proportion of Population 65 and Over...... 141

42 Violent Crime Index 1967...... 142

Ix ABSTRACT

Previous political-ecological analysis of the Louisiana Electorate has demonstrated the existence of a persistent cleavage structure along socio-economic,religious, and racial lines. This dissertation takes these factors into account as an attempt to measure differential environmental effects on voting behavior. Focusing on the 1968 presi­ dential election, survey and aggregate data are combined systematically to link the individual voter to his iranedlate social context in an t explanatory model of electoral behavior.

A factor analysis of parish level data was conducted as a preliminary step to determine the State's cleavage structure, and three distinct socio-political dimensions were extracted and mapped geographi­ cally: (1) an incipient Urban-Republican base; (2) an emergent state-wide

Black coalition; and (3) a solid Cathollc-Moderation base in south

Louisiana. Parish factor score profiles were used as input in an hierarchical grouping procedure, and a set of eight political-ecological subregions were delineated, confirming Howard's Louisiana Voter Type Areas.

Least squares analysis of variance and path analysis techniques were used to test the effects of selected contextual variables in several three-candidate choice models.

In the least square analysis of variance the main effect of the voter's race was highly significant in the Humphrey, Nixon and Wallace models. Religion only approached significance in the Wallace model. Independent main effects of the racial and religious contextual variables were statistically Insignificant in all three models, but there was a significant first-order Interaction effect of voter's race and his racial context In the Nixon and Humphrey models. The racial composition of the voter's ismedlate environment played a role in the voter's decision to support Nixon or Humphrey, and the racial environ­ mental factor was more significant for Blacks than for Whites. A first- order interaction effect was found for race and religion at the Individual level in the Wallace and Humphrey models. Residence was significant only in the Nixon and Humphrey models, and a race-region interaction effect in the Humphrey model revealed that the racial difference in mean candidate support was greater in the north than in the south.

A causal model combined three exogenous contextual variables

(Black, Catholic, and urban), the respondent's perceived primary group support (PGS)» and his S.E.S. PCS was construed as an intervening variable between the three contextual variables and S.E.S., and controls were made for individual-level race and religion. The hypothesized causal directions in the three candidate-choice models were borne out. The voter's PGS variable had the strongest intact on candidate choice, followed by his S.E.S. and the three main contextual variables.

In the Nixon model, S.E.S. had a significant positive direct and indirect effect (through PGS) on candidate choice. The Black context had a low negative impact on the Catholic subsample and a relatively high positive impact for the Protestants.

In the Wallace model, S.E.S. was inversely related to candidate choice, directly and indirectly through PGS, with a stronger effect for

xl Protestants than Catholics. The Black contextual effect was positive for the Catholic subsample and negative for Protestants. The Catholic context had a relatively high, indirect, dampening effect on Wallace support.

The S.E.S. effect in the Humphrey model was negligible for the white subsample, as a whole, but the direct effect of class was negative for Catholics and positive for Protestants. The total Indirect effect of the Urban context on the white subsample was negative and relatively high. The Catholic subsample exhibited a relatively high negative effect from the Black contextual effect, compared to the low positive beta for the Protestant segment. Catholics had twice as much primary group support for Humphrey than Protestants.

The relatively high contribution of perceived primary group support to the total explained variance in the causal model tested in this study lends support to the notion that such indicators are an essential ingredient in a valid model of the political decision-making process.

In general, the contextual analysis substantiated the relative importance of the racial, religious, and economic factors Inferred in previous political-ecological research, even when such effects were not apparent at the aggregate level.

xii INTRODUCTION

For three decades the Louisiana Electorate has been the subject of extensive aggregate ecological analysis (Heberle, 1952, 1954, 1954;

Howard, 1971; Key, 1949; Sindler, 1956). More recently the Louisiana voter was the subject of a statewide survey analysis in connection with the 1968 presidential election (Howard in Kovenock and Prothro, forth­ coming) . Typically, aggregate and survey research are conducted indepen­ dently of each other, as has been the case in Louisiana, .and each approach has a different focus. Aggregate analysis aims at explaining the political behavior of "electorates,11 and survey research focuses on individual level political behavior. Focusing on the 1968 presidential election, this study combines both levels of information in an effort to link the individual voter to his immediate socio-political context, not simply to overcome the limitations of survey or aggregate analysis, but to systema­ tically take the social structure into account as an important factor in the explanation of voting behavior. Direct measurements of phenomena on different levels makes explanation more interesting and realistic, and there is good reason to believe that multivariate, cross-level analysis will become more common within the near future (Scheuch in Dogan, 1967).

Looking at the 1968 election results in terms of the long-run trend of national Democratic party disaffection, Table I and Figure 1 leave little doubt as to the direction and magnitude of the conservative realignment taking place in Louisiana politics. National Democratic 2 FIGURE 1

POPULAR SUPPORT OF NATIONAL DSOGRATIC PARTY CANDIDATES AS A PERCENTAGE OF THE TOTAL VOTE FROM 1932 to 1968

1001

50%..

1932 1968

TABLE I

PERCENTAGE OF TOTAL STATE VOTE

Combined Non-Democrat Year Democrat Republican Other Support

1932 93% 7% 0.3% 7.0

1936 89 11 0.1 11.0

1940 86 14 0.2 14.0

1944 81 19 0.2 19.0

1948 33 18 49.0 69.0

1952 53 47 0.0 41.0

1956 40 53 7.0 60.0

1960 50 29 21.0 50.0

1964 43 57 0.0 57.0

1968 29 23 48.0 71.0 Party support in Louisiana averaged 87% during the 1932-44 period, but in 1948 there was a dramatic reversal of national party support, marking the beginning of an emerging protest movement which culminated in a massive drain from the Democratic ranks in the 1968 Humphrey-

Nixon-Wallace election. In that election the Democratic vote was reduced to 29% of the total vote (with white voters contributing but

13%), the lowest since the turn of the century, compared to a combined

Republican-States Rights total of 71%. While the third party movement failed by a substantial margin at the national level, it was a predic­ table and striking success in Louisiana— when the State joined with only five others to give full support to the Wallace ticket.

Ecological analysis of voting tendencies have demonstrated the existence of persistent cleavages among the electorate on a regional basis which have been explained in terms of religious, racial, and economic factors. However, in view of the overwhelming Wallace victory, and the long-term conservative trend in the State, some of the tradition al polaritities become problematical, at least in terms of national level politics.

In 1968 we witnessed what seems to be a break-down in the tradition al French south Louisiana— Anglo Saxon north Louisiana cleavage. On the other hand, increased Black registration and participation became an important force behind the opposing liberal Democratic aspirant,

Humphrey, and this growing coalition made a significant impact in the

1972 Gubematoral election. Increasing industrialization and diversifi­ cation has altered the socio-economic structure of the State, diluting the impact of the traditional rural-urban polarity. The political ecology of the state is a complex and changing phenomenon!. This 4 dissertation attempts to explain certain aspects of the voter realign­ ment process by bringing the underlying social structure systematically into an electoral behavioral model, with the guiding sociological principle that the individual and his behavior cannot be studied in isolation from his social context. It presents a multi-level, multi­ variate strategy designed to combine survey data and aggregate data in a systematic way to investigate the problem of differential environ­ mental effects on electoral behavior of the individual. Logically the first step would be to map the social terrain— to discover the basic structure of the State's political-ecological domain. Multivariate analysis of aggregate level properties consisting of the racial, ethnic, religious and economic components was performed to (1) get an overview of the State's socio-political context, and (2) to provide a basis for generating hypotheses for the contextual analysis. The contextual analysis consists of several multivariate candidate-choice models which combine certain environmental variables and individual level variables aimed at measuring differential environmental effects on individual candidate choice.

CONCEPTS AND STRATEGIES

In spite of increased use of sample survey research since the early forties, the greatest proportion of studies of electoral behavior continues to rely on aggregate data. Furthermore, until recently (and for various good reasons), these two strategies generally have been used exclusively of each other. Scarcity of studies com­ bining both survey and ecological data in electoral analytical research can be attributed not only to their separate intellectual traditions, 5 but also to the fact that each has been able to Individually make signl- ficant contributions In the field and are valid approaches in their own right. No attempt will be made here to elaborate on the rationale and justification of straight aggregate or survey designs, but since a section of this study involves an analysis of cross-level relationships, a few words on research strategies is offered at this point as frame of reference. An important requisite for political-sociological research and sociological research in general is a sensitivity to the conceptual problem of levels and directions of analysis, and the general problem of aggregation and disaggregation. There are essentially four research strategies open to a political ecologist shown below (Dogan, Rokkan,

1969:3-11). He can focus on explained variance at the level of the individual, or at the level of the territorial unit.

FOCUS OF ANALYSIS

Interaction One of two Level Levels

Individual I III LEVEL OF DEPENDENT VARIABLE Territorial II IV Unit

He can limit his analysis to one level at a time, or he can combine properties of more than one level and focus on either indivi­ dual behavior or territorial behavior. This study focuses on Types II and III in the Table above. Aggregate data is used to delineate political ecological areas within a defined territorial unit (II), and second, individual level properties are combined with ecological variables (III) to explain individual level relationships in what can 6 be considered a "contextual analysis of electoral behavior". Type

III strategy focuses on the individual but tries to take his social milieu into account. Type IV focuses on aggregates or collectivities and tries to take the individual and other sub-categorical character­ istics into account. In both Type III and IV the focus is different and in opposite directions, which suggests a fifth kind of approach-- a systems strategy. Since political ecology can be subsumed under the rubric of human ecology, the problem of dealing with different levels of social phenomena and their interrelationships becomes an important consideration. If a goal of political ecology is to explore and analyze the interrelationships of political collectivities and individual voters and their larger socio-cultural and political environment, a meaningful research strategy would ultimately have to deal systematically with the problem of cross-level effects and sub-systemic relationships.

In that sense, the approach used in this study must be considered only a partial solution, and to a great extent exploratory. Neverthe­ less, it is hoped that it might provide some empirical foundation for further research. Conceptualization and operationalization of the kind of environmental effects conceived in this investigation are admittedly speculative. In this analysis, the social and political environment will be used synonymously with group level effects which can be conceptualized as any level of abstraction above that of the individual. Aggregate effects are made up of properties of individuals which describe aggregates, such as levels of education, disposable income, racial composition, racial attitudes, etc. Global effects, like aggregate effects, are subsumed under the concept of environ­ ment, but unlike aggregate effects they will be treated as properties 7 properties that exist for all units Independent of composition of the group and are not considered derivatives of properties of individuals, ie, population, urbanization, etc. (McKinney, Bourque, 1972: 230)

THE GENERAL PROBLEM— TWO STAGES OF ANALYSIS

The general orientation of this analysis is to investigate the social contextual dimension of Louisiana electoral behavior, and the analysis falls into two stages which correspond to Categories II and III in the above typology, The first stage is concerned with the general problem of measuring and mapping the State's socio-political contextual profile, and the second stage deals with the problem of analyzing contextual effects.

Problem 1: Aggregate data is used to extract a set of contextual dimensions from the general milieu, and to empirically delineate a set of political-ecological areas which can be used to test Howard's voter type areas, as well as to provide a framework for further analysis.

An effort was made to collect and analyze data that reflect the major social, cultural and political dimensions of the milieu, using multi­ variate techniques to extract and construct a latent political disposition profile of the State. Multi-variate techniques are used to test hypotheses, explore the underlying dimensions of the political process, and to derive indexes of such configurations for use in subsequent analysis. An Important part of this process involves the detection and identification of various hypothesized contextual dimensions, as well as mapping these dimensions for descriptive purposes. The mapping technique used is based on a gravitational model and avoids the constraints of conventional mapping techniques. 8

Problem 2: Using both survey and aggregate data, the second stage of the investigation is basically concerned with detection and measure­ ment of differential environmental effects on electoral behavior.

The data management aspect involved collating survey and aggregate data in such a way that each respondent is located in an empirically determined and measurable socio-cultural milieu. A limited segment of a potentially large number of possible hypothesized effects are examined, and several research strategies are presented for consideration.

The techniques used in the analysis are aimed at statistically measuring differential environmental effects, and several alternative cross­ level inference models are tested. Both strategies are variations of multiple regression analysis. First, a least squares analysis of variance and co-variance is used to measure and test individual level effects, environmental effects, and joint individual-environmental effects. And second, several cross-level causal models are tested, as an alternative, focusing on the relative strengths of direct and indirect contextual effects on candidate choice.

Analysis Flow Chart; The overall design in this investigation is summarized in the flow chart on page 9. Essentially there are two general stages of analysis, and two levels of information--aggregate

(ecological) and survey data.

Stage (A) utilizes ecological data exclusively, and consists of five steps, beginning with a factor analysis of parish level properties and ending with a political ecological areal analysis. Steps Al, A2, and A3 are the tasks of Chapter II, and step A5 is the topic of Chapter

III. 9

FIGURE 2

ANALYSIS FLOWCHART

Aggregate Data Survey Data Pariah Level Properties Individual Level Properties

(6) Collate Analysis Survey and Agg. Data Agg. Data

(2) (4) (7) (8) Factor Cluster Contextual Contextual Scores *3 Parishes Effects: Effects: for each by Score Parish Profiles Statistical Path Interaction Model Model

---- St---- ___ ^ r-. (3) (5) Contour Map Map of Political* Factors Ecological Area 8

V A B 10

The second general stage (B) deals with the problem of Isolating and measuring contextual effects. Survey and aggregate level information are combined (B6), locating each respondent In terms of his respective areal unit and the associated measured aggregate properties. Steps B7 and B8 represent two alternative approaches to the problem of analyzing differential environmental effects on electoral effects and appear In

Chapter IV.

ECOLOGICAL ANALYSIS

Starting from a collection of ecological variables which reflect

the state*8 socio-cultural, political and demographic make-up, the initial goal is an attempt to systematically extract from the total configuration a set of distinguishable contextual dimensions which could conceivably represent the political milieu in Louisiana. An index of standardized values for each dimension is derived and used to map

the dimensions separately to facilitate a descriptive analysis. The final stage of this phase of analysis consists of the recombination of the several contextual dimensions into a set of parish clusters in what shall be called a latent disposition profile analysis.

POLITICAL ECOLOGICAL VARIABLES AND TECHNIQUES

The goal at this stage of analysis consisted of (1) uncovering and measuring underlying socio-cultural and political distensions for the ✓ State as a whole, and (2) systematically cluster homogeneous parished

in terms of measured values of each parish's factor profile. A factor analysis and a hierarchal clustering technique is used, in that order, and are discussed more fully in the analysis section. The delineation of regional sub-systems is a prerequisite for a meaningful political 11 ecological analysis. The procedure outlined is highly amenable to replication, and it is hoped that the findings can serve as a base point for a temporal analysis of structural change.

Factor Analysis: The general areal unit of analysis is the State of Louisiana and the data are measured in terms of properties of the parish sub-units. Variables measured are empirically relevant to a political ecological study of Louisiana. In addition to the standard socio-economic and political variables and demographic indicators

(Allardt, Sweetser, Cox), several indexes are constructed to reflect certain structural characteristics peculiar to Louisiana. Following

Allardt (1967), a set of political indicators, such as candidate support and voter participation are incorporated as a separate component to measure the ideological milieu. The election results for the last three presidential elections are included to give the factor analysis a historical dimension.

The choice of variables is guided to a great extent by Allardt's study, "Cleavages in Finnish Politics," and Howard's Political Tendencies in Louisiana. Approximately 30 variables are used in the final analysis and are categorized below:

Election Returns: Percentage of total votes cast by parish for Kennedy, Nixon, States Rights, 1960; Johnson and Goldwater, 1964; Humphrey, Nixon, and Wallace, 1968.

Political Equality Index: a voter-registration index based on the relationship of eligible black to eligible whites for 1960.

Economic Equality: ratio of mean wages and salaries of blacks to those of whites, 1966.

Urban: per cent of persons living in urban areas, 1970.

Black: ratio of non-white to total population, 1960. 12

Catholic: per cent of total population reported as Catholic, 1971.

Age: percentages of total population 14 years and under, 15-65, and over 65 years, 1970.

Education: median school years completed by persons 25 and older, 1960.

Occupation: per cent of total labor force classified as labor; craftsman-operative; clerical-sales; manager- off ic ial-proprie tor ; crafts; professional and kindred, 1960.

Industrial Types: per cent of total labor force engaged in trade, manufacture, agriculture, and government, 1968.

Union Membership: per cent of total labor force who are members of organized labor unions, 1967.

Income: per cent of total families with disposable annual income of $3000 or less; per cent of total families with disposable income greater than $10,000, 1967. Parish per capita Income, 1968.

It is recognized that the validity of both dimensional analysis and cluster analysis are constrained by the nature of the variables

in the original matrix and by the criteria for their inclusion.

The third element of this phase of analysis consists of identifying

the basic dimensions produced by factor analysis. Variables included

in this analysis are hypothesized to conform to three main political-

ideological dimensions— racial, religious, and rural-urban. Several conventional criteria for selection and inclusion of the specific dimensions retained for analysis are used and explained in Chapter II.

Cluster Procedure: The final multi-variate analytical procedure

Involves grouping and mapping parish clusters in terms of contextual dimensions derived from the initial factor analysis. Each parish was assigned a profile of index values derived for each dimension in

the final factor matrix and a statistical clustering technique was used 13

which compares index score profiles for each of the 64 parishes and

progressively associates them into groupings in such a way as to

minimize an overall variation within clusters, and to maximize variation

between clusters.

Preliminary to the cluster analysis, each contextual dimension

derived from the factor analysis is mapped to facilitate a descriptive

analysis of spacial distribution of each separate dimension across the

State. The contour (or isoline) mapping techniques used for this

purpose are based on a gravitation model and can be viewed as a three

dimensional display. Ecological studies typically categorize observed

phenomena in terms of arbitrarily bounded territorial units, such as wards or parishes. The contour mapping technique used is intended to

provide a more realistic visual image of the way measured socio-cultural phenomena are distributed spatially across a given territory. A computer program used in this analysis was made available through the Laboratory

for Computer Graphics, Harvard University.

A contour map consists of closed curves known as contour lines which connect all points having the same numeric value or height. The

contour lines emerge from a datum plane at selected levels which are determined from the scale of the map and the range of data. Between any two contour lines, a continuous variation is assumed, which means

that only continuous information can be used as input, such as per capita

Income and per cent Black. The computation takes as given the coordin­

ates of a set of data points and values associated with them. The surface

is smooth (continuous), passes through the given points and represents

trends or patterns which those points imply. The method of calculation Is a weighted average of slopes and values of nearby data points.

According to the basic model, the value at a given point P would be a weighted average of the values at data points 1, 2, 3, 4,...n, with the weighting based on the Inverse square of the distance.

n 1 1 1 1 ^ ■ Zl ———- + ■ Z2 + »• • + — .— Zjj 1-1 (?±)2 (Pa )2 (P2)2 (Pn)2 Zp------n 1 1 1 1 + + + ----

■ ■ — 2 —2 — 2 — 2 1-1 (P1) (P1) (P2) (Pn )

Where P^ is the distance from P to data point 1

h is the distance from P to data point 2, is the value of data point 1 zi is the value at data point 2, etc., and Z2 Z is the computed value at the point P P

The finished product Ignores artificial boundaries and enhances a realistic visual Interpretation of the three contextual dimensions generated in this study. (Holberg and Cloyd, ASR. February, 1972,

"Definition and Measurement of Continuous Variation in Ecological

Analysis" for a technical application of a variation of the contour map approach.)

The analysis of the aggregate data basically procedes through three simple steps. First, a Factor Analysis is performed on the variables described above. Second, indexes for each of the derived dimensions are determined. Third, the index scores are used as input in the counter mapping and parish clustering techniques. 15

Factor analysis of data from areal unita has been used to delineate

regions for different purposes (Berry, 1966, for a review; Carl-Gunnar

Janson, Ch. 12, Dogan and Rokkan, 1969). It is used in this analysis

for both analytical purposes and for areai classification.

The factor extraction algorithm follows the principal axes method

and both orthogonal and oblique rotations are performed on the factor

matrix in the final solution. Factor scores for each dimension

retained for the oblique solution are generated and transformed to

exclude positive and negative values in order to meet the assumptions

of techniques used in subsequent analysis.

CONTEXTUAL ANALYSIS

This section of the analysis will deal with one of the central

ideas of political ecology— the influence of differential soclo-cultural

and political environmental configurations on electoral behavior.

This concern is Implicit in the political ecologist's Interest in

detecting and isolating political ecological areas. The ideal situation

for an ecological analysis of electoral behavior is one in which the

unit of analysis contains well-defined geographic subregions, with

reasonably distinct homogeneous soclo-cultural subregions. A second and

equally Important ingredient consists of empirical evidence of signifi­

cantly related patterns of political behavior regionally distributed.

This second ingredient has been thoroughly developed by Howard and

Heberle, mainly on the aggregate level of analysis. Aggregate research

of this nature is a valid methodology in its own right. Previous

ecological research provided the necessary background information to the

extent that It pointed out relevant variables, suggested hypotheses, and provided a framework for validation of the survey analysis. 16 Scholarly studies of electoral behavior have depended mainly upon two types of data— survey samples and aggregate level data In the form of election returns and census information. In this analysis we combine survey data and aggregate data in what is being called a contextual ecological analysis of electoral behavior in Louisiana. It is an

Investigation of the determinants of individual candidate choice in the

1968 presidential election controlling for differential environmental effects.

The contextual analytical approach in political ecology has been inhibited until fairly recently for various reasons (Scheuch, Linz,

Alker, in Dogan and Rokkan, 1969, Ranney, 1962), and are summarized as follows: (1) Exclusive aggregate and exclusive survey research continue to make valid and significant contributions. Aggregate data for political research is inexpensive and highly accessible and falls in the category of "hard" data. It is amenable to replication. Sample surveys are efficient and can accurately describe individual character­ istics of large populations and has high statistical reliability *

(2) There has been a general lack of conceptualization and operation­ alization of the notion of levels of sociological phenomenon and their interrelationships. (3) Development of appropriate research methods to handle multi-level, multi-variate analysis has been slow.

(4) Difficult to obtain good survey data that correspond to areas with high regional contextual variation. Ecological surveys are expen­ sive and time-consuming; and (5) There is a general logistical problem of collating survey and aggregate data even when both are available.

A contextual analysis, very broadly, is concerned with the detection and measurement of differential environmental effects on 17

Individual behavior. It Involves the explanation, specification, and interpretation of individual level relationships by systematically controlling for relevant aggregate and global level effects.

Explanation refers to testing for spurious relationships; inter­ pretation involves taking into account intervening variables in causal relationships; and specification is a process of determining relative strengths of direct and Indirect effects. Within this framework, an attempt is made In Chapter IV to construct and test some alternative multivariate, multi-level explanatory models of electoral behavior.

A basic concern of the political ecologist is the differential effects of various socio-political environments on electoral behavior.

But this is a complex process. The way in which a particular environmental factor influences specific behavior patterns may depend on other environ­ mental factors, on other individual characteristics, or on both. This concept is usually thought of in terms of interactions between variables conceptualized at several levels of aggregation. In a simply hypothetical case with two levels of aggregation a Catholic might be more likely to vote for (the moderate, Democratic Catholic) than a Protestant and, in addition, he might be even more likely to vote for him if he were living in a parish with an 80% Catholic population.

In this case, the candidate choice is dependent on an individual char­ acteristic such as being Catholic— plus the joint effect of being a

Catholic in a Catholic environment. Multl-varlate statistical techniques allows one to measure the separate effect of a number of individual level main effects and environmental effects, as well as the Interaction effects of different combinations of individual and environmental 18

variables. The number of independent variables of either level in

a given model is arbitrary and environmental level variables can

logically be of an aggregate or a global nature.

Interaction Effects: Over the past 20 years a number of analytical

innovations have been developed to put social structure back into the

picture (Barton, 1968, for a review). This study will rely principally

on the concept of statistical interaction as a means of isolating and measuring environmental effects and a least squares analysis of variance and covariance (LSAOV) is used in the initial analysis. And also, since

the concept of interaction is bound up with causal analysis, a path model is presented to test the direct and indirect effects of environ* mental effects.

The concept of interaction is Implicit in a large number of findings

in the sociological literature that give support to this approach. For example, Stauffer found evidence that while promoted soldiers were less critical of the military promotional system than non-promoted soldiers, criticism was greater among both promoted and non-promoted in units with higher rates of promotion (Stauffer, 1949). Berelson, Lazarsfeld, and

McPhee (1954; 100-101) showed that among friendship groups the percentage voting Republican Increased with the proportion in the group whose party affiliation is Republican, both for individuals who are Republicans and

for those who are Democratic. At the face-to-face level, Pelz (1956)

found a relationship between a scientist's performance and the frequency of contact with his colleagues, but the value orientation of partici­ pants, as an environmental factor, had an effect. Scientists who inter­

acted with colleagues that held similar values exhibited decreases in performance as contacts Increased, while Increased contact with colleagues of contrasting value orientation resulted in increased productivity.

Paris and Dunham (1939; 110-123), in their 1939 study of the ecological distribution of home addresses of psychotics in Chicago, found that certain psychosis rates were higher for Negroes living in white areas and whites living in Negro areas than for the same races when living in areas where they comprised the majority.

The concept of interaction is relevant for explanatory theory of differential environmental effects and the LSANOV technique used in this analysis facilitates a systematic analysis of such effects. An alternative approach to the analysis of environmental effects can be accomplished in cross-level Inferential path models, which, like LSANOV, is a form of multiple regression analysis. In the final section of the contextual analysis a path model is tested focusing on the measurement of direct and indirect effects of several hypothesized environmental level variables on candidate choice.

In addition to the high statistical utility of both the statistical interaction model and the causal model, these techniques have high descriptive qualities, especially path analysis. Also, path analysis forces the user to make assumptions explicit, however tentative, and accordingly makes any model highly amenable to constructive criticism.

A path model was included along with the interaction model, because these approaches are complementary and have a strong potential for future use in cross-level analysis. Among other things, according to

Sonquist (1970), it is possible to develop models where interactive raw variables are combined into single terms and included in additive causal schemes using regression techniques. And as an alternative to multiple linear regression, Wilber (1967) has demonstrated that causal 20 models containing Interaction terms may be couched In probability

statements of dependency. However, no attempts to go beyond conventlal path analysis are made in this study. It is used here as an explana­

tory device.

SURVEY LEVEL VARIABLES

The Dependent Variable: This study analyzes political partisanship in Louisiana within the larger social political framework of a democracy, which is tentatively defined here as "... a social mechanism for resolving the problem of cleavages and integretion with the minimal use of force and maximal emphasis of consensus of values" (Llpset: 165-92). Within the context of Western majoritarlan systems of political representation, it is theoretically feasible to consider the vote a significant political act— and is conceptualized here as "... an inherent disposition or inclination to move in a given direction" conditioned by the underlying social structure (Howard, 1971:xiii). It is essentially a dichotomous social act--a unit of behavior that reflects a decision to support or reject a given candidate or amendment. For the purpose of contextual analysis, a conservative vote will be interpreted as a manifest expression of a reslstence to change or a predeliction for the status quo anti. A liberal vote will have the opposite connotation. In terms of the specific candidates, it is reasonable to assume tentatively that Humphrey was the liberal candidate and that Nixon and Wallace were conservatives.

Electoral behavior provides the most convenient and precise measurement of underlying political alignments in the State and will be used here for the most part as the dependent variable. The dependent variable was dichotomized in order to make use of statistical techniques which 21 assume interval level measurement. Since the election Involved three choices it is necessary to have three dependent variables, which means

that is is necessary to have three versions of any model constructed in

the analysis. In the Nixon model, a vote for Nixon is assigned a value of one, and all others a value of zero; in the Humphrey model, a vote for Humphrey is scored as a one and all others a zero, etc.; and in the

Wallace model a vote for Wallace is assigned a one and all others a zero. Candidate choices for the 1960 and 1964 elections are also dichotomized. The fact that there is some dependency among the dependent variables is recognized and taken into consideration in the analysis.

Independent Variables: Individual level determinants consist of the conventional items, such as sex, race, religion, age, education,

Income, occupation, and S.E.S. A socio-economic status index comprises respondents' occupational prestige score, Income, and education.

A type of "significant other" variable was used in an effort to measure the perceived relative support of each of the three candidates by the respondent's primary group environment, consisting of friends, fellow workers, and spouse or family (if single).

Survey data for this project consists of the University of North

Carolina's 1968 Comparative State Election’s sub-sample of the Louisiana

Electorate, conducted within 10 days of the November IS presidential election. II

A LATENT POLITICAL DISPOSITION PROFILE ANALYSIS

The existence of historically persistent political tendencies

among the Louisiana electorate, particularly in terms of rural-urban

and religious and racial dimensions has been adequately demonstrated

(Howard, 1971). Howard's and Heberle's research makes it clear that

the political ecology of Louisiana is a complex, multi-dimensional

phenomena. Using these studies as a jumping-off point, one of the

initial goals of this analysis was to empirically isolate and measure

the underlying SQclo-politlcal dimensions that make up the State's

political-ecological structure and to describe the variations of these

dimensions geographically.

In order to facilitate interpretation of the spatial distribution

of these dimensions, each is depicted graphically with the use of

contour mapping techniques. The delineation of political ecological

sub-regions will follow in Chapter III. The measurement and description

of the multi-dimensional nature of the political ecology of the State, and the delineation of regional subsystems based on these dimensions,

is necessary for a meaningful and thorough analysis of political behavior.

Among other things: (1) It should enhance the explanation of the 1968 presidential election as well as future national level and gubernatorial

22 23 elections; (2) A graphic description of latent disposition profiles provide a basis for interpretation of potential voter alliances and cleavages in terms of the structural components; (3) It suggests hypotheses for further research and isolates Important variables; and,

(4) It provides a basis for successive longitudinal studies of struc­ tural change. For example, inclusion of gubernatorial election returns during the sixties was found to produce no significant alteration in the original disposition profile typology, or the voter type areas*

(See Hunter's "The Ecology of Chicago: Perslstance and Change 1930-

1960," ASR, 1971, for a factor analysis of changes in ecological structure.)

FACTOR ANALYSIS PROCEDURES

In terms of the nature of the problem, the factor analytical approach is logically the most appropriate solution, and it has prece­ dence in a number of human and political ecological studies (Allardt,

Cox, Sweetser in Dogan, 1967). In broad terms, the problem was to determine a set of variables that represent the soclo-cultural, political, and economic characteristics of the State, and reduce them to a smaller number of underlying factors that represent the basic structure of the domain under consideration.

