Citation
Racial composition and racial threat : the anatomy of contemporary racism

Material Information

Title:
Racial composition and racial threat : the anatomy of contemporary racism
Creator:
Horn, Randolph Claiborne
Publication Date:
Language:
English
Physical Description:
x, 184 leaves : ill. ; 29 cm.

Subjects

Subjects / Keywords:
African Americans ( jstor )
Conservatism ( jstor )
Mathematical variables ( jstor )
Political attitudes ( jstor )
Political candidates ( jstor )
Prejudices ( jstor )
Psychological attitudes ( jstor )
Racism ( jstor )
Voting ( jstor )
White people ( jstor )
Dissertations, Academic -- Political Science -- UF
Political Science thesis Ph. D
Race relations -- History -- United States -- 20th century ( lcsh )
Racism -- History -- United States -- 20th century ( lcsh )
Genre:
bibliography ( marcgt )
non-fiction ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 1993.
Bibliography:
Includes bibliographical references (leaves 177-183).
General Note:
Typescript.
General Note:
Vita.
Statement of Responsibility:
by Randolph Claiborne Horn.

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright [name of dissertation author]. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Resource Identifier:
030390991 ( ALEPH )
31063988 ( OCLC )

Downloads

This item has the following downloads:


Full Text












RACIAL COMPOSITION AND RACIAL THREAT:
THE ANATOMY OF CONTEMPORARY RACISM
















BY


RANDOLPH CLAIBORNE HORN
















A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY



UNIVERSITY OF FLORIDA 1993


























Copyright 1993

by

Randolph C. Horn















ACKNOWLEDGMENTS



I wish to express my gratitude to some of those who

helped make this dissertation possible. A complete catalogue of all who have inspired, assisted, cajoled, or otherwise pushed me into action is not possible. Let me try to thank a few of you.

I am indebted to the members of my doctoral committee

for their helpful comments at various stages of this project. To Peggy Conway, I must express particular gratitude for her encouragement, criticism, inspiration, and administrative heavy lifting.

I owe special gratitude to my family. My parents have inspired me through their dedication to service. my wife, Frannie, has provided unwavering support and timely distractions, both of which were essential for the completion of this project.














TABLE OF CONTENTS


ACKNOWLEDGMENTS ................................... :iii

LIST OF TABLES ...............................................vi

LIST OF FIGURES o........ ............. .......viii

ABSTRACT ........... .... ............ ....................ix

CHAPTER 1 ....................................................1

INTRODUCTION .................. .......... ....... ... 1
Recent Trends in the Study of Prejudice................... 1
Perceptions of Threat.................................2
Symbolic Racism ....................................... 4
Synthesis..................................................6

CHAPTER 2 ............... ..... .. ...... .. .... ...... 10

PERCEPTIONS OF RACIAL THREAT: THEORY AND MEASUREMENT ....... 10
Foundations........................... ..... 11
The Contemporary Significance of Race at the Macro
Level ....................................................15
Aggregate Analyses....................................... 19
Individual Analyses ................................24
Reconciliation....... ... .... .......... ..... 32

CHAPTER 3 ..................... .......o.o.. .. ...... 36

SYMBOLIC RACISM: THEORY AND MEASUREMENT .................... 36
Symbolic Politics in a Nutshell....................... 37
Theoretical Underpinnings of Symbolic Racism............. 39
Newness or Distinctiveness from Old Fashioned Racism..... 42 Combination or Multi-dimensionality ..... ........ 45
Abstraction vs. Self Interest.............................47
Methods and Measurement.................................. 49
Self Interest: The Alternative Hypothesis................ 53
Data..................................................... 58

CHAPTER 4 ................................................... 61

SYNTHESIS ................................................... 61
Multi-dimensionality ...................................... 63
Defining the Nature of Interests ....................... 71
Synthesis ...................... .... ........ ............73


iv








Context and the Changing Meaning of Racial
Composition ........................................... 75
The Archaic Significance of Racial Composition ........... 83

CHAPTER 5 .................................................. 89

DATA ....................................................... 89
Data ..................................................... 89
Dependent variables ....................................... 91
Contextual Variables ..................................... 95
1980 Census ........................................... 96
1900 and 1940 Censuses ................................ 98
Endogenous variables .................................... 104

CHAPTER 6 ................................................. 107

NATIONAL MODELS OF PREJUDICE AND ITS EFFECTS .............. 107
Procedures .............................................. 107
Models Explaining the origins of Racial Attitudes ....... 111 Racial Attitudes and Other Political orientations ....... 121

CHAPTER 7 ................................................. 134

REGIONAL ANALYSIS ......................................... 134
Southern Models of Racial Attitudes ..................... 135
Northern Models of Racial Attitudes ..................... 140
Regional Differences in Attitudinal Unity ............... 146

CHAPTER 8 ................................................. 152

CONCLUSIONS: FARTHER ALONG ................................ 152
Racial Threat ........................................... 152
Symbolic Racism ......................................... 154
The Synthetic Theoretical Position ...................... 155
Assessing the Derived Racial Threat Hypotheses .......... 157 Derived Symbolic Racism Hypotheses: ..................... 160
Synthetic Hypotheses: ................................... 162
Conclusions ............................................. 165

APPENDIX .................................................. 172
Racial Threat scale items: .............................. 172
Symbolic Racism scale items: ............................ 173
Coding of other items: .................................. 174
Data Derived from the Census ............................ 175

BIOGRAPHICAL SKETCH ....................................... 184










v














LIST OF TABLES

Table 5-1. T-test for race of interviewer bias on
whites .................................................. 93

Table 5-2. Factor analysis for racial items among
whites with white interviewers .......................... 94

Table 5-3. Black Concentration in 1900, 1940, and 1980
and the levels of Perceived Racial Threat and Symbolic
Racism in 1986 for white Respondents only ............... 102

Table 6-1. Symbolic Racism Regressed on Contextual and
Individual Level Variables for White Respondents
only .................................................... 112

Table 6-2. Perceived Racial Threat Scale Regressed on
Contextual and Individual Level Variables for whites
only .................................................... 118

Table 6-3. Regression of Evaluation of Candidate and
Group objects on Racial Attitudes for White
Respondents only ........................................ 123

Table 6-4. Logistic Regression of Decreased Federal
spending on Racial Attitudes for White Respondents
Only .................................................... 129

Table 7-1. Symbolic Racism Regressed on Contextual and
Individual Level Variables for White Southern
Respondents Only ........................................ 136

Table 7-2. Perceived Racial Threat Scale Regressed on
Contextual and Individual Level Variables for White
Southerners Only ........................................ 139

Table 7-3. Symbolic Racism Regressed on Contextual and
Individual Level Variables for White Northern
Respondents Only ........................................ 141

Table 7-4. Perceived Racial Threat Scale Regressed on
Contextual and Individual Level Variables for White
Northerners Only ........................................ 143

Table 7-5. Standard deviation of Symbolic Racism
Regressed on Contextual Variables in Each County ........ 147




vi








Table 7-6. Standard Deviation of Perceived Racial
Threat Regressed on Contextual Variables in Each
County .................................................. 149




















































vii













LIST OF FIGURES

Figure 4-1. Matrix of racial attitudes ................... 65

Figure 4-2. Quadratic function with increasing slope ..... 78 Figure 4-3. Parabolic quadratic function ................. 80

Figure 4-4. Quadratic function with gradual decay ........ 82







































viii













Abstract of Dissertation Presented to the Graduate
School of the University of Florida in Partial Fulfillment of
the Requirements for the Degree of Doctor of Philosophy


RACIAL COMPOSITION AND RACIAL THREAT:
THE ANATOMY OF CONTEMPORARY RACISM

By


Randolph Claiborne Horn


December, 1993

Chairman: M. Margaret Conway Major Department: Political Science

While the use of thinly veiled racial issues in recent elections serves as a reminder of the salience of race in American politics, scholars disagree about the causes and effects of white racial attitudes. The proponents of symbolic racism and group conflict approaches have engaged in a debate over which single approach best explains racism. This dissertation demonstrates that symbolic attitudes and perceptions of racial threat occupy different dimensions in the construction of white racial attitudes. Racial attitudes do not exist in a vacuum; the components and the effects of those attitudes are tempered by historical factors and the contemporary racial context. Further, each dimension has a differential impact on political behavior.

This project combines composition data from the 1900,

1940, and 1980 censuses with survey data taken from the 1986



ix








American National Election Studies (ANES) to assess the influence of racial composition and other contextual variables on perceptions of racial threat and symbolic racial attitudes. These orientations, in turn, are used to predict issue positions and evaluations of candidates.

The analysis are conducted on two levels. First, the general hypotheses of interest are tested at the national level. Factor analysis confirms that the two racial attitudes occupy different dimensions. However, both perceptions of threat and symbolic racism are influenced by contextual measures like the local percentage black. The racial measures differ in their ability to predict issue positions or evaluations of groups or candidates; perception of threat is a better predictor of candidate and group evaluations, while symbolic racism is a consistent predictor of federal spending preferences. A second level of analysis applies the same techniques to compare the South with other regions; the same factors which explain differences in racial attitudes between the regions do not explain marginal differences within the South. The dissertation demonstrates empirically the importance of local history and political culture in explaining contemporary political and racial orientation.