The selection of variables for the profile analysis was based on the following criteria: (1) their relevance in terms of general voting behavior theory; (2) their relevance in terms of voting behavior theory peculiar to Louisiana's unique political history (Howard et al.

1969). In this connection emphasis was placed on the racial, religious, rural-urban and economic dimension. Two Indicators of racial Inequality were constructed, and updated Catholic population figures secured; 24 (3) Allardt's factor analysis of the Finnish political structure and

Cox's Investigation of contextual effect on voting behavior In London provided some good clues for relevant socio-economic and demographic such as occupation and industrial compositions, age distribution, unemployment, etc.; (4) the variables were Intended to reflect the

1960-70 period; and (5) the availability of the data, most of which was obtained from census reports and election returns.

The general procedure begins with dividing the State up into the

64 unit areas of observation (parishes), and for each of the unit areas values of the variables used in the analysis are recorded. These raw data serve as input for the factor analysis which first transforms the raw data matrix into a correlation matrix. Through an iterative process, subsequent calculations yield a matrix of factor loadings representing correlations with each of the factors extracted in the solution.

The next step involves the mode of computing factor scores, or indexes, for each of the derived factors to meet the assumptions of other techniques used in subsequent analysis. Factor scores are used in this analysis as a basis for the construction of contour maps as well as the hierarchial clustering technique in the regionalization analysis to be described in Chapter III. The scores for this analysis are computed for each data case as the regression estimates of each factor on all variables, normalized with a mean of 0 and a standard deviation of 1.

For various technical reasons the positive and negative values of the scores were eliminated by a transformation to an arbitrary positive scale with a mean of 50 and a standard deviation of 10 (Harmon, 1967;

358).

The factor score values indicate how each parish ranked in terms of each of the separate dimensions produced In the Initial factor analysis. They are determined roughly be weighting each variable for each parish proportionately to Its contribution to a given factor pattern.

In a crude sense they are the product of the values of the factor load­ ings for a set of variables and the standardized values of the corres­ ponding raw scores associated with each of the 64 parishes. The correla­ tion matrix, the factor analysis matrix, and the factor scores for each parish provide a powerful set of data for interpretation.

Several criteria were used in the factor selection process. The number of factors extracted in the initial principal component analysis was based on Kaiser's "eigenvalue-one" criterion, which retains factors with values greater than unity (Runmell, 1970: 362-63). Cattell's scree text (Runmell, 1970: 361) was applied to the unrotated solution, isolating four significant factors which contributed 69.0 percent of the total variance in the solution. Three of the factors correspond very highly with the hypothesized existence of (1) a rural-urban dimension, (2) a racial dimension, and (3) a religious, north-south cleavage dimension.

The fourth factor was apolitical, consisting of several transitory demographic and economic effects which were traced to the accelerated activity at an army base in Vernon Parish, and lack substantive meaning in terms of the theoretical problem at hand. The three political factors, explaining 60.1 percent of the total variance in the original solution, were rotated to an oblique solution for the final analysis.

FACTOR ANALYSIS RESULTS

The matrix for the rotated three-factor solution is presented below in Table II. Correlations between any of the three factors in the matrix are all less than .10 indicating that there is a very high degree of 26 TABLE II

A Rotated Factor Matrix of Selected Soclo-Cultural and Political Indicators Representing the 1960-70 Period

Variable Variable Urban Catholic Racial- Number Republi- Modera- Radical canlsm tlon

3 Humphrey 1968 -.049 .280 .793 4 Nixon 1968 .746 .018 -.049 5 Wallace 1968 -.408 -.272 -.692 9 Kennedy 1960 -.060 .948 -.048 10 Nixon 1960 .313 -.714 -.069 11 States Rights 1960 -.164 -.728 .119 12 Johnson 1964 .051 .861 -.066

18 Catholics .121 .903 .075 19 Blacks -.280 -.318 .796 20 Political Equality Index .026 .849 -.261 22 Economic Equality Index -.313 .090 -.445 30 Family Income Less than $3,000 -.719 -.382 .210 31 Family Income More than $10,000 .634 .374 .049 32 Urban Population .849 .048 .163 35 Median Education .788 -.386 -.262 37 Percent Manufacturing Employees . .176 -.031 .069 38 Percent Agriculture Employees -.700 -.082 .262 39 Percent Government Employees -.156 -.406 -.326 40 Percent Workforce in Trades .391 -.021 .116 41 Percent Workforce in Unions .605 -.009 .158 42 Per Capita Welfare Expenditure -.624 -.391 .106 46 Density .540 -.002 .245 47 Age Less than 14 -.103 .636 .329 48 Age 15-65 .347 -.223 -.435 49 Age Greater than 65 -.308 -.104 .169 50 Income Per Capita 1968 .679 .143 -.173 56 Percent Laborers -.758 -.044 .325 57 Percent Professional .655 -.391 -.141 58 Percent Managers-Proprletors .655 .391 -.319 59 Percent Clerical & Sales .853 -.045 .045 60 Percent Craftsmen .059 .343 -.341

Total Contribution of the factors to the variance of all the variables 39.0 34.6 18.4 TABLE III

DIMENSION I - URBAN-REPUBLICANISM

Factor Loadings

VARIABLES LOADINGS

Political

Nixon 1968 .746 Wallace 1968 -.408 Nixon 1960 .313 States Right8 1960 -.164 Kennedy 1960 -.060 Johnson 1964 .051 Humphrey 1968 -.049

Socio-Cultural

Proportion Clerical & Sales .853 Urban Population .849 Education .788 Proportion Laborers -.758 Family Disposable Income Less Than $3,000 -.719 Income Per Capita 1968 .679 Proportion Professional .655 Proportion Managers-Proprietors .655 Family Disposable Income More Than $10,000 .634 Per Capita Welfare Expenditure -.624 Proportion of Workforce in Unions .605 Density .540 TABLE IV

STANDARDIZED*

FACTOR SCORES FOR THE ROTATED FACTOR SOLUTION

I II URBAN- CATHOLIC- PARISH REPUBLICANISM MODERATION ______DIMENSION DIMENSION

Acadia 47 63 Allen 47 56 46 Ascension 51 63 47 Assumption 39 64 58 Avoyelles 43 56 48 Beauregard 51 50 41 Bienville 45 39 53 Bossier 62 42 47 Caddo 70 40 55 Calcasieu 65 56 47 Caldwell 42 42 46 Cameron 44 67 25 Catahoula 40 42 42 Claiborne 50 37 54 Concordia 48 42 55 DeSoto 45 41 63 East Baton Rouge 75 47 51 East Carroll 44 43 76 East Feliciana 42 37 57 Evangeline 41 61 48 Franklin 38 40 49 Grant 39 43 40 Iberia 56 63 53 Iberville 47 60 60 Jackson 56 43 46 Jefferson 74 56 45 Jefferson Davis 53 60 51 Lafayette 65 57 50 Lafourche 54 68 45 LaSalle 50 42 33 Lincoln 63 34 45 Livingston 44 51 28 Madison 44 40 72 Morehouse 53 40 55 Natchitoches 47 42 54 Orleans 76 50 68 Ouachita 66 40 51 Plaquemines • 50 42 38 Polnte Coupe 39 58 62 Rapides 59 47 51 Red River 36 37 Richland 40 41 Table IV (cont'd) I II III URBAN- CATHOLIC- RACIAL - PARISH REPUBLICANISM MODERATION RADICAL DIMENSION DIMENSION DIMENSION

43. Sabine 45 47 40 44. St. Bernard 66 56 38 45. St. Charles 56 64 50 46. St. Helena 32 48 55 47. St. James 48 67 66 48. St. John 49 67 62 49. St. Landry 43 61 58 50. St. Martin 39 64 53 51. St. Mary 59 61 56 52. St. Tamnany 56 52 44 53. Tangipahoa 46 48 47 54. Tensas 36 39 66 55. Terrebonne 60 65 49 56. Union 45 45 47 57. Vermillion 47 62 44 58. Vernon 50 41 20 59. Washington 51 49 48 60. Webster 60 40 47 61. West Baton Rouge 47 61 56 62. West Carroll 36 38 42 63. West Feliciana 43 39 60 64. Winn 48 43 44

Z Scores Standardized Around a Mean of 50 and a Standard Deviation of 10. 30 Independence between the three dimensions. The values in each of the

three columns are the factor "loadings" for the 31 variables in the

solution and measures the correlation of each variable with the three

separate dimensions and can be Interpreted as ordinary correlation coefficients. Each of the three factors are discussed in terms of

(1) the factor loadings in each dimension as presented in Table II;

(2) the factor scores which shew how each parish ranks on each dimension; and (3) the contour maps of each dimension. These maps were generated from the factor scores in Table IV and can be interpreted in terms of the five-level distribution of scores shown at the base of each map.

For purposes of interpretation, levels one and two can be considered low: level three, medium: and levels four and five, high.

A correlation matrix of all aggregate variables used in this study is provided in Appendix III, and a set of contour maps for selected variables in the factor analysis solution are presented in Appendix II.

ANALYSIS OF FACTORS

Factor I— Urban-Republlcanism: This factor held no surprises

(Table III). It correlates highly and positively with Nixon support in 1968 (.745) and all of the classical correlates of Republicanism, such as urban population (.849), education (.788), Income (.758), profession (.655), owner-managers (.655), and clerical-sales (.853).

It is a bi-polar factor correlating inversely with Wallace support in

1968 (-.408), welfare expenditures (-.624), and the poverty index (-.719) proportion of families with disposable Incomes of less than $3,000.

Column 1 in Table IV shows how each parish ranks on this dimension, keeping in mind that these indexes are normalized around a mean of 50.

The parishes that rank the highest are Orleans (76), East Baton Rouge 31 •—

IMII h2 +jlooo: 14604 0G00I )QO( 1 J+2( 166661 000002 J++-I 000001 ■♦♦♦♦I 10002 166641 ►♦♦♦♦I -♦+ + ♦ 2*00001 H++2+4 I0( ►♦♦ + ♦♦♦♦■( )003( L+++- J0002 + + +2-W** 1 • •/+ ++•< 1 ♦+++++++++++++++++^ ♦♦♦♦♦♦+2++♦+♦♦♦+♦^ ♦+++++++++^r*l}+++H ♦2+++++++++7T++++++- ♦♦♦♦ ♦♦♦♦♦♦♦ ^ 000000000++++++< 00000B«00q+ + < 30oooooooogk-++++^ 000000000000(^+42 + +H,______00000000004T+++++++++++++++i IDOUDt 0000<♦♦♦< 16697 Y3QT >+♦♦•<

•+« SYMAP

frequency distribution of DATA PQlINT values jn each level LEVEL 1 2 3 4 S sssustizsasaisnHatuaasausaHaaanaaaaauanaai ••••••• +♦++♦♦+♦+ OOOOOOOOO 666666666 666666666 ••••••• +++♦+++++ OOOOOOOOO 666666666 666666666 SYMBOLS •l**** ++++2++++ 000030000 666646666 666696666 ••••••• +++++++++ OOOOOOOOO 666666866 688666866 ••••••• ++♦♦+++++ OOOOOOOOO 666666666 888686866 FREQ* 12 25 . _i*_ 10 4

MAXIMUM IS*22 40*80 49*60 58*40 67*20 MAXIMUM 40*80 49*60 58*40 67*20 76*00

LATENT DISPOSITION PROFILE I: URBAN REPUBLICANISM

FIGURE 3 32

(75), and Jefferson as one would expect, and St. Helena (32), West

Carroll (36), and Red River (36) are among the lowest.

For a visual Interpretation of the Urban Republicanism dimension,

see Figure 3. The factor scores from Table IV were used to construct

a contour map of this dimension which produces a generalized effect of

the spatial distribution of the factor, and can be conceptualized a s

the latent disposition profile of the Urban-Republlcanlsm dimension.

It can be viewed as an elevation map with the darkest areas representing

the greatest density of the measured phenomenon and the lightest areas

reflecting the lowest. The darkest areas correspond to Orleans, Caddo, and East Baton Rouge Parish and represent the areas highest on this

dimension. Conversely, the light shaded areas are low on this dimension and can be viewed as the agrarian base of State. This dimension of

the State's political ecology also represents the latent Republican base for Louisiana. In general, this profile conforms to the hypothe­

sized distribution and makes the potential Republican base explicit.

It is particularly Interesting to see that medium-high score on the

Urbanism phenomenon had encroached Into the eastern Floridas and along

the Gulf, stretching from the Morgan City area to Lafayette and to

Calcasieu. The effects of increased economic activity associated with

Fort Polk brought Vernon Parish into the crescent shaped ridge connecting

Calcasieu and Alexandria. In the north the urban-republican base is reflected in the crest running from Caddo Parish to Ouachita. The parishes lowest on this dimension are Red River, Tensas, Franklin,

Richland and West Carroll, Polnte Coupe, St. Helena and Tangipahoa.

Cathollc-Moderation; This factor (Table V) can logically be

identified as Catholic (.903) and is associated with high loadings 33

TABLE V DIMENSION II - CATHOLIC-MODERATION

Factor Loadings

VARIABLES LOADINGS

Political

Kennedy 1960 .948 Johnson 1964 .861 States Rights 1960 -.728 Nixon 1960 -.714 Humphrey 1968 .280 Wallace 1968 -.272 Nixon 1968 .018

Socio-Cultural

Catholics .903 Political Equality .849 Age Less Than 14 .636 Proportion of Workforce in Government -.406 Proportion Professional -.391 Proportion Managers-Proprietors .391 Education -.386 Family Disposable Income Less Than $3,000 -.382

for Kennedy support In 1960 (.948), and Johnson support In 1964 (.861).

There Is a significantly high Inverse correlation with Nixon in 1960

(-.714). The moderation aspect of this dimension is reflected in the high loading on political equality (.849) and the high Inverse correlation with States Rights in 1960 (-.728) and Goldwater in 1964 (-.861). The relatively high loading on the fertility index (.636— proportion of the population less than 14 years old) corresponds with the high Catholic population. The moderately high Inverse loading on per capita welfare

(-.391) and education (-.386) is an interesting aspect of this dimension.

The factor scores In Table IV rank French Catholic parishes Lafourche (68), 34

, Jj ♦ • • • • • • • ! ♦ ♦♦ « • • • • 1****1 4 4 2 4 | * l* * * * * * * * ► 44-V********.** • • i v t ••••••••• I******* • A* • • ••••••••*••• »•« i«»m I • • «|2m • f ■ m •• • • • • • l * * ♦ 4 » a * • •••••••• *1*4424 4444** ••••< • •••#* *♦ ♦ ♦ ♦ ♦♦♦♦♦■♦V* i • • i ♦ ♦♦«» • • 1 • • • • • i ++ + «{•••• ••*/£+++ 2 + + + +♦+++(«•••, 4444► 44^* . . . * * £ 44444444444444^.., •4444444444444444444444441 ^44444442444444444424424444 K424444444444444444444442 V4 44444444444444444444444 ■ 444 44444 4 4 44 4444»QdiQ»444i \j>44 4 4 ♦ 44 44 4 4_2JOOOOOO ^4 4 4 42 4 4 4 4 4J0OQOOg0OQ00Qeeeeeeee • • • • *>44 4(00000000001 >000000908600000000 1 • • • 1 • 4 4(3000000030C >000000010000040000000© 44 ■♦•4440000000000 < )ooo3ooogte4eeeewee4eea 000 ♦ 44*0003OOOOOOOl >OOOOOOQ|60©0O0O*i|0000flfl000 44000000000000V )OooooM00000fl|i«flBeepiirte40 O3OOOOOQ0O3O( LoagsBBeeeeoeaBaeoeepMaaiiBBe 00< 14000080400005000 0( lOfiflfiepiiiiiiinQaeeelBaaniiiiaMsaMB »aaaaaaaaaaaaalBBfl©gasaaaaaaaaaaaaaa iaaaaBB5aaaaaaaaa5aiBiaaaaaaasasaaaaaaa IBBBBaBBBBaaaBBBBBBBBl laaaaaaaa i b b b b i

4 *

FREQUENCY DISTRIBUTION OF DATA PQJNT VALUES IN EACH LEVEL LEVEL 1 2 3 4 5 •«••••••• 4 4 4 4 4 4 4 4 4 OOOOOOOOO 0 0 0 0 6 0 6 0 0 aaaaaaaaa !«••••••» 4 4 4 4 4 4 4 4 4 OOOOOOOOO 000000000 aaaaaaaaa SYMBOLS ••••I**** 4 4 4 4 2 4 4 4 4 0 0 0 0 3 0 0 0 0 0 0 0 0 4 0 0 0 0 aaaasaaaa 4 4 4 4 4 4 4 4 4 OOOOOOOOO 000000000 ••••••••• 4 4 4 4 4 4 4 4 4 OOOOOOOOO 000000000 :ssas3ss=s; IS33S3333 3: I S S H » 3 t I 3 FREQ. 14 16 13 13 MINIMUM 34.00 40.80 47 * 6 0 54*40 6 1 * 2 0 MAXIMUM 4 0 * 8 0 47.60 54*40 61*20 68*00

LATENT DISPOSITION PROFILE II: CATHOLIC MODERATION

FIGURE 4 35 St. James (67), and St. John (67) at the top, and Lincoln (34),

Claiborne (37), and East Feliciana (37) at the bottom of this dimension.

The contour map of the Catholic-Moderation profile makes this dimension explicit (Figure 4). It conforms to the classic French

Louisiana demarcation except for Plaquemines Parish and the eastern

Floridas. Plaquemines is low on this dimension as a result of the right- wing conservative voting behavior under the Perez dynasty during the

1960's. The pattern of solid Catholic moderation base in the lower half of the southern parishes is interrupted only to some extent by the urban influences of Orleans and Lafayette.

Racial Radicalism— Dimension III: This is a racial-radical dimension

(Table VI) and gets its identification from the high loadings on black population and support of the left wing candidate Humphrey (.793) on one hand, and from the high negative loadings on Wallace support (-.672), the right wing candidate, and economic equality (-.445) on the other.

Referring again to the factor score on this dimension, it is no surprise to see Madison (72), East Carroll (76), and Orleans, (68) at the top. Among the lowest were Vernon (20), Livingston (28), and

Cameron (25), which can explain for the most part by their very low relative black populations— between 5 and 15 percent. For a visual interpretation of the geographical distribution of this dimension, see

Map (Figure 5). The darkest areas represent the high peaks of the racial-radical profile, and the lightest areas point out the parishes that are very low on black population and very high on Wallace support.

It is more diffuse than the other two profiles and to some extent concentrates in the parishes with high black populations.

These three factors are statistically Independent and each can be 36

TABLE VI DIMENSION III - RACIAL-RADICALISM

Factor Loadings

VARIABLES LOADINGS

Political

Humphrey 1968 .793 Wallace 1968 -.692 States Rights 1960 .119 Nixon 1960 -.069 John8on 1964 -.066 Nixon 1968 -.049 Kennedy 1960 -.048

Socio-Cultural

Blacks .796 Economic Equality -.445 Age 15-65 -.435 Age Less Than 14 .329 Proportion Laborers .325 Education -.262 Political Equality -.261

conceptualized as a separate dimension of a given parish's overall latent disposition profile. However, referring again to Table IV, the profile of any one parish is made up of the factor scores for each of three dimensions, which adds another dimension to a comparative analysis of the individual parishes. For example, northern urban parishes are high on the Urban-Republicanism dimension and below average on Dimensions

11 and III, while Orleans, for example is very high on I, average on II, and very high on III. Some Catholic parishes are low on I and high on both II and III and others are high only on Dimension II and low on I and 37

J. lOOOCOOOOOO )OOOOOOOOOfB6pO 1000000000 )00003000qB4pOi )OO3OO3OOOd0DOOOOOOOOaMOO2 flei^oooocooooooooooooooooOTooj e^goooooocooooajooooooooo 8989000000000000000030003d69< 999995900030000000000000 99999999€lp00000300000a00( I999996999€0000000000000003 999499949€000000000030000C 9999999999 000003Qflfl000000C 000001 lOOOOf )0000f j o o o a ot ioooool ♦4 0000000000 -«oa30ooaoooQ( Wooaoooooooac 0000000030L______ooocoooogg£eeeeeeeeee988bocd06fl0c 000000000^999994999499940000000301 OOOOO3OOQ99999499999990OuOOOOOOOOO> 00300000C 4999999999 00000300000001 ftJCUUUU OOOOOOOOOOI 999999999 ooff77%poooooooV 00000000000000000 999999949 ooW t♦ Vo o o o o o 3 o o n 0000000000000000000 9999999 00030000003000030000 99949 _00009<*Q£)000000000300 999999 4< ipooooooooooo 9999999 90j 30000000000 99999999 91 30030000001 99094909 00030f coooot >9999999991 OOOOOOOI ooooc 1499 o o o o o o o o o * 000001 00000003000( OOOOOOOOOOOOf 0030000Qf300f OOOOOOOl 00(

frequency distribution of data point values in each level LEVEL 1 2 3 4 5

• • • • • • • ♦ ♦ o o o o o o o o o 9 6 8 9 9 9 9 9 9 ■ ■ ■ ■ ■ ■ ■ ■ ■ • • • • • • * ♦ ♦ ♦ ♦ ♦ ♦ o o o o o o o o o 9 6 8 8 9 8 8 8 8 8 8 8 8 8 8 8 8 8 SYMBOLS • • • 1 * . • ♦ ♦ ♦ ♦ 2 + ♦ ♦ 0 0 0 0 3 0 0 0 0 868648889 ■B99SB999 • • • * * * • ♦ ♦ ♦ ♦ ♦ ♦ ♦ ♦ o o o o o o o o o 6 6 8 9 9 9 8 6 6 8 8 9 9 8 8 8 8 9 • • • •• • • ♦ ♦ ♦ ♦ ♦ ♦ o o o o o o o o o 6 9 6 6 6 9 9 9 6 8 8 8 8 8 8 8 8 8 FREO* 8 32 17

MINIMUM 20*00 31*20 42*40 53*60 64*80 MAXIMUM 31*20 42*40 53*60 64*80 76*00

LATENT DISPOSITION PROFILE III: RACIAL RADICAL

FIGURE 5 38 III. There is a pattern to the composite parish profiles and it is possible to cluster parishes on the basis of similar profiles. This will be the task of Chapter IV.

VALIDITY TEST OF THE LATENT DISPOSITION PROFILE ANALYSIS

To test the validity of the three contextual dimensions of the

State's political ecology recent gubernatorial election returns were used in a multiple regression analysis aimed at determining the statis­ tical significance and relative strength of each of the three dimensions.

Using Parish per cent candidate support for Edwards as the dependent variable, and the Parish index scores for each contextual dimension as the predictor values, the regression analysis yielded a coefficient of determination (R) of .864. This implies a very high relationship with the independent variables, and implies a very high level of explained variance (R^ * .745). Standardization of the raw regression coefficients indicates the relative strength and direction of each of the three dimensions for both candidates and substantiate the validity of the latent disposition profile analysis.

Urban- Catholic- Raclal- Candidate Republicanism Moderation Radical Support Dimension Dimension Dimension

Edwards -.225 +.814 +.235

Treen +.225 -.814 -.235

Edwards, the liberal Democrat, received the majority of his support from the Catholic-Moderation electorate, and also to a significant extent from the parishes high on the Racial-Radical dimension. As expected,

Edwards support was Inversely related to the Urban-Republican influence.

A similar analysis of Johnson (Conservative) support in the November, 1971 39 Democratic gubernatorial primary yielded standardized regression coefficients that followed the Treen pattern with values of +.477 on the

Urban-Republican dimension, -.504 on Cathollc-Moderatlon, and a +.149 on Racial-Radicalisms dimension.

These results were encouraging and lend some credence to the validity of the latent disposition profile analysis. It is expected that a regression of the Presidential candidate support in November, 1972 on these three dimensions will provide similar results.

A FACTOR COMPARISON

The original factor analysis solution (Table II) included a political component makeup of variables on election returns from presi­ dential elections, inasmuch, as the focus in this analysis was on the

1968 presidential election. A general model should include election returns from a series of recent gubernatorial elections along with the relevant socio-cultural and demographic variables. When election returns for gubernatorial elections from 1960 to 1972 were Incorporated into a subsequent factor analysis (see Table VII) the original factor structure was preserved, and the solution demonstrated very adequately the stability and validity of the original model (Grenier and Howard, forthcoming).

Focusing on the candidates, an examination of the factor loadings in parentheses reveals the predicable location of the various gubernatorial aspirants on the three dimensions of the original matrix. Johnson, the conservative, had a positive loading on the Urban-Republican dimension and a negative loading on Catholic-Moderation; Edwards and Morrison, the liberal Democrats, yielded high positive loadings on Cathollc-Moderatlon; and Lyons the right-wing conservative, had a positive loading on the

Urban-Republican dimension and a slightly lower on the Racial-Radical 40 TABLE VII

Rotated Factor Matrix of 36 Selected Gubernatorialt Presidential, and Socio-Cultural Variables

FACTORS

I II III URBAN CATHOLIC RACIAL REPUBLICANISM MODERATION RADICAL

Nixon 1968 (.75) .02 .03 Nixon 1960 .32 (-.76) -.02 Lyons 1964 (.58) -.28 (.53) Johnston Nov. 1971 (.49) (-.58) .30 Education, median (.83) (-.41) -.14 Clerlcal-Sales (.82) -.03 .10 Trade .34 .00 .06 Urban (.79) .08 .22 Income Per Capita 1968 (.71) .16 .05 Income More Than $10,000 (.64) .33 .18 Manuf ac tur ing .18 -.09 .15 Manager-Official-Proprietor (.74) .14 -.23 Professional (.65) -.37 -.08 Union Membership (.56) -.02 .22 15-65 Years of Age .41 -.17 -.35

Edwards 1972 -.27 (.88) .21 Edwards Nov. 1971 .15 (.83) -.12 Kennedy 1960 -.02 (.94) -.04 Morrison 1960 .13 (.92) .11 Johnson 1964 .09 (.84) -.05 Political Equality .10 (.84) -.26 Catholic .13 (.92) .08 14 Years Age & Less -.13 (.59) (.30) Craft smen-Operatives .17 .23 -.28 Government Employment -.11 -.37 -.34

Humphrey 1968 -.21 .32 (.79) Wallace 1968 -.26 -.31 (-.74) States Rights 1960 -.23 (-.68) .07 Black (-.45) (-.29) (.77) Laborer (-.82) -.02 .24 Welfare (-.65) (-.41) .01 Income Less Than $3,000 (-.77) (-.35) .08 Agriculture (-.75) -.07 .15 Economic Equality -.23 .07 (-.48) Violence .22 .26 .33 65 Years Age Plus -.06 -.14 .18 41 dimensions.

The latent disposition analysis presented in this chapter, along with the evidence that some validity can be attached to the findings, provides a good starting point for future analysis along these lines.

It is highly amenable to replication, and provides a basis for a

longitudinal analysis of the changing political ecological structure of the State.

For the purposes of this dissertation, the description of factors and mapping of factor scores accomplished the original goal of systematically measuring the political-ecological landscape, extracating the substantive contextual dimensions, and mapping the separate dimensions geographically to define the relative Impact of each. In the following chapter, these three contextual dimensions are used as a basis for systematically delineating an empirically based typology of political ecological sub-regions.

Finally, the findings presented here provide the necessary empiri­ cal foundation for the contextual analysis in Chapter IV. Ill

DELINEATING POLITICAL ECOLOGICAL AREAS

In Chapter II three socio-political dimensions of Louisiana's

"Cultural landscape" were extracted and mapped. The task of this chapter is to Identify and delineate a set of homogeneous political ecological subregions, based on the three dimensional latent disposition profiles derived in Chapter II. The overall procedure of delineating regions involved two steps. The first was the general problem of processing a set of raw data that represented socio-cultural, political, economic and demographic characteristics that describe the 64 unit areas, and then extracting a set of profile scores for each parish. The second step was to systematically identify and delineate a set of sub-regional areas.

In this analysis one of a variety of multivariate grouping techniques is used to deal with the problem of profile similarity. The goal is to systematically compare the latent disposition profile scores of the 64 parishes and progressively associate them into groupings in such a way as to minimize an overall estimate of variation with clusters.

PARISH CLUSTERING PROCEDURES

The computer program for this analysis was taken from Donald

Veldman's FORTRAN PROGRAMMING FOR THE SOCIAL SCIENCES <1967; 308-317) and was chosen because it is one of the simplest and most successfully applied (King, 1969: 165-125, for a review and application). The HGRQUP

42 program (hierarchial grouping) is classified as a centroid grouping technique and is based on the principle of the so-called 0 (distance) statistic used in statistical geography. The algorithm (See Veldman,

1967: 308-317; King, 1969: 194-204), simply put, is a type of iterative, step-wise, discriminant analysis which utilizes a generalized distance function based on the concept of error sums of squares, and determines the optimal grouping of units by maximizing the average inter- group distances between scores and minimizes the average intra-group distances. The solution does not necessarily yield contiguous regions when the analysis is used on a set of areal units as it did in our case.

The clusters are based on the similarity of a set of three derived factor scores on each of the 64 parishes. Since this technique treats each of the variables constituting the profile as equally important in determining the single distance index between pairs of groups the variable input should be standardized. Our transformed factor score profile meets this criterion.

The criteria for choosing the most desirable level of a given number of groups from the HGROUP analysis is based on (1) a type of within-group error value associated with each of the successive reductions in the step-wise grouping process, (2) a subjective inter­ pretation based on theoretical considerations relevant to the particular problem, and (3) a test of hypotheses. All three criteria entered into the group level selection process in this analysis. The nine level option was chosen to facilitate a comparison with Howard's nine-group

Voter Type areas. In the final analysis, an eight-level grouping solu­ tion was used for two reasons: (1) the nine-level solution produced only one insignificant change in the overall cluster configuration 44 compared to the eight-level solution, and this minor distortion would have detracted from a meaningful comparison, and (2) the statistical error, indicating the cost of a further reduction, would have been substantially large.

The delineation of the political ecological subregions for the 64

Louisiana parishes was derived in the following sequence: (1) A factor analysis was performed on 31 variables ascertained to reflect the political, social, economic, and demographic dimensions of the state's social structure during the 1960-70 period. Three factors were extracted— and identified— the Urban-Republican dimension, the Cathollc-

Moderatlon dimension, and the Black-Radical dimension; (2) A factor score was determined for each of the three dimensions (see Table IV) and assigned to the 64 parishes on the basis of their contribution to the original factor matrix; and (3) The 64 unit areas were grouped by the procedure (HGROUP), on the basis of each parish's three-factor profile score, resulting in essentially eight sub-regions. The clusters take on distinct regional patterns when cartographlcally illustrated

(see Figure 6).

EIGHT POLITICAL ECOLOGICAL AREAS

Except for a few isolated urban parishes and the "Emerging-

Transitional" subcategory, there is a high degree of congruity among adjacent parishes and the clusters conform remarkably well to Howard's

Voter-Type areas (Howard, 1971: 2). Beginning at the lowest level of aggregation the HGROUP analysis revealed that there is a definite regional "north-south" dimension in the state. At two levels, the state divides up almost exactly on the classical French-Louisiana boundary indicated by the heavy black line in Figure 6. At the next 45

HOWARD'S GRENIER'S POLITICAL

m f M E f s 8

I

URBAN FRENCH

a g r a r i a n f r e n c h

URBAN SOUTH

URBAN NORTH

PINEY WOODS-HILLS

BLACK BELT

EMERGENT-TRANSITIONAL

ORLEANS FIGURE 6 highest level this congruity Is preserved, except that the urban centers, north and south, come to the surface. At the 4th level the region labeled Black-Belt appears, and at the 5th level the so-called Emerging-

Transitional cluster breaks out, the only non-urban cluster that is not congruous. At the 6th level the Rural-Catholic cluster emerges out of the solid French-Catholic substructure, leaving what appears to be contiglous, crescent-shaped string of parishes we are naming the Urban-

French cluster. At the 7th level of aggregation, Calcasieu, Lafayette,

Jefferson, St. Bernard, and Orleans Parishes become a separate urban cluster, providing empirical justification for including East Baton

Rouge Parish with the Northern-Urban parishes. Finally, at the 8th level of aggregation, seen in Figure 6 , Orleans Parish buds off as a separate political ecological entity.

These eight political-ecological areas can logically be analyzed in terms of the factor score profiles that make up each sub-region.

The factor scores presented in the following series of tables are stan­ dardized around a mean of 50 with a standard deviation of 10, and vary roughly between 20 and 80. Like the individual parishes, these clusters have a factor score profile reflecting their strengths on the three dimensions. At this level of aggregation, the focus is on the mean factor score for each dimension as well as the overall pattern that distinguishes the clusters. The tabular presentation of parish factor scores shown in the following description allows one to see how individual parishes contribute to the overall cluster profile.

Urban-French— Cluster I; This group is made up of the 12 parish crescent-shaped areas surrounding Lafayette and the Rural-French cluster on the east (Table VIII). As expected, the paramount feature of this 47

TABLE VIII

URBAN FRENCH - CLUSTER I

Factor Scores

I II III Republican- Catholic- Racial Parish Urbanism Moderation Radical Dimension Dimension Dimension

1 Acadia 47 63 46 2 Allen 47 56 46 3 Ascension 51 63 47 5 Avoylles 43 56 48 20 Evangeline 41 61 48 23 Iberia 56 63 53 27 Jefferson Davis 53 60 51 29 Lafourche 54 68 45 45 St. Charles 56 64 50 51 St. Mary 59 61 56 55 Terrebonne 60 65 49 57 Vermilion 47 62 44 Rounded Means 51 62 49

cluster Is its high rank on Dimension II--Catholic-Moderatlon— and it

shares this distinction with the Agrarian-French cluster. However, it

is distinguished from the Agrarian-French cluster because of the

relatively higher scores on the Urban-Republican dimension and, also,

its relatively low score compared to the Agrarian-French on the Racial-

Radical dimension.

The urban character of the Urhan-French cluster is reflected in the modest positive loadings on high family income (.374) and proprietorship

(.391). The urban dimension is bi-polar and it suggests a slight

tendency for Republican support in contrast to the Rural-French cluster. 48

These factor score profiles should be Interpreted In terns of

their corresponding factor loadings in the original factor matrix on

page 28. The relatively high mean score for the Urban-Catholic cluster

on Dimension II relates specifically to the high positive factor loadings

reflecting Kennedy support in 1960, and a high Catholic population

(.903). This cluster can also be characterized as racially liberal in

terms of its high positive loadings on the political equality index

(.849). And it is not surprising that there is a significantly

high loading on the proportion under 14 years old index (.636).

Agrarian-French— Cluster II: Like its sister cluster, this

eight-parish cluster is very high on the Catholic-Moderation dimension

(Table IX). The inportant difference in its profile is its contrasting high score on the Racial-Radical dimension and its low score on the

Urban-Republican dimension relative to the Urban-Catholic cluster.

The relatively high average of the Agrarian-French cluster on the

Racial-Radical dimension— a standard deviation higher than the mean—

ranks it among the top three clusters on this factor with Orleans and

the Black Belt (an indicator of the historical presence of a plantation

economy). From a political viewpoint, this is an interesting and impor­

tant dimension. This cluster has a relatively high Black population, and the racial cleavage is reflected in the negative factor loadings

on politicAl and economic equality, and in the high Humphrey, high

Wallace, and low Nixon candidate support. The significance of the

Racial-radical dimension was made explicit in the path analysis of

contextual effects in Chapter IV, parltcularly in the Wallace and

Humphrey models. 49

TABLE IX

AGRARIAN-FRENCH - CLUSTER II

Factor Scores

I II III Republican- Catholic- Racial Parish Urbanism Moderation Radical Dimension Dimension Dimension

4 Assumption 39 64 58 24 Iberville 47 60 60 39 Pointe Coupe 39 58 62 47 St. James 48 67 66 48 St. John 49 67 62 49 St. Landry 43 61 58 50 St. Martin 39 64 53 61 West Baton Rouge 47 ~61 56 Rounded Means 44 '53 59

Pinev Woods-Hills--Clu8ter III: The distinctive aspect of this 13-

parish cluster is its consistently below-average scores on all three

dimensions- (Table X). It cannot be distinguished by any one dimension

as is the case in the other clusters. This cluster included most of the

traditional "Hill,, parishes where a farmer class predominated. While

black population has Increased in these areas in contrast to half a

century ago, for the most part the proportion approximates the state

average. While traditionally a bulwark of Democratic support in the

State as well as of Long support, since the advent of the second reconst­

ruction in the 1950's these parishes have tended to be among the leaders

of the Democratic disaffection. 50

TABLE X

PINEY WOODS-HILLS - CLUSTER III

Factor Scores

I II III Republlcan- Cathollc- Racial- PariBh Urbanism Moderatlon Radical Dimension Dimension Dimension

6 Beauregard 51 50 41 11 Caldwell 42 42 46 13 Catahoula 40 42 42 21 Franklin 38 40 49 22 Grant 39 43 40 42 Richland 40 41 51 43 Sabine 45 47 40 52 St. Tammany 56 52 44 53 Tangipahoa 46 48 47 56 Union 45 45 47 59 Washington 50 49 48 62 West Carroll 36 38 42 64 Winn 48 43 44 Rounded Means "44 ~45 ~45

Black Beit— Cluster IV: This 13-parish cluster ranks one standard

deviation above the mean on its most prominant factor--the Racial-

Radical dimension. These are the high black population parishes which

exhibited a strong Humphrey-Wallace polarization in 1968. In terms of

candidate support, these parishes are high on the economic inequality

loading (-.445), but in a negative direction because of the impact of

the contextual effect of these formerly plantation parishes. This

cluster is one standard deviation below average on both the Urban and

the Catholic dimensions. 51

TABLE XI

BLACK BELT - CLUSTER IV

Factor :Scores

I II III Republican- Cathollc- Racial- Parish Urban! sm Moderatlon Radical Dimension Dimension Dimension

7 Bienville 45 39 53 14 Claiborne 50 37 54 15 Concordia. 48 42 55 16 Desoto 45 41 63 18 East Carroll 44 43 76 19 East Feliciana 42 37 57 33 Madison 44 40 72 34 Morehouse 53 40 55 35 Natchitoches 47 42 54 41 Red River 36 37 56 46 St. Helena 32 48 55 54 Tensas 36 39 66 63 West Feliciana 43 39 60

Rounded Means 40 40 60

Urban-North— Cluster V ; This cluster is made up of all of the

urban parishes north of the French Louisiana boundary line including

East Baton Rouge Parish; which Includes a large proportion of racially

conservative white Protestant Anglo American migrants from north

Louisiana and Mississippi (Havard et al, 1963: 8). This cluster ranks

significantly high on the Urban-Republican dimension, about one and

one-half standard deviations above the mean. These parishes have

competed with those of the Piney Woods-Hills cluster for first place 52 in Democratic disaffection (Howard, 1971). It is about average on the

Black-Redical dimension and almost one deviation less than the mean on

the Catholic dimension.

TABLE XII

URBAN-NORTH - CLUSTER V

Factor Scores

I II III Republican- Catholic- Raclal- Farish Urbanism Moderation Radical Dimension Dimension Dimension

8 Bossier 62 42 47 9 Caddo 70 40 55 17 East Baton Rouge 75 47 51 25 Jackson 56 43 46 31 Lincoln 63 34 45 37 Ouachita 66 40 51 40 Rapides 59 47 51 60 Webster 60 40 47 Rounded Means 64 42 ‘49

Urban-South--Cluster VI; This 4-parlsh cluster (Table XIII) la

almost two standard deviations above the mean on the Urban-Republican

dimension and more than one-half deviation above on the Catholic

dimension which was about what we expected. A comparison of the factor

score means on the three dimensions with the Orleans means (Table XV)

explains why Orleans is considered a separate urban entity in the South.

Orleans is the highest on the Urban dimension but it is also the highest

of all clusters on the Racial-Radleal dimension, whereas the Urban-

French cluster is about average on the Black-Radical dimension. The

Urban-French cluster is slightly higher than Orleans on the Catholic

dimension. 53

TABLE XXII

URBAN-SOUTH - CLUSTER VI

Factor Scores

I II III Republican- Cathollc- Racial- Parish Urbanism Moderatlon Radical Dimension Dimension Dimension

10 Calcasieu 65 56 47 26 Jefferson 74 56 45 28 Lafayette 65 57 50 44 St. Bernard 66 56 38 Rounded Means 68 56 45

Emergent-TransItIonaI~-CIuster VII; These five parishes are not

contiguous but they have a great deal in common (Table XIV). This

cluster is about average in Dimensions I and II, but extremely low on

the III. This low score on Dimension III is explained by this cluster's

very low black population and high exclusive support for Wallace in 1968.

In addition to the two qualities they share with respect to the low

black population and a conservative voting history, they are all in a

state of transition which can be traced to: the increase in military

activity at Fort Folk in Vernon Parish during the 1960's (population

increase, per capita income increase, etc.); economic development in

Cameron; breakdown of the Perez dynasty in Plaquemines; and the move to

the suburbs reflected in the Livingston Parish economic indicators. 54

TABLE XIV

EMERGENT-TRANSITIONAL-- CLUSTER VII

Factor Scores

I II III Parish Republican- Catholic- Racial- Urbanlsm Moderation Radical Dimension Dimension Dimension

12 Cameron 44 67 25 30 LaSalle 50 42 33 32 Livingston 44 51 28 38 Plaquemines 50 42 38 58 Vernon 50 41 20 Rounded Means 48 49 29

1— l ■! . -I . L.LE-S ■ ■ ' in — — * ■ * * * , ! I. II ......

Orleans— Cluster VIII; This entity is distinguished by its unique

profile, with extremely high scores on both the Urban-Republican

dimension and the Racial-Radical dimension and the average score on the

Cathollc-Moderatlon dimension (Table XV). It shares the Urban-

Republican characteristics of the other urban cluster, but it is

different to the extent that it exhibits the strong bi-polar

characteristics of the clusters with high black populations. Politicians

have successfully built coalitions made up of blacks and whites in New

Orleans since the rise of Chep Morrison after World War II. In terms

of the 1968 presidential election, its pattern of support for the

three candidates was different from the other urban areas. Orleans gave 55 Humphrey a significant plurality, and the other urban areas— north and south--gave their support to Wallace predominantly.

TABLE XV

ORLEANS - CLUSTER VIII

Factor Scores

I II III Republican- Catholic- Racial- Parish Urbanism Moderation Radical Dimension Dimension Dimension

36 Orleans 76 50 68

HOWARD'S VOTER TYPES REVISITED

The final task of this chapter Is to make a brief comparison of the political-ecological subregions derived in the proceeding analysis with Howard's Voter-Type areas (Figure 7 )• Prof. Howard's Typology was derived from an historical-geographic analysis of land use patterns, and is based on the sociological assumption that differential land-use patterns produce differential social-structural configurations. The underlying assumption is that "the economy determines the economic class structure. . . and ultimately, political tendencies." (Howard* x:lll) 56

LoauUma Voter Type Areas

FIGURE 7 57 The Voter-Type areas (LVTA) proved to be a reliable frame of

reference for an extensive historical quantative analysis of the State's

political tendencies and its underlying structure has been confirmed

empirically by the political ecological sub-regions delineated in this

study.

A careful comparison of the two maps will show that the main differ­

ence is in the accentuation of several urban clusters in the sub-regions

derived in this study. Whereas Howard's typology has distinguished only

New Orleans as an important urban entity, three types of urban configur­

ations were determined significant in the present study: the Northern-

Urban, the Southem-Urban, and Orleans. Superimposition of these three

clusters on the Louisiana Voter-Type Areal (LVTA) map reveals this

basic distinction.

The second alteration is in connection with what is referred to as

the South Louisiana Bayou area in the LVTA map. A close inspection will

show that the areas derived in this study extend the South Louisiana

Bayou area into Southwest Louisiana to include St. Mary, Vermillion,

Acadia, Jefferson Davis, Allen, and Evangeline, connecting Avoylles

Parish. The South Louisiana Bayou area has been extended to include

St.Landry, St. Martin and Assumption.

The third major alteration consists of an accentuation of what is

referred to as the Black-Belt Cluster in this analysis, extending

influence into Howard's Hills Parishes. It modified the LVTA northern

planter configuration also by extending the North Louisiana planter

parishes along the Mississippi River to include East and West Feliciana.

The transparency in Figure 6 numbered one to nine, corresponds to

Howard's Voter Type Areas (Figure 7). With a few qualifications about 58 several underlying characteristics that distinguish the parishes, the superimposition of the LVTA profile reveals a very good fit between

the two typologies.

Taking the two French areas together, 14 parishes In Howard's

typology match up in this analysis. Seven of his Southwest Louisiana

Cluster became part of the Incipient Republican base.

Comparing the Florida parishes (4), the underlying factors that produced the variation in this area were race and urbanization. Over

50 percent of the population in the Black-Belt parishes are Black, while Livingston (Emergent-Transitional) has a Black population of less than 10 percent. East Baton Rouge parish was classified with the Urban-

North in the latent political disposition profile analysis. Analysis of survey data for the white subpopulation for the Floridas as a whole revealed a high degree of unanimity in terms of strong Wallace support, and for the most part these parishes take on the characteristics of the whites in the North Louisiana and Central Pine Hills areas.

There was a very good correspondence In terms of Howard's Central

Pine Hills area (8). The whites in this area gave relatively strong support to Wallace. Like the Floridas, the main factor that contributed to the variation within the Central Pine Hills area cluster was the wide racial difference between LaSalle and Vernon (Emergent-Transitional), and Natchitoches (Black Belt).

Several of Howard's Planter parishes (6) were determined to be urbanized in the profile analysis, but if they can be included for comparison purposes, nine out of the ten parishes corresponded with the Black Belt cluster derived in this study.

Thirteen urban parishes were delineated, eight in the north and 59 five in the south, and the latent disposition analysis demonstrated empirically that Orleans was sufficiently unique to remain a separate entity. IV CONTEXTUAL EFFECTS-MULTI-LEVEL ANALYSIS

INTRODUCTION

Using aggregate data exclusively, Chapters II and III dealt with

the problem of Identifying and measuring various dimensions of the

socio-political context of the State, and delineating a set of political-

ecological sub-regions. This chapter Is concerned with the problem of

Identifying and measuring differential environmental effects on voting behavior using (combining) both survey data and aggregate data.

The basic sociological assumption of this section is that an

Individual and his behavior cannot be studied In isolation from his social context. To paraphrase Linz (1967: 107), persons in apparently the same objective situation— in terms of race, religion, occupation— will think differently about their position and react differently, depending on their social context. Contextual effects for the purpose of investigation, will be defined In terms of such things as: whether or not an individual group member is in a majority or a minority In his imsediate community; to what extent he is Integrated into a rural or urban environment. In this connection, the analysis focuses on the three most significant components of the profile analysis in Chapter II— the racial, the religious, and the rural- urban. These and several other such environmental elements are operationalized In terms of ratio level measurement at the Parish level

60 61

of Aggregation.and are conceptualised here as analytically distinguish­

able components of the socio-political milieu. These selected, theory­

relevant variables are used In combination with survey data In an

attempt to systematically take the social context Into account as an

intervening factor In electoral behavior.

In addition to these more diffuse environmental effects, an attempt

will be made to measure effects of the voter's more immediate social

environment, In terms of primary group support. The significance of

primary group influence as a mediating effect on individual behavior is

well documented, and the importance of peer group influence as an

environmental factor in voting behavior has been recently demonstrated

by Shaffer (1971).

The research strategy used to Isolate and measure various

hypothesized environmental effects on candidate choice Involved two

basic analytical techniques, the statistical interaction model and the

causal model. The first model provides a calculus for Isolating and measuring interactions between variables conceptualized at different

levels of aggregation, and the second allows one to analyze direct and

indirect effects of environmental factors while controlling for spurious

effects. Both are explained below.

THE VARIABLES

The dependent variable at the individual level is the respondent's

candidate choice in the 1968 presidential election and in order to meet

the interval level measurement assumption of the techniques used

the vote was operationalised by dichotomising the voter's response as

either a one or a zero. Because it was a three-candidate race, it was

necessary to run three separate regressions for each model, and to 62

accomplish this each respondent is assigned a one or a aero for each

of three candidate choices dependent variables.

The Independent variables measured at the individual level

consisted of (1) a set of categorized personal characteristics, such as

race, religion, sex, occupation, etc.; (2) a set or ordinally measured

attitudinal responses related to the racial, economic, and law-and-order

issues of the period; and (3) several interval level attributes includ-

ing soclo-economic-status and primary group support.

Excluding the primary group support index, environmental level

variables are measured at the Parish level of aggregation and include

such properties as per cent black population, per cent urban population,

per cent Catholic population, per capita income, and per cent unemployed.

The variable used to measure primary group environmental effects was determined as follows (from Kovenock, forthcoming): each respondent

provided information concerning his perceived primary group support for each of the three candidates— his friends, his fellow workers, his

spouse if married, and his family (if unmarried). For perceived

friend^ support for candidate X he was assigned a value of zero for none,

1 for a few friends, 2 for some friends, 3 for most, and 4 for all

friends. For fellow workers perceived candidate support the same scale

is used. If his spouse voted for candidate X he is assigned a 4, and if

she did not, a zero. If unmarried, the score for perceived family candidate support for X is determined by assigning a zero if most of the

family voted for some other candidate, a one if he perceived that the

family was divided on this candidate, and a 4 if most of the family was perceived to support X. A mean absolute score for each of the three candidates for each respondent Is calculated by adding up the total values for each (which range generally from 0 to 12; 0 to 8 If unemployed) and dividing by 3 If employed( or 2 If unemployed. The relative primary group score for each respondent Is determined by subtracting the mean absolute score for candidate X from the mean of the mean absolute scores of the other two candidates and multiplying by 10. For example,

if a respondent's mean absolute score for candidate X was 4.0, and the average of the mean absolute scores for candidates Y and Z was, say

1.5; then this respondent was assigned a score of +25, (4.0-1.5*2.5x10).

CONTEXTUAL EFFECTS— A STATISTICAL INTERACTION MODEL

The investigation of contextual effects was accomplished through the use of least-squares multiple classification technique, a variant of ordinary multivariate regression analysis and analysis of variance.

Each regression model included variables conceptualized at several levels of aggregation— principally individual properties and environmental properties, and was aimed specifically at isolating and measuring

(1) the main effects of independent individual level variables on candidate choice, holding all other effects constant, (2) the independent environmental level effects, and (3) the joint (interaction) effects of hypothesized individual level variables and their related environmental level variables. Since, for the most part, the predictor variables are categorical, we are more realistically analyzing the main effect of each explanatory class separately and jointly, while testing for signifi­ cance. The dichotomized dependent variables can be conceptualized in terms of the binomial distribution and in effect represent proportion scales ranging from 0 to 100, so that if a catetory mean for candidate 64

choice in a given model Is determined to be 0.65, It Is appropriate to

Interpret this as the expected value of the dependent variable and can be

expressed as a percentage. The multiple classification analysis

solution to the models presented in this section are couched in the

regression format.

Yij - a + B ^ j + b2X tj + bOCjXjj) + e where Y^j is the expected value of the dependent variable of an individual

in the area j and having the value X^j of the independent variable.

In our case X j j could represent the category Black or White Race;

X could represent the proportion Black population; and the term

(Xj + X^j) represents the joint (interaction) of race and Black popula­

tion density. The constant "a" represents the mean (or intercept), and

the "b's" represent the regression coefficient for the separate predictor effects. The regression coefficients vary in terms of magnitude and sign.

It defines the nature of the relationship between a classificatory variable and the dichotomized dependent variable and represents essentially the profile of means (slopes) of Y (candidate choice) C v over the categories of X^ (classes of race). Specific empirical hypotheses are stated in terms of the related coefficients. The hypothesis that race is a determinant of candidate choice is B ■ 0, or

that the means of the subsets of scores of candidate choice for the

two categories do not differ significantly from each other. If the coefficient for the main effect race (and all other main and interaction effects) were 0, Ycc would simply equal the grand mean. If the variance ratio is statistically significant, then the conclusion would be that

B 4 0, which states that the means of the subsets are significantly different. The convential F value (variance ratio) is used to test significance for each effect in the model.

In terms of the basic problem of measuring contextual effects,

we are interested in joint effects, or the interaction of Individual

level variables and hypothesised intervening aggregate level variables.

Referring again to the regression equation above, these are cases where

the environmental variable has an effect on the relationship between

the individual level predictor and the dependent variable, aside from

its own main effect, which is the same thing as saying the individual-

level slopes of regression of Y-^j on X^ are not equal in all areas.

Following our example, this condition would reflect say the reality

that blacks in areas of high black population density vote differently

than blacks in all white environments, b(XjXij), separate from the fact

that blacks vote differently than whites for a given candidate. An

auxiliary goal in this section includes the construction of voter-type

profiles which include interaction effects when such effects were

significant in a given regression model.

HYPOTHESES

Inasmuch as it was necessary to generate three subsets of each of

three candidate choice models, and since in each regression there are

at least two main effects and one Joint effect, the potential combinations

of hypotheses make a detailed list unwieldy. Furthermore, this is an

exploratory study, and given the current state of theoretical and

empirical knowledge about relationships of the variables under

consideration (especially the interaction effects) any hypotheses would be highly tentative. In the path analysis section which follows,

it is possible to be more explicit about hypothesized multivariate 66 direct and Indirect effects. Nevertheless, in connection with the immediate problem, a few general hypotheses are made with regard to several key individual and aggregate level relationships in order to structure the analysis.

In terms of main effects at the individual level, it would be reasonable to hypothesize that there would be racial differences especially in terms of Humphrey and Wallace support, the left and right-wing candidates respectively. The importance of the racial dimension was apparent in the preceeding ecological analysis, and has support in other studies (Howard in Kovenock, forthcoming). In terms of religion, it would be expected that a moderate candadate would attract a Catholic constituency more than a Protestant one, a characteristic implicit in Factor II of the latent disposition analysis.

Since the Wallace support was so diffuse geographically, findings at the ecological level would suggest that regional hypotheses would have to be highly tentative. However, there was an apparent clear-cut urban-rural pattern in terms of aggregate level support for the three candidates, which makes It possible to hypothesize that the urban voters are more likely to support a Republican candidate, rather than either the more conservative or liberal candidates, all things equal. Support for the other two candidates was expected to be more highly related to racial and rural dimensions.

With regard to the second general level of hypotheses— the inter­ action effects of individual and environmental properties--it was expected that certain individual properties such as race and religion would interact significantly with certain contextual level variables

(black density, Catholic density, and industrialization). By way of a 67 general hypotheses for example, one would expect that the minority group candidate support for a liberal candidate would be higher in areas of high black population density than in lower areas, in addition to the main effect of race Itself. On the other hand, the dominant white would be expected to support the most conservative candidate in areas where the black-white ratio Is higher than in areas where the reverse is true, under the assumption that the dominant ethnic group would be less threatened economically or otherwise. The Catholic milieu would be expected to have some moderating effect in terms of candidate choice for all voters but is also expected to have a joint effect for Catholics living in Catholic areas, and Protestants living in Catholic areas.

"It makes a great deal of difference to the cultural and political behavior of an individual whether he or his household is in a minority or a majority position with a given social context: the Catholic tends to behave in one way in a predominantly Protes­ tant milieu, in another in a thoroughly secularised environment, and again in another in a homogeneous context of like-minded people; similarly, an office employee's behavior is quite different in a working- class district from his behavior in an upper-middle class area." (Dogan & Rokkan, 87)

The latent disposition analysis made Louisiana's strong regional contrasts explicit in terms of the racial, religious, and urban-rural dimensions, and the interaction effects tested are confined, for the most part, to these environmental factors.

In terms of general hypothesis testing the conventional .05 level of significance will hold in this investigation; however, several theoretically relevant relationships that approach significance at the

.10 level are also analysed for explanatory purposes.

FINDINGS

A variety of models of varying degrees of complexity were tested 68 in the preliminary investigation that were of interest, but for our immediate purposes a fairly simple and manageable model of selected racial and religious factors would be more appropriate and is offered as a point of departure.

A three-candidate choice model was subjected to a multiple- classification, least-squares analysis of variance and is summarized below. It consisted of three regressions with candidate choice for

Humphrey, Nixon, and Wallace as the dependent variable, and selected racial and religious factors as Independent variables.

The independent variables are abbreviated as follows for exposition purposes:

R ■ Individual's race

L ■ Individual's religion

7 - ERACE - the individual's racial context— the proportion black population in the respondent's parish of residence

L ■ EREL ■ the individual's religious context— the pro­ portion Catholic population in the respondent's parish of residence.

In the regression format used here, the notation R and L, It and E by themselves imply an independent main effect, and the notation (RxK) represents the interaction effect of Individual's race with his racial context. The racial and religious contextual variables are also abbreviated as ERACE and EREL, and these variables were dichotomized in terms of their mean percent scores to facilitate the analysis of interaction effects, for statistical and discrlptive purposes. The multiple classification analysis is summarized in Table XVI.

This table summarizes (1) the overall significance of each regression and the explained variance (R^); (2) the statistical significance of the independent main effects of individual and 69

TABLE XVI

LKAST-SQUARES ANALYSIS OF VARIANCE FOR SELECTED INDIVIDUAL LEVEL AND AGGREGATE LEVEL ATTRIBUTES AND THEIR FIRST-ORDER INTERACTIONS

Dependent Explained Significance Ratio for Selected Independent Variable Variance Variables; Main and Interaction Effects 1 2 3 4 5 6 7 8 9

Candidate R2 Proba­ Race Religion ERACE EREL Race Race Race Choice bility by by by of ERACE EREL Rel. F R L K r (RxK) (LxL) (RxL)

Humphrey .582 F-80.3 0001 .522 .212 .880 .102* .708 .005** 0001** **

Nixon .079 F-4.9 0001 .226 .941 .831 .044** .790 .869 0001** **

Wallace .306 F-25.4 0001 .095* .500 .989 .623 .913 .0178 0001** ** **

**Indicates Significance ^ .05 * Indicates Significance ^ .10 70

aggregate effects (columns 3-6); and (3) the statistical significance

of the interaction effects (columns 7-9). Overall there is high

significance in the regressions for this model in terms of the F values, and explained variance is high especially for the Humphrey and Wallace solutions. Fifty-eight percent of the variance in the

Humphrey solution is explained by the assumed linear dependence on the

set of independent variables R, L, £ and T, and also on the first-order

interactions of R and L with the intervening environmental variables

Ti and L. The multiple correlation coefficient of .761 for this regression suggests a strong degree of association among the variables defined in the model.

Race: Racial differences for all three regressions were highly significant (the means of the arrays of candidate choice for the two categories white and black are statistically different at the .0001

level) and are summarized for purposes of comparison in Table XVII.

TABLE XVII

CANDIDATE CHOICE MEANS FOR THE THREE CANDIDATES FOR BLACK AND WHITE VOTERS

Humphrey** Nixon** Wallace** Whites .122 .289 .588 Blacks .914 .059 .017

Religion: The main effect of religion only approaches significance in the Wallace regression and is not statistically significant in the

Humphrey and Nixon solutions, which generally refutes the notion that one's religion was an important factor in the 1968 presidential election.

Integration and civil rights were volitlle state-wide issues in the late sixties, and both public and parochial school systems were being 71 threatened, affecting both Catholics and Protestants alike, so this finding should have been anticipated. The unadjusted mean candidate choice for Wallace and the other two candidates are summarised below:

TABLE XVIII

CANDIDATE CHOICE MEANS OF THE THREE PRESIDENTIAL CANDIDATES FOR PROTESTANTS AND CATHOLIC VOTERS

Humphrey Nixon Wallace* Protestants .364 .185 .446 Catholics .316 .295 .387

Race Interaction: The main effects on candidate choice were significant neither for ERACE nor EREL, which was consistent with the aggregate level analysis. However, the interaction (joint) effects of race and ERACE (R x 10 was significant for both Humphrey and Nixon, and is summarized below in terms of unadjusted candidate choice means.

TABLE XIX

CANDIDATE SUPPORT MEANS FOR CATEGORIES OF INTERACTION BETWEEN RACE AND VARIATIONS IN MINORITY GROUP POPULATION DENSITY (ERACE)

RACEERACE HUMPHREY NIXON** WALLACE

WHITE Low .135 .259 .604 High .102 .333 .564

BLACK Low .871 .100 .014 High .978 .000 .021

The interaction effect in the Nixon regression, significant at the

.05 level, can be sumnarized further in the form of a contlgency table 72 and facilitates a more elaborate explanation of the contextual effect of ERACE.

TABLE XX

MEAN NIXDN SUPPORT BY RACE

ERACE High Low White .333 .259 RACE Black .000 .100

In the Nixon regression, not only is there a statistical difference between Black and White support, there is a statistically significant contextual (interaction) effect which is made explicit by a visual inspection of the mean Nixon support in the contingency table.