CHAPTER 1
INTRODUCTION

Contemporary studies of white racial prejudice have

tended to concentrate exclusively on one of two factors in explaining racial attitudes. Proponents of the racial threat explanation look to contemporary demographic and economic conditions, while proponents of symbolic explanations argue that values learned early in life explain current levels of prejudice independent of the individual's current demographic or economic context. This dissertation makes a case for augmentation and synthesis of these two schools. It is argued that individuals may simultaneously hold more than one kind of racial orientation. Further, racial attitudes are not only the product of either psychological traits or the current economic context. Rather, racial attitudes result from both of these factors and are conditioned by local historical and cultural conditions.


Recent Trends in the Study of Preiudice


Events in recent years have spelled disaster for the hard fought gains of the civil rights movement. Supreme


Racial threat may be defined as the belief on the part of whites that black ascendance reduces the chances of white success in competition for jobs, admissions to schools or other areas of achievement; it may also represent fears on the part of whites that blacks pose challenge the security of whites' persons or property.

1






2


Court decisions have made it easier to discriminate and have whittled away at affirmative action programs. Extremist groups, such as those led by Tom Metzger and manned by skin heads, seem on the rise. Indeed, the number of hate groups nation-wide increased 27% between 1990 and 1991. (Southern Poverty Law Center 1992) The qualified electoral success of

David Duke in Louisiana seems to scare everyone, except of course his supporters.

while these events point to an increasingly engaged white backlash, other events seem to offer contradictory signals. Douglas Wilder was elected Governor of Virginia--a feat which required substantial white support. Throughout

America the number of African-American elected officials has been increasing steadily since 1970, largely through the efforts of black candidates and voters. (Joint Center for Political Studies 1988)


Perceptions of Threat


What explains these seemingly contradictory trends? One starting point is the white mind. Both Wilder and Duke have been supported by white voters. This support may spring from the same pattern of racial orientation. The notion of racial threat is exemplified by whites who feel that their jobs, admission to schools, or lifestyles are in jeopardy because

of African-American advances. More generally, one may define the perception of racial threat as the belief on the part of whites that blacks:






3


1) as a class or as individuals may detract from whites,
economic well being by providing competition for
jobs that would otherwise go to whites;

2) present challenges to physical well being (i.e. that
criminal activity engaged in by blacks will endanger
their persons or property); or that

3) present a serious challenge to whites' political
standing (e.g. that black officials or officials
elected by blacks will wrest control of the ship of
state from whites and use the state apparatus to the
disadvantage of whites).

Perhaps low levels of perceived racial threat allowed some Virginia whites to vote for Wilder. Similarly, high levels of perceived racial threat fueled support for Duke in Louisiana. Studies by Hertzog (1991) and Giles and Buckner (1993) support this notion by finding a relationship between local racial composition and support for candidates at the district level. These studies echo the findings of earlier scholars (Key 1949, Blalock 1957, Allport 1954, Pettigrew 1959, and Giles 1977).

While both Hertzog (1991) and Giles and Buckner (1993)

claim the relationship between racial composition and support for these candidates is a function of racial threat, they are unable to substantiate this because they lack individuallevel data; they do not demonstrate that whites actually feel threatened. While from a behavioral standpoint it is only necessary to demonstrate that individuals perceive a threat (or feel threatened), the position adopted by Giles and Buckner and others posits that the material interests of whites actually are threatened; the necessity of this contention is assessed in the next chapter. using






4


individual-level data to construct a racial threat scale would facilitate the study of the relationship between perception of racial threat and political behavior. A number of other studies support the salience of white perceptions of racial threat and the effects of local racial composition. (Fossett and Kiecolt 1989, Giles and Evans 1985, 1986)

The difficulty encountered by those who hold theoretical stock in the racial threat concept is understandable. Relatively few of the national surveys contain items that would allow measurement of such a notion. Still, the 1986 American National Election Study (ANES) contains an extensive battery of racially oriented questions. Many of these items may be used to measure perceptions of racial threat. (See the Appendix for actual text and marginals). The theoretical ground staked out by the proponents of the racial threat framework may be measured, surveyed or tested with individual level survey data which is widely available.


Symbolic Racism


Despite a number of papers indicating the salience of

perceptions of racial threat and local racial composition in influencing white attitudes and voting behavior, the lion's share of research on white racial attitudes has situated itself in the symbolic racism paradigm (Kinder and Rhodebeck 1982, Kinder and Sears 1981, 1985, McConahay 1982, McConahay and Hough 1976, Sears, Hensler, and Speer 1979). These






5


authors share a common perspective based on a few simple premises. First, the changes wrought in America in the 1950s and 1960s have changed the nature of racial attitudes held by most Americans. It is no longer socially acceptable to hold openly racist positions, especially positions based on segregation or white supremacy. So, old fashioned racism, defined as the belief in segregationist values, cannot be advocated openly.

The second premise maintains that racism has a new form of expression. Whites now use their adherence to traditional, conservative values to legitimize anti-black positions. This allows them to explain prejudicial feelings and beliefs in terms of acceptable ideological stances. The use of these values is independent of the individual's demographic or economic context. In other words, one's class or location and proximity to blacks have no effect on one's racial attitudes.

This orientation is commonly termed 'symbolic racism,, a combination of anti-black affect and traditional values which (according to its proponents) has replaced the old racism rooted in segregationist values (Kinder and Sears 1981). According to Kinder and Sears, symbolic racism refers to a sense of disdain for those who fail to adhere to the Protestant work ethic and thus seek handouts, lack thrift or punctuality, fail to repress sexual urges or delay gratification to work for distant goals. A symbolic racist is likely to blame the victims of past injustice by saying






6


that if they just tried harder they could 'succeed on their own like the Italians, Irish and Jews did.' Kinder and Sears call this sort of racism symbolic because it is abstract, based on ideological positions and attitudes rather than experience. They argue that perceptions of threat and actual vulnerability are ineffectual predictors of behavior while symbolic measures excel.

Despite the growing literature on symbolic racism, more research is needed to refine our understanding. While these perspectives have drawn some criticism, chiefly from Sniderman and his co-authors, important elements of the theory have not been tested. Further, the elements that have been tested use local samples which might yield results with little external validity, or validity only if we throw out the South. Additionally, insufficient effort has been made to test the symbolic racism theses against racial threat hypotheses. There are exceptions to be sure. (Green and Cowden 1992, Kinder and Sears 1981, Sears, Lau, Tyler and Allen 1980) The theory and measurement of this school of thought are treated in detail in chapter three.


Synthesis


One need not choose sides in such a debate, for there is no a Priori reason to suspect that the two attitudes are mutually exclusive. The proponents of symbolic racism theory may be correct in asserting that an abstract sense of prejudice is held by many whites, but many may feel






7


threatened as well. Indeed, the two orientations may exist independently of each other in the same individual. These constructs may reflect different dimensions of contemporary racial attitudes. Individuals may evince high levels of symbolic prejudice, or feel racially threatened, or some combination of the two.

Similarly, one need not buy the entire theoretical package of either camp. For example, one might easily disagree with racial threat theorists who maintain that the cause of racial conflict is the incompatibility of the interests of whites and blacks. A political economist could probably demonstrate that black ascendance poses little threat to most whites. what is important from the behavioral perspective is the perception of conflict or threat. Still, the material context can influence the perception of threat.

For example, Green and Cowden's (1992) finding that

respondents with children who would be bused were more likely to protest the busing plan in Boston implies that the perceptions of threat may influence behavior dramatically, even if it does not influence one's tendency to articulate symbolically racist responses to survey questions. From this literature, one may gather that the perception of threat and symbolic racism may each have an independent force in influencing different kinds of positions on issues and evaluative behaviors. Rather than trying to supplant one perspective with another, the argument presented emphasizes recognizing the possibility that the two perspectives






8


represent different dimensions of racial attitudes among whites.

Further, one may argue that current racial attitudes, be they the result of perceived racial threat or symbolic racism, are not merely either the result of current demographic conditions or isolated psychological traits. Indeed, both may factor into the equation. If the current material context can influence an individual's racial attitudes, couldn't the cultural context have an influence as well?

In Chapter Four, a synthetic theoretical position is staked out which allows for the simultaneous influence of personal characteristics, contemporary material context and local cultural factors. Chapter Five addresses measurement issues and issues related to multi-dimensional racial attitudes. National level models predicting the racial attitudes and their effects are presented in Chapter Six. Chapter Seven explores regional variations in the sources of prejudice. A review of the central findings and the conclusions are found in the final chapter.

Before ending this introductory chapter a cautionary note concerning the motives of the author is in order.

The purpose of this project lies not in finding more ways to brand people as racists. Rather, the goal of this research lies in seeking a better understanding of the anatomy of prejudicial attitudes. These prejudicial attitudes may or may not have implications for political life






9


in America; likely as not, they do. It is instructive to recall R. K. Merton's conception of the relationship between prejudice and discrimination (1957). In a nutshell, Merton pointed out that the two concepts represented different dimensions; imagine a contingency table with prejudice on one axis and discrimination on the other. It is possible, Merton

hypothesized, that one can be highly prejudiced and not discriminate, never acting on that prejudice. Similarly, nonprejudiced individuals may engage in discriminatory activities.

At times provocative expressions are used. For example, the expression (not originated by this author) symbolic

racism carries some unnecessary shock value. The expression is really a misnomer, for the concept refers to a particular

theory of prejudice and need not carry the pejorative implications of racism. Branding people as racists may have few positive results, like kicking the dog. Nonetheless, this term is used in this research because it is used by other scholars in the field.