This means we cannot explain Nixon support solely in terms of a respondent's race; his decision to support this candidate also depends on whether he lives in a high or low ERACE area. In addition to being white, there is also a significant tendency for a white in high ERACE areas to support Nixon, and conversely a significant tendency for a black in a white (low ERACE) area to support Nixon. Looking at this from another perspective, ERACE by itself has no effect, but becomes significant in terms of the interaction effect it has when combined with the individual's race. Keeping in mind that there is no significance for the independent effect of ERACE, further elaboration of interaction effects can be accomplished by a visual interpretation of the difference which obtains between corresponding subsets of means in the table. For example, it is interesting to note that the differences between blacks and whites are greater in areas of high black population 73 density than the difference between the two races in areas where the black-white ratio is low.

Furthermore, the difference in the effects of ERACE is more significant for whites than blacks. Blafeks are relatively more affected by the white environment than whites are by the black environment.

Whites in high ERACE areas have a significant tendency to support the conservative Republican, while blacks in the Black areas completely reject the same candidate. On the other hand, it can be shown that when blacks are in high ERACE areas, they demonstrate a strong tendency for mutual support of the liberal Democrat Humphrey. A visual inspection of the interactions effects for the Humphrey regression summarized in the contengency table below makes this relationship explicit.

TABLE XXI

MEAN HUMPHREY SUPPORT

ERACE High Low White .102 .135 RACE _ . .978 .871

Again, race is significant by itself, but a full explanation should take the interaction effects of race and ERACE into account. Also, as in the Hlxon regression, ERACE by itself has no independent main effect, but becomes significant as intervening effect in combination with individual's race. Blacks living in high ERACE areas have more of a tendency to support the left wing Democrats than blacks in low ERACE areas, and the reverse is true for whites.

The fact that blacks in black areas are more likely to support the

liberal candidate probably can be explained in terms of primary 74 group support and group identity, but the white tendency to

support the same candidate in white areas is more difficult to explain.

All other things equal, it could be that they support Humphrey for

economic reasons, and since they are a dominant majority in their

respective geographic areas the consequences of increased civil

(black) rights would not be as threatening if their candidate was

successful.

Also, this interaction contingency table brings out the fact that

the difference between blacks in high and low ERACE areas is greater

than the difference for whites, which suggests a potential emerging

race consciousness on the part of the black electorate.

Looking at the differences between the two races in terms of high

and low ERACE, it is significant that the two races are more in alignment

on Humphrey support in the low ERACE areas than in the high ERACE areas.

This is a particularly interesting finding in terms of the problem of

identifying potential political realignments. It deserves additional

investigation in terms of the recent gubernatorial election, where the

liberal candidate, Edwin Edwards succeded in winning the contest by

effecting a coalesence of the hard core French-Catholic vote and the

crucial black vote.

The greater racial differences in high ERACE areas compared to low

ERACE areas could conceivably be due to the uneasy balance of power held by the dominant white minority in these areas especially in times

of civil rights agitation and potential federal intervention.

Religion by EREL Interaction: Referring to the summary table on

page the contextual effects of the Catholic environment variable was

not statistically significant by itself in any of the three regressions, 75 and the independent main effect the individual's religion was significant only in the Wallace regression (44 percent of Wallace supporters were Protestant, and 39 percent were Catholics.) A summary table of the interaction means for candidate choice by religion and

EREL is presented here to point out two potentially significant contextual effects. First there is an apparent tendency for Protestants in high Catholic areas to support Humphrey more than those in Protestant areas, and second, there is an apparent tendency for Protestants in low

Catholic areas to support Wallace more than Protestants in Catholic areas.

TABLE XXII

CANDIDATE SUPPORT MEANS FOR CATEGORIES OF INTERACTION BETWEEN RELIGION AND VARIATIONS IN CATHOLIC POPULATION DENSITY (EREL)

RELIGIONEREL HUMPHREY NIXON WALLACE

Protestant Low .337 .190 .467 High .526 .157 .315

Catholic Low .294 .308 .397 High .337 .283 .387

Race by Religion Interaction: A final interaction effect of interest in the model under discussion is the Individual race by individual religion effect on candidate choice, which, according to the summary table (XVI) is highly significant for both the Humphrey and the Wallace regression.

The interaction means table is presented below for consideration: 76 TABLE XXIII

MEANS FOR INTERACTION EFFECTS OF INDIVIDUAL LEVEL CATEGORIES OF RELIGION AND RACE

Race Religion Humphrey Nixon Wallace

White Protestant .067 .259 .672 Catholic .205 .333 .461

Black Protestant .934 .043 .010 Catholic .840 .120 .040

On the surface, referring the the above table, there is a significant tendency for white Protestants more than white Catholics to support Wallace, and black Protestants are more likely than black

Catholics to support Humphrey. A closer examination of the Wallace interaction effects reveals that the Catholicism has a mediating effect in the sense that there was less difference between the races for the

Catholics than the Protestants. And the same was true in Humphrey interaction. This supports the general assumption that Catholicism manifests a higher degree of cohesion than Protestantism, and for our purposes it lends support to the notion that the religions factor is an important component of the socio-political milieu.

A PROBABALISTIC VOTER PROFILE

The explanatory power of the proceeding least squares analysis of varaince can be expanded by utilizing the regression coefficients that define the relationships of the independent main and interaction effects with the dependent variable, candidate choice. Couched in the conventional regression format, it is possible to construct a probaba- listic voter typology that includes contextual effects when they are 77 significant, and several models based on the findings is presented here for consideration.

Using the Nixon regression in Table XVI as a point of departure, the analysis produced statistical significance for independent main effect of race (R) and the Interaction effect of race and ERACE (R x TC), and can be expressed following regression format

Y - a + b (R) + b(R x R) + E N where represents candidate choice, Nixon; R represents racial categories black and white; R represents high or low black population density; "a" represents the mean; and "b" represents the regression coefficient for each predictor. The standard error of the estimate E represents the confidence limit for a particular multiple regression prediction. In this particular regression for Nixon, the variance in

Yty is explained by the individual's race and also in addition, by the joint effect of race and ERACE. Applying the regression coefficients to the appropriate categories of each predictor in the model it is possible to typify the Nixon voter in a set of four possible configura­ tions as defined by the variables that were statistically significant

In the original regression.

LEAST SQUARES MEAN: YN - a + b(R) + b(R x K)

1. .2663 - .1923 + .1192 (1) + (-.0452) (1,1)

2. .3567 - .1923 + .1192 (1) + .0452 (1,2)

3. .1183 - .1923 +(-.1192) (2) + .0452 (2,1)

4. .0279 - .1923 +(-.1192) (2) + (-.0452) (2,2) The (1) and (2) in the R column correspond the whites and blacks effects respectively, and the (1,1) in the last column represent the effects of a white in a low ERACR area, and (1,2) represents the effects of a white in a high ERACE area, etc. This analysis elaboration can be considered an extension of the visual analysis of contingency tables in the proceeding section. It is much more explicit and takes two levels of effects into account simultaneously. The probability of a voter with a given set of individual and contextual characteristics appears in column one and is derived by adding the constant to the main effect of race and the interaction effect of race and ERACE. All things equal, the most likely to support Nixon was a white living in a high ERACE area with an expected mean of 35.67 percent. The next most likely supporter is a white in a low ERACE area (26.63 percent)— the effect of living in a low ERACE area had a dampening effect of -.0452. A black, living in a low ERACE area is third with 11.83 percent and a black, living in a high ERACE area is the least likely to support the moderate conservative candidate.

In the Wallace regression, race and the interaction of the individual's race and the individual's religion (R x L) were significant and the simplified corresponding model is as follows:

LEAST SQUARES MEAN: Yy ■ a + b(R) + b(R x L) 1. .6710 .2934 + .2658 (1) + .0632 (1,1)

2. .4472 .2934 + .2658 (1) + (-.0632) (1,2)

3. .1394 .2934 + (-.2658) (2) + .0632) (2,1)

4. .0000 .2934 + (-.2658) (2) + (-.0632) (2,2) 79 Being white had a positive effect and being a white Protestant had an additional positive effect In the Wallace candidate choice model.

This type was most likely to support the right wing candidate with a

67.10 percent probability, followed by white Catholics at 44.72 percent.

Least likely to support Wallace was a black Catholic (.0000).

The Humphrey profile Is more complex. Race, and the interaction effects of race and ERACE, and race and religion were significant, which multiplied the number of possible sub-type configurations to eight. The probabilities of Humphrey support, based on the above effects are summarized as follows, in rank order:

98.1 Black Protestants in Black areas 92.4 Black Protestants in White areas 86.9 Black Catholics in Black areas 81.3 Black Catholics in White areas

For the whites:

21.8 White Catholics in White areas 15.7 White Catholics in Black areas 10.3 White Protestants in White areas 5.6 White Protestants in Black areas

Taking into account the main effects of race, and the interaction effects of race and ERACE, and the individual level interaction effect of religion and race, there was significant tendency for Black Protes­ tants in Black areas and White Catholics in White areas to support the

liberal Democrat in 1968. The least likely supporters were the Black

Catholics in White areas and White Protestants in Black areas.

It would be quite possible to test the validity of such voter type models against actual election returns if the model were based on

individual level socio-cultural characteristics that had a counterpart

in census information at the precinct level, and could very well be used as a prediction model. To conclude this section, the following summary tables of mean candidate support, in terms of the interaction of racial, urban-rural and north-south categories are presented here as a point of reference.

TABLE XXIV

CANDIDATE SUPPORT MEANS FOR CATEGORIES OF INTERACTION BETWEEN RACE AND REGION

RACE REGION HUMPHREY** NIXONWALLACE

WHITE South .170 .296 .533 North .059 .265 .675

BLACK South .900 .066 .022 North .961 .038 .000

TABLE XXV

CANDIDATE SUPPORT MEANS FOR CATEGORIES OF INTERACTION BETWEEN RACE AND RESIDENCE

RACERESIDENCE HUMPHREYNIXON**WALLACE

WHITE Urban .145 .365 .489 Rural .097 .150 .752

BLACK Urban .910 .871 .015 Rural .922 .057 .020

These summary tables clearly show how the candidate support was distributed in terms of the several categories of residence and region and race. Aside from the obvious racial bi-polarity, there were several significant main and interaction effects for race, region, residence. The main effect of residence (urban-rural) was significant at the .01 level for both Nixon and Wallace, which was consistent with aggregate data findings. Wallace received substantial rural support and Nixon the urban support. The cross-classification tables above show the tendency for Wallace support to come from whites in the north and whites in the rural sector. Nixon's support was significantly urban for the white sector, but there was no significant residential effect for blacks. The tables indicate that Humphrey support among the whites tended to come from the south. Also, his support was split in terms of the urban rural categories, with a slight edge on the urban side, owing to the relatively high white support in the Orleans and

Urban South clusters (Table XXVI). The race-region interaction effect for Humphrey further specifies that there is a statistically significant difference between the races and candidate support by region. The black-white difference in the north is greater than in the south. There was more of a tendency for whites in the south to support Humphrey than in the north, and the white voters were less unified regionally than the black vote, which came as no surprise.

A breakdown of the candidate support by parish-cluster for the white subsample reveals that Humphrey support was strongest in the

Orleans and Urban-South clusters, and as predicted, very low in the

Emergent-Transitional, Black-Belt and Urban-North clusters (Table XXVI).

Nixon support by the white sector was highest in the Orleans Area, and he did better in the Urban-South than in the Urban-North cluster, where interestingly enough the black population is the lowest in the State, followed by the Agrarian-French, Plney Woods-Hills, and the Urban-North clusters in that order.

The inordinatly high Wallace support by the Agrarian-French was 62 TABLE XXVI

MEAN CANDIDATE SUPPORT FOR WHITE RESIDENTS IN THE EIGHT POLITICAL-ECOLOGICAL AREAS

PARISH CLUSTER HUMPHREY NIXON WALLACE

Urban French .100 .325 .575

Agrarian French .125 .125 .750

Plney Woods-Hllls .107 .178 .714

Black Belt .031 .312 .656

Urban North .081 .225 .662

Urban South .288 .311 .400

Emergen t-Tradit iona1 .000 .222 .778

Orleans .200 .457 .342

Total Whites .130 .280 .590

somewhat of a surprise, especially when compared to the relatively low

support by Its sister cluster the Urban-French. However, It can be explained, at least on the surface, In terms of the latent disposition profiles I and III. The Urban-French cluster context Is distinctly

Urban-Republican, and the Agrarian-French context Racial-Radleal. The

Urban-French are part of the emerging Republican base as reflected in

the profile analysis, and its corresponding support for Nixon helps

explain the difference between the two clusters.

The difference in Wallace support between the two clusters might

be due to the changing composition of the workforce attracted to the

two areas. The industrial complex along the Mississippi River had a

larger influx of the conservative element from nearby northern clusters

and Mississippi. A CAUSAL MODEL OF ENVIRONMENTAL EFFECTS 83

This section employs the use of a causal model scheme as an alternative approach to the problem of measuring environmental effects on electoral behavior, still focusing on the 1968 presidential election.

Since this is an exploratory study, the path model presented here will be used simply as an hypothesis testing device, and is aimed at examin­

ing the relative impact of two levels of environmental influence on the presidential candidate choice process: (1) the more diffuse effects of the aggregate level racial, religious, and urban contexts, and (2) the voter's more immediate soclo-psychological environment, in terms of his perceived primary group support for his candidate choice. In this section the focus is on investigating the direct and indirect effects of these environmental factors, while controlling for the relevant individual level characteristics. The main concern here is not so much with statistical significance, but with the isolation of specific environmental

factors and the measurement of the independent, relative impact each has

on the endogenous variables incorporated in the model.

Figure 8 is conceived as a recursive, asymetric, multi-level model of individual voting behavior. As in the preceeding section, three regressions were determined, one for each candidate choice (Humphrey,

Nixon, and Wallace). Also, since the racial and religious environmental variables are considered aggregative, (made up of the properties of the subjects under investigation) it was necessary to control for race and religion. The individual level race effect was controlled-by restricting the analysis to sub-samples of white Catholics and white Protestants.

The general procedure, then, was to determine the path coefficients for each of the separate candidate choice models, for subsamples of whites, white Catholics, and white Protestants, making a total of nine regressions 84 (three for each candidate choice model).

In order to facilitate a visual interpretation of the three separate

candidate choice models and their three corresponding subsamples,

it was decided to enter the path coefficients for the three subsamples

in each of the general candidate choice models. Each path connecting any two variables contain three path coefficients: the first represents

the all white subsample effects; the second, the white Catholic sub-

sample effects; and the third, the white Protestant subsample effects,

in that order, and will appear as follows: .252. (.145). (-.2341.

This format provides a reasonable efficient way of looking at the

relative differences and magnitudes for each path, and for each sub­

sample, in succession.

THE GENERAL HYPOTHESES

The model presented here was based on the premise that the explana­

tion of voting behavior is more meaningful when the individual's immediate

social context is taken into account. And to that extent, the main

concern here is with the problem of isolating.and measuring differential

contextual effects on voting behavior. Referring to Figure 8 the assumption is that individual candidate choice is to some extent a

function of (1) the racial, religious and urban-rural environmental

factors, (2) his socio-economic status, and (3) the primary group

support for a given candidate. It is also hypothesized that the

individual's perceived primary group support is conditioned by his

immediate environment, in terms of the racial, religious and urban

contexts, as well as his social class.

Controlling for race and religion, it is hypothesized that there is

a substantive difference in environmental effects between the white HYPOTHESIZED RELATIONSHIPS FOR A GENERALIZED CONSERVATIVE CANDIDATE CHOICE MODEL

cc(c)

C Per cent Catholic population

B Per cent Black population

U Per cent Urban population

S Respondent's S.E.S.

PGS(C) Respondent*8 perceived primary group support index

CC(C) Candidate choice: Conservative

FIGURE 8 86 Catholic and white Protestant subsamples. Technically, In addition to testing the null hypotheses of no environmental effects, we are testing the null hypotheses of no difference between subsamples. Since this

Is a three-candidate choice model Involving subsystems of relationships, no attempt will be made here to specify all of the possible alternative hypotheses. However, referring to the model, the positive and the active signs shown in the causal scheme will facilitate the elaboration of a general set of hypotheses as a point of departure. Rather than assume a specific candidate choice model with Humphrey, Nixon, or

Wallace candidate choice as the dependent variable, the hypotheses can be stated in terms of a "generalized" right-wing conservative candidate.

That being the case it is reasonable to assume that there would be a negative relationship between Catholic context and CC(C) based on the latent disposition analysis in Chapter II. (Catholic-Moderation).

Based on the assumption that whites in a numerical minority would vote conservatively, a positive relationship is hypothesized between Urban context (U) and CC(C). Individual primary group support PGS(C), based on the same assumptions, is hypothesized to be related to the three contextual variables the same way. The relationship between PGS(C) and

Respondent's SES should be positive, and Respondent's SES should be positively related to the urbanization variable.

The postualted causal relations among the variables are represented by undirectional arrows. The two-headed curvilinear arrows between the exogenous variables represent the non-causal, or unanalyzed, correlations.

The residual paths are introduced to account for the variance unexplained by the measured anticedent variables, and are represented by undirec­ tional arrows. The residual variables are assumed to be uncorrelated with the determinants of the varlabLes under consideration. The values 87 entered on the causal paths are standardized regression coefficients; the values entered on the curvilinear arrows are zero-order correlations; and the residual paths represent the coefficient of alienation

( 1 “ R2 )• The correlation matrixes for the three candidate choice models and their respective path diagrams are offered here for interpretation.

Since the general hypotheses were couched in terms of a "generalized- conservative", candidate choice model the analysis should begin with the

Nixon model, followed by Wallace and Humphrey models, in that order.

THE NIXON MODEL

All of the general hypotheses concerning the direction of the relationships between the variables in the model (Figured) were confirmed for the white subsample except for the indirect path from

Catholic context (C) to PGS(N). The variable that produced the greatest impact on CC(N) was FGS(N), the perceived primary group support for the respondent's candidate choice. This variable is conceived of here as the primary environmental factor in the model, and was expected to be higher than the more diffuse aggregate (and global) effects of the three contextual variables. The FGS(N) variable, measures the net effect of the influence of the respondent's co-workers, family and friends support for all three candidates. This path averaged around

.60 for the three candidate choice models, but there were several significant variations in this relationship in certain subsamples and they shall be pointed out in the comparative analysis section. The respondent's socio-economic status was positively related to CC(N), as anticipated, but SES also had a more important, positive, indirect effect on CC(N) through the PGS(N). In essence, this says that the higher a 88

ZABLE XXVII

INTERCORRELATIONS OF VARIABLES FOR A SUBSAMPLE OF WHITE LOUISIANA VOTERS IN THE 1968 PRESIDENTIAL ELECTION

VARIABLES S CBC P(N) P(H) P(W) WN

SES S -

CATH C .104 -

BLACK B .091 ■-.402 -

URBAN U .301 .213 -.326 m

PGS(N) P(N) .239 .172 -.065 .143 -

PGS(H) P(H) .042 .214 -.152 .183 .080 -

PGS(W) P(W)- .225 ■-.278 .152 -.232 -.777 -.563 -

WALLACE W - .254 ■-.167 .129 -.246 -.500 -.350 .634 -

NIXON N .246 .090 -.042 .160 .609 -.061 -.464 -.755 -

HUMPHREY H .041 .124 -.133 .146 -.087 .601 -.307 -.453 -.241 89

TABLE XXVIII

INTERCORRELATIONS OF VARIABLES FOR A SUBSAMPLE OF WHITE CATHOLIC VOTERS IN THE 1968 PRESIDENTIAL ELECTION

VARIABLES S CB U P(N) P(H) P(W) W N

SES S -

CATH C .124 -

BLACK B -.043 -.320 -

URBAN U .336 -.242 -.192 -

1 - to PGS(N) P(N) .156 .117 s -.024

PGS(H) P(H) .012 -.010 -.078 -.000 -.278 m

PGS(W) P(W) -.147 -.098 .089 .023 -.686 -.056 -

WALLACE W -.170 -.020 .157 -.132 -.448 -.381 .690 m

NIXON N .203 .037 -.019 .037 .677 -.198 -.457 -.654 m

HUMPHREY H -.206 -.017 -.171 .119 -.237 .702 -.318 -.470 -.359 90

TABLE XXIX

INTERCORRELATIONS OF VARIABLES FOR A SUBSAMPLE OF WHITE PROTESTANT VOTERS IN THE 1968 PRESIDENTIAL ELECTION

VARIABLES SC BU P(N) P(H)P(W) W

SES S -

CATH C .135 -

BLACK B -.160 -.249 -

URBAN U .289 .306 -.327

PGS(N) P(N) .302 .177 -.037 .250 -

PGS(H) P(H) .060 .127 -.092 .199 .094 -

PGS(W) P(W) -.281 -.212 .078 -.302 -.879 -.556 m CM r-> a WALLACE W -.313 -.051 .012 • -.541 -.210 -.550 -

NIXON N .268 .057 -.012 .232 .552 .044 -.499 -.848

HUMPHREY H .117 -.002 -.002 .056 .047 .315 -.191 -.386 FIGURE 9 Per Cent Catholic Population Per Cent Black Population Per Cent Urban Population S.E.S. PGS(N) Perceived Primary Group Support Nixon Candidate Choice

NIXON CANDIDATE CHOICE MODEL 92 person's socio-economic-status, the more likely he is to support

Nixon, the conservative candidate. Furthermore, the higher a person's

SES, the more likely he is to perceive primary group (net) support for

Nixon, and to vote accordingly. The direct effect of FGS(N), the direct effect of respondent's SES, and the indirect effect of SES are the three most substantive factors in this model. The relatively high po8sitive relationship between urban population and SES further supports the underly urban influence on Nixon support.

The direct effect of C on CC(N) is a small negative one, but the indirect effect of C through PGS(N) is positive and relatively high.

The summary measure of the indirect effect of religion was a positive

.117. This effect is consistent with the aggregate level, simple correlation of Nixon support and Catholic population .242. As anticipated, of the three exogenous environmental variables in the

Nixon model, the strongest direct influence on CC(N) was exerted by the urban context (U). While the direct effect of U was only .059, the total indirect effect of this variable was a positive .152. The racial contextual effects were not an important factor in the Nixon model.

In order to get a more meaningful evaluation of the contextual effects of the three exogenous environmental variables, a comparative analysis of the path coefficients of a subsample of Catholics and

Protestants was conducted.

The direct effect of PGS(N) for the Catholic subsample was substantially higher than the Protestant's. Protestants were affected by the dampening effect of C slightly more than the Catholics which is consistent with the general hypothesis. Protestants were affected positively by the black population contextual effect, while Catholics 93

exhibited a very small negative influence from this effect, demon­

strating a slightly higher racial tolerance threshold (in this

particular model). The urban contextual effect was virtually insigni­

ficant for the Catholics, but it exerted a relatively high direct

effect on CC(N) for the Protestant subsample.

THE WALLACE MODEL

The candidate in this model represents the third party movement, and

his attraction to the Louisiana electorate was tied up with his pronounced

racial conservatism and his defense of state sovereignty in the face of

increasing civil rights activities around the Country. Overall, in the

absence of controls for religion, the directions of the path coefficients were about what was expected for this candidate, except for the small positive Catholic contextual effect of CC(W). However, there is some compensation for this in terms of the negative indirect effect the

Catholic milieu has on the individual's primary environment PGS(W).

Another compensating factor is found in the camouflaged total-

indirect-effect of the Catholic milieu, (rcc - Pcc) which was determined

to be a -.201. This summary measure of aggregate effect of this variable on CC(W) through all possible direct and indirect paths can be inter­ preted to mean that the Catholic milieu has an overall dampening effect on Wallace support.

As predicted, the positive path from B to CC(W) reflects the reaction to high black population environments by whites. The direct effect is admittedly weak, but the total indirect effect of high

black-white population ratios is a redeeming factor (.112) in this

case. FIGURE 10 Per Cent Catholic Population Per Cent Black Population Per Cent Urban Population S.E.S PGS(W) Perceived Primary Group Support CC(W) Wallace Candidate Choice

CC(w)

PGS(w)

Rc R b WALLACE CANDIDATE CHOICE MODEL 93 The direct effect of the Urban Context was negative both

directly and indirectly through PGS(W). The negative path from SES

to CC(W) and PGS(W) demonstrates the tendency for lower class support

for the radical conservative.

A comparison of the Catholic and Protestant subsample path

coefficients indicate that PGS(W) is substantially higher than the

Protestant effect, perhaps reflecting a higher degree of cohesiveness

among the Catholic sector. Returning the direct effect of Catholic

Context on CC(W), the relatively low positive direct Influence on both

Catholics and Protestants should not be too surprising when one considers

the diffuse support Wallace had State-wide. In fact the aggregate level

correlation between Catholic population and Wallace support is only -.39, which explains only 16 percent of the variance in candidate support for

Wallace at the parish level. Furthermore, by controlling for the

Catholic subsample, we are in effect confining the analysis of

differential environmental effects of Catholicism to South Louisiana.

There is virtually no correlation between Wallace support and Catholic

population at the aggregate level in South Louisiana which can be

confirmed by a visual inspection of Table XXX in the Appendix 1.

And at the Individual level the correlation between the respondent's

candidate choice and the intensity of the Catholic milieu (C) was -.02.

Consequently, with these mixed findings, the influence of the Catholic

context is still highly speculative.

HUMPHREY MODEL

The direction of the effects in the model are not too surprising

after reviewing the Nixon and Wallace models, and holds a few interesting

relationships worth mentioning. First, the PGS(H) for the all white FIGURE 11 Per Cent Catholic Population Per Cent Black Population Per Cent Urban Population S.E.S Perceived Primary Group Support Humphrey Candidate Choice

CC(h )

Ra

HUMPHREY CANDIDATE CHOICE MODEL 97 subsample ia about average (.595), but there is a aubatantlal difference between the Catholic and Proteatant subsample (.696 vs. .320) which again seems to suggest the higher individuality mean score for all subsamples, the .320 for the Protestant subsample means that these people were acting against considerable pressure to vote for Wallace or

Nixon.

No connection with social class was apparent in the Humphrey model.

The beta for social class on Humphrey support was virtually nonexistent, and the simple correlation for all three subsamples between SES and

CC(H) are all quite low, which destroys any illusions about a potential

"Atlanta Solution" (a coalition of upper-middle class whites and blacks) to the race problem statewide. (Howard and Brent, 1966).

Comparatively, Catholics were positively affected by the Urban contextual effects on CC(H), while the Protestant segment showed virtually no influence (.130 vs. -.016). It should be pointed out however, that the urban effects were relatively high in terms of thair indirect effects, yielding an overall negative indirect effect of -.167

(compared to the direct effect of -.027.)

The Catholic subsample exhibited a relatively high negative effect from the black contextual effects (B), compared to the lew positive beta for the Protestant segment.

These findings can be summarized as follows: (1) Overall, the general hypotheses concerning the expected direction (positive or negative) of the paths in the three models were confirmed. (2) Generally, the primary group support variable had the greatest impact on candidate choice in three models and in all subsamples, and the path for Catholics 98 was generally higher than the Protestant subsaaple. Social class was the strongest In the Nixon model and the Wallace model, and washed out

In the Humphrey model. (3) The direct effects of the contextual variables, Catholic, Black, and Urban, were weak but for certain subsamples their total indirect effects, via primary group support, were significantly high, particularly in terms of the Catholic and Urban

Influence. (4) These findings were in general consistent with the hypotheses generated from the latent disposition profile analysis, and provide reasonable support for findings at the ecological level in general. V

SUMMARY

For the past three decades the Louisiana electorate has been the subject of extensive aggregate ecological analysis, and these studies have demonstrated the existence of a number of persistent political tendencies and cleavages among the electorate on a regional basis.

(Howard, 1971; Heberle and Howard, 1934; Heberle 1952; Howard, et al,

1963). Howard (1971; 23-24) has systematically accounted for a cata­ logued comprehensive variety of historical polarities, many of which are highly relevant to the current political ecology of the State and played an Important role in this study.

This dissertation has attempted to extend these works in a multi­ variate, multi-level research strategy, designed to link the individual voter with his socio-cultural context, in an effort to isolate and examine the impact of several hypothesized socio-cultural environmental factors on individual voting behavior.

This study departed from the traditional political ecological method, which characteristically analyzes data at some level above the individual, and focuses on Individual level relationships in varying social contexts. Aggregate data were examined independently and in combination with survey data in what could be considered a synthetic ecological survey analysis. A comprehensive multivariate analysis of aggregate level data was made as a preliminary step to provide a frame of reference for the contextual analysis. A latent political disposition profile

99 100 analysis was performed and a set of eight political-ecological subregions were delineated that conformed highly with Howard's Voter Type Areas.

This was an Important finding in itself, inasmuch as Howard's Voter Type

Areas were derived from an historical-geographical analysis, and the sub-regions derived in this study were based on measured aggregate level political, economic, and cultural variables representing the 1960-1970 period.

Although electorates, no less than individual voters, are significant units for political analysis (Ranney, 1963: 99) the use of straight aggregate data sometimes leads to fallacious or misleading findings when generalizations are made about individual level relation­ ships. The problem of the ecological fallacy has come under considerable scrutiny since the Robinson article (1950), although in practical research this error continues nearly unabated. However, at least three alternative solutions to the problem of making inferences from straight- ecological data have been developed (Duncan and Davis, 1953; Blalock,

1965; and Goodman, 1959). Partly as a reaction to this problem, since the early fifties there has been increasing emphasis on sample survey research, which has lead to the opposite problem--the individualistic fallacy. Now, as Dogan and Rokkan put it, "... it is time we react with equal insistence against the individualist fallacy, the tendency to generalize findings for individual behavior without controlling for the characteristics of the Immediate social, and political contexts". (1969: 88).

In this sense, the research strategy of this dissertation can be considered a partial solution to the ecological fallacy as it applies to recent political behavior in Louisiana. It is based on the socio­ 101 logical premise that the analysis of individual behavior must take his social context into consideration. A corollary goal of the cross-level inference approach in this study was to investigate the fit of the findings made here with the inferences being made in on-going aggregate level research of Louisiana political behavior.

Matthew Schott, an historian, has recently drawn attention to another related problem to which this dissertation is addressed. Taking on Howard, Key, Slndler, T. Harry Williams, et al., he contends that the preoccupation of these students of Louisiana politics with class conflict and economic considerations has obscured "... the relationship between non-economic cultural factors and political ideology..." (1971: ISO), as if they were unrelated. He seems to have missed the point of The

Louisiana Elections of 1960 (Havard, et al., 1963) which he cites, and apparently did not have the benefit of the revised edition of Political

Tendencies in Louisiana (Howard, 1972).