This study provides neither apology for nor an

admonition of prejudice per se. Rather, the author hopes to provide insights into the causes and effects of racial

prejudice. It is hoped that the development of a more sophisticated understanding of prejudice may contribute to

the amelioration of social conditions which have ill effects for whites and blacks alike; if we are to defeat the forces of ignorance and prejudice, we must first understand them.













CHAPTER 2
PERCEPTIONS OF RACIAL THREAT: THEORY AND MEASUREMENT

The claims of some scholars that control of the symbols of the race debate represents the central struggle of American politics (Omi and Winant, 1986) may be somewhat over stated. Still, race seems to represent a fundamental cleavage in American society. Even thirty years after the heyday of the civil rights movement, the United states continues to be a highly segregated society. This social and geographic segregation has implications for the political sphere as well. Race or race baiting continues to be used as a trump card in elections throughout the nation. To be sure, the race card is not always effective, but in many contests it continues to be decisive.

The role of race in the literature of political science is not unified by a common theoretical orientation, unit of analysis, or methodology. Two important albeit divergent streams are easily identifiable. The first tends to focus on the effects of race at some level of aggregation. Scholars in this stream may examine the effect of race or the local racial composition on party systems, class differences or tendencies within an electorate in particular contexts or environments. The second stream addresses the psychology of racial attitudes in individuals, apart from any specific




10





11


context. This second stream will be taken up in Chapter Three.

In this chapter, two examples of the first school will be examined in some detail. The first is a classic with a regional focus, and the second an important contemporary effort which explores the significance of race for electoral and party politics nationally. This will be followed by a close examination of the current work in this school which uses aggregate and individual level data. While there are some notable debates among the proponents of this general area of scholarship, the similarities deserve more emphasis than the points of disagreement.


Foundations


V. 0. Key's (1949) analysis in Southern Politics is

typical of the first and older line of argument. Key argued that understanding the role of race was essential to understanding Southern politics (p. 5). The means through which blacks were socially and politically subjugated thwarted the political articulation of class differences among whites. In particular, the one-party system common throughout the region prevented the empowerment of both blacks and poor whites through the political arena. Blacks were excluded from the system altogether. Extra-legal sanctions against those who sought change were severe. Blacks who braved those sanctions faced a maze of legal





12


obstacles to participation, including poll taxes and literacy tests.

Poor whites were stymied by the one party system in two ways. First, the barriers designed to exclude blacks from participation if.evenly applied would have also excluded many whites. Key dismisses the effects of legal barriers to white participation. This view is upheld by Harold Stanley's analysis of participation patterns since the revocation of those barriers (Stanley 1987; for contrary views see Kousser 1974; and Rusk and Stucker 1978). More important, according to Key, were other factors: low education, a lock by elites on real power, the punishment of dissenters, the obfuscation of factions caused by the one party system and race baiting. Indeed,

[t]he presence of the black provides a ready instrument for the destruction of tendencies toward class division
among the whites. The specter of Reconstruction, of Negro government can still be used to quell incipient
rebellion by discontented whites. (Key 1949, p. 655)



Key paints a picture of the South in which Democratic single-party rule operated to the benefit of upper class whites, planters and industrialists. Cheap black labor was exploited and blacks had little opportunity to participate in the political arena. Poor whites (rednecks, peckerwoods and peckerheads) enjoyed a similarly degraded position. They farmed inferior lands and had little hope for meaningful participation. Poor whites who did join the fray met with stiff retribution or faced near impossible odds in the





13


pursuit of their goals. The solidarity of elites within the Democratic party stifled class-based or populist efforts to address the agenda of the have-nots, be they white or black. The one-party south and the techniques of control used by its elites proved effective in preventing the development of class based alliances between blacks and poor whites.

In addition to this basic set of relationships, Key stresses the importance of the local racial composition. Recall Key's statement that, "[w]hatever phase of the southern political process one seeks to understand, sooner or later the trail of inquiry leads to the Negro.- (p. 5) The desire on the part of whites to control blacks is greatest, so the argument goes, in the those areas in which blacks make up a larger proportion of the population, namely the blackbelt. Planters sought to maintain the system which provided cheap labor for their estates. whites outside the black-belt were, hence, less concerned about the maintenance of white supremacy. So, it was in the black belt that the subjugation of blacks and the enforcement of a unified front on the race question by white elites was most effective. In some states, Key uses maps to demonstrate the strong correlation between those counties with more than 50% black and strong support for particular candidates and high levels of turnout. The political impact of the local racial composition was potent.

While this description bears some resemblance to the

broad view of Key's canvas, there were of course variations. Some populists, like "Big Jim" Folsom in Alabama, enjoyed





14


tremendous success. In some elections, outcomes had more to do with the county from which the candidate hailed than his platform or the black population. Still, as a rule the analysis is compelling.

This analysis makes much of the differences in the

interests and attitudes between poor whites on one hand and the planters and industrialists on the other. Indeed, Key makes much of the claims of the "haves" that race relations was a problem for the backward rednecks. Despite their divergent interests and claims, the trigger mechanism of their racial attitudes was often the same. Both groups thought they had something to lose from black ascendance. Planters were concerned about the loss of cheap labor which threatened their economic well-being and losing control of the state government through which they maintained their advantaged position. Key's argument implies that whites outside the black-belt had less to lose from black ascendance and even felt less threatened than other whites. Still, these whites were sensitive to race-baiting. This sensitivity grew out of the fear of a return to the poverty and degradation of the Reconstruction era. Key takes great pains to indicate that these poor whites acted in a way that was contrary to their class interests because, among other things, they feared black ascendance.

Some may argue that the usefulness of Key's work has faded as a result of the remarkable changes that have occurred in the South and the nation since its publication.





15


For example, even before Key started the project, dependence on cheap black labor began decreasing as a result of mechanization and lessened dependence on mono-culture (McAdam 1982). The civil rights movement has changed the face of racial politics. Electoral barriers have largely been erased. Not only have blacks earned the vote, but they have used it to make government and its policies more

representative. Reapportionment has changed the balance of power in state governments away from rural bourbons. Key (1949) himself pointed out the rapid migration of blacks out of the South.

Despite the changes, Key provides useful insights into

the mechanisms of politics generally. At the very least, Key's work provides a useful starting place for political

analysis. often his theses are upheld. For example, Stanley (1987) endorses Key's estimate that the relaxation of voting restriction would have a negligible effect on white turnout. This hypothesis is upheld by many subsequent studies (e.g. Allport 1954; Blalock 1957; Pettigrew 1957; Wilcox and Roof 1978)


The Contempgrary Significance of Race at the Macro Level


Key's analysis of the effect of race on the class

composition of the Democratic party has not fared as well in the changed environment. The threat (real or imaginary) posed by black ascendance was used as a mechanism for maintaining the one party system. This system forced lower





16


class whites into an unnatural coalition with upper class whites while excluding blacks altogether. After the civil rights movement, black allegiance to the Democratic party became almost universal. Now the perception of racial threats is the basis for another unnatural combination of classes in not merely the Southern party system but the national party system (Huckfeldt and Kohfeld 1989). According to Huckfeldt and Kohfeld's argument, the Democratic party plays host to middle and lower class blacks and upper middle and upper class white liberals. The Republican party combines socially conservative working class whites with rich, economically conservative whites.

The crux of Huckfeldt and Kohfeld's argument is that the nature of this unnatural coalition undervalues the contribution of blacks since their Democratic votes are taken for granted by both parties. If the parties had to compete for their votes then, the argument goes, blacks would be better represented because more elected officials would have to be more responsive. Further, such diversity could produce more stable, class based coalitions in the two major parties.

The conclusions drawn from this argument are provocative and in some way reinforce the importance of the project at hand. But a keener focus on some of the interior elements of the argument provide helpful insights. Huckfeldt and Kohfeld argue that the current coalitions are uncomfortable for a number of reasons. First, they affirm that the ideological base of American politics has shifted, that the labor-





17


management dichotomy no longer represents the sole axis of ideological placement. New left issues such as environmental concerns and social change issues (e.g. feminism, gay rights, etc.) don't fit on the old continuum and represent a different dimension.1 So, the Democratic coalition of the New Deal era is complicated by the addition of a new ideological dimension, and old-style labor liberals find they have less and less in common with new-left-lifestyle liberals.

The influx of black voters into the Democratic party after the passage of the Voting Rights Act of 1965 has further complicated the task of coalition building in the

party system. In particular, working class whites have become less supportive of the party. When one confronts this issue in terms of electoral coalitions, it becomes clear that the inclusion of blacks displaces working class whites disproportionately. In other words, the larger the

proportion blacks voters comprise of the Democratic vote, the smaller the proportion of whites in that same Democratic electorate. This would seems merely tautological but for the

fact that the reduction in the white contribution (in terms of Democratic votes) comes disproportionately from the





It should be noted that while Huckfeldt and Kohfeld
identify the salience of the changing landscape of American ideological orientations, they distance themselves from those
who would couch those changes in terms of the development of "post-material values." Indeed, they demonstrate many of the shortcoming of post-materialist theory.





18


working class. Indeed, there is a strong inverse relationship.

So, according to Huckfeldt and Kohfeld (1989), middle class and Working class blacks make strange Democratic bedfellows with new left upper class whites and some residue of the working class whites who do not convert or demobilize. The Republican coalition is similarly uncomfortable with its combination of upper class fiscal conservatives and working class whites. The authors also provide anecdotal data of the splintering effects of race in particular contests. Again the assumption concerning the "unnaturalness" of the coalitions lies in the class differences between the various groups.

Theoretically, working class whites and working class blacks have compatible material interests. Similarly, according to Huckfeldt and Kohfeld (1989), middle class whites and blacks are presumed to have the same material interests. Still, working class whites tend to flee the Democratic column in inverse proportions to black contributions to that column, even though it is in their material interest to pursue like policies and candidates. They flee because rightly or wrongly they feel as though their interests are incompatible. They flee because they feel that black politicians or politicians whose agenda gains obvious black supports pose some threat if elected.

Despite the potent effect race can have on electoral

coalitions, the argument put forth by Huckfeldt and Kohfeld





19


does not indicate that race must always have an effect. Indeed, by pointing to a specific contest in which race baiting was particularly effective, one is pushed to recognize that race baiting is less effective in other contests or was not employed at all.


Aggregate Analyses


Huckfeldt and Kohfeld (1989) identify the macro level relationship and pose potential solutions for the disequilibrium afflicting the party system. They do not, nor do they claim to, nail down the operative mechanism driving the white flight from the Democratic coalition. some recent papers have added support to this line of inquiry.

Two analyses of recent elections shed some light on the aggregate effects of local racial composition in electoral contests. First, Hertzog (1991) made creative use of county level returns in his analysis of African-American Douglas Wilder's 1989 election as governor of Virginia. At first it seems counter intuitive that the first elected black governor would come from a Southern state, especially the Southern state with the lowest percentage of black population of all Southern states. Hertzog's (1991) analysis, influenced by Key (1949), argues that this low percentage black actually facilitated Wilder's success. He presents compelling evidence to substantiate an inverse relationship between the percentage of African-Americans and the proportion of the white vote falling in Wilder's column.





20


of course, many other factors contributed to Wilder's success (Jones and Clemons 1993). wilder, the former Lieutenant Governor, had risen through a series of influential Virginia offices, so there were no doubts about his qualifications. He was endorsed by many prominent Democratic leaders in the state. Wilder's reputation and campaign statements placed him clearly as a "social moderate and a fiscal conservative" (Jones and Clemons 1993, p. 140). Wilder's pro-choice stance allowed him to benefit from the urgency lent the abortion issue by the U.S. Supreme Court's 1989 Webster decision. Wilder was perceived as tough on crime and most thought his opponent responsible for negative campaigning. Jones and Clemons (1993) claim that Wilder intentionally distanced himself from the black community and from racial issues, running what they term a 'deracialized, campaign.

The traditionally perceived relationship between racial composition and racial attitudes and white political behavior is born out by Hertzog's (1991) analysis. Of course, the percent black does not automatically determine victory in electoral contest. Much more is at play. In the Wilder case, one sees a highly qualified moderate Democrat running a professional campaign, endorsed by important 'pols', on the popular side of a divisive issue, and intentionally deemphasizing race. Wilder's performance was optimal. still the pattern of white support remains. Hertzog (1991) argues that those whites living in areas with fewer blacks felt less





21


threatened and could more easily cast their votes for the candidate they preferred for some reason other than race. Those who felt threatened had more difficulty in casting a vote for Wilder, other things being equal.

Of course, Hertzog (1991) is not able to substantiate the intervening variable, or mechanism, of threat without individual level data. It is entirely possible that the relationship he observes, albeit a strong one, is spurious. It is possible, but unlikely, that any other unmeasured demographic or social characteristic would correlate well enough with racial composition to produce such a pronounced effect. Indeed, Wilder's extremely narrow victory, despite wider margins in opinion polls, implies the importance of latent prejudices of which individuals were loath to speak.

Other scholars bring a similar sort of analysis to bear on a most different case. Giles and Buckner (1993) examined the pattern of support for David Duke in the 1990 Louisiana race for a seat in the U.S. Senate. Duke, widely perceived a racist, ran as a Republican and found himself in a run-off with Bennett Johnston, a Democrat. Duke, former leader of the Ku Klux Klan, has been characterized by some as a purveyor of thinly veiled racist jingoism. Less benevolent observers call him an overt racist. Duke tends to deny that he is a racist, saying he has changed since his days with the Klan. Still he does consider himself an advocate for white people's rights; he even founded an organization called the National Association for the Advancement of White People





22


(NAAWP). There is little doubt about the racial content of Duke's appeal.

Giles and Buckner's (1993) findings are consistent with those of Hertzog. They find a strong relationship between local racial composition and the level of white support for Duke2. Hertzog's estimates of white support for Wilder were complicated by that candidate's significant black support. Giles and Buckner assume that Duke garnered a negligible number of black votes. This reasonable assumption allows them to more fully examine the central relationship and to

introduce important controls in a multivariate analysis.

Despite the fact that they have only aggregate level

data, Giles and Buckner (1993) include a number of controls by including census data at the SMSA and parish level. These controls provide a convenient opportunity to assess the current impact of theoretically important variables from earlier work. Included as controls are two measures of social status, median white income and the percentage whites with a high school education. Additionally, the percent of whites unemployed, the percent living in urban areas, the percent of white immigration, and the percentage of the population coming of age after 1961 are also included.

Even with all these controls, their measure of racial composition proves efficacious in predicting both the
2 Operationally, Giles and Buckner use the percentage of all registered voters who are listed as black. They argue that this better represents the threat to white political hegemony than a population based figure. Of course, the two figures are so highly correlated that it makes no difference in the analysis.





23


percentage of registered whites voting for Duke and the percentage of eligible whites voting for Duke. Further, the direction of some of the significant controls lends support to the notion that the mechanism generating Duke's support is related to perception of threat. For example, median income is negatively correlated with Duke support, indicating that the most economically vulnerable localities were more likely to support Duke. The percentage of whites unemployed was significant and positive in the equation predicting the percentage of eligible whites voting for Duke. This implies that Duke's appeal was strong and had the effect of mobilizing voters who were particularly vulnerable and most likely to feel threatened by black ascendance. Urbanization and immigration had the expected ameliorative effects.

While both Hertzog (1991) and Giles and Buckner (1993) demonstrate support for the relationship between threat and candidate support in the aggregate and Giles and Buckner (1993) find some support that implies that threat is one of the mechanisms driving this relationship, both studies suffer from limitations imposed by their data. There is only so much one can learn from even the creative use of aggregate level data. Individual level data are required to establish the nature of the mechanism driving the behavior recognized by these scholars.





24




Individual Analyses


Despite the well-developed literature linking the size of minority populations and various measure of racial inequality, there are only a few studies using survey data which seek substantiate the connection between local racial

composition and the attitudes and behaviors of individuals. Really, only a handful of recent studies have systematically confronted the assumptions emanating from the aggregate literature discussed above (Fossett and Kiecolt 1989; Giles and Evans 1985, 1986; and Giles 1977). All of these studies performed secondary analysis of nationally sampled data collected by the two main academic research centers, the General Social Survey (GSS) conducted by National Opinion Research Center (NORC) and the American National Election Studies (ANES) by the Institute for Social Research (ISR). Despite the obvious attraction of testing hypotheses with a nationally drawn sample, the use of such data is not without

costs. Finding valid measures for the theoretical concepts of interest usually proves difficult and is often impossible. The creative operationalizations used by Giles and Evans and Fossett and Kiecolt represent serious, if not altogether successful, attempts to overcome the limitations imposed by the data. Though none satisfactorily address the intervening threat mechanism, they do merit discussion. A brief discussion of these recent studies follows. Three of





25


the four studies find support for the threat hypothesis (Giles 1977; Giles and Evans 1986; and Fossett and Kiecolt 1989). Giles and Evans's 1985 analysis marks the only exception.

The Giles (1977) study attempts to analyze the effect of racial composition on racial intolerance or hostility. using data from the 1972 ANES, Giles constructed a racial attitude score for each respondent. The six questions used to construct the scale covered a wide range of topics related to race including the federal government's role in guaranteeing equal opportunity, the federal government's role in overseeing school desegregation, the role of "the government in assuring equal access to public accommodations, the pace of the civil rights push, the trade off between whites, right to maintain segregated neighborhoods and blacks, right to live where they wish, and whether the respondent favors desegregation, strict segregation or something in between. He found that the expected relationship held, but only for respondents in the South. The regional differences may have been attributable, in part, to the heightened sensitivity of Southerners at the time to federally mandated desegregation programs; mandated desegregation of northern schools, for example, did not come until later. Half the items in the scale address the role of the government in mandating civil rights reforms. Of course, even this heightened sensitivity on the part of Southerners would follow from their greater sense of threat from the government and ascendant blacks.





26


Unfortunately, he was not able to tap perceived threat directly. Still, his findings support the notion that the perception of a threat is an intervening step between the context, individual attributes and political attitudes and behaviors. In this case, the articulation of intolerance provides an additional clue of the existence of perceived threat at the individual level.

The Giles and Evans study of 1985 uses data from the

1972 ANES and marks a departure from the earlier findings in two ways. First, they find what is on first appearances a viable measure of threat. In that study, respondents were asked if they thought the civil rights people were pushing too fast, too slow, or at the right pace. Additionally, they were asked if black people had too much, too little or the right amount of influence (See the Appendix for actual text). These two items were combined to make an additive scale of threat. So, respondents who thought the movement was going too fast and thought blacks had too much influence scored highest on the threat scale.