After accusing the above scholars of using biased samples of aggregate data to make their case, Schott makes the observation that

"Louisianians in recent elections have . . . reverted to voting habits which have typically expressed not class conflict but traditional and historic concerns reflecting sectional economic and cultural differences between the northern Protestant and southern Catholic parishes."

Although the north-south, Protestant-Catholic Cleavage structure was obvious in the 1960 and 1964 presidential elections and in most guber­ natorial elections, it certainly was not apparent in the 1968 presidential election. Strangely enough Schott tries to use aggregate data to show the underlying regional-religious cleavage behind Wallace support at the parish level, which was not a very strong point from which to argue the 102 reality of such a cleavage structure, especially in view of the overwhelming statewide support of the right-wing conservative candidate, even in the most solid Catholic parishes. There is a religious element in Wallace's case, but it is not quite as simple as aggregate percentile rankings suggest. The simple correlation of Wallace support and percent

Catholic population was only a modest -.39, and when controls were made for black population and urban population, race became a dominant factor.

Furthermore there was considerable variation among the Catholic parishes and Wallace support, and the simple correlation for this relationship vanished when controls for region are made. Survey findings show that the whites in the Agrarian French parish cluster voted overwhelmingly for Wallace, and overall the difference between white Catholics and white Protestants only approached statistical significance. Yet in the multivariate contextual path model tested in this study, the Catholic

Context did have a modest overall dampening effect on Wallace Support through the individual's primary group support.

Using multivariate analysis techniques the religious dimension was made explicit in this study, and its importance was placed in balance with two other major dimensions, the economic and the racial.

Three dimensions were extracted in the latent political disposition profile analysis: The most Important was the strong confirmation of the North-South split— the Catholic Moderation dimension— which had been previously inferred in Political Tendencies and Louisiana Elections of I960: second, the delineation of an Urban-Republican dimension made explicit the direction of the Republican trend taking place in

Louisiana; and the third factor— the Racial Radical dimension— produces a clear picture of the emerging black coalition. This study shows that tht socio-economic atructural determinants of political behavior should not be underestlasted, they are crucial to a meaningful explanation of political behavior at all levels of analysis

It would be unrealistic to play down the significance of social class la an appraisal of the racial cleavage In the State, as Schott would have us do. Findings In this analysis suggest that the Individual's social class, and the urban contextual environment are significantly related to candidate choice for both blacks and whites. Furthermore,

Llpslt and Rokkan's cross-cultural and historical analysis of cleavage structures and the voter alignment processes make it clear that these factors are Interrelated, complex, and changing, and at different points in history certain factors become manifestly more apparent than others (1967).

This study focused on the 1968 presidential election which marked the end of a long run awing of national Democratic party disaffection

In the state. The overwhelming victory by Wallace, which reduced total

Democratic support to 29 percent of the total vote, clearly Indicated the Impact of the conservative trend In the State, particularly in view of the fact that the white sector accounted for only 13 percent of the total Democratic candidate support. Behind these events there has been a change In the socio-political cleavage structure of the

State, and some of the traditional, apparent polarities, have become highly problematical, at least in terms of presidential level politics.

Among other things, in 1968 we witnessed what seemed to be a break­ down In the traditional South Louisiana French-Catholic--North Louisiana

Anglo Saxon Protestant cleavage structure In a three-candidate presidential election between e liberal and a radical conservative, and 104 moderate conservative. Lending to this phenomenon, increasing

industrialization and diversification since World War II has

altered the socio-economic structure of the State, mediating traditional

urban-rural polarity and broadening the Republican base. Increasing

black participation resulting from Federal Civil Rights legislation was

a significant factor in support for the left-wing candidate, and this

growing coalition made a critical impact in the 1972 State Gubernatorial

election, the first populist coalition of blacks and whites In the

history of the State.

In view of these complexities, the strength of the preliminary

ecological analysis presented in this study is in the ability to

account for and measure the relative impact of a wide range of relevant

political, economic, and socio-cultural characteristics in a systematic way. At the aggregate level, the multivariate analysis procedures were

essential given the complex set of interrelated factors that makes up

the State's political ecology, and as a consequence, it was possible

to avoid the trap of Ignoring the influence of the socio-cultural

factors at the expense of the economic ones.

A latent political disposition profile analysis confirmed the

socio-economic and socio-political cleavage structure in the State,

Inferred in previous studies. As a follow-up, gubernatorial election

returns for 1972 were used in a multiple regression analysis to test

the reliability of the factor structure with very good results. The

factor structure proved stable when gubernatorial election returns for

the 1960-1972 period were included in subsequent analysis, and a

comparative profile analysis over two election periods has already

been made in conjunction with this study. The profile analysis and 105 the clustering procedure used in this study provides a foundation and

framework for a systematic longitudinal analysis of changes in the

State's political-ecological structure.

Over 80 aggregate level variables were Investigated in connection with this study and they are summarized in an intercorrelation matrix

in the appendix for consideration in future studies.

The contour mapping technique used in this study to display the

factor score profiles geographically was original. Nothing In the

literature goes beyond mapping the Z scores within artificial boundaries.

The technique used in this study presents the phenomena as continuous

data, providing a more realistic visual interpretation.

The cross-level inferential path models and the statistical inter­

action model were employed in this investigation in an effort to

contribute to the general solution of the problem of micro-macro

linkages in sociological model building. These approaches are consistent with the basic sociological concern with the problem of Isolating and measuring differential environmental effects on individual behavior.

It was a deliberate attempt to systematically "bring the social

structure back in" as Barton puts it (1968). The contextual analysis

strengthened and validated'several prevailing aggregate level general­

izations concerning la the racial, religious, and urban-rural dimensions.

These three factors were paramount in the profile analysis and

reflected the three major structural dimensions of the State's

political ecology. Consequently, they were used as independent exogenous

variables in several cross-level models to represent the main contextual

dimensions comprising the environment within which candidate choice was

made in the 1968 election. In addition to these three aggregate level effects, individual survey respondent's social class and primary group support were incorporated into a causal model and tested, control­ ling for race and religion. In the model tested for this dissertation the relative primary group support variable was construed as an intervening variable between the three exogenous contextual variables and the respondent's socio-economic status. In all three candidate- choice models the direction of the effects (positive or negative) was generally consistent with the general hypotheses, and in all of the subsamples tested relative primary group support had the highest impact, followed by SES and the three contextual variables, in that order.

It is recognized that the primary group support variable had little competition in the model except for SES. However, previous survey research has shown that in this State, party identification, and perceived class, and other standard individual level explanatory variables are relatively weak and unpredictable (Howard in Kovenock, forthcoming). The relatively high contribution of primary group support to explained variance in the causal model tested in this study was consistent with findings in the Comparative State Elections

Project at Chapel Hill (Kovenock, et al, forthcoming). Furthermore, it confirms Shaffer's recent findings in his 1972, Computer Simulations of Voter Behavior. Shaffer has empirically tested several of the leading voting behavior models in a series of computer simulations and has developed a revised model of voting behavior in which primary group interaction plays a critical role (1972: 143-147). The importance of primary group interaction and the "significant other" as mediators of individual behavior is central to sociological theory, and the findings presented In this study lend support to the notion that such

Indicators are an essential Ingredient in a valid model of the political decision making process.

General findings for both contextual models provide support for inferring the importance of the racial, religious, and economic factors in the voting process in Louisiana even when such effects have not been apparent at the aggregate level, both in terms of individual characteristics and in terms of their environmental effects. APPENDIX I

PER CENT CANDIDATE SUPPORT IN THE 1968 PRESIDENTIAL ELECTION AND SELECTED ECOLOGICAL VARIABLES

108 109

T A B U XXX

PER CBIT CANDIDATE SUPPORT IN 1968 PRESIDENTIAL SUCTION AND SELECTED ECOLOGICAL VARIABLES

Par Cant Par Cent Per Cent Par Cent Per Cent Par Cent Candidate Candidate Candidate Black Cethollc Urban Pariah Support Support Support Population Population Population Humphrey Nixon Wallace 1960 1971 1970 1968 1968 1968

I. Acadia 24.1 18.7 57.2 19.7 77.6 56.8 2. Allen 27.9 13.8 58.3 24.8 20.5 35.1 3. Aacenalon 30.4 12.7 56.9 31.9 55.2 32.0 4. Assumption 33.6 19.7 46.7 41.2 67.4 - 5. Avoyelles 24.4 20.2 55.4 27.8 S4.8 26.3 6. Beauregard 22.2 22.2 55.6 22.4 2.9 35.1 7. Bienville 28.7 15.2 56.1 49.4 0.0 18.5 8. Boaaler 17.6 23.8 58.6 24.9 3.9 64.9 9. Caddo 26.2 31.6 42.2 36.5 14.0 85.1 10. Calcasieu 32.9 21.5 45.6 20.9 37.8 75. 11. Caldwell 26.2 13.2 60.6 27.8 2.9 - 12. Cameron 20.6 15.6 63.8 6.4 53.5 - 13. Catahoula 18.3 18.0 63.7 35.2 0.0 23.5 14. Claiborne 25.4 18.7 55.4 50.3 0.6 44.3 15. Concordia 26.4 13.0 60.6 46.3 2.9 47.7 16. DeSoto 39.7 11.4 48.9 57.5 5.1 28.3 17. E. Baton Rouge 27.8 27.5 44.7 31.8 23.7 86.9 18. E. Carroll 45.7 13.9 40.4 61.2 3.6 48.0 19. E. Feliciana 34.4 11.2 54.4 54.0 1.4 26.6 20. Evangeline 22.6 13.3 64.1 26.8 64.2 40,6 21. Franklin 9.6 14.7 75.7 46.6 0.9 22.3 22. Grant 16.9 20.2 62.9 24.2 2.8 - 23. Iberia 29.0 28.5 42.4 28.7 75.8 63.5 24. Iberville 41.6 14.4 44.0 49.0 51.8 33.3 25. Jackson 23.2 16.8 60.6 32.4 1.1 31.8 26. Jefferson 22.0 32.3 45.7 15.3 57.8 95.8 27. Jefferson Davis 27.1 22.7 50.2 21,2 57.9 62.7 28. Lafayette 26.3 35.1 38.6 24.0 73.3 71.6 29. Lafourche 26.0 22.6 51.4 12.2 78.3 38.8 30. LaSfelle 12.2 21.5 66.3 12.7 1.7 - 31. Lincoln 22.6 29.8 47.6 41.8 4.4 64.4 32. Llvlngaton 11.5 7.7 80.8 15.0 8.3 18.5 33. Madison 46.8 11.4 41.8 64.9 1.8 64.0 34. Morehouse 20.1 19.8 60.1 46.9 1.8 45.3 35. Natchitoches 33.4 19.9 46.7 43.7 15.8 45.4 36. Orleans 41.7 26.6 32.8 37.4 40.9 99.7 37. Ouachita 20.4 31.8 47.8 32.2 5.8 78.5 38. Plaquemines 13.4 11.3 75.3 28.8 51.0 28.3 39. Polnte Coupee 41.9 11.3 46.8 53.6 57.9 17.9 40. Rapides 25.0 28.9 46.1 30.5 19.4 52.2 41. Red River 24.2 10.1 65.7 47.6 0.0 - 42. Richland 15.7 16.0 68.3 44.4 0.8 .3JL5 43. Sabine 17.0 16.5 66.5 23.6 16.8 16.7 44. St. Bernard 13.1 18.3 68.6 7.5 72.3 91.3 45. St. Charlte 33.6 18.4 48.0 27.1 50.7 27.2 46. St. Helena 40.1 6,5 53.4 55.5 1.5 - 110

Par Cant Par Cant Par Cant Per Cent Per Cent Par Cant Candidate Candidate Candidate Black Catholic Urban Parish Support Support Support Population Population Population Humphrey Nixon Wallace 1960 1970 1970 1968 1968 1968

47. St. Ji m s 45.7 12.0 52.3 49.3 65.4 32.8 48. St. John 43.7 12.6 43.7 51.6 71.4 51.8 49. St. Landry 36.0 13.9 50.1 43.0 67.3 39.1 SO. St. Martin 34.2 16.8 49.0 37.2 87.7 37.3 51. St. Mary 31.9 27.5 40.6 45.7 30.6 65.2 52. St. Tauaany 21.4 23.4 55.2 27.5 31.6 36.6 53. Tangipahoa 22.8 17.3 59.9 33.9 16.6 35.5 54. Tensas 32.0 19.1 48.9 65.0 1.1 55. Terrebonne 24.8 27.9 47.3 20.5 68.4 52.6 56.Union 19.8 16.5 63.7 36.8 0.0 18.5 57. Vermillion 25.0 21.6 53.4 12.9 71.7 38.4 58. Vernon 17.4 18.3 64.3 13.3 2.8 60.4 59. Washington 19.2 10.8 70.0 33.9 2.9 52.3 (0. Webster 20.5 17.8 61.7 34.5 1.5 51.3 61. W. Baton Rouge 38.4 12.7 48.9 49.3 35.9 38.9 62. W. Carroll 8.7 11.8 78.5 22.4 0.9 — 63. W. Feliciana 48.7 11.2 40.1 66.1 2.9 -- 64. Winn 19.5 16.7 66.8 31.2 0.0 43.6

28.2 23.5 48.3 32.0 36.3 66.1 APPENDIX II

MAPS OF SELECTED VARIABLES USED IN THE LATENT POLITICAL DISPOSITION PROFILE ANALYSIS

111 112

£-444444444*1 ♦ ♦ ♦ ♦ ♦ ♦ ♦ ♦ ♦ 444442444442-4/5^

|03}4444444444 p o k * * * * * * 4244444444A4- ooooot^jmoooook♦♦♦♦♦ z * < 000000000030004 4 4 4 4444 4tV 0000000009442+ 4±4 4< 000004♦♦♦♦> 000|r 4 ♦♦♦♦♦! 0 0 0 0 4 4 4 2 4 4 44•' O0OQQOOO^4♦■•■♦♦♦♦♦’♦♦♦*♦^ OOOOHDOOi* + 4 46"* • •'• ••14 4 < 4 4lk0 akbo 0» 4 4 4{» • • 1 • 4 4 H 44 4 4 4 ^QQb 0 g|f 44 2k« • • • «J 4 4 ^ 4ift24 4 4^000(44444T?Wj%jf4 44 44 4 44 44 44 44 4 4 4TT 44-1(000( [444 44-444 44 44 4 4 44 44 44fOO( V44 4 4 44 4 4 4 4 •MO’Sy-4 4 44 «6 O O f V4444244 44400f4 44 44P00^ -44444444«004444?20( i-4 4 4 4 4 4i4j)o |4 4 4 4Ooog8MBMibee0efleaDO|4 4 4 4 4 ^ 1444 4 4 4000004 4 +cooQeeH0S8iB4ea646OooP4 4424-1 h44444-Hoooool»2 4boocee9s«aBeeeoooaooQ4 4 4 4 4 +v •444244-Joo3qoW4 4pof»eeea— Be8QPoooQ4T2 444444-1 )OOOOOOOOOOOO0*4C) 08066686688930^44444 444444- 100000000000000000008666666461 t244\ (0000000000000000000001666001 000400000030000£7?*p0Weee< 1000000000000O 0 ok 4 420001689666861 }oooooo_QQgoooaoo64 44poooo5eee9ee< '000000*44^ 1000a±iJ «0030060000066*1)00 * )00P^4424 44 44- ioo3oooaoooooo)B4e4eeeoe6a0oj •4444+44444- JOOOOOOt 100C IOC L4- 100000/ 1Q300000/ rooooooi I • • • • I )0000( 1000000000031 lOOOOOOOOOC )00030000( 10000000C lOOOOC

frequency distribution of data point values in each level

LEVEL 1 2 3 4 5

••••••••• 44 44 44 44 4 OOOOOOOOO 668666606 ■■ • • • • • • • • • 444444444 OOOOOOOOO 868086880 66 SYMBOLS • * * * 1 * * * « 444424444 0 0 0 0 3 0 0 0 0 8 6 0 6 4 6 0 6 0 6 • ••• • • • ■ • 444444444 OOOOOOOOO 068886668 66 • • • • * • • • • 444444444 OOOOOOOOO 800888606 66 6 6 0 0 6 6 ISISSl FREQ* 2 3 16 11 8

MINIMUM 8*70 16*70 2 4 * 7 0 3 2 * 7 0 4 0 * 7 0 MAXIMUM 16*70 24*70 3 2 * 7 0 40*70 4 8 * 7 0

PER CENT CANDIDATE SUPPORT FOR HUMPHREY 1968

FIGURE 12 113

* r — ----1-----♦----2-----♦-----3--— 4----4-----♦----S---- OOOOGOOOOO 30000

^TTFf4 4 4 4244 4 44 1J44444 44 4 44424 44 . 1000000000000000044 *0300000000003003*444 10000000300000004444 3000QOQnnngooon44424 ^ 4 444poooooaeae&ipoo44 4 4 4 )OOOOQB0000aOOaV444 )oooooeee468enooo)»4 )oo3 ooooteeees0 ooooffr4 00000000088881000030*4 4 OOQOOOgW)OQaOOOOi /OOOOOQ* 4444444444444 30000004 4444 4244 444 4 3000300044244 4 444 4244 300000000*444 444444444441 |00000000005&44 4fi0 oooooooooooooOaa 1000300000030000 3000000000000000 300000000000000 4 /oooaq*TT>opaoooo e >42444^0000031 ___ 50|pOJf4444 1444 4444 4444 A O O O O O O O aapoooooooooouoBaaoo/f4♦♦♦ *0000000, 84^000000000000800*»4444 lOOOO eeeeBB0Bd&3oooo/f44 eeeeaeeipooo 6004000010000 eeeeeeaT QOir-n v***« 88 QT O v >•••

4*

FREQUENCY DISTRIBUTION OF DATA POINT VALUES IN EACH LEVEL LEVEL 1 2 3 4 5 :»ai«l«M3»HSni*SMStmi3M»aMMlMSX3l ••■•••••• 444444444OOOOOOOOO 688888888 68868 ••••••••• 444444444OOOOOOOOO 868888888 66666 SYMBOLS ••••!•••• 444424444 000030000 888848888 66665 ••••••••• 444444444 OOOOOOOOO 688888888 86868 • • • • • • • « • 444444444OOOOOOOOO 868888888 88666 Z23a3I=»=»II33XX32£;3IlXl»X«2333»32SIZ:Xl FREQ* It 23 19 7 5

MINIMUM 6*50 12*22 17*94 23*66 29*38 MAXIMUM 12*22 17*94 23*66 29*38 35*10

PER CENT CANDIDATE SUPPORT FOR NIXON 1968

FIGURE 13 114

* ------1—— ♦------2------♦------3------♦------4------♦------5------+♦ HoooooodoooooodjjttSSaooOl------(OOOO OOOOO 3 OO0000400000380030* 4j ]030BTgftOOOQOOOdflaOOOO< >00000000 ooi^>i)ooaoo oof el >oooooooot±2i 3 T m |00 OOOOOOOOOOOODOCM240 0000003000000000001 C0fic00000000300000 z r . oaiiaooooqgggooooi V+* ocioQOoooeee«eeeBeoee jooooaooooooqeeeeeeeeee aO&++] jodeeeeeeeeee 00( 00300000400400000I \eeee€lpd*±^Qode^3BOBeeaeefooi vee4e^ooaoooaooaoeeeeeeD3o oooooooooodtroOUooooj OOOQ/fXJOOOOOOQOOj 0004+2+T4000000j 00'4++++(000Q00( 0004*^00030L ------0000000000000000004+♦♦♦♦$0000001 leeei 000000000000/ffBp0004+++*1+++Q000300< 100401 0000000000000460004+++2+++++++to00000( ___ nnnn3nnnnmnonflataog++++++ + + ++++eogBto30000000) 00000000000000Q4++++ ++++++2-tflMiBBaOOOOOOv ♦+mrr+oooooooaoo( +++++++yr»poooooq/f ++++++++++++4 ♦ ♦ ♦ 2 ♦ + ♦ ♦ 4 ♦ 2^ 4003/+ +|f*|4 2+♦++ 2 ♦ 3001 J+ + + + -f+Qtm+ + + 4000(++ll«l»++ + + ++ » + + P Q 0 0C /+♦+♦♦4100000000 Cff4*+U* ^♦♦♦♦♦♦♦♦♦♦+13Q<*4**A2♦♦♦♦+♦♦♦♦ ♦♦♦♦ ♦2 ♦< foaooo( 10000003V++++M+++++2+2++ ♦ ♦ ♦ ♦ ♦ ♦ J looooc IOOOOOO0|+4+yWW>++++++ ♦ ♦ ♦ ♦ ♦ ♦ < )OOOOOOl+p ♦♦♦♦♦♦ ♦ + + + + J ++ I l O O O O Q ^ v— \W^44+++ ♦ ♦♦ 200( >*+♦♦♦♦ ♦♦♦♦jooof + ♦+2 ♦♦♦♦pooee\ kOO( ► ♦♦ ♦ ♦ > ' +/

>♦4

FREQUENCY DISTRIBUTION OF DATA POINT VALUES IN EACH LEVEL

LEVEL ri3>i»xa3iz33xx3XH«asai»iaiixiaxiiasiIxaxaxalaxI 1 2 3 4 5 ••••••••• ♦♦♦♦♦♦♦♦♦ ooooooooo eeeeeeeee 0 0 0 0 0 0 0 0 0 • • • • • • • • • ♦♦♦♦♦♦♦♦♦ OOOOOOOOO 000000000 000000000 SYMBOLS ****1 **** ♦♦♦♦24+++ 000030000 000040000 000050000 • • • • • • • • • ♦♦♦♦♦♦♦♦♦ OOOOOOOOO 000000000 000000000 • • • • • • • • • ♦♦♦♦♦♦♦♦♦ OOOOOOOOO 000000000 000000000 X3X 333 XX 3a X X r 3XX 3XXXX 3X 3 XXXX=Z:X=XXaXX 3 XX3XXXXXX 3XX FREQ* 7 22 18 14 4

MINIMUM 32*80 42*40 52*00 61*60 71*20 MAXIMUM 42*40 52*00 61*60 71*20 ao*ao

PER CENT CANDIDATE SUPPORT FOR WALLACE 1968

FIGURE 14 115

000000300 0300000000 000000000003 00000000000 0000 0030 00000 o o o o o o o 000000000000 OOOOOOOOO00 003000000000 oooooooooooo 0000000030 oooooooooooo 0000000000 o joooooooo^boo ioooo3ooo» 43000 ooooooooot^Qoao 0000000000$+ OOOOOO0»4 80400000030 00000004444 00000000000/44 4 44 ooooooooooff+4+4++took++2+-mpooHiB9eepo3peeeee |00ioocoqarrrW+++++2*«4t*+4^TadKj)ae0§boolB4e0e4 b0imnaOjLdL£jfc.44 n 44 >444 4Jjr $$4/Da*jq00000pOOO0QB000 <4444444441 O3OOGPB000PO00000000(00 '^44444 00000000003000001303 0000000POOOOOOQOOO ©840

* ------+------| - 4- _2 —— — 4- 4 ------5------44 SYMAP

FREOUENCY DISTRIBUTION OF DATA POINT VALUES IN EACH LEVEL

LEVEL 1 2 3 4 5 sssacassaaamasMssanMaMMHtsauBsauasanaasaa ••••••••• 4444 44444 OOOOOOOOO 000000000 000000000 • • • • • • • • • 4 4 4 4 4 4 4 4 4 OOOOOOOOO 000000000 000000000 SYMBOLS • *••1 » •*• 4 4 4 4 2 4 4 4 4 000030000 000040000 000050000 • • • • • • • • • 444 4 4 4 4 4 4 OOOOOOOOO 000000000 000000000 • • • • • • • • • 4 4 4 4 4 4 4 4 4 OOOOOOOOO 000000000 000000000 zxssxssscssssKaMszrnsaasisiisasasxasassisssssfsa FREQ* 5 12 17 17 14

MINIMUM 62*60 66*24 69*60 73*52 77*16 MAXIMUM 66*24 69*88 73*52 77*16 60*80

VOTER PARTICIPATION IN THE 1968 NATIONAL ELECTION

FIGURE 15 116

4* 0000 0300 000 0000 00000 00000000000000 00000000030000 o o o o o o o o 00000000 0300000 0000000442 o o o o o o o q * 44 0000000000(44444 0 3 0 0 0 * * * * *****-* 0000(1* 444^00*4444 *

0 0 00 30 30 0

4* SYMAP

FREOUENCY DISTRIBUTION OF DATA POINT VALUES IN EACH LEVEL

LEVEL 1 2 3 4 5 • • • • • • • • • ********* ooooooooo eeeeeeeee oooooeooo ••••••••• ♦+♦+4++++ ooooooooo eeeeeeeee ■■■■■■■■■ symbols ****i**** 444424444 000030000 eeeo4eeee ■■RBseeee 444444444 OOOOOOOOO 6 0 0 9 0 0 0 0 0 ■■■■■■■■■ • • • « • • • • • 444444444 OOOOOOOOO eeeeeeeee OOOOOOOOO

FREQ* 15 26 10 8 6

MINIMUM 3*50 13*44 23*38 33*32 43*26 MAXIMUM 13*44 23*38 33*32 43*26 53*20

PER CENT BLACKS REGISTERED AS A RATIO TO TOTAL REGISTRATION 1968

FIGURE 16 117

••••••••••• ■ «•••••«» ^424 ■ • • • 1***** 1** •••• 1**1• • *\44 ••••••••••••a •••• ••••••••••• >•••• ♦ • • • • */F 44 4 4 2444 ______444 ■I- 4 4 ♦JKTt±± + ♦ ♦ ♦ ♦ ♦ 2 4 • • • • • **V444 ♦ ♦♦ -fjffO0 OOOf ♦ 44421000300*444 4444C00000 44 4 4 4 *\pCQQ C 444 444 444^0C444 4444444)3 444 4 4 44 4 4 ■#04444 4 442444444444 44444444444 ♦tf-Lx 444444444444 0000 4444444444424

03000

oooaooo ooooo BH8S

SYMAP

FREQUENCY DISTRIBUTION OF 0ATA POINT VALUES IN EACH LEVEL

LEVEL c=S3s$s3axs:sss:ss3s:338SflisisassfssssBmusasaas: 1 2 3 ♦ S • • • • • • • • • ♦♦♦♦♦♦♦♦♦ OOOOOOOOO 660800000 000000000 ••••# •+ •• ♦♦♦♦♦♦♦♦♦ OOOOOOOOO 860000000 000000000 SYMBOLS ••••!•••• ♦ 000030000 000040000 000050000 4444 44444 OOOOOOOOO 600060660 lllllllll • • • • • • • • • 44 4444444 OOOOOOOOO 660666060 600060000

FREQ. 14 16 10 13 12

MINIMUM 32*00 43*00 54*00 65*00 76*00 MAXIMUM 43*00 54*00 65*00 76*00 67*00 PER CENT CANDIDATE SUPPORT EDWARDS 1972

FIGURE 17 .5---- +* (■■■■■■■■■■Ml i i i m m lim niiinsii l 5 B B M r a i ■■115115111111 IIIISM! ■5nm iiaiiaii i n i m i i ■■■■■■■■■■■■■■i i i n i i a m is im s m m h i

lasaiaaal Isnasiniiai IIN5inillllUI (■■■■■■■■■■■■■■■■■■■■■■■■■I ■■■■■■■Ml ■■■■■■■■■■■I \eee? IB5MM II5H5HI [eeeee ■■■■■■■■[ 164666 ■HM5M tagaaaaaaaaeaaaaa iQOQOOoQeeaaaaaea JOOOOOOOOOeflASOQOOO *0000300000000000000 *000000000000000003001 qoooqoooqqoopttt eeeaooooooot oooootf> ♦♦♦♦♦♦♦♦+-aooo 5fii506e*aoooooo3oc 000000|f ^♦ ♦ ♦ 2 ** + *b000 eeeaaeeaegoooooooo < 0003000>«- 42 ♦•*■♦♦♦♦ ♦gftQgOOaZJBtJOO 0030000001 OOOOfiggA* ♦♦+♦♦♦♦♦ ♦♦♦♦+?'0OOOOO3OO o o o o o o o o c ♦♦♦♦♦+♦♦ ++++++++♦ 3000000300000000301 ♦ ♦» + + + »»»»+ ++ + » + + ♦ ♦ » + ♦ ♦ ♦ + 4TFH0 0 Q ♦♦♦2++++++2*»t+2*4*+*2+*++2t+++2f*+>t / ♦ ♦ ♦ ♦ ♦ ♦ ♦ -f ♦ ♦♦ +±±JL&±+ 2 ♦ ±+ ♦♦ ♦ ♦ ♦ ♦♦♦ ♦ ♦ ■i- ♦♦♦♦ + ♦♦ ♦ ♦ + «T o o o o O ± ± o o £ ♦ -f ♦♦♦♦♦ +(1 • •• a+ ■»■♦♦♦♦ + ♦♦♦♦♦ +I000000000003+ + ♦++++♦?)•••• 42 Jooool JQQ0 Q Q + ♦ ♦ ♦ 2+ 2 +■♦{••• • *)♦ 100031 leeeeipojyig^*-*- ♦♦ ♦♦♦ ♦♦♦(**«/♦♦ )0( loiaeeeedir \j++2 ++++++++tt+++ 1666661 '^♦♦♦♦♦♦♦♦♦♦2 ++ !■■!