The second way in which this study marks a departure from previous ones lies in its findings. Giles and Evans (1985) argue that there is no significant relationship between local racial composition (percentage black) and perceived threat. Additionally, this non-relationship holds up even when regional differences are accounted for. This would seems to represent a chink in the armor of the proponents of the threat hypothesis. But appearances can be





27


deceiving. The Giles and Evans (1985) finding should not be accepted uncritically. Later efforts, discussed below, revealed shortcomings in their modeling procedures. still, they served the academy by stimulating discussion of this issue.

Additionally, Giles and Evans (1985) established a

relationship between the respondent's racial attitudes and his or her level of identification with white people as a group; the more hostile the respondents were, the greater their attachment to whites. In other words, the sense of racial polarization was evident not only in terms of people's positions on about the civil rights people, or black people, but also in terms of how they saw themselves. Similarly, those who identified British ancestry, as opposed to those who identified no specific ethnic heritage, felt warmer toward white people. So, identification with whites as a reference groups was enhanced by both the Giles and Evans (1985) racial threat scale and by identification with some other group, in this case the British. One caveat about these findings is in order. Giles and Evans (1985) did not include controls for region in the model estimating attachment to whites. This may present a significant source of bias if a greater proportion of Southerners trace their ancestry to Britain than do Northerners, a not altogether remote possibility.

The scale developed by Giles and Evans (1985) represents the most serious attempt by political scientists to date to






28


measure perceived threat. It was later adopted by Fossett and Kiecolt (1989). Fossett and Kiecolt (1989) reanalyze the same data used by Giles and Evans (1985), but their findings differ. While Giles and Evans (1985) find that racial composition is not generally linked to their threat scale, the more recent study finds the expected relationship between local racial composition and the threat scale, and not only for Southern respondents. Fossett and Kiecolt (1989) found the percent black to predict their measure of threat in the total sample and in both the Southern and non-Southern subsamples. The difference in their findings results from differences in model specification, discussed later in this chapter.

Despite its apparent usefulness, the scale is not without problems. First, one might question the face validity of the items. One could think that the civil rights people are pushing too fast because of a strategic orientation; perhaps a sympathetic respondent believes that more economic power should be amassed before continued political maneuvers proceed. Similarly, one may think that any civil rights push is too fast regardless of one's vulnerability to threats from ascendant blacks. A second and related problem lies how the items have been perceived and utilized by other scholars; this issue will be taken up in the next chapter.

Fossett and Kiecolt (1989) also look for regional

differences in respondent's support for integration, but are





29


unable to support Giles, (1977) Southern exceptionalism explanation. Fossett and Kiecolt's (1989) results differ because of a slight difference in the way they specified their models. Using data from the 1976 and 1977 GSS, they find that support for integration was inversely related to the percentage black in all regions. They included a measure of the respondent's community size, because 'urbanness' should produce greater tolerance (Anderson 1962; Jackmnan 1978; Thomlinson 1969). Indeed, their data support the ameliorating effect of 'urbanness.' Moreover, the inclusion of urbanness in the models reveals the relationship between racial composition and perceived threat. Presumably, respondents from urban areas are so much more tolerant in general than those residing in rural areas that the predicted linear effect of racial composition is obscured unless community size is controlled.

Fossett and Kiecolt's (1989) rebuff of Giles, (1977)

lies in their finding that the direction of the relationship is the same for both southern and non-southern sub samples. This they take as support for "Lieberson's (1980) hypothesis that regional differences in white racial attitudes result in part from regional differences in the distribution of certain contextual variable." (Fossett and Kiecolt 1989, p.829) This may be true in part, but their own data indicate that even with several control variables a South dummy variable remains statistically significant in the prejudicial direction. Similarly, the effects of percentage black and a measure of




30


the respondent's Southern origins are significantly different in models estimated for Southern and non-Southern subsamples. So, in addition to "differences in the regional distribution of certain contextual variables," Southernness make a significant and unique contribution. The regional difference is one of degree not of kind, but it remains a significant difference.

As might be expected, the various measures of

Southernness have their effect in the more prejudicial direction. This points in the direction of another

unrealized finding from the data presented by Fossett and Kiecolt (1989). The effect of Southern origins is insignificant in the non-southern sample. This insignificance may be the result of a paucity of cases. Alternatively, the insignificance of this variable may be like the dog that did not bark of Sherlock Holmes fame. The fact that Southern transplants in the North are not significantly more prejudicial that their neighbors may point to the importance of context in structuring political attitudes. This interpretation is supported by [or

supports] contextual arguments made by Wright (1977) and others, namely that the most pronounced and important contextual effects are mediated through one's primary group

processes. In other words, "contextual characteristics operate by influencing the types of people with whom one is likely to interact." (Wright 1977, p.505) The behavior of




31


the Bubbas in Babylon, like that of a fish out of water, is different than it would be 'Down Home.'

Using the same data (1972 ANES) exploited by Giles and Evans (1985), Fossett and Kiecolt (1989) re-estimate the effects of several contextual and individual predictor variables on their shared measure of perceived threat. They find that local racial composition influences their measure of threat in the expected direction for all samples and subsamples. Again, their specification differs from that of Giles and Evans (1985) by the inclusion of the community size control. Community size is negatively related to perceived threat in all samples, so urbanness appears to have an ameliorating effect on prejudicial attitudes. What they do not find is a significant relationship between either a South dummy variable or southern origins and their measure of perceived threat. This runs counter to what one would expect. Indeed, no variable predicting their measure of threat has a significant regional difference.

Fossett and Kiecolt (1989) go further and examine the

effect of the threat scale on support for racial integration. Here they find, as expected, that the threat scale has a significant inverse relationship with support for integration. Even with the threat scale in the model, the percentage black has the expected (inverse) effect in all samples. But in these models there are important regional variations. First the threat scale has a more pronounced effect in the Southern model. Second the Southern origins





32


variable, as above, is significant in the prejudicial direction only in the Southern sample; Southerners in exile are not less likely to support integration that their neighbors. Further, the difference of these two variables alone accounts for a 13.4% increase in the amount of variance explained in the Southern sample over the Northern sample. These difference reinforce the importance of context in allowing both access to socialized values and in allowing the use of those values in determining policy preferences.


Reconciliation


This discussion of the interpretation of regional

contextual variable points to another question deserving further discussion. Giles and Evans (1985, 1986) treat racial composition as a measure of external threat. This interpretation is consistent with some of the sociological literature on 'ethnic conflict, or competitive ethnicity (see Giles and Evans 1985, 1986 for a review of this literature and citations). In the 1985 article they make the distinction between external threat and perceived threat. Perceived threat was measured by asking the respondents questions while the percent black represented the sole measure of external threat. Giles and Evans (1985) may be correct in making this assumption. In this way, they distinguish themselves from Marxian and modernization theorists who characterize race as an ascriptive identifier and of diminishing salience in developed societies.





33


The discussion of the role of race in politics by Key

(1949) and Huckfeldt and Kohfeld (1989) implies that race is not primordial, that it does not represent an unresolvable cleavage. For Key (1949) the fundamental struggle of Southern politics was class based. Race was used, in part, as an instrument to thwart insurgent actions by the havenots. Key (1949) characterizes the net effect of racial politics as anti-democratic because the system prevented the recognition of objective interests and action based on those interests. In this sense race is an unnatural basis for the formation of group cleavages. Rather, tenant farmers and yeoman farmers had common interests regardless of race. Race baiting prevented the recognition of these common interests and other forms of repression discouraged participation not only of blacks, but of poor whites as well. Huckfeldt and Kohfeld's analysis is similar. Race represents an unnatural cleavage that prevents the formation of class based political parties.

While Giles and Evans (1985,1986) do not test the validity of their assumption that whites actually are threatened by local blacks or cite empirical studies which support it, there may be reason to think that there is something to the notion. For example, the social change movements of the 1960s and 1970s created an environment in which racial and gender discrimination was made more difficult. As a result, women and blacks entered the job market and began taking some of the jobs that would have





34


either gone to men or for which they would have previously received substandard wages. This represents a good deal for those who had been discriminated against before, because they were getting better treatment and wages than they had previously. White male workers faced increased competition for positions and wages. This depressed real wages. So, white men lost the noncompetitivee advantage they had previously held.

of course, some would argue that working class whites

are right to feel threatened because the inclusion of blacks in the mainstream work force had a more immediate effect on those on the lower end of the job market. Neither upper management nor the owners of capital have much to fear from the rapid expansion of the labor market after three hundred years of systematic and costly discrimination. Demonstrating the actual effects of the desegregation of the work place is less important for this study than demonstrating whites, belief that such desegregation poses a threat to them and that perception of threat influenced behavior.

This sort of supply and demand analysis of the effects of the relaxation of color and gender-based barriers to employment on whites makes the Giles and Evans assumption seem less far-fetched. Of course, the changes described above could have been a short term shock which later resulted in a new competitive equilibrium. This may be an empirical question, but it is one that need not be answered in the present study.





35


Whether or not local racial composition represents an actual or imagined threat to whites will not be examined in this study. The question taken up in this study centers on the operative cognitive mechanism driving racial politics among whites. In this sense, one need not answer the question of whether or not the political, economic or cultural status of whites is objectively threatened or if such a specter is merely the result of the political conjuring of opportunists, entrepreneurs and extremists. This central difference between scholars in the threat school is less important than their area of agreement, namely that the perception of threat can be a potent political force. The potential reconciliation of these perspectives will be discussed in more detail in Chapter Four. As the second major school of political research into the racial attitudes of whites is discussed in the next chapter, this common ground will come into even sharper focus and the points of difference will fade into the background.














CHAPTER 3
SYMBOLIC RACISM: THEORY AND MEASUREMENT

The role of race in politics has stimulated much controversy in political science. In the last chapter, discussion focused on approaches which point to the perception of threat as the operative mechanism of white racial antagonism. While scholars disagree as to the verity of the threat, many see its perception as instrumental in explaining, in part, larger political phenomena. This school of thought finds its recent roots in the literature of Southern politics and related changes to the party system since the civil rights era. Its methods are largely aggregative, although a handful of individual level studies exist. Some of these were discussed.

In this chapter, discussion centers on different

explanations of the role of race in politics. Symbolic

racism represents perhaps the dominant paradigm in the political science literature on white racial attitudes. This


1 "Symbolic racism" refers to a school of thought in the literature on American racial prejudice. Other names for this school have been bowed, new racism and modern racism. These alternatives have all the flaws and none of the strengths of the original term. Symbolic racism represents the most universally recognized moniker for this particular theoretical tradition. Many have attacked the expression on a number of grounds. The use of the term in this study is not meant as an endorsement of its content or its appropriateness as a label for that content. Rather, its use is adopted, for the purpose of clarity only, to refer to this school of thought.

36






37


school of thought grows out of the more general orientation of symbolic politics. Since this theoretical tradition is psychological, its method is exclusively individually oriented. Although a few experimental designs have been conducted, most work in this area has exploited survey data.

The chapter opens with a brief discussion of symbolic politics and the role of racial attitudes in that theory of politics. This is followed by a description of the development of the essential elements of the theory and its landmark studies. The chapter closes with an account of the major critiques of the approach and its methods.


Symbolic Politics in a Nutshell


The symbolic racism orientation grows out of the

symbolic politics tradition. A thumbnail sketch of this orientation will provide the foundation for the later treatment of symbolic racism. The symbolic tradition argues that political behavior is driven by cognitive and affective processes. Pre-adult socialization and later resocialization creates a number of affective orientations or symbolic predispositions which provide an enduring standard for the evaluation of new stimuli. Among these affective orientations are numbered racial attitudes, ideology and


2 The brand of symbolic politics advocated by Sears (see Sears 1988 for a summary), in particular, shares some common ground with Edelman's (1971) articulation, but features one dramatic point of departure. While Edelman left room for the influence of self interest, Sears and his colleagues dismiss such influences.






38


partisan identification. These enduring predispositions determine political attitudes, like issue positions and candidate evaluations, and behaviors, like voting or participation in demonstrations.

The impact of the symbolic predispositions attitudes or behavior occurs to the exclusion of an individual's current personal or economic situation. In this perspective, the material context has little influence on the individual. Indeed, the effects of actions based on self-interest (or what people think their self-interest is) is largely relegated to the private sphere.

of course, this caricature of this theoretical

traditions may seem a bit far-fetched. A less exaggerated example from the economic voting literature may provide a more favorable reading. Some have thought that voters in America engaged in pocket-book voting, making their electoral choice in favor of the candidate most likely to benefit them financially (Bloom and Price 1975; Tufte 1978) or closest to them on the content of issues (e. g. Page 1977; Riker and Ordeshook 1973). Students of symbolic politics have argued that this is not the case (Kinder and Kiewiet 1979; 1981). Rather than making their decision based on their individual interest, voters cast their ballots based on how they think the country as a whole is doing, sometimes called 'sociotropic voting'. Thus the choice revolves around which candidate is best for the country, not around which candidate is best for the individual. Indeed, Kinder and Kiewiet found






39


support for this proposition in their 1981 study, even when measures of satisfaction with income and satisfaction with current financial status were controlled.

of course, this literature is not without critics, but neither is it without fans. In this case discussed above, the symbolic politics approach contributes to our understanding of the calculus of the vote choice. At the same time, its usefulness in understanding politics in general may be reasonably questioned. The approach constitutes a black box of sorts; the symbolic predispositions go in one side and the vote comes out the other. One might imagine that the context in which this process occurs might have some influence on it. How does the voter evaluate what is 'good for the country, economically vis-a-vis what is 'good for the country, militarily or racially or educationally? Do voters project their preferences as the most benevolent path for the country? Despite these and other obvious criticisms, one can see the possibility that an individual might suppress his or her economic self-interest in making the vote choice.


Theoretical Underpinnings of Symbolic Racism


The theory of symbolic racism was introduced about 20 years ago (Sears and Kinder 1971). It has not remained static since that time but has evolved. In particular, the claims made by its proponents early on have been tempered. For example, The assertion that traditional prejudice, "can






40


no longer be a major political force," in American politics (Kinder and Sears, 1981) is no longer defended by either of its authors. The proponents of symbolic racism theory are not of one mind. There are disagreements about many aspects of the theory. What follows is a portrait of a symbolic racism that may not match any one scholar's description exactly but which attempts to capture the mean or essence of all its proponents; this is meant to provide a basis for a general critique of the theory and the generation of hypotheses. Unlike a straw man, the composite presented below is likely stronger than any one formulation of the theory presented to date. This portrait is followed by some attention to the elements of disagreement and unity within the symbolic racism camp over central themes. The discussion in later sections should make the limits of the derivative statement of symbolic racism apparent and point to specific places in the literature where disagreements occur. In subsequent sections, attention is paid to controversies surround data measurement and methods.

The logic of symbolic racism mirrors the logic of

symbolic politics outlined above. Pre-adult socialization and adult resocialization result in the creation of a set of racial predispositions or attitudes which provide an immutable standard for the evaluation of new issues, candidates or other stimuli. Current racial attitudes are not the function of one's situation in life. Rather, they are abstract. These attitudes or predispositions provide the






41


source for positions on issues, the evaluation of candidates and other external stimuli. Ultimately the predispositions result in behaviors such as voting.

There are two central elements to the racial attitudes. The first is anti-black affect, a generalized distaste for or disapproval of blacks. The second essential element is found in a commitment to traditional values, not limited to but epitomized by the Protestant work ethic. Thus Kinder and Sears (1981, also quoted in Sears 1988) list the ingredients of symbolic racism as

a blend of antiblack affect and the kind of traditional
moral values embodied in the Protestant Ethic . a form of resistance to change to the racial status quo
based on moral feelings that blacks violate such
traditional American Values as individualism and selfreliance, the work ethic, obedience and discipline. (p.
416)

These components work in 'conjunction, together to form symbolic racial predispositions.

The proponents argue that this symbolic attitude is a new form of racism, that it is distinct from redneck racism, "open bigotry of the sort frequently associated with workingand lower-class (Southern) whites" (McConahay 1982, p. 705), in either or both of two ways. Redneck racism has either 1) disappeared or is of dramatically reduced salience in contemporary society or 2) become old fashioned. That is to say, it "is no longer fashionable to express these beliefs or to support these practices openly in the elite circles of our society.- (McConahay 1982, p. 705)






42


Finally, like other symbolic predispositions in the

symbolic politics paradigm, symbolic racial attitudes were thought to be completely abstract. While older forms of racial attitudes had been thought to be related to the context within which the individual found him or herself, this new set of attitudes was thought to have no relationship to the individual's context or material interests.

In sum, the essence of symbolic racism, as argued by its proponents, is characterized by three points. First, it is a new racial attitude, distinct from previously examined prejudicial attitudes. Second, it is comprised of two central elements, antiblack affect and traditional values. Third, it is wholly abstract, completely unrelated to an individual's context or material interests. These three characteristics seem to represent the best and most widely accepted definition, of the concept. The proponents of symbolic racism have stood by these elements in the face of some of their more trenchant critics (see the Sniderman and Tetlock (1986a) exchange with Kinder 1986, or the chapters by Sears and Bobo in Katz and Taylor, 1988). These exchanges provide a more detailed history of the concepts development and weaknesses than space permits here. A more detailed discussion of the three central elements follows.


Newness or Distinctiveness from old Fashioned Racis


Early articulations of symbolic racism maintained that the new combination of anti-black affect and traditional






43


values had replaced the old racism (Kinder and Sears 1981). Lately, both Kinder (1986) and Sears (1988) have backed of f from that position and now argue that the old fashioned racism still exists and is different from symbolic racism and that both have anti-black affect in common. All proponents defend the distinctiveness of symbolic racism from previously recognized forms of prejudice.

The new racism was distinguished from two other

recognized for-ms of racism: 1) generalized (in)egalitarianism or 'redneck racism' which fostered "legalized or formalized racial discrimination, [and] inherent racial inferiority" (Sears and McConahay 1973, 138) among other things; and 2) personal racial threats, characterized by threats to an individual's job, housing, or safety.

Kinder and Sears (1971, 1981) dismissed the importance of redneck or old-fashioned racism and were joined in this dismissal by associates McConahay and Hough (1976) and various constellations of co-authors (examples follow). They point to declining levels of support in surveys for segregationist positions as evidence of the declining significance of old fashioned prejudice3. So by 1981, Kinder


3 See Schuman, Steeh and Bobo (1985) for an extensive discussion of the declining levels of support for segregation over time. In that study, they indicate a period of decline in prejudicial attitudes throughout the civil rights era before a new equilibrium (at relatively low, but not inconsequential levels) is reached. Few respondents to segregationist survey questions indicate the most prejudicial positions. But as Sniderman and Tetlock point out (1986a, also Sniderman, Tetlock, Piazza, and Kendrick 1991) it is not evident why few choose the prejudicial responses to those questions. Sniderman and Tetlock (1986a) pose one






44


and Sears could declare that America had become essentially racially egalitarian.