?♦♦♦

•5 ♦ * SYMAP

FREQUENCY DISTRIBUTION OF DATA POINT VALUES IN EACH LEVEL

LEVEL s«ss*a*s:asssas3*ssssss***a*s***issi«**»******,3,] I 2 3 A 5__r„, ••••••••• +♦♦♦♦♦♦♦♦ ooooooooo eeeeeeeee ■■■■■■■■■ «•••••••• ♦♦♦♦♦♦♦♦♦ ooooooooo eeeeeeeee ■■■■■■■■■ SYMBOLS ****1**** ♦♦♦^♦♦♦+ 000030000 606646666 M M 5 M M ••••••••• ♦♦♦♦♦♦♦♦♦ ooooooooo eeeeeeeee ■■■■■■■■■ sssss3ss;s==ssss:sias*ssss*s*ss*ssssss=*s=asss*sss:••••••••• ♦♦♦♦♦♦♦♦♦ ooooooooo eeeeeeeee ■■■■■■■■■ FREQ* 3 16 15 6 25

MINIMUM 20*00 34*00 48*00 62*00 76*00 MAXIMUM 34*00 48*00 62*00 76*00 90*00

PER CENT CANDIDATE SUPPORT FOR GOLDWATER 1964

FIGURE 18 119

• • • • • i + + +A 1* « • 1 • • • • 1 • • • •••••«»• •••••••••••• •••*••••• >•••••••••• •••••••I* * * I * • • • • • •••l****** »•***•« 1••••••••• •• •••••••••a* \44444T 4444444 I******** \44444 4 4244 44 444-m** 1** ! • • • l4 4 44 4444 4 444442-*U******** ►4 24 4 440l>fc±444444¥\t4 « *** ■ ♦ ♦ ■h3doooo^+ 4 4 4 4 4 lOOOOOOOkfr 444 4 4 4 444 ♦ ♦I ]oooooQoao»tii4o n n n n OQ003QOOOOOOTOOOOOO 000000000000000 3000 OOOQOOOOOQOQOieia ooo 4 4440000000C Joooooeeeeeeeeeee ooo ♦ 24(000000301 >50000000680004000 00034 44 + 44 +44 4-*J0000000( fiooo3OOQ8e40eeeeee4 OOOO000000o\t-*/DO3OOOOOO( 100000000686801 )000003000yjj000000000t leeeeeeeeeeeeeee&c I300000030QQQQQP030C 6666660686000606606006008399000]; 0064000000406884000004006040039400000 JQ888888888888888884888888888880I i88880888880aOOOO9nie0Pa088868800|reeeei a6B688689699flOQOOOQQOOO3te6886808990000l»0l 10000004000000O^437pOOpOq000040400i»if|B61 IflflflaafiflflflSpOfT444 4D O O n Q 3 8 O Q O 0 3 0 6 0 9 3 0 M B9C ■44444> 1B646066000090B06C * 4 4 4 4 4J Ae0000000000S00f )6600000«B0C )B0046600fi£6f |0006600f 10001

SYMAP

FREQUENCY DISTRIBUTION OF DATA POINT VALUES IN EACH LEVEL LEVEL 1 2 3 4 5 :32ss33::3:a*333x:ss2siM:ss3s»s33i>tsxssmia**a< ••••«•••• 444444444 OOOOOOOOO 000800000 ■•••*•••• 444444444 OOOOOOOOO 000000000 SYMBOLS ••••1•••• 444424444 000030000 000040000 •■•■500B0 ••••••••• 4444 44444 OOOOOOOOO 000900000 ■■■■■••■■ ••••••••• 444444 444 OOOOOOOOO 000000000 ==asss=2sssss53xsss*xa3=3=*sasa*ssaa*aaaasa*s»3*a=s FREQ* 23 8 _J4 16 4

10*00 24*00 38*00 52*00 66*00 MAXIMUM 24*00 38*00 52*00 66*00 80*00

PER CENT CANDIDATE SUPPORT FOR JOHNSON 1964

FIGURE 19 120

•••••••••••a********** I* * * * * * !* * * * * 1 *|444< 1 • • l**************** #w+ 2 *j •••••••••••«••!>>«>«••> !•••*••««•• 1« » t|» * ^ ••«••• ■ ♦ ♦♦ 2+' tfll • i ••■ ••• 1 • • *f4 44444444 ‘♦Vi •• • < • *1 ♦ • • • 444W •• • • • • • J4 44 44444444 42:$* •• 4 42J**« 1 «***J44444442444 44i ♦ « * y4 4444 2 44 44 444 4 44)** • li 444 44 4 4 T 4 44 44 44 444444 44 • « •• j S4 + 444444 44444 44 44j£*"*T*8> V4 44444424 444 44 4«(**l** 1 • ••*. loool><*± ♦ t ♦ ♦ 12 ♦ • • • • • I 1003000000^44 4 44 44444-iy • * 1 ml VPOOOOOOOOd* ♦♦♦♦ 4 mj ,oaooQaoooo*+±±jfooooo\ loooooooooot V0000300000C o o o o o o o o o m o o o o o o o c 100001 i++4>t+++t+«dbi3tidtii]oi 'OOOOOt 1000001 t2±44tfV4 4 2 400000030C )OOOOOt IS000000I t o 4444 4000000000( J000300C JOOSOUOOO30000000] 5000( IOOOOOOOOOOOOOOV 100 10000001 103Q000000030C 00000000000 100000001 0 0 0 4 0 030606461 6 0 1051 IS000050I JoflgggSli o i i i i 10001 10000000000051 aoit i i a00000000000 1006000^0000000500001 70000000000000000000000000 1060400000000000000! tooool |0000000500000050000000005001 >eeec 150500000001 J0003C F44J «00000000 ■ ■ ■ ■ ■ ■ ■ )0e6i000000000i )0/f4l 100000001 1000000001 J0H-444J 10000051 36980000001 16641 leeei

FREQUENCY DISTR16UT10N OF DATA POINT VALUES IN EACH LEVEL

LEVEL I 2 3 4 5 isssssKMSisassasssssssSHnzanssssscansaasssm 444444444 OOOOOOOOO666666666 000000000 • * • • • • • • • ♦♦♦♦♦♦♦♦♦ OOOOOOOOO 666666666 000000000 SYMBOLS ****1 **** 444424444 000030000 666646600 000050000 • • • • • • • • • 444444444 OOOOOOOOO 660666666 000000000 • • • • • • • • • 444444444 OOOOOOOOO 666666666 000000000 SSaMXX33S3MSS33S3SSSSMUSSISM3XUS33MKXISSSII FREQ* 18 11 12 7 17

MINIMUM 12*00 26*20 40*40 54*60 68*80 MAXIMUM 26*20 40*40 54*60 68*80 83*00

PER CENT CANDIDATE SUPPORT FOR KENNEDY 1960

FIGURE 20 121

♦* OOO 000 003 o o o o o o o o o 30000000 00000030 o o o o o o

00+0 o o o o o 0300 o o o o o o 00000000 o o o o o o o o oooooooooooo 000000000030 000030000000(7+ oooooooooop4++ 0000004+ + y# •■• •• + + + + +U •• • • • • • 000j+++|<# ••••••• ♦ + 2 + <(*l*** 1** OOH- + +U l******** ♦ +TT+T** 002 + 4j» ••••! « • • • • « QQ4+++U**•••*••••• OOOOOOOOQ+ + + + +U •••••••••• 00000000++++++U* 004ZTTT+++2+++A1

SYMAP

FREOUENCY DISTRIBUTION OF DATA POINT VALUES IN EACH LEVEL

LEVEL 1 2 3 4 5 ssastisxsssssssxsssssassKMsasaaxiiissxsssaaHina •••*•*••• +++++++♦+ ooooooooo eeeeeeeee ••••••••• +++++++++ ooooooooo eeeeeeeee ■■■■■■■■■ SYMBOLS ****1 « **» ++++2++++ 000030000 666040060 I I I I 5 I I I I ••••••••• +++++++++ ooooooooo eeeeeeeee ■■■■eeeee ••••#••*• +++++++++ ooooooooo eeeeeeeee b b b h m h 3£=sss3333:s=S3aissxsass=j:ssxxsssxa33aESSsss3SXSSSsas*a FREO. 22 12 17 . 6 B

MINIMUM 10*00 19*00 28*00 37*00 46*00 MAXIMUM 19*00 26*00 37*00 46*00 55*00

PER CENT CANDIDATE SUPPORT FOR NIXON 1960

FIGURE 21 122

♦ — -s» i-----1---- 4.---- 2— — - 3— ---- ♦ — — ♦— — 5- — 4*

OOOOOO Aeaoooo., ______0300300 6 0 0 0 0 0 4 4 44 4 4-tXIO3604 O O O O O 66 OOOOOQBOOOO 86 oogCTee&Doo 66 000606094 eeeesa oooaeemeeeeao 444^0000600 ooo OCH-44 4 4 *60000306666 900 OObfefeLAO30000000 900 00003000000000000030 000 OOOOOOOa^TMOOOOOO 000000444444)0000 O O O O O 000300 OOOOOO 00

0000000 0000030 0000000 03000000 o o o o o o o g o

■56

SYMAP

FREOUENCY DISTRIBUTION OF DATA POINT VALUES IN EACH LEVEL

LEVEL 1 2 3 4 5

• • • • • • • • • 444444444 OOOOOOOOO 666666686 66666666 • • • • • • • • • 444444444 OOOOOOOOO 666660660 66666666 SYMBOLS ••••l#### 444424444000030000 666646666 6 6 6 6 5 6 6 6 • • • • • • • • • 444444444 OOOOOOOOO 660666666 66666666 • • • • • • • • • 444444444 OOOOOOOOO 666666666 66666666

FREQ# 22 17 14 8

MINIMUM 3# 00 15# 40 27*80 40# 20 52#60 MAXIMUM 15.40 27*80 40#20 52*60 65# 00

PER CENT SUPPORT FOR STATES RIGHTS CANDIDATE I960

FIGURE 22 123

2---- ♦---- 3 m r T T m r n

• •••••••»•*• »«« •• • • • • • • • • • > * *JF + ♦♦ ,,«•••••• ^H+44)UOOll*+ • 1«*« §A + + t *00000 000^00 • •••••• «^t t w« u i D a 8a 00 •••■•• *(♦♦♦00800000 ******* ,*♦++p0*4000000 • • • 1 • • » 0)0000400000 • ••■ *fRHlOOB0O0OflOfl0flfl ♦ ♦♦♦♦♦♦ntfootfeeeiBiiBBirteB a o a o ♦tfJUJCIU o o o o o o o 0000000 o o o o o o 00©000»3B000 ooooo 00000006040 o o c o o 060000666066066 oeoaiuiueeeeoee OB0000|i60050l0006 060i6B6«66m00qOO 0004a661116«B0OqOOO ooooodpf/^Boe) ^vooooo\ 6006nr ibbt ^ d o o or 10/ ^o

SYMAP

FREQUENCY DISTRIBUTION OF DATA POINT VALUES IN EACH LEVEL

LEVEL 1 2 3 4 5

...... ♦♦+♦+++++ OOOOOOOOO 000060066 ■060B6666 ...... ♦♦ + ++♦♦ + ♦ OOOOOOOOO 000666606 ■■■■■■■■■ SYMBOLS , , , , 1, , , , ++++2+++* 000030000 666046006 B6BB56BB6 • » • « • • • • « ♦♦+♦♦♦♦♦♦ OOOOOOOOO 606606660 BBBBBBB60 • • • • • • • • • ♦♦♦♦♦♦♦+♦ OOOOOOOOO 660000660 BBB060B06 FREQ, 34 5 6 II _ _9

MINIMUM 0*0 17,54 35,08 52*62 70,16 MAXIMUM 17,54 35,08 52*62 70*16 07*70

PROPORTION CATHOLIC POPULATION, 1971

FIGURE 23 124

C D D T H B B O T mroo 00000468 000300 03000088 OOOOO 00000806 OCOOOO 0000008888 30000000 0000000 00000300 o o o o o o o o o o 8888888800003000 88888 BBC OO 00Q * T 4♦ ■■■5(8884 808(00000Off♦ ♦ 2 888888(000034 ♦ ♦♦♦♦ 8886888800004 ♦♦+*++ OOOCf+***fm »^|+ oocoooaarooaa4+ + -ni* i.k 5T?>ppooood+ ♦ 2 ■» +\_aJ+ •f + 2* + T > 0 0 0 0 4 ♦♦ + ♦♦♦♦♦

o o o o o 000030 a o o a o o 0300000

0300 o o o o o

o o o o o o o 0003030 o o o o o o o 30000 ♦ ♦♦♦ ■f 42 + + ♦ ♦♦ ♦ ♦ ♦♦

♦* SYMAP

FREOUENCY DISTRIBUTION OF OATA POINT VALUES IN EACH LEVEL LEVEL ____ 1 2 3 4 5

OOOOOOOOO 888888888 ■■■■■■■■■ * • • ♦ ♦♦♦♦■»■♦♦ + o o o o o o o o o eeeeeeeee ■■■■■■■■■ SYMBOLS • 1. ♦ ♦ + ♦ 2++* + 000030000 888848888 ■■■■5U«B . . . o o o o o o o o o eeeeeeeee ■■■■■■■■■ . . . o o o o o o o o o eeeeeeeee ■■■■■■■■■ =="sss*sas: FREQ. 8 18 19 14

MINIMUM 6.40 18.34 30.28 42.22 54. 16 MAXIMUM 18.34 30.28 42.22 54.16 66. 10 PROPORTION BLACK POPULATION I960

FIGURE 24 125

i—— ♦— -----1 —— -z -----♦------3—

* 1 * * * * * * * * * • •• */t 4/tf* • •• • • •• • • • • • • ) ••••••••• 24*••••••••••••< ••«•*•••• 1*** I*******! • • • • ••••l**** •••••• 1 • • « • •( ««*! •••••••1 4 444*+ + 444 ( 4 4 ___ >■444 000 >444444 000 >442442 3f> 4 444 44 oooooog/> 4 444 4 4| 0000(^444 4 4 44 0000(4 4 4 44444 4 000V 4 2 4 4 4j/5\> 4 4 0G0Ot4J>€00OytJk,+ eeeeeaoooaooooooo3oai flfleeeegBOOUQyflBeaooow • • • • UU0UUT44'44TT? BneeeegfVliVilV 40000>4424j jebcpeeafnaaeasaaa OOOOOt»4 44 4( eeo4eeaii5ee«e«ee5»0eeDoov4oO\>4 4 44000003 0 0 5 5 5 S c Beeeeeeaaaaaaaaaaaa ooo^taooQfSetaoocoooot eeeeeeeqaaaaaaaaaaaa QQ 30Q00Q(B4eMJ OOOOO 30( e e e e e e e e e aaaaaaaaaaeee oooooqeeeeeeda v)0 0 0 ( e e e ^ e e e e t saaaasaBBee400100300^64060 e e e e e e e e e aaaaaaaaa^eeeeespggBeeee ieeeeeef aaaaaaaaaQfifieeeeeeBeeee4e jeeeee« eaaaeaBaaaaaaapAeeeeeeeeeeeejHeSfcee me e t eeeee4B4eeeegaeaate eeeeeeeeeeeeaaewee eeAeeeeeeeeeaaflee eeeeOoaeeeepjBe ooo OO3OO&0 oooooctempo o o o o >• *<

FREQUENCY 0ISTRI BUT IJJN QF DATA POINT VALUES IN EACH LEVEL

LEVEL

444444444 OOOOOOOOOe e e e e e e e e ••••••••• 444444444 OOOOOOOOOe e e e e e e e e aaaaaaaae SYMBOLS ••••I**** 444424444 000030000 e e e e A e e e e aaaasaeae • • • •#£••• 444444444 o o o o o o o o o e e e e e e e e e aaaaaaaae ••••••••• 444444444 o o o o o o o o o e e e e e e e e e aaaaaaaaa : 3 S S H 8 X S 3 ! : u t a s * M t : : 3 i i r z i 2 » FREQ* 20 8 11 14 12

MINIMUM 0*0 1 9 * 0 0 3 8 * 0 0 5 7 * 0 0 7 6 * 0 0 MAXIMUM 1 9 * 0 0 3 6 * 0 0 5 7 * 0 0 7 6 * 0 0 9 5 * 0 0

RACIAL POLITICAL EQUALITY INDEX: BASED ON THE RELATIONSHIP OF ELIGIBLE REGISTERED BLACK TO WHITES IN 1960

FIGURE 25

» ♦ * SYMAP

FREQUENCY DISTRIBUTION OF DATA POINT VALUES IN EACH LEVEL

LEVEL 1 2 3 4 5

••••••••• +♦+++♦♦♦+ ooooooooo eeeeeeeee ■■■■■■■■■ ••«••••«• +++♦+++++ ooooooooo eeeeeeeee ■■■■■■■■■ SYMBOLS «***i***« 2 ♦♦ ♦ ♦ 000030000 eeeeAeeee ■ ■ ■■ su m ••*•••••« +♦+♦++♦♦♦ ooooooooo eeeeeeeee ■■■■■■■■■ ••••••••• ♦♦♦♦♦+■»•++ ooooooooo eeeeeeeee

FREQ. 23 18 8

MINIMUM 1265*00 1752*20 2239*40 2726*60 3213*80 MAXIMUM 1752*20 2239*40 2726*60 3213*60 3701*00

PER CAPITA INCOME 1968

FIGURE 26 * ---- +---- 1---- +---- 2--- -♦— — 3 4 9----- SYMAP

FREQUENCY DISTRIBUTION OF DATA POINT VALUES IN EACH LEVEL

LEVEL 1 2 3 4 5

••••••••• ♦♦♦♦♦♦♦♦♦ ooooooooo eeeeeeeee ooooooooo ••••••••• ♦♦•#■♦♦♦♦♦♦ ooooooooo eeeeeeeee ■eeaeeeee SYMBOLS • • • • !• • • • ♦♦++2*+++ 000030000 860040000 000050000 ••••••••• ♦♦♦♦♦♦♦♦♦ ooooooooo eeeeeeeee ooooooooo ••••••••• ♦♦♦♦♦♦♦♦♦ ooooooooo eeeeeeeee 0 0 0 0 0 0 0 0 0 9issss«aHara»ssssass3uaaaaaasMaasaBanscs3ass« FREQ* 41 19 2 2 1

MINIMUM 20*00 68*20 116*40 164*60 212*60 MAXIMUM 68*20 116*40 164*60 212*80 261*00

CHANGE IN PER CAPITA INCOME 1960-1968

FIGURE 27 128

4* H44444444 00000000000000096 44 44 4 4 4 4 4 1103000000300000 444 42442 44MOOOtoOOOOOOOgB6*M5 4244 444444>k)OOA4TrM)0000666 ♦♦♦♦♦♦♦♦^Oooo»»2-»4»yrKi6ee 09001X0000000004444442409*8000000 000a0G00003Q0q444444j>4XJ§666666566 OOOlf 424 4 4xJ0aoo(Be 0 0 0 3 0 6 0 0 0 3 0 0 0 0 0 0 o o o o o o o o o o o o o o ooaoit-42 00030(4 44 030 o o o o o M F Q o oooaaooooo ooooooooogfieeee 0000^1000600000 OOOOOOO00600000 00000000666*666 0000000666668866686686866 0 6 0 0 0 0 0 0 00000000000666666666666*690*6 8 6 0 0 0 0 3 0 000000000066* 66686666*6660 OOOOO 00030000036066866*66866600*4 0 0 3 0 0 0 0 0 0 0000000606666015916(1000014 4(1 *14 4 4 4 4>cgoooooo0eooooooooof24 +(^^4 4 4444 490000000000600000004444 4444 444 0 3 0 0 0 0 3 0 444444 O O O O O O O 42444444 444444 O O O O O O O 4444444 O O O O O O 4442444 0 0 0 0 3 0 0 00000000 oo o o o o o 0000

4* SYMAP

FREQUENCY DISTRIBUTION OF DATA POINT VALUES IN EACH LEVEL

LEVEL 1 2 3 * 5 ssaassaaasassssasxxsssassMSBsasasusaaaHNnsaasi • • • • • • • • • 444444444 OOOOOOOOO 666666666 660666066 • • • • • • • • • 444444444 OOOOOOOOO 666666666 000066666 SYMBOLS • • • • ! • • • • 444424444 000030000 6686*6666 066050000 • • • • • • • • • 444444444 OOOOOOOOO 666666666 000000000 • • • • • • • • • 444444444 OOOOOOOOO 666666666 000000000 ss33S333s«MM«ssza3r*3**3itas»ssasssaaaa3M3ssasi FREQ. 9 17 21 11 7

MINIMUM 12*30 22*1* 31*98 *1*82 51*66 MAXIMUM 22*1* 31*98 *1*82 51*66 61*50

PROPORTION OF FAMILIES WITH DISPOSABLE INCOME LESS THAN $3,000

FIGURE 28 129

- 2- — + - — 3---- ♦---- ♦------s--- 000000000p03000011»24 2 ooo

0000

BB

SYMAP

FREQUENCY DISTRIBUTION OF DATA POINT VALUES IN EACH LEVEL

LEVEL | 2 3 A ®

• • • • • • • • • ♦♦♦♦♦♦♦♦♦ ooooooooo eeeeeeeee ■■■■■■■■■ ♦♦*♦♦♦♦♦♦ ooooooooo eeeeeeeee ■■■■■■■■■ symbols .m i*!!* * * * * 2 * * * * 000030000 eeeeteeee 1BBBSBBM symbols • • • • i# .# # w ooooooooo eeeeeeeee ■■■■■■■■■ !!!!!!• • • ♦♦♦♦♦♦♦♦♦ ooooooooo eeeeeeeee eeeeeeeee ssass*xs=s**=**=s**»*=**=»******■*»**********»=**»*s FREQ# 21 20 10 7 7

MINIMUM 4#70 J2*Sf 23?24 lllSS 35^60 MAXIMUM I0#8B 17#06 23#24

PROPORTION OF FAMILIES WITH DISPOSABLE INCOME GREATER THAN $10,000

FIGURE 29 130

-5- ----♦ •

♦♦♦♦♦♦♦♦♦♦♦♦♦< ♦♦♦♦♦♦444444< ♦♦♦♦♦4TT2*44 4-i 10 4 ♦24-MO300001 ♦♦♦-tbooooot f*± ± , 0000004♦♦♦< OOOOO] ♦♦♦«•••/ 000030p£44< 0003 oooag<44444 oooo 00000^44424 < oooooo ♦ 44^ 100000 00 4/ \00000 0000000424 4(3000 loo 00 Qf l W OOO* 4 40000 103000 ooaBoooo d#6ooo3 ooooooooooooooooooooaoo oooocAoooooooooooooooo 44444^10003000000000 4444244^000000000000 ♦ 4 4♦4 49000000000030 [♦ 44 44 4 0 0 0 0 ___ OOOOOOO (4 ♦♦ 4Wooooof seeeseie 03001 ♦ 4 4 4-*{00000(184688861 44100000 444244000300066664 4444(0000301 4 444 444000000066666 ♦4244(0000001 4 444 4 4400000000000666 r»4 490660001 *_4><5lM;44TmX2pOOOO O O O O O 0003(^4444 42^60031 0 3 0 0 0 0 3 0 r4>44 4444 4 44 44 40044 000000000^43 ♦♦♦♦♦♦♦♦♦♦♦44 4 4 444V•>£♦ 006(000(42 4 444 4 ♦♦♦♦♦♦♦44444444 4444 O W 4 4 44 44\v4 4 < •♦♦♦♦♦424 44 4444 4424 44^ 5®30P44J» • • • J24442frj 144 44444 4 444 44 4 444 444J ♦ •••••••••••• t.4444444^ ► 444-1 ►••••••••••••••I )•••!•••••••••< >•••••• V' * •• i >••••<

>4*

4 FREQUENCY DISTR16UTI0N OF DATA POINT VALUES IN EACH LEVEL

LEVEL s3xxKxxssxx3»uaii£iai»aiaax«»*xii«««it»M«aia] 1 2 3 4 5 • • • • • • • • • 444444444OOOOOOOOO 666666666 666666866 • • • • • • • • • 444444444 OOOOOOOOO 666666666 666666666 SYM60LS • • • • ! • • • • 444424444 000030000 666646866 666656666 • • • • • • • • • 444444444 OOOOOOOOO 666666686 668008688 • • • • • • • • • 444444444 OOOOOOOOO 666666666 000006000

FREQ* 15 20 20 5 5

MINIMUM 2*50 4.08 5.66 7*24 MAXIMUM 4*08 5*66 7«24 8*82 10»4Q

PER CENT UNEMPLOYMENT 1968

FIGURE 30 131

♦ ♦♦ ♦♦♦■MOO ♦ ♦♦♦♦♦■MOO ♦ ♦♦^l)f+poo ♦2 +*>++ ♦ ♦ ♦ ♦ ♦ ♦♦ 00000030000 OOOOO oooooooooooo 0300

o g a o 0600 000 6606466 oo o o o e e e e e e e e e e e O O O O O O O o o o o o o 0 0 0 3 0 OOOOOOOOO 0000 C003C0000 O 3 0 0 0 0 * ♦ 0000 OOOOOOOOO*+/1 .*«C)OOOOCK* +++ 000 ooaoooo^-Hff^iKaoooofr ♦*+2 ♦ ♦ ♦*pooooaoav 3 2 £3«ooo ♦♦♦03000030000300000 ♦ ♦♦ 2 * j»OOqOOQOOOQ>F*30»0 ♦♦♦♦++TT^Q000004+2+ ♦ ♦♦♦♦♦♦2 ♦PQOOOok*♦♦♦ ♦OZ ♦ 2 ♦♦♦♦*.♦♦♦♦ 3 0 ♦ ♦ ♦ ♦ * 1 ♦ ♦ ♦ ♦> w 4 ± ± * ♦ ♦ ♦ ♦ ♦ ♦ ♦♦A^ ♦^ ♦ ^ *** * ♦ ♦■w • • • • • 1 •

SYMAP

FREQUENCY DISTRIBUTION OF DATA POINT VALUES IN EACH LEVEL

LEVEL 1 2 3 4 5

••••••• • * ♦♦♦♦♦♦♦♦♦ OOOOOOOOO 666666660 006000000 • • 444444444 OOOOOOOOO 666666666 000000000 SYMBOLS ••••!•• • • 444424444 000030000 666646666 000050000 ••••••• • • 444444444 ooooooooo eeeeeeeee 000000000 ••••••• • • 444444444 ooooooooo eeeeeeeee 000000000 = = s FREQ* 10 14

MINIMUM 14*00 32*00 50*00 68*00 86*00 MAXIMUM 32*00 50*00 68*00 86*00 104*00

PER CAPITA PUBLIC ASSISTANCE 1967

FIGURE 31 132

— t - -l - »- r ; - *z z ~ ~ —♦----♦----♦----8----♦* >000000060000000000 ►♦♦■( ' ioo3ooggoo30oooo 30001 lOOOOOQfiedQOOOOOOOOO ♦2( )ooogfeeeBiQQogo ♦ ♦♦•< joooeeesee _ loaooooaaaaseeA ♦♦♦♦♦♦< >00000^300000 __ ♦ ♦♦ ♦ ♦ 2 +#> OQOOQ+TT++40000030000 +♦♦♦++♦♦4000000000 2 4 -f ♦ 2 ♦ ♦♦ -HOO0 0 0 O O O 3i ► ♦♦♦♦<< 103000001 ►♦♦♦I ♦ ♦ ♦ ♦ ♦ ♦ ♦ ♦ ♦ ♦ ♦ ♦ 4400000000 ♦ ♦H +++♦ + ♦+♦♦♦♦{000000001 ■♦♦♦♦ - !• + ++♦ + +2 + + + +P 0 Q 0 0 0 3 0 !♦+♦♦< )0000y+ + + + ++ + 0 0 3 0 0 0 0 0 ?0300CT*+++/00 OOOOOOOOO joooooznsoooooooooooooo >0000000000000000 )000000000003000f+++ + ++ >000030000000000++++++ \OOOOOOOOOOOOOQf+++2 + ++ )OOOOOOOOQrrrr»++++++»++ ♦ ♦ ♦ ♦ ♦ ♦ ♦ ♦ ♦ 0066661061 JOOOOOO/++ + + T** ••••♦ 00000301 looooooo ++++-41 •••«•••• 11+ oooooooooc I0003000C+2 ♦♦•4****I***** 00030000001 IOOOOOOOC ♦++♦++»• >«>«♦««« oooooooooc 0 0 0 ♦♦++♦+ OOQ0O3OC Q + + + + + + + + 00( 0 +2++++2 + + + f 00 ♦♦ + ♦♦♦♦ ♦ ♦♦ ♦ ♦ ♦ 2 +! 00c ♦♦♦♦+♦+ ♦ ♦♦ ♦ 2 + + 4 0 0 » + + ++ + l»+>0>¥++2 + +++{00 ♦♦♦(er*Tv+♦♦♦♦♦♦ ♦ ♦ 1QQQQQ + + + +++JL* • • .\fct4rv+++ ♦♦ ♦♦♦♦ 0 0 c ♦FTV?++++2+ 00003C ♦ ♦♦♦♦♦♦♦•♦♦•FVQOOOt ♦♦♦2 ++++il+++++{ Voyoooi ♦♦♦♦♦♦y\+++^ ♦ ♦♦♦V V r ♦ l

■♦*

FREQUENCY DISTRIBUTION OF DATA POINT VALUES IN EACH LEVEL LEVEL 1 2 3 4 5 ssssssssstsnssissxssssssiCMiacsuKsisunnsasai m i ♦♦♦♦♦♦♦♦♦ ooooooooo eeeeeeeee ■■■■■■■■■ •••••*••• ♦♦♦♦♦♦♦♦♦ ooooooooo eeeeeeeee ■■■■■■■■■ symbols ****i**** ++++2++++ 000030000 eeee4eeee misiaii ♦♦♦♦♦♦♦♦♦ ooooooooo eeeeeeeee ■■■■■■■■■ ••••••••• ♦♦♦♦♦♦♦♦♦ ooooooooo eeeeeeeee ■■■■■■■■■ 33S3333::s:s3ts3ss3:8:ssBxs«iiasM«isa»ssHaMssi FREQ* 8 24 25 _ 3 5