The most provocative assertion of the newness of

symbolic racism is found in Kinder and Sears' 1981 study. "Since the explicitly segregationist, white supremacist view has all but disappeared, it can no longer be a political force . . what has replaced it . is symbolic racism (p. 416). McConahay, too, makes claims that the new racism in now dominant while old fashioned racism is a thing of the pre-civil-rights-era past (1982).

while some studies made claims about symbolic racism

replacing other racial attitude, others point to the concept as simply one form of racial attitude among any number of racial attitudes. Sears and Kinder (1985) back off from the replacement hypotheses to a position recognizing the distinctness of symbolic attitudes as opposed to other forms. Others point to common elements across different types of racial attitude. One landmark study which portrayed symbolic racism as new and emergent in suburban America points to its similarity to older prejudicial attitudes by saying that symbolic racism is an "expression of some of the negative


explanation based on the changing language of prejudice. They argue, for example, that the content of racial stereotypes has changed over the years, but that the underlying reasons for the application of prejudicial stereotypes has not, i.e. bigotry. In this sense, they imply that the measures evoke less response not because America has become a land of racial egalitarianism. Rather, the old questions no longer resonate with the underlying prejudicial attitudes. In other words, the changing context has rendered the old measures useless in tapping persistent prejudicial attitudes.






45


feelings underlying old fashioned or red-neck racism, but it differs from them in its other psychological roots and in many of its specific forms of expression." (McConahay and Hough 1976, p. 24) A similar claim is made in a later study by McConahay (1982, p. 705) A 1979 study's factor analysis found three racial dimensions, but since all were correlated with anti-busing attitudes they were treated as one dimension (Sears, Hensler and Speer 1979, p. 382n).

The original claims of newness have been softened

somewhat. The proponents do maintain that this expression of racial attitudes is a post civil rights era phenomenon, that it is distinct from other racial attitudes, and that it is more important.


Combination or Multi-dimensionality


While most proponents of symbolic racism agree that

traditional values and anti-black affect are ingredients of symbolic racism, there is some dissension as to the exact nature of the relationships between the concepts. Kinder recently claimed that symbolic racism is the 'conjunction' of these values and racial prejudice. (1986) The precise relationship between the two elements has never been specified. McConahay and Hough (1976) stress this facet of symbolic racism by calling attention to a number of constituent elements. For McConahay and Hough (1976), symbolic racism has blacks as a group as its object, a moral tone (p. 36-7), addresses issues surrounding the violation of






46


the Protestant ethic (p. 41; also McConahay 1982), and has as it -most important psychological factor . political and economic conservatism . ." (p. 37)

A similar approach is taken by Sears, Hensler and

Speer's (1979) articulation of symbolic politics, "anti busing sentiment as primarily derived from long-standing political predispositions, specifically racial prejudice and political conservatism, which in turn can be traced to political socialization in preadult life." (p. 372). Kinder

and Sears (1981) point to symbolic racism's "blend of antiblack affect and the kind of traditional American moral values embodied in the Protestant ethic.,, (p. 416; also endorsed in Kinder and Sears 1985; Kinder 1986; Sears 1988) The formulation by Sears and Citrin is similar, "the conjunction of two separate underlying dimensions: a specifically anti-black attitude, and conservative value priorities, which stress individual effort and responsibility rather than government activism." (1982, p. 168)

Nowhere in the literature on this topic is there a clear articulation of how the elements fit together, why they should not be treated as distinct dependent or independent variables or how they are different from garden-variety prejudice and.conservatism. one gathers that the

relationship is mutually reinforcing. A dedication to the values embodied in the work ethic reduces sympathy with the unemployed, underemployed, or otherwise less successful folk one might hear about--perhaps poor blacks on the television






47


or a street corner. The residue of anti-black affect reduces the level of sympathy resulting in the assignment of blame for the outcast group. So the perceived failings of the out group to live up to expectations and the attendant, but abstract, disdain for that out group result in the symbolic" racial attitude.

Despite criticism over the combination of analytically and (potentially) empirically distinct elements (note especially the frontal assault by Sniderman and Tetlock 1986a and a less successful effort by Bobo 1983), the proponents stand by the 'conjunction, premise.


Abstraction vs. Self Interest


The concept of symbolic racism was introduced in Sears and Kinder's 1971 study of the 1969 Mayoral election in Los Angeles. This election pitted a conservative white candidate, Yorty, against a liberal black candidate, Bradley. The concept developed inductively from their analysis of survey questions asked during the election (Kinder 1986, McConahay 1976). From the responses to the questions the authors inferred that a new form of racism had taken root. This new racism consisted of "abstract moral assertions about blacks, behavior as a group .-(Sears and McConahay

1973)

This new racism was thought to be abstract, not based in the individual's current material conditions, and hence symbolic. If not based in material conditions, where does






48


the attitude have its roots? Like other predisposition in the symbolic politics paradigm, symbolic racism has its roots in pre-adult and adult socialization. Let us leave aside for the moment issues arising from the effects of material conditions on socialization.

All the studies of symbolic racism draw the distinction between the abstract, or value driven, nature of symbolic politics and the material context in which the individual finds him- or herself. This abstractness principle is generally presented in juxtaposition to the notion of threats to personal lives (Kinder and Sears 1981, p. 416, Sears and Citrin 1982) or the effects of personal experiences (McConahay and Hough 1976, p. 37; McConahay 1982, p. 705n; Sears, Lau, Tyler and Allen 1980, p. 671; Sears and Lau 1983, p. 224) The articulation found in Sears, Hensler and Speer (1979) is particularly behavioral in its emphasis on the stimulus-response mechanism:


Whether or not the issue has some tangible consequence
for the adult voter's personal life is irrelevant.
One's relevant personal stake, in the issue is an
emotional, symbolic one; it triggers long-held, habitual
responses. (1979, p. 371)






4 The position articulated in a more general study of symbolic attitudes is inconsistent with the statement. In Kinder and Rhodebeck's analysis of the 1972-76 ANES panel study (1982) they state, "These stability coefficients . imply both that racial beliefs are comparably durable . and that they are far from fixed. Racial socialization does not end with adolescence.- (p. 205)






49


Two statements by Kinder and Sears (1985) reinforce this point.

Resistance to change in the racial status quo would
seem, there-fore, to have little to do with the direct
tangible threats blacks might pose to personal life, and
a great deal to do, instead, with prejudice and values.
(p. 1141)


We say that personal life is usually compartmentalized
away from political life; that any real or imagined
threats posed by blacks to whites' private lives seldom spill over into antiblack political positions. (p. 1143)

Given the evidence of regional variation presented in

the previous chapter, it is difficult to accept uncritically the assertion that racial attitudes are divorced from context. Still, the proponents of symbolic racism (speaking for themselves) make their position abundantly clear.


Methods and Measurement


Given its ad hoc development, symbolic racism did not start out with a set of theoretically derived measures. Rather the concept was deduced from measures meant to tap general racial attitudes in the Los Angeles studies (Sears and Kinder 1970, 1971; Kinder and Sears 1981). Since that time, efforts have been made to develop better measures. These efforts have met with varying degrees of success. While critics of symbolic racism point to shortcomings in the measures used in a number of studies employing national level data, Kinder (1986) and Sears (1988) argue that these studies were not designed to test symbolic racism per se. Rather, they argue that the studies were examinations of general






50


symbolic politics hypotheses which happened to include racial measures. For example Sniderman and Tetlock (1986) criticize Sears, Hensler, and Speer (1979) and Sears, Lau, Tyler and Allen (1980) for combining measures of old fashioned and symbolic racism in these studies. Kinder 1986 and Sears 19885 respond that those studies do not measure symbolic racism but symbolic politics generally. Rather than dwelling on contested measures, what follows is a discussion of the current concepts and measures most widely accepted by the proponents of the theory, setting up a rationale for the measures used in the present study.

while symbolic racism is the conjunction of antiblack

affect and traditional values, Sears argues that it manifests itself in two ways: "(a) antagonism toward blacks' "pushing too hard' and moving too fast, especially (though not exclusively) through the use of violence and (b) resentment toward special favors for blacks . ." (Sears 1988, p. 56) These manifestations, more or less, reflect the constituent theoretical parts. For example, a belief in self-reliance combined with antiblack affect may make one more disposed to latch onto stereotypes concerning welfare. Table 3.1 below list examples of the operationalization of these concepts in a series of symbolic racism studies conducted between 1976 and 1985. This list, compiled by




5 Sears (1988) claims that these two studies did not measure symbolic racism on page 64 after crediting those same studies as examples of the concept's efficacy on page 59.