MAXIMUM 6*54f #2° 6#5A7*88 79*22»S® 10*589»22 1 11*90? * t $

EDUCATION: MEDIAN SCHOOL YEAR COMPLETED BY PERSONS 23 YEARS OLD AND OVER, 1960

FIGURE 32 133

♦- ♦» 1 2 ► ►--•—A——4—--—5— —4* e e e e a o o o oS t O O O ^ T f 500000300?V + 303000QG00 lOOOOOOOOO OGGI lOO/RQOOO Jet ^»4jrnooo 348f ♦ ♦T T ? ♦♦♦442 4444444-w**** 144421 1 • • *J4 ♦ m • 1 • • • 4 4 ♦P 3 f 4 • • • • • 094 44V* »**••• ♦ ♦|T« • • • • 1 • > O k 4 4L * !«••••• F4 IG00v 4-W« •••••••■! h4 4 OOtfc 4 ♦ ♦ * 4 4 4 4 4 4 4| O O Q ^44 0 0 0 0 0 0 0 ^44 44-4 4444oooo 4 4 4 4 4 • • • lttJf24V> 1 300 3 00J ►4444444 4 44\Q0 3 4 4 4 4 JOOOO( 3t3\fc424 444 42 4 4 > 00 _ 4 4 • M y 7 w t p m i / r 1 ** t jooooooOOQOOemcoooo 4 44 44 ^G SBBbjaU 4 4 4 ►444V lOQOOOOOOOOOOOOOOOO 4 4 4 4 4 2D a a n b W 4 4 4 4j 300 0 0 0 0 0 0 00CM0 44 444^000^4444 iee4eedSdi4d6d03Bee 44442444^244 ?ea0CTgO000Dtfo0fee4 C o 50»4 444 4 444 4 }0QflgD44>Q0000Q00O8eee 000(44 444424 44 ► 4 4 4 4*Vfr 47 ^ 0 0 0 0 0 0 0888 00 0 * 4 4dfS*l4 4 4 4 4 ► 4 4 + ^{#1 * * • W^ Q ^ + S p O O 88(0000)24a ••14444 14 4 H a m ■««!»■« 4 4 4 4V0 30044^/74444 mOQ44444444Doa e e g ooc 30000000 00000$ 30000000 44444H 300030000 4 444/" \j*444< 30000000, 4444) \t***4V 3000 4 V \44 4J 30 •4*

FREOUENCY DISTRIBUTION OF OATA POINT VALUES IN EACH LEVEL LEVEL 1 2 3 4 5

444444444 OOOOOOOOO 888888888 •••••• 44-4444444 OOOOOOOOO 888888888 ■■ 888888 SYMBOLS 444424444 000030000 888848888 8 5 8 8 8 8 •••••• 444444444 OOOOOOOOO 888888888 888888 •••••• 444444444 OOOOOOOOO 888888888 SS = 3 3 = FREQ< 15 22 13 10

MINIMUM 0*0 19*94 39*88 59 * 8 2 7 9 * 7 6 MAXIMUM 19*94 39*88 59*82 79*76 99*70

PER CENT URBAN POPULATION 1970

FIGURE 33 *— — <1 — 1 — — ♦---- 2 SYMAP

FREQUENCY DISTRIBUTION of data p o in t values in each le vel 1 2 3 4 5 LEVEL ■xxxaxaaai■X1X33X3

MINIMUM 0*0 9*96 19*92 2 9*88 39*84 MAXIMUM 9*96 19*92 29 * 8 8 39*84 49*80

PROPORTION OF THE POPULATION CLASSIFIED RURAL FARM 1960

FIGURE 34 * 1 2 3 4. . . . 4.----- 5----- +0 SYMAP

FREQUENCY DISTRIBUTION OF DATA POINT VALUES IN EACH LEVEL

LEVEL 1 2 3 4 5 ••••••• +♦++♦♦+♦♦ ooooooooo eeeeeeeee ••••••• +♦♦+♦♦♦♦♦ ooooooooo eeeeeeeee SYMBOLS ••i**** ++++2 ++*+ oooosoooo eeeeAeeee eeeeseaae ••••••• +++♦++++♦ ooooooooo eeeeeeeee aaaaaaaaa ••••••• ♦♦♦♦♦♦♦♦♦ ooooooooo eeeeeeeee aaaaaaaaa ss:s:3:3:ssa3Ss3S3sxs3»siz*axMM33sssMMstaat FREQ* 31 20 9 4 1

MINIMUM 0*80 11*40 22*00 32*60 4 3 * 2 0 MAX IMUM 1 1*40 22*00 32*60 43*20 5 3 * 0 0

PROPORTION OF THE TOTAL WORKFORCE IN MANUFACTURING 1960

FIGURE 35 136

* ♦ 1 + 2 - •5»— 444*T444MJUUUjmn • ••••••rr444244444MQ|4444i • ••• ,1****<44444444444 44444< *b •••••]♦ 444 4444 42 4 4 4444 ^DDDDD»*♦♦♦♦♦♦2♦ C0000300P*44 4 4444V o H D Q Q 0030( 24 444 DOOO oofr ♦ ♦ + ► 2 4 * 0 0 0 0 0 0 ♦Pqojj4 ♦♦♦♦♦/ 300000004+44T*4?442 44444 +4 4 4 4 4 4 4 4 4 4 4 ♦ 4 FQ 0 0 0 C 0 0 0 0 C M 4 4U * *> 4 44 4 4 44 4 4 4 H 4 44 44 4 4 444 4 4* 0 0 0 0 0 0 0 3 * *fc* • *J&4*♦ 2 * H ♦ ♦ ♦^Cod^ ♦ ♦ *♦ «pflpooo ±jP±yy**+'*- ♦4A• 44 444 4 44 444lk*QQp000( i* *^ 44444444444^ 444*001 ♦ 4'4 44 4 44 4 44 4 4 4 4 * 2 «O0( [•!••• l«» *j 4 4 4 44 44 44 4 4 4444i roogfrl&K^H*! ► 4444-1 4 4 4 2 4 4 4 4 4 - 44 444 44 4 4 4444

•4* freouency distribution of data point values in each level

LEVEL 1 2 3 4 5 M 3 M S * t « i H H i * 3 tuaa*ii*isiU M siauiaM susiini 4 4 4 4 4 4 4 4 4 ooooooooo eeoeeeeoo 000000000 ••••••••• 444444444 O O O O O O O O O 006660060 060660660 SYMBOLS ****1**** 4 4 4 4 2 4 4 4 4 000030000 6 6 6 6 4 6 0 6 0 660050606 ••••••••• 444444444 O O O O O O O O O 666666600 000000000 • • • • • • • • • 4 4 4 4 4 4 4 4 4 OOOOOOOOO 666666666 000000000

FREQ* 10 25 __ 17 9 4

MINIMUM 5*90 9*76 13*62 17*4S 21*3* MAXIMUM 9*76 13*62 17*46 21*34 25*20

PROPORTION OF THE TOTAL WORKFORCE CLASSIFIED AS LABORER 1960

FIGURE 36 137

4* > • • • • • • • • »••••••••••] ••• 1••1 * I •••••• 4444444

4444444 4 444 44444 4 4 4444

4444 444 4 444444424 44444 44 44 44 4 4 4444Jfc444 44 4444 4444 44 4 444-KBOOOO»44 4444444 4 44 4 l44444t300000()dkf4424 42444 t« >>444)0000000304 444 44 44 0000044 4 4 44 44 44CQ(BaeeaoooQir44444 44444444 2444444444] eetaas

0/^4 44 4 4444 4 +jf a a a a 444 444 444ja a 1 a a 42 4444 42j* a a a a a 4 44 444/C* • •• • • • ••••••••••••a laaaaaalaaaala aaaaaaaaaaaaaa.

k W a a a a a a a

— 4- 4* SYMAP

frequency distribution of oata point values in each level

LEVEL 1 2 3 4 5 SS3SMSMXMSX*>SMKSt*SSH*SI***S**a*S****M>aaw •aaaaaaaa 444444444 OOOOOOOOO 6B866B668 IHIIIIM ...... 444444444OOOOOOOOO 888888888 888888888 SYMBOLS aaaalaaaa 444424444 000030000 888848888 888858888 •••••

MINIMUM 6a30 I5a06 23a82 32*58 *i*f± MAXIMUM 15a06 23a82 32a5Q 41a34 50a10

PROPORTION OF THE TOTAL WORKFORCE IN GOVERNMENT 1960

FIGURE 37 138 & 1 1 1 1 1 1 1 t 1 •

> > < X1 (A 1 (A i'O-ft < pi r > m s s in c > ID < r Tl > x o 9 c z ■n o O •< PI 9 o r% w S o z O •«* O 'PI

m < m r r

H N Ul N N N N N N N N N M N * N ft ft + tN+ oou |s- ft >+ SBB" * a (A r o ft ft X o (A ft ft ftft ft ♦ ♦ ♦ 4 ♦ N ♦ 4 H ♦ ♦ ♦ ♦ ♦ ♦ ♦ ♦ OOOOO N OOOOO ♦ + ♦ N OOOOO N OOOOO 0 0 * 0 0 ooooi ♦ ♦ ♦ ♦ ♦ N OOOOON OOOOON OOOOON N OOOOO N OOOOO N OOOOO N OOOOO QOOOON N OOOOO U OOUOON OOOOON OOOOON OOOOON SSSSS OOOOO N OOOOO N • • N lift ft lift II IS) 9 T» • * N »■ IK O •

I o ** XX XX o o ■go ■go (DU o « o o x z o o o o MO cc XX OS UIO ft ft utv ft out UIS o o ft • o • • • o o ft ft ft ** u • • •

POPULATION CHANGE 1960-1970 139 >46 0000300000030 966940 30000000*0000B806 660000000044*000906666 eeeloooooo»42>oaG seeeeeee eeeeaoooooaatjoooG *6664681 I009\ 66666(000300000000016866668 10501 6666600000000030001 6666666600000000000006664 666486646000000000003 8666666000000003000000(0060 0501 666669(00000000000000018666 66C0Q0000000000000000666 6000000300000000003001*6 000000000000300000 30004T>OQOQB6QOOO 1050 aog/t 4 4 4 4oo cbeeeeBBe 4 6Dol6eeeeeeee f** • • • i o ueeeeeeeec leeeeeeeec >eeeeeee4e< leeeeeeeeec k*44^ 0000 ieeeeeeeec 150000000501 OOP reeee864eei 666766666466068 5061 161 66666666666009601 >466C I6666( 6660666666066666 1661 66666668006666080 6664066660*6660466^ 00500005000 66606666696666666644000000000 >060080900 eeeeeeeec 0000000000000500001 >666666 eeeeeeec 00050000000000000051 >66666 W000Bee6646( 10000000505001 t 6 0 6 4 0 B 0 8 f >6600 leeeeeeee 00000000000000001 >666666/ 166666661 1050000000000000438666661 >6661 >6646< I6C 100000001 10000/

>44 SVMAP

FREQUENCY DISTRIBUTION OF DATA POINT VALUES IN EACH LEVEL

LEVEL 1 2 3 4 5

• • • • • • • • • 444444444OOOOOOOOO 6 6 6 6 6 6 0 6 6 0 0 0 0 0 0 0 0 0 • • • • • • • • • 444444444OOOOOOOOO 066666666 0 0 0 0 0 0 0 0 0 SYMBOLS ****1 **** 444424444000030000 666646666 0 0 0 0 5 0 0 0 0 • • • • • • • • • 444444444OOOOOOOOO 666666666 0 0 0 0 0 0 0 0 0 * • • • • • • • • 444444444OOOOOOOOO 6 0 6 6 6 6 6 6 6 0 0 0 0 0 0 0 0 0 »3MMIt3SH33SaSUSSS*l*>Sa3MSS3M4*SX3ISE::SS» FREQ* 2 1 1 5 22 25

22*38 26*26 30*14 34*02 MAXIMUM 22*38 26*26 30*14 34*02 37*90

PROPORTION OF POPULATION UNDER 15 YEARS

FIGURE 39

O *■» a a i f u u u i f t H z V a o o n o > < o S < a > ffl z r c ffl (A ffl HI ? O > X m ffl m o p < p ■c

< r p ffl ni

N N n n n ft N ♦ ♦ N ♦ ♦ -f 0®0 r < o a (Dan) it it (Dan) ® OO®®® II oooaan OO®®® ® » o » on ® 00 ®ao®o 0 • ♦ ♦ ♦ ♦ ♦ H ♦ ♦ ♦ ♦ ♦ ooo®o II ♦♦ N M N HOOOOO H OOOOO M N OOOOO H N h uoaoo® H ♦ ♦ ♦ ♦ N N ♦ H ♦ ♦ ♦ N OOOOO M OOOOOH N N OOOOO II OOOOOII U N N oaxDaa II it h N N h N N OOOOO N OOOOO N !!■»■■■ ))■■■■■ II M■■■■■i ))■■■■■II H■■■■■N ■■■■■NM■■■■■N H■•■■ON II H■■■■■N U U N N N N ffl a rv 30 M H M H M U IiaiU I * N OOUOO N U N OOUOO N * ffl• z ** m m ** O o

* n >UI z z >*■* xz cc z z (ffUl (ffUl O'** o>o 01 • • • 0 SOI ••o ••o MO' O'O' a*- ♦ a ♦ sa ® • • • MO ss SM wa a s • • • O* Oft • • • • • •

PROPORTION OF POPULATION 15 to 65 YEARS OLD 141 o o o n o o o m C o o o o o o o o o 4 t k C O O O 4 0 0 0 0 ♦ ♦ 4 ♦ tpnQ4+ 4 Q n p t 4 U O O o o o o o o o o o o o o 4*^4 4 4 ^ * 4 4 0 0 0 4 4 4 4 / 4 0 4 0 0 0 0 + / o n +

I + + g rtO • o o o o o o o o o o o o o OI+ 0 - - *k> □ q-*- 4 ♦ Pfc ♦ 4 4 4 7 4 0 0 4 0 4 / 0 4 0 4 0 0 4 J

0 O O O O O O O O O O of o o o n 4 + ('!♦♦*♦ + 4 N ('!♦♦*♦ 4 ♦ ♦ ♦ c\i + + + + + + + + + + + 444 4 4 ♦ Q / 0/4 4 4 4 4 o o o O o o o O o o O o o O a ^ i i t O O 4 + 4 0 0 0 o o o o ♦ ♦ ♦ ♦ ♦ ♦ ♦ ♦ ♦ ♦ ♦ /1 • + + 4/014 4 ♦ ♦ 4 ♦ 4 4 ♦ 4N4 ♦ 4 4 r + + + QI + + + + + + n +ry+ 4\» • • * ® 4 V 4 4 401 4 440O

♦ ♦ ♦ ♦ ♦ ♦ ♦ ♦ ♦ ( © 4 ♦ 4 4 ♦ ♦ > 4 4 4 4 ♦ 4 ♦ 4 4 4 4 ♦ 4 ♦ 4 4

♦ 4 4|o|4 4 4 4 4/0000 0 4 0 0 4 0 4 / 4 V • «I4 4 4

4 4 #k4N4

4/#«* • ♦ 4 4444 4 4 o ♦ 4 oo oo 4 • •

14 a UI 0 ui o 0 z u i o ui lil J > O a o UI Ui < M O H a o z o u. < 0. o ro M z > o < -I o vt > UI UI - _l > J N H II N 000 j 9000009 o 9 N o M O n OOOOO o o O 1 n - 9 1 O 0 0 O 9 0 O 0 0 9 N 8 K 0 N 8 0 8 II 0 4 0 0 0 8 laaaaan 9 H 9 O O O O O I I N O O II O O O O 0 O O O N O 0 ll 0 8 ® 0 0 4 N 0 8 H $ 0 0 0 li 0 0 0 M 0 4 H8 0 N 0 U OOOOON II u ooooai N n a B a a a M 0BBBBBH iiin Niaaaalt a o B B R 9BBBBBN II II N n ■■■■■■! n i a a a B H 000009 B B B B B 9 N 0 0 0 0 4 401 4 4 4 4 4 44 4 44 44 4 4 4 44 44444 44444 44444 44444 • • • • • • 4444 • • • • • • • • • « 4 • • • — • (A z 0 > 8 0 -1 • • • • n • • • • • • • • • • • • • •no* *11 > N >11

II 11 H M N H 11 0 II n n n 9 II H II 4 K ti­ n N

n

ui IL a 0 •

N O OO— —n O N — O 00 • '• «4 «4 «« o» #0 #0 00

1 On § M 0 2 OF POPULATION 65 AND OVER 142

•••••••••• T n r r r m

SYMAP

FREQUENCY DISTRIBUTION OF DATA POINT VALUES IN EACH LEVEL

1 2 3 4 5 LEVEL xssssssssssMstassssusassaasissMaaasunssaaaaBM *•••••••• ♦♦♦♦♦♦♦♦♦ ooooooooo e e e a a e e e e aaaaaaaaa ••••••••• ♦♦♦♦♦♦♦♦♦ ooooooooo a a a a e e e e e aaaaaaaaa SYMBOLS ••••i**** +*++2++++ ooooaoooo e e e e t e e e e aaaasaaaa ••••••••• ♦♦♦♦♦♦♦♦♦ ooooooooo e e e e e e a e e aaaaaaaaa •••*••••• ♦♦♦♦♦♦♦♦♦ ooooooooo e e a e e e a e e aaaaaaaaa SaSSSaSltSltlSaXMSSMSMIMMSXaSIMMSIKSKliM1 FREQ* 33 2A 5 I 2

MINIMUM 1*40 22*70 44*00 65*30 MAXIMUM 22*70 44*00 65*30 86*60 107*90 VIOLENT CRIME INDEX 1967

FIGURE 42 APPENDIX III

AN INTERCORRELATION MATRIX OF SELECTED AGGREGATE LEVEL VARIABLES FOR THE 64 LOUISIANA PARISHES

143 TABLE XXXI AN INTERCORRELATION MATRIX OF SELECTED AGGREGATE LEVEL VARIABLES FOR THE 64 LOUISIANA PARISHES

VARIABLES 3 4 5 6 7 8 9 10 11 12

3 Humphrey 68 1.000 4 Mixon 68 -.199 1.000 5 Wallace 68 -.798 -.429 1.000 6 Desecrate 48 -.090 -.062 .128 1.000 7 Republicans 48 .329 .572 -.654 -.003 1.000 8 States Rights 48 -.099 -.212 .215 -.869 -.483 1.000 9 Democrats 60 .271 .053 -’.286 .255 .377 -.405 1.000 10 Republicans 60 -.332 .403 .070 .023 -.090 .034 -.703 1.000 11 States Rights 60 -.103 -.406 .339 -.357 -.455 .522 -.805 .153 1.000 12 Democrats 64 .263 .101 -.304 .221 .394 -.386 .856 -.542 -.737 1.000 13 Republicans 64 -.250 -.096 .289 -.218 -.387 .380 -.852 .546 .729 -.991 18 Catholic .260 .242 -.392 -.065 .515 -.190 .874 -.638 -.683 .756 19 Black 60 .665 -.395 -.366 -.260 -.093 .256 -.343 .048 .424 -.333 20 Political Equality .063 .145 -.149 .208 .357 -.363 .849 -.541 -.722 .814 22 Economic Equality -.150 -.172 .243 .219 -.193 -.126 .099 -.108 -.052 .033 23 Mean White Wage .281 .211 -.386 -.039 .384 -.137 .197 -.101 -.188 .278 24 Mean Black Wage -.079 .170 -.028 .020 .172 -.094 .246 -.121 -.246 .202 25 Increase In Black Wage .070 -.189 .036 -.251 -.146 .294 -.293 -.078 .473 -.319 26 Change X in Black Regis. .437 -.309 -.216 -.322 -.214 .386 -.613 .227 .662 -.600 27 Proportion Black Regis. .880 -.390 -.570 -.230 .070 .146 -.075 -.153 .223 -.068 28 Unenployment 68 .118 -.451 .168 -.009 -.249 .127 .107 -.240 .057 .004 29 Change in Uneaploynent .154 .039 -.180 -.256 .089 .204 .139 -.235 -.015 .107 30 Incase Less than $3000 .115 -.515 .208 -.056 -.484 .274 -.313 .073 .374 -.429 31 Income More than $10,000 .085 .459 -.355 -.053 .557 -.219 .289 -.097 -.334 .357 32 Urban Population .049 .611 -.418 -.059 .379 -.121 .088 .104 -.191 .155 33 Rural Parser -.110 -.377 .341 -.013 -.401 .196 -.184 .073 .193 -.329 34 Rural Non-Farmer .026 -.483 .261 .038 -.237 .077 .002 -.165 .112 .036 35 Median Education -.351 .506 .015 .154 .058 -.161 -.329 .446 .100 -.172 37 Manufacturing Employment .101 -.119 -.003 .102 .049 -.106 -.075 .180 -.042 .021 38 Agricultural Employment .138 -.416 .129 -.002 -.250 .103 -.103 -.087 .213 -.244 39 Government Employment -.114 -.100 .151 -.008 -.272 .131 -.248 .096 .261 -.168 VARIABLES 3 4 5 6 7 8 9 10 11 12

40 Trade Employment -.108 .350 -.119 -.156 .114 .100 .007 .051 -.046 -.018 41 Union Membership .088 .217 -.204 .145 .168 -.201 .025 .136 -.137 .099 42 Welfare Index -.006 -.446 .289 -.006 -.460 .226 -.343 .263 .249 -.333 43 Dependency Index .260 -.331 -.034 .008 -.130 .043 .015 -.113 .064 -.037 44 Violence Index .315 .043 -.317 .063 .237 -.165 .229 -.263 -.106 .200 45 Population Change 70/60 -.218 .2lb .055 .071 .061 -.091 .214 -.098 -.227 .301 46 Population Density 70 .177 .339 -.353 .028 .290 -.166 .059 -.001 -.084 .118 47 Age to 14 .236 -.058 -.185 .099 .190 -.192 .462 -.450 -.280 .384 46 Age 15-65 -.176 .205 .029 -.054 .032 .040 -.056 .101 .002 -.001 49 Age Over 65 .047 .092 -.084 -.062 .069 .023 -.143 .210 .023 -.094 50 Income Per Capita 68 -.076 .379 -.166 .203 .293 -.311 .160 -.055 -.172 .205 51 Income Per Capita 68/67 -.175 .313 -.031 .044 .288 -.174 .313 -.059 -.387 .373 52 Income Per Capita 68/29 .183 -.120 -.096 .268 .199 -.328 .435 -.275 -.387 .400 53 Income Per Capita 68/59 .132 -.209 .000 .164 -.028 -.134 .114 -.153 -.037 .066 54 Farm Area .153 -.165 -.038 .081 -.006 -.064 .025 -.079 .066 -.144 55 Percent Black Regia. 60 .472 -.068 -.386 .147 .361 -.328 .711 -.486 -.593 .698 56 Laborers .282 -.549 .078 -.128 -.207 .206 -.078 -.159 .226 -.158 57 Professionals -.155 .508 -.166 .127 . .112 -.165 -.238 .447 -.034 -.119 58 Manager-Proprietor -.299 .749 -.194 .059 .348 -.202 .144 .208 -.358 .133 59 Clerical-Sales -.119 .589 -.247 .019 .264 -.138 .007 .163 -.133 .036 60 Craftsmen-Operators -.200 -.020 .204 .284 .026 -.252 .191 .034 -.297 • .317 61 Farmer 6 Farm Laborer .160 -.471 .141 -.107 -.339 .239 .000 -.210 .169 -.190 62 Change, White Wage -.002 -.315 .181 .096 -.206 -.012 -.099 -.187 .290 -.119 63 Change, Black Wage -.138 -.030 .141 -.101 -.117 .149 -.200 .119 .173 -.139 64 Latitude -.213 -.072 .253 -.055 -.422 .261 -.789 .749 .476 -.712 65 V.P., 71, Dec. Glib. -.089 -.017 .098 .279 .127 -.283 .469 -.204 -.461 .460 66 V. P., 72, Nov. Gub. -.155 -.188 .261 .123 -.073 -.068 .262 -.190 -.207 .208 67 Johnson - December -.411 -.107 .448 -.006 -.450 .231 -.785 .623 .556 -.648 68 Edwards - December .415 .092 -.444 -.030 .428 -.188 .811 -.643 -.578 .677 69 Edwards - November .072 .165 -.176 .051 .297 -.190 .759 -.597 -.528 .672 70 Johnson - November -.060 .244 -.090 -.045 .031 .027 -.475 .474 .260 -.289 71 Davis - November -.251 -.505 .547 .304 -.485 -.039 -.390 .156 .405 -.379 72 Long - November .331 -.307 -.105 -.021 -.166 .092 -.102 -.013 .128 -.073

Ul VARIABLES 3 4 5 6 7 8 9 10 11

73 Aycock - November -.165 .008 .142 -.369 .181 .315 -.085 -.143 .229 74 Bell - November .691 -.246 -.487 -.282 .109 .185 .003 -.208 .153 75 Speedy Long - November -.351 .042 .298 .077 -.172 .022 -.178 .369 -.069 76 Schvegmann - November -.022 .370 -.205 -.100 .223 -.014 .108 -.088 -.081 77 Morrison 60 .325 .233 -.447 .033 .549 -.303 .901 -.661 -.709 78 Morrison 64 .425 .262 -.553 .043 .575 -.319 .875 -.620 -.708 79 Lyons 64 .153 .594 -.499 -.125 .365 -.058 -.236 .453 -.035 80 Edwards 72 .495 -.143 -.376 -.122 .309 -.052 .768 -.781 -.409 81 Blacks 70 .715 -.374 -.423 -.248 -.053 .224 -.302 .017 .397

m 13 18 19 20 22 23 24 25 26

13 Republicans 64 1.000 18 Catholic -.755 1.000 19 Black 60 .340 -.327 1.000 20 Political Equality -.811 .799 -.539 1.000 22 Economic Equality -.021 -.066 -.224 .248 1.000 23 Mean White Wage -.293 .314 .046 .113 -.624 1.000 24 Mean Black Wage -.195 .275 -.318 .301 .398 .246 1.000 25 Increase In Black Wage .311 -.237 .220 -.362 -.133 .089 -.003 1.000 26 X Change in Black Regis. .598 -.527 .751 -.755 -.229 -.030 -.310 .411 1.000 27 Proportion Black Regis. .080 -.078 .892 -.257 -.138 .146 -.203 .199 .682 28 Unemployment 68 .013 .013 .192 .057 .153 -.328 -.319 -.148 .074 29 Change In Unemployment -.117 .161 .066 -.026 -.277 .298 -.067 .266 .118 30 Income Less than $3000 .442 -.436 .522 -.374 .158 -.611 -.546 .091 .481 31 Income More than $10,000 -.365 .429 -.227 .252 -.224 .589 .439 -.057 -.308 32 Urban Population -.150 .271 -.202 .178 -.268 .430 .148 -.037 -.155 33 Rural Farmer .330 -.247 .233 -.166 .176 -.506 -.339 .034 .164 34 Rural Non-Farmer -.041 -.156 .098 -.094 .183 -.148 -.009 .047 .103 35 Median Education .163 -.269 -.338 -.158 -.094 .299 .187 -.060 -.130 37 Manufacturing Employment -.034 -.097 .089 -.052 -.147 .406 .270 -.153 .102 VARIABLES 13 18 19 20 22 23 24 25 26

38 Agricultural Employment .245 -.152 .438 -.212 .131 -.502 -.434 -.121 .220 39 Government Employment .190 -.376 .032 -.164 .254 -.296 -.181 .134 .125 40 Trade Employment .023 .157 -.189 .109 -.227 .061 -.149 .197 -.032 41 Union Member ship -.102 .040 -.036 .000 -.226 .554 .357 -.144 -.047 42 Welfare Index .351 -.467 .415 -.332 .115 -.537 -.435 -.036 .305 43 Dependency Index .039 .014 .446 -.080 .022 -.202 -.139 .068 .149 44 Violence Index -.185 .194 .070 .134 -.105 .338 .247 .100 -.067 45 Population Change 70/60 -.304 .164 -.455 .344 .205 -.046 .118 -.129 -.398 46 Population Denelty 70 -.121 .140 -.051 .061 -.046 .218 .166 -.058 -.067 47 Age to 14 -.397 .481 .038 .372 .000 .154 .115 -.046 -.252 48 Age 15-65 .011 -.097 -.280 .006 .016 .106 .079 .027 -.076 49 Age Over 65 .095 -.200 .118 -.119 -.058 -.032 -.054 -.037 .090 50 Incaew Per Capita 68 -.215 .227 -.364 .198 -.086 .437 .358 .051 -.265 51 Inccaw Per Capita 68/67 -.363 .342 -.361 .451 .101 .102 .263 -.288 -.477 52 Income Per Capita 68/29 -.393 .294 -.097 .314 .088 .101 .110 -.191 -.181 53 Income Per Capita 68/59 -.062 -.011 .103 .004 .058 -.069 -.215 -.075 .070 54 Farm Area .137 .016 .280 -.074 -.266 -.016 -.451 .067 .207 55 Percent Black Regia. 60 -.683 .595 .087 .671 .122 .219 .158 -.302 -.494 56 Laborers .175 -.154 .552 -.218 .019 -.314 -.308 .204 .330 57 Professionals .142 -.197 -.190 -.075 -.000 .146 .143 -.131 -.097 58 Manager-Proprietor -.133 .258 -.615 .239 -.017 .186 .347 -.067 -.351 59 Clerical-Sales -.032 .177 -.296 .123 -.187 .323 .171 -.178 -.220 60 Craftsmen-Operatora -.338 .139 -.379 .270 .067 .227 .343 -.384 -.342 61 Farmer & Farm Laborer .191 -.077 .431 -.130 .129 -.458 -.441 .137 .274 62 Change, White Wage .111 -.193 .073 -.070 .313 -.136 -.203 .263 .123 63 Change, Black Wage .137 -.182 -.035 -.135 -.176 .023 -.224 .029 .051 64 Latitude .714 -.759 .369 -.750 -.102 -.308 -.355 .061 .529 65 V.P., 71, Dec. Gub. -.463 .340 -.341 .437 .054 .048 .033 -.295 -.397 66 V.P., 72, Nov. Gub. -.221 .178 -.240 .226 .063 -.061 -.008 -.310 -.201 67 Johnson - December .642 -.832 .104 -.633 ' .084 -.223 -.093 .153 .291 68 Edvards - December -.671 .852 -.119 .671 -.061 .207 .100 -.154 -.289 69 Edvards - November -.673 .844 -.455 .745 -.048 .306 .284 -.142 -.525 70 Johnson - November .384 -.405 .237 -.469 -.234 .208 -.063 -.058 .297 71 Davis • November .364 -.585 .180 -.360 .286 -.401 -.252 -.021 .250 VARIABLES 13 18 19 20 22 23 24 25 26

72 Long - November .097 -.157 .370 -.088 -.054 -.045 -.217 .016 .238 73 Aycock - November .084 .012 -.060 -.102 -.102 .130 .221 .542 -.056 74 Bell - November -.002 .037 .693 -.153 -.109 .015 -.204 .056 .552 75 Speedy Long - November .191 -.280 -.280 -.073 .225 -.373 -.082 -.147 -.172 76 Schvegmann - November -.209 .314 -.259 .224 -.146 .388 .286 -.112 -.201 77 Morrison 60 -.786 .916 -.280 .797 -.006 .316 .272 -.197 -.582 78 Morrison 64 -.809 .895 -.201 .737 -.081 .420 .301 -.232 -.488 79 Lyons 64 .178 -.047 .159 -.229 -.398 .249 -.085 -.093 .241 80 Edvards 72 -.631 .792 .003 .600 .022 .168 .173 .041 -.195 81 Blacks 70 .298 -.288 .983 -.495 -.219 .078 -.298 .221 .753

28 29 30 31 32 33 34 35 37

28 Unemployment 68 1.000 29 Change in Unemployment .160 1.000 30 Income Less than $3000 .557 -.078 1.000 31 Income More than $10,000 -.582 .104 -.874 1.000 32 Urban Population -.436 .078 -.526 .499 1.000 33 Rural Farmer .495 -.177 .755 -.645 -.444 1.000 34 Rural Non-Farmer .189 .085 .171 -.201 -.801 -.095 1.000 35 Median Education -.449 -.127 -.544 .362 .534 -.425 -.352 1.000 37 Manufacturing Employment -.188 -.217 -.261 .249 -.046 -.221 .137 .130 1.000 38 Agricultural Employment .546 -.100 .719 -.569 -.519 .637 .177 -.547 -.362 39 Government Employment .135 .191 .281 -.328 -.121 .004 .292 .077 -.317 40 Trade Employment -.116 .073 -.062 .024 .540 -.048 -.534 .186 -.315 41 Union Membership -.412 -.040 -.471 .407 .447 -.432 -.247 .465 .557 42 Welfare Index .418 -.231 .773 -.643 -.488 .607 .222 -.423 -.064 43 Dependency Index .052 -.328 .412 -.229 -.262 .422 .007 -.494 -.196 44 Violence Index -.236 .205 -.338 .362 .203 -.295 -.094 .120 .011 45 Population Change 70/60 -.221 .134 -.324 .203 .289 -.297 -.005 .254 -.149 46 Population Density 70 -.151 .099 -.258 .221 .469 -.231 -.412 .279 -.017 47 Age to 14 -.140 -.122 -.170 .243 .065 -.015 -.136 -.264 -.129 VARIABLES 28 29 30 31 32 33 34 35 37

48 Age 15-65 -.025 .376 -.250 .129 .235 -.327 .006 .387 .087 49 Age Over 65 -.106 -.175 .080 .038 -.044 .093 -.001 -.084 .015 50 Incone Per Capita 68 -.531 .043 -.656 .536 .569 -.555 -.280 .542 .191 51 Incone Per Capita 68/67 -.122 -.058 -.347 .217 .129 -.277 .017 .098 .018 52 Incane Per Capita 68/29 -.043 -.015 -.154 .247 -.131 -.165 .324 -.235 .211 53 Incone Per Capita 68/59 .164 .235 .162 -.051 -.107 .036 .222 -.157 .049 54 Farm Area .246 .021 .436 -.322 -.028 .347 -.174 -.255 -.126 55 Percent Black Regia. 60 .173 -.040 -.189 .261 .057 -.137 .013 -.273 .027 56 Laborers .382 .028 .642 -.459 -.649 .494 .444 -.712 .028 57 Professionals -.255 -.154 -.284 .180 .481 -.219 -.364 .714 .039 58 Manager-Proprietor -.470 .029 -.597 .470 .580 -.412 -.431 .468 -.085 59 Clerical-Sales -.408 -.098 -.514 .423 .825 -.399 -.690 .694 .019 60 Craftsmen-Operators -.312 -.181 -.468 .383 -.090 -.323 .272 .105 .351 61 Farmer & Farm Laborer .554 .048 .809 -.626 -.459 .757 .111 -.632 -.312 62 Change, White Wage .165 .102 .258 -.257 -.237 .090 .345 -.108 -.281 63 Change, Black Wage .113 .134 .119 -.194 -.135 .005 .188 -.016 -.288 64 Latitude .045 -.290 .466 -.414 -.146 .368 -.050 .197 .109 65 V.P., 71, Dec. Gub. .197 .195 -.106 .040 -.194 .006 .252 -.211 .146 66 V.P., 72, Nov. Gub. .045 -.063 .008 -.060 -.375 .032 .413 -.222 .195 67 Johnson - December -.058 -.127 .180 -.223 -.073 .146 -.001 .447 .079 68 Edwards - December .084 .116 -.168 .206 .082 -.147 -.002 -.461 -.090 69 Edvards - November -.016 .110 -.387 .238 .275 -.239 -.177 -.141 -.116 70 Johnson - November -.277 -.208 -.169 .207 .280 -.214 -.214 .613 .238 71 Davis - November .272 -.159 .517 -.481 -.421 .405 .236 -.060 -.000 72 Long • November .258 -.179 .320 -.289 -.129 .130 .117 -.154 .228 73 Aycock - November -.233 .245 -.213 .260 .073 -.082 -.048 -.054 -.159 74 Bell - November .096 .163 .305 -.025 -.144 .097 .131 -.441 .023 75 Speedy Long - November .122 -.000 .146 -.213 -.356 .155 .331 -.048 -.077 76 Schvegmann - November -.331 -.019 -.571 .459 .537 -.455 -.340 .481 .207 77 Morrison 60 .025 .161 -.431 .430 .260 -.281 -.131 -.217 -.125 78 Morrison 64 -.036 .191 -.498 .523 .332 -.353 -.176 -.127 -.050 79 Lyons 64 -.323 -.176 -.204 .281 .562 -.201 -.545 .458 .085 80 Edwards 72 .113 .168 -.104 .165 -.028 -.104 .085 -.578 -.097 81 Blacks 70 .186 .083 .503 -.216 -.165 .206 .076 -.337 .110 VARIABLES 39 40 41 42 43 44 45 46 47

39 Government Employment 1.000 40 Trade Employment -.031 1.000 41 Union Menibershlp -.235 .042 1.000 42 Welfare Index .299 -.068 -.336 1.000 43 Dependency Index -.254 .000 -.257 .433 1.000 44 Violence Index -.154 -.056 .260 -.340 -.090 1.000 45 Population Change 70/60 .407 .005 .038 -.253 -.436 .056 1.000 46 Population Density 70 -.103 .250 .340 -.278 -.235 .363 .021 1.000 47 Age to 14 -.558 .118 .016 -.201 .662 .208 -.182 -.099 1.000 48 Age 15-65 .416 -.091 .191 -.298 -.884 .048 .561 .142 -.711 49 Age Over 65 -.094 -.041 -.172 .159 .176 -.040 -.185 -.033 .008 50 Income Per Capita 68 -.166 .073 .454 -.632 -.440 .337 .538 .345 .001 51 Income Per Capita 68/67 .040 .076 -.038 -.192 -.170 .129 .126 .048 .036 52 Income Per Capita 68/29 .032 -.303 -.089 -.021 -.083 -.031 .359 -.177 .023 53 Income Per Capita 68/59 .234 -.320 -.132 .073 -.311 -.017 .406 -.082 -.312 54 Farm Area -.100 .116 -.186 .220 .277 -.235 -.397 -.227 .083 55 Percent Black Regis. 60 -.153 -.133 .067 -.077 .223 .298 .070 .080 .464 56 Laborers .037 -.353 -.392 .562 .462 -.104 -.294 -.336 .049 57 Professionals .318 .171 .342 -.145 -.377 -.064 .279 .362 -.411 58 Manager-Proprletor -.172 .270 .102 -.617 -.409 .060 .369 .227 -.023 59 Clerical-Sales -.170 .499 .484 -.497 -.298 .215 .238 .504 .030 60 Craftsmen-Operators -.259 -.156 .249 -.201 .008 .101 .132 -.145 .282 61 Farmer & Farm Laborer .057 -.022 -.451 .529 .441 -.308 -.292 -.265 .075 62 Change, White Wage .545 .008 -.257 .078 -.000 -.105 .073 -.074 -.044 63 Change, Black Wage .353 -.048 -.268 .161 -.074 -.192 .033 -.108 -.206 64 Latitude .164 .008 -.014 .597 .158 -.368 -.300 -.189 -.399 65 V.P., 71, Dec. Gub. -.002 -.150 -.039 -.015 -.201 .021 .101 -.117 -.045 66 V.P., 72, Nov, Gub. -.069 -.256 -.068 -.015 -.059 -.032 .010 -.209 .030 67 Johnson - December .308 .011 .052 .325 -.109 -.096 .012 -.034 -.445 68 Edwards - December -.305 .060 -.047 -.297 .094 .079 -.011 .032 .435 69 Edwards - Novenber -.321 .159 .072 -.545 -.066 .087 .162 .090 .405 70 Johnson - November .044 .073 .309 -.099 -.179 .079 -.091 .078 -.216 71 Davis - November .230 -.099 -.079 .424 .182 -.347 -.150 -.248 -.112 72 Long - November .189 .070 .027 .495 .098 .050 -.079 .183 -.166 73 Aycock - Novenber -.167 -.095 -.110 -.277 -.016 .131 .029 -.037 .067 VARIABLES 39 40 41 42 43 44 45 46

74 Bell - November -.077 -.213 -.086 .112 .333 .221 -.269 -.045 75 Speedy Long - Novenber .347 -.113 -.273 .430 -.112 -.193 .120 -.160 76 Schvegmann - November -.267 .303 .382 -.535 -.317 .341 .278 .452 77 Morrison 60 -.304 .051 .082 -.476 -.016 .338 .211 .228 78 Morrison 64 -.339 .099 .193 -.524 ,003 .388 .188 .298 79 Lyons 64 -.188 .303 .264 -.182 -.073 .084 -.112 .238 80 Edvards 72 -.320 -.101 -.131 -.292 .200 .200 -.020 -.014 81 Blacks 70 .039 -.171 -.005 .389 .399 .083 -.473 .042

49 50 51 52 53 54 55 56

49 Age Over 65 1.000 50 Incone Per Capita 68 -.139 1.000 51 Incone Per Capita 68/67 .011 .186 1.000 52 Income Per Capita 68/29 .016 .315 .176 1.000 53 Income Per Capita 68/59 -.062 .317 -.005 .690 1.000 54 Farm Area .055 -.263 -.204 -.016 .074 1.000 55 Percent Black Regis. 60 -.017 .000 .301 .286 .043 -.004 1.000 56 Laborers .086 -.514 -.264 .099 .190 .309 .101 1.000 57 Professionals -.071 .380 .163 -.077 -.042 -.172 -.138 -.502 58 Manager-Proprietor .082 .545 .246 .033 -.092 -.240 -.195 -.659 59 Clerical-Sales -.153 .543 .141 -.233 -.216 -.114 -.022 -.670 60 Craftsmen-Operators .055 .126 .241 .247 -.137 -.398 .075 -.300 61 Farmer & Farm Laborer .038 -.570 -.374 -.065 .202 .500 .025 .635 62 Change, White Wage -.031 -.166 -.005 .037 .190 .046 -.044 .053 63 Change, Black Wage .000 -.212 -.055 -.056 -.061 .058 -.162 .007 64 Latitude .148 -.322 -.352 -.293 -.074 .175 -.538 .169 65 V.P., 71, Dec. Gub. -.121 -.035 .382 .346 .218 .004 .249 .119 VARIABLES 49 50 51 52 53 53 55 56 57

66 V.P., 72, Nov. Gub. -.082 -.138 .178 .240 .153 -.064 .045 .159 -.416 67 Johnson - December .076 -.040 -.234 -.337 -.068 -.200 -.515 -.099 .352 68 Edvards - December -.076 .019 .246 .322 .049 .188 .527 .076 -.348 69 Edvards - November -.263 .284 .272 .165 -.045 .098 .387 -.224 -.163 70 Johnson - November .007 .195 .054 -.114 -.058 -.051 -.238 -.313 .466 71 Davis - November -.010 -.386 -.343 -.159 .037 .186 -.305 .209 -.141 72 Long - November .023 -.163 .008 .061 .069 -.030 .202 .308 .091 73 Aycock - November .253 .197 -.030 -.137 -.041 -.148 -.085 .018 -.188 74 Bell - Novenber .148 -.248 -.236 .030 .143 .163 .156 .407 -.353 75 Speedy Long - November .065 -.311 .036 .035 -.002 -.239 -.155 .088 .054 76 Schvegmann - November -.200 .593 .144 -.104 -.113 -.361 .037 -.466 .275 77 Morrison 60 -.144 .270 .318 .299 .035 -.040 .735 -.139 -.116 78 Morrison 64 -.120 .306 .271 .298 -.015 -.062 .712 -.197 -.057 79 Lyons 64 .184 .235 -.019 -.138 -.182 .163 -.154 -.329 .411 80 Edvards 72 -.103 .046 .158 .341 .111 .136 .544 .212 -.471 81 Blacks 70 .123 -.349 -.346 -.094 .113 .286 .103 .518 -.173

59 60 61 62 63 64 65 66 67

59 Clerical-Sales 1.000 60 Craftsmen-Operator s -.026 1.000 61 Farmer & Farm Laborer -.520 -.477 1.000 62 Change, White Wage -.265 -.050 .231 1.000 63 Change, Black Wage -.226 .005 -.041 .238 1.000 64 Latitude -.073 -.203 .207 -.075 .103 1.000 65 V.P., 71, Dec. Gub. -.244 .278 -.024 -.028 .009 -.301 1.000 66 V.P., 72, Nov. Gub. -.249 .364 .064 -.008 -.015 -.155 .541 1.000 67 Johnson - December .069 -.061 -.104 .058 .082 .678 -.384 -.246 1.000 68 Edvards - December -.072 .056 .123 -.051 -.086 -.683 .371 .242 -.977 69 Edvards - November .192 .162 -.091 -.119 -.043 -.722 .286 .215 -.780 VARIABLES 59 60 61 62 63 64 65 66 67

70 Johnson - November 395 >.058 -.317 -.057 -.055 .503 -.347 -.260 .454 71 Davis - November 261 .Q84 .378 .371 .035 .448 -.102 .178 .471 72 Long - November 078 -.219 .115 .094 -.082 .180 -.082 -.205 .104 73 Aycock - November 106 -.140 -.045 -.073 -.003 -.227 -.187 -.169 .034 74 Bell - November 264 -.201 .325 -.012 -.136 -.009 -.108 .070 -.160 75 Speedy Long - November 254 .166 -.047 .025 .297 .199 .197 .169 .311 76 Schvegmann - November 633 .159 -.479 -.187 -.223 -.286 -.073 -.149 -.071 77 Morrison 60 165 .046 -.090 -.144 -.212 -.823 .287 .098 -.769 78 Morrison 64 249 .117 -.158 -.200 -.247 -.781 .254 .042 -.732 79 Lyons 64 523 -.194 -.280 -.379 -.116 .338 -.297 -.375 .039 80 Edwards 72 206 -.017 .175 -.048 -.162 -.734 .282 .204 -.863 81 Blacks 70 283 -.403 .428 .083 -.047 .330 -.312 -.229 .052

69 70 71 72 73 74 75 76 77

69 Edwards - Novenber 1.000 70 Johnson - November -.444 1.000 71 Davis - November -]435 .067 1.000 72 Long - November -.316 .098 -.017 1.000 73 Aycock - November .057 -.241 -.248 -.306 1.000 74 Bell - November -.175 -.045 .017 .014 -.030 1.000 75 Speedy Long - November -.355 -.237 .043 .050 -.152 -.278 1.000 76 Schvegmann - November .296 .225 -.398 .066 -.043 -.194 -.342 1.000 77 Morrison 60 .755 -.406 -.572 -.094 .102 .055 -.286 .267 1.000 78 Morrison 64 .708 -.278 -.557 -.068 .004 .115 -.372 .351 .949 79 Lyons 64 -.117 .654 -.299 .061 -.180 .044 -.341 .319 -.090 80 Edwards 72 .721 -.567 -.453 -.080 .209 .329 -.385 .010 .783 81 Blacks 70 -.412 .202 .146 .394 -.074 .707 -.294 -.246 -.239 BIBLIOGRAPHY

Alker, Hayward R., Jr. 1966 "Causal Inference and Political Analysis", in Joseph L. Bernd (Ed.), Mathematical Applications in Political Science II. Southern Methodist University Press. 7-43. 1969 "A Typology of Ecological Fallacies", in M. Dogan and S. Rokkan (Eds.), Quantitative Ecological Analysis in the Social Sciences. Cambridge: M.I.T. Press.

Allardt, Erik 1969 "The Merger of American and European Traditions of Sociological Research: Contextual Analysis," Social Sciences Tnfnrnmtlnn. 7 (1), 151-168.

Allardt, Erik and Pesonen, Pertti 1967 "Cleavages in Finnish Politics", in Lipset and Rokkan's Party Systems and Voter Alignments. Free Press. New York.

Barton, Allen H. 1968 "Bringing Society Back In: Survey Research and Macro- Methodology," American Behavioral Scientist. (November- December). 1-9.

Berelson, B. R., Lazarsfeld, P. F., and McPhee, W. N. 1954 Voting: A Study of Opinion Formation in a Presidential Campaign. Chicago, Illinois, University of Chicago Press.

Blalock, Herbert M., Jr. 1961 Causal Inferences in Nonexperlmental Research, Chapter 4, Chapel Hill, North Carolina, The University of North Carolina Press.

Blalock, Hubert M. 1967 "Status Inconsistency, Social Mobility, Status Integration,* and Structural Effects," American Sociological Review. Volume 32. 790-801.

Blau, P. M. 1957 "Formal Organization: Dimensions of Analysis," American Journal of Sociology. 63, July. I960 "Structural Effects," American Sociological Review, 25. 178-193.

Boudon, R. 1969 "Secondary Analysis and Survey Research: An Essay in the

154 155 Sociology of the Social Sciences," Social Sciences Information. 8 (6), 7-32.

Campbell, Angus, Converse, Phillip E., Miller, Warren E., Stokes, Donald E, 1966 Elections and the Political Order. New York: John Wiley and Sons.

Campbell, Ernest Q., and Alexander, C. Norman 1965 "Structural Effects and Interpersonal Relationships", American Journal of Sociology. 71. 284-289.

Cartwright, D. S. 1969 "Ecological Variables," in E. F. Borgatta (Ed.) Sociological Methodology. San Francisco: Jossey-Bass, Inc.

Coleman, James 1961 "Relational Analysis: The.Study of Social Organizations with Survey Methods" in Amatai Etzioni's A Sociological Reader on Complex Organizations. Holt, Rinehard and Winston. 517-528.

Converse, P. E. 1969 "Survey Research and the Decoding of Patterns in Ecological Data," in M. Dogan and S. Rokkan (Eds.) Quantitative Ecological Analysis in the Social Sciences. Cambridge: M.I.T. Press.

Cox, Kelvin R. 1969 "Voting in the London Suburbs: A Factor Analysis and a Causal Model" in Dogan and Rokkan's (Eds.) Quantitative Ecological Analysis in the Social Sciences. 343-370.

Davis, J. A., Spaeth, J. L., and Huson, C. 1961 "A Technique for Analyzing the Effects of Group Composition." American Sociological Review. 26.

Dawson, Richard E., and Prewit, Kenneth 1969 Political Socialization. Boston. Little, Brown and Co.

Dogan, Mattel and Rokkan, Stein 1969 Quantitative Ecological Analysis in the Social Sciences. Cambridge: M.I.T. Press.

Duncan, 0. D., and Davies, B. 1953 "An Alternative to Ecological Correlation", American Sociologi­ cal Review, 18.

Duncan, 0. D., and Cuzzart, R. P. and Duncan, B. 1961 Statistical Geography: Problems in Analyzing Areal Data. Illinois: The Free Press of Glencoe.

Durkheim, Emile 1950 The Rules of Sociological Method (1895). (Translated by S. A. Solovay and J. H. Mueller). C.E.G. Cat1in (Ed.). Glencoe, Illinois: The Free Press. 156 Durkheim, Emile 1951 Suicide. Spaulding and Simpson (trans.). Glencoe, Illinois: The Free Press.

Faris, R. E. L. and Dunham, H. W. 1939 Mental Disorders in Urban Areas. Chicago: The University of Chicago Press. 110-123.

Fenton, John H. 1966 People and Parties in Politics. Glenview, Illinois.

Goodman, L. A. 1953 "Ecological Regressions and the Behavior of Individuals". American Sociological Review. Vol. 15. 1959 "Some Alternatives to Ecological Correlation." American Journal of Sociology. 44 (May) 610-625.

Green, A. A. J. 1964 Aggregation in Economic Analysis. Princeton, New Jersey. Princeton University Press.

Grenier, C. E. and Howard, Perry H. "La Coalition Cafe au Lait: La Puissance Acadian et Noir," The Louisiana Review, forthcoming.

Hardon, T. and Pappi, F. U. 1969 "Multiple-level Regression Analysis of Survey and Ecological Data." Social Science Information. Vol. 8, No. 5. 43-67.

Havard, William C., Heberle, Rudolf and Howard, Perry H. 1963 The Louisiana Elections of 1960. Louisiana State University P ress.

Heberle, Rudolf 1945 From Democracy to Nazism: A Regional Case Study on Political Parties in Germany. Louisiana State University Press. 1951 Social Movements. New York: AppletonCentury Crofts. 1952 "On Political Ecology," Social Forces. Vol. 31, No. 1, (October) 1-9.

Heberle, Rudolf and Bertrand, Alvin L. 1949 "Factors Motivating Voting Behavior in a One Party State." Social Forces XXVII. 343-350.

Heberle, Rudolf, Hillery, Jr., George and Lovrich, Frank 1953 "Continuity and Change in Voting Behavior in the 1952 Primaries in Louisiana," Southwestern Social Science Quarterly. XXXIII. 328-342.

Heberle, Rudolf and Howard, Perry H. 1954 "An Ecological Analysis of Political Tendencies in Louisiana: The Presidential Election of 1952," Social Forces XXXII. 344-50. 157

Hoiberg, Eric 0. and Cloyd, Jerry S. 1971 "Definition and Measurement of Continuous Variation in Ecological Analysis," American Sociological Review. Vol. 36, No. 1, (February).

Howard, Perry H. 1952 "An Analysis of Voting Behavior in Baton Rouge," Proceedings of the Louisiana Academy of Science. XV, 84-100. 1971 Political Tendencies in Louisiana (revised and expanded). Baton Rouge, Louisiana: L. S. U. Press.

"Louisiana: Resistence and Change," a chapter in William C. Havard (ed.), The Changing Politics of the South, L.S.U. Press, forthcoming.

Chapter on Louisiana in David M. Kovenock, Prothro and Associates, Explaining the Vote: Presidential Choices in the Nation and States. 1968. University of North Carolina Press, forthcoming.

Howard, Perry H. and Brent III, Joseph L. 1966 "Social Change, Urbanization, and Types of Society," Journal of Social Issues. XXII.

Hunter, Albert 1971 "The Ecology of Chicago: Persistence and Change, 1930-1960," American Journal of Sociology. Vol. 77, No. 3, (November) 425-444.

Irwin, 6. A. and Meeter, D. A. 1969 "Building Voter Transition Models from Aggregate Data." Midwest Journal of Political Science. XIII.

Janson, Carl-Gunnar 1969 "Some Problems of Ecological Factor Analysis," in M. Dogan and S. Rokkan (eds.) Quantitative Ecological Analysis in the Social Sciences.

Kendal, P. L. and Lazarsfeld, P. F. 1955 "Problems of Survey Analysis" in R. K. Merton and P. F. Lazars- feld (eds.) Studies in the Scope and Method of the American Soldier. Free Press.

Key, V. 0. 1949 Southern Politics in State and Nation. New York.

King, Leslie J. 1969 Statistical Analysis in Geography. Prentice-Hall.

Kish, L. 1962 "A Measurement of Homogeneity in Areal Units," Bulletin of the International Statistical Association. Vol. 4, 201-210. 158

Kovenock, David M., Prothro, James, et al. Explaining the Vote: Presidential Choices in the Nation and States, 1968, University of North Carolina Press, forthcoming.

Lane, Angela 1968 "Occupational Mobilization in Six Cities," American Journal of Sociology, Vol. 33, No. 5 (October) 740-749.

Lazarsfeld, P. F. and Menzel, H. 1961 "On the Relation Between Individual and Collective Properties," A. Etzioni (ed.), Complex Organizations. New York: Holt, Rinehart, and Winston.

Linz, J . J . 1969 "Ecological Analysis and Survey Research," in M. Dogan and S. Rokkan (eds.), Quantitative Ecological Analysis in the Social Sciences. Cambridge: M.I.T. Press.

Lipset, S. M., Trow, M. A. and Coleman, J. S. 1951 Union Democracy. Glencoe, Illinois: Free Press.

Lipset, S. M. 1960 Political M a n . New York: Doubleday. Ch. 6-8. 1965 "Political Sociology," Sociology Today. Robert Merton, Leonard Broom and Leonard S. Cottrell (eds.). New York: Harper and Row.

Lipset, S. M. and Rokkan, Stein 1968 Party Systems and Voter Alignments. New York: Free Press.

McKinney, John C. and Bourque, Linda Brookover 1972 "Further Comments on 'The Changing South: A Response to Sly and Weller.1" American Sociological Review, Vol. 37, No. 2 (April)

Mathews, Donald R. and Prothro, James W. 1966 Negroes and the New Southern Politics. New York: Harcourt, Brace and World.

Parenton, J. Vernon 1948 The Rural French-Speaking People of Quebec and South Louisiana. Unpublished dissertation. Harvard University.

Pelz, D. C. 1961 "Adjustment of Performance Measures to Hold Constant Some Background Factors," Analysis Memo No. 16, Study of Scientific Personnel, Institute for Social Research, University of Michigan.

Price, Douglas 1967 "Micro-Macro-Politics: Notes on Research Strategy," in Oliver Garceau (ed.), Political Research and Political Theory. 124-134. 159 Ranney, Austin 1962 "The Utility and Limitations of Aggregate Data In the Study of Electoral Behavior,*' In Ranney Essays on the Behavioral Study of Politics. 91-102.

Riley, Mathilda White 1964 "Sources and Types of Sociological Data," in Robert FarIs* Handbook of Modern Sociology. Chicago: Rand McNally.

Robinson, W. S. 1950 "Ecological Correlations and the Behavior of Individuals," American Sociological Review. XV. 351-357.

Rokkan, Stein 1962 "The Challenge of Comparative Microanalysis," in Austin Ranney (ed.) Essays on the Behavioral Study of Politics. Urbana, Illinois: University of Illinois Press. 49-55.

Runciman, W. G. 1963 Social Science and Political Theory. Cambridge University Press.

Scheuch, Erwin K. 1969 "Social Context and Individual Behavior," in Dogan and Rokkan (eds.) Quantitative Ecological Analysis in the Social Sciences. Cambridge: M.I.T. Press.

Schoenberger, Robert A. 1968 "Conservatism, Personality and Political Extremism." American Political Science Review. No. 3, Vol. LXI1 (September).

Schott, Matthew J. 1971 "Class Conflict in Louisiana Voting Since 1877: Some New Perspectives," Louisiana History Quarterly. Vol. XII, No, 2.

Schuessler, K. 1969 "Covariance Analysis in Sociological Research," in E. F. Borgatta (ed.), Sociological Methodology. San Francisco: Jossey-Bass, Inc.

Segal, David and Mayer, Marshall 1969 "The Social Context of Political Partisanship," in Dogan and Rokkan's (eds.) Quantitative Ecological Analysis in the Social Sciences. Cambridge: M.I.T. Press.

Selvin, H. C. 1958 "Durkheim's 'Suicide' and Problems of Empirical Research," American Journal of Sociology. 63.

Shaffer, William R. 1972 Computer Simulations of Voting Behavior. Oxford University Press. 160 Shevky, E. and Bell, W. 1955 Social Area Analysis. Stanford: Stanford University Press.

Shugg, Roger W. 1939 Origins of Class Struggle In Louisiana. Baton Rouge, Louisiana.

Sindler, Allan P. 1956 's Louisiana: State Politics. 1920-1952. Baltimore.

Singer, J. David 1961 "The Level of Analysis Problem in International Relations." World Politics. 14 (1). 77-92.

Slatin, 6. T. 1969 "Ecological Analysis of Delinquency: Aggregation Effects." American Sociological Review, Vol. 34, No. 6, (December) 894- 906.

Smelser, Neil J. 1968 "Personality and the Explanation of Political Phenomena of the Social-System Level: A Methodological Statement." The Journal of Social Issues. XXIV, 3 (July) 111-126.

Sonquist, John A. 1971 Multivariate Model Building: The Validation of a Search Strategy. Survey Research Center, Institute for Social Research: University of Michigan.

Stouffer, S. A., et al. 1949 "Studies in Social Psychology in World War II," The American Soldier During Army Life. Vol. 1, Princeton: Princeton University Press. 1950 Measurement and Prediction," Vol. IV. Studies of Social Psychology. Princeton University Press.

Sweetser, F. L. 1965 "Factor Structure vs. Ecological Structure In Helsinki and Boston." Acta Sociologlca 8. 205-225.

Tannenbaum, A. S. and Smith, C. G. 1964 "The Effects of Member Influence in an Organization: Pheonmenology vs. Organization Structure," Journal of Abnormal Social Psychology.

Valkonen, T. 1969 "Individual and Structural Effects in Ecological Research," in M. Dogan and S. Rokkan (eds.) Quantitative Ecological Analysis in the Social Sciences. Cambridge: M.I.T. Press.

Wilbur, G. L. 1967 "Causal Models and Probability," Social Forces. Vol. 46, No. 1.

Williams, T. Harry 1966 Romance and Realism in Southern Politics. L.S.U. Press. VITA

Charles E . Grenier was born In Poughkeepsie, New York, on

September 10, 1930. In 1935 he moved with his family to Connecticut where he completed his elementary and high school education, graduating in 1948 from Windham High School. After completing two years of liberal arts studies at Mitchell College and the University of Connecticut, he enlisted in the U.S. Navy, serving four years in the Medical Corps.

Honorably discharged in 1954, he entered Southeastern Louisiana

University where he received a Bachelor's Degree in 1956. The following

September he enrolled in the Graduate School at Louisiana State University, and in 1960 he received an M.B.A. Degree. The candidate has also done advanced work in economics at the University of Connecticut on a summer- sabbatical basis. After nine years of teaching— two years at Arkansas

A & M, and seven years at Southeastern Louisiana University, the candidate took a four-year sabbatical leave in 1968 to work full-time on the

Ph.D. in Sociology at Louisiana State University, and has finally met all of the requirements for the degree.

161 EXAMINATION AND THESIS REPORT

Candidate: Charles G. Grenier

Major Field: Sociology

Title of Thesis: An Analysis of Contextual Effects on Electorial Behavior

Approved:

Major Professor and Chairman

Dean of the Graduate School

EXAMINING COMMITTEE:

Date of Examination:

May 5, 1972