51


Table 3.1 Sears's two dimensions of symbolic racism measures

Antagonism Toward Blacks' Demands

Blacks are getting too demanding in their push for equal rights. (Agree)b,c

Blacks shouldn't push themselves where they are not wanted. (Agree)a

Some say that the civil rights people have been trying to push too fast. Other feel they haven't pushed fast enough. (Trying to push to fast)d

It is easy to understand the anger of black people in America. (Disagree)b,c

Resentment About Special Favors for Blacks

Over the past few years, the government and the news media have shown more respect to blacks than they deserve. (Agree)b

Over the past few years, blacks have gotten more economically than they deserve. (Agree)b

The government should not make any special effort to
help blacks and other racial minorities because they should help themselves. (Agree)e

Do you think blacks who receive money from welfare
programs could get along without it if they tried, or do you think they really need the help? (Could get along)a

Do you think the Los Angeles city officials pay more,
less, or the same attention to a request or complaint from a black person as from a white person? (More)a

a Kinder and Sears (1981). b McConahay (1982)
c McConahay and Hough (1976) d Sears and Allen (1984) e Sears and Citrin (1982)



Source: Sears (1988, p. 57)






52


Sears (1988)6, contains none of the contested measures and is drawn from a wide range of studies. one must conclude that these items have some measure of face validity for Sears, Kinder, McConahay and the other advocates of the concept. Upon review of Sears, list of the currently accepted measures for symbolic racism, one observes that the questions have one fundamental thing in common. They all have blacks as their objects. This marks a point of theoretical consistency. If the concept of symbolic racism has at its core the notion that blacks violate traditional American values, then it follows that survey questions meant to tap that core should have blacks as their focus. Beyond this, some of the questions in this list clearly indicate specific values, in particular self-reliance, or getting ahead without government assistance. Other questions not on this list show remarkable face validity in terms of their apparent ability to capture both antiblack affect and values such as self-reliance. The "blacks could get ahead like the Jews, Irish and Italians', question is a particularly good example of this (see the appendix for the text of questions used in this study). These and other questions will be discussed at greater length in the chapter on methods.







6 Kinder (1986, pp. 156-7) provides a more exhaustive list. The shorter list showcased here is adequate to give the flavor of the measures.






53




Self Interest: The Alternative Hypothesis


The series of symbolic racism studies go to great

lengths to test the various symbolic racism scales against a single alternative hypotheses. The primary alternative is represented by some measure of self interest. The proponents may be justified in their selection of this hypothesis, for it is compelling. But there has been relatively little response to criticisms of their operationalization or choice. For example, the theoretical traditions alluded to in the previous chapter are given little play in the writings of the symbolic advocates. Their discussion of the self interest alternative sometimes draws clear distinctions between its narrow construction and more general threat hypotheses and sometimes that distinction is more obscure. The alternative hypotheses they explore represent a most narrow construction of 'interest.' On the other hand, this narrow construction may provide clarity and assure that this particular alternative is well explored.

One must ask how was this alternative hypothesis

measured. While methodological cheap shots do not provide a basis for the formulation of a theory, they can help one critically evaluate the extent to which the evidence marshaled in support of a theory is supportive. The proponents of symbolic racism go to great lengths to test






54


this (self-interest) alternative hypothesis. Unfortunately, the extensive tests provided leave lingering questions.

First, a couple of general methodological issues should be addressed. In a number of studies, the proponents use a number of tests to gauge the relative explanatory power of self-interest and symbolic racism (Sears, Hensler and Speer 1979; Sears, Lau, Tyler and Allen 1980; Sears and Citrin 1982; Kinder and Sears 1981; McConahay 1982). Some of the tests consistently present difficulties in interpretation. while there is no reason to suspect an intent to obfuscate, one finds a consistent pattern of modeling procedures which could underestimate the effects of alternative hypotheses. These three tests are discussed below. The first two types of test involve problems associated with the comparison of a scale to a collection of individual variables and the third involves questions concerning the proper comparison of populations.

First, a symbolic racism scale is compared with a number of measures for self interest. For example, in Kinder and Sears 1981 study, the effect of the symbolic racism scale is compared with that of a handful of self interest indicators in a regression equation predicting support for one of the mayoral candidates in the 1973 election in Los Angeles. The scale has a consistently significant effect in the predicted direction while the other measures tend not to have significant effects. Of course, comparing a scale to a collection of, presumably, correlated measures presents some






55


difficulties. Assuming the measures for self-interest were valid, they might not appear as significant while appearing in an equation with other measures of the same concept; the explained variance may have been spread thinly over the multiple indicator so thinly that no individual measure had a statistically significant effect. In other words, each of the parameter estimates for the measures of self interest report the effect of that measure of self interest while controlling for the effects of other measures of the same concept. Finding even a small effect may have theoretically significant ramifications (see for example, King 1986), namely that self-interest makes some contribution.

A second kind of test is used to compare the predictive abilities of the two concepts. For example, in their table 3, Kinder and Sears (1981) estimate a regression equation with the symbolic racism scale as the sole independent variable. They note the adjusted R2, the percentage of variance in the dependent variable explained by all the independent variables adjusted for the total number of variables in the model. Next they add the several measure of self-interest to the model and note the adjusted R2 for the total model including the symbolic racism scale and the measures of self-interest. By subtracting the first R2 from the second they can gauge the additional amount variance explained by the measures of self-interest. But herein lies the rub. In the calculation the adjusted R2, symbolic racism is treated like a single variable while the collection of






56


self-interest measures, also measuring a single concept, are treated as multiple unrelated measures. This comparison presents a bias against the unsealed items, a bias that is amplified by colinearity among the unsealed items.

A third test breaks the sample into smaller groupings

based on the responses to one of the measures designed to tap self-interest. The complete model is then re-estimated (without the measure used to divide the sample) for each subgroup of the total sample. As above, the symbolic racism scale is found to have significant predictive power while the measures of self-interest fail. This interpretation is problematic. First, the division into the groups represents just another way of controlling for the effect of the variable used as the divisive criterion. Such a division is reasonable under a number of circumstances, for example if nonlinear effects are expected for different levels of the criterion variable or if the criterion variable was shown to empirically or has a theoretically significant interaction with another variable in the model. The presentation by Kinder and Sears (1981) is not much of an improvement over having the criterion variable as another independent variable.

A second problem associated with this test is that it rests on a comparison of the marginal effects of the independent variables across 'populations, or sub-samples rather than on the different absolute levels within each subsample. In this sense, as above, one expects the marginal






57


effects to be the same as when the criterion variable was controlled statistically in an earlier equation. If there were differences in the absolute levels of the dependent variable, in this case the evaluation of a mayoral candidates, how would one know? The marginal effects of the independent variables should not be very different from the equations where the criterion was included as a control, but one would find differences in the constant of the equation. Technically, the constant is the value of the dependent variable when all the independent variables are zero. In a sense, the constant gives a measure of the dependent variable before the contributions of the independent variables are included. In this light, the differences between the subsamples could reside in the constant. For example, those living in the area where busing is occurring may be so much more prejudicial than those in other areas that the marginal effects of the other independent variables is negligible; in such a situation, comparing constants would expose or disconfirm. this possibility. Kinder and Sears (1981) do not report the constant, so the comparison across sub-samples is incomplete.7 Of course, the constants may not have been


7 In general, the omission of the constant is not an egregious transgression. Generally, when one or more of the independent variables lacks a meaningful zero point (i.e. zero falls outside the accepted range for the independent variable or the models in general) the constant may be omitted. For example, if year is a predictor of automobile cost, the value of the constant (the cost of an automobile in the year zero) is of little interest. The trend in many journals is to report the constant even when it is meaningless, for no apparent reason. Here Kinder and Sears






58


different. Kinder and Sears could have strengthened their argument by calling attention to this fact.


Data


With only a one exception (McConahay 1982; Louisville), all the data used in the examinations of symbolic racism are drawn from non-Southern samples. There are some instances of national level data set being used to measure general symbolic politics hypotheses, but none address symbolic racism specifically.

This leaves a number of city-wide studies conducted in

the north and west and some studies conducted in colleges and seminaries. The McConahay article of 1982 does feature data from a Louisville sample collected by the Harris organization; some consider Louisville a Southern city. Still, the fact that regional differences in racial tolerance are often found (see examples in the previous chapter) gives one cause to expect either that regional differences would exist in the new racism or that the prevalence of the old racism would follow some regional pattern. The significance of the second point depends on the theoretical expectations concerning the relationship between symbolic and old fashioned racism: if symbolic racism is a submerged redneck racism then one would expect either a greater prevalence of symbolic racism where redneck racism had existed previously



provide an interesting counter example to this trend by not reporting the constant when it may have been meaningful.






59


or a decreased level of symbolic racism where redneck racism prevails because people there would not need to veil their attitudes. This would be especially useful if tests were provided comparing the two forms of racism.

Of course, given the framework of symbolic politics, there is no expected relationship between context and symbolic predispositions. In this sense, the theoretical problems caused by the loci of the various studies only apply to the alternative hypotheses. Still, the use of national level data or comparisons of absolute level of symbolic racism across regions might provide a useful test of the degree to which the new concept is actually symbolic. If it were truly abstract then there should be no differences across regions. Previous studies have not tested this. A fuller elaboration of this hypotheses is outlined in the next chapter.

The framework of symbolic racism rests on three pillars: distinctiveness from known forms of prejudice; the combination of antiblack affect and traditional values; and the divorce of symbolic attitudes from context and self interest. The concept has been tested against a narrowly constructed self-interest alternative hypothesis. The data used for those tests have been drawn from municipal and state-wide samples, primarily collected outside the South.

In this chapter and the one before it, discussion has focused on two major theoretical traditions in the study of white racial attitudes in political science. in the next






60


chapter, an alternative approach will be outlined. This approach is more synthetic than innovative. Indeed, innovation is difficult to come by in a literature as "sprawling" as that on racism (Kinder and Rhodebeck 1982). This new approach attempts to build on the strengths of the' two outlined below. For example, the claims made by proponents of symbolic politics concerning the enduring effects of pre-adult socialization to the exclusion of the material or social context must be tempered. Similarly, the assumption that realistic group conflict, is necessarily realistic will be relaxed; even if such is the case, an individual must perceive the conflict as real to act upon it. Of course, not all questions raised by this discussion will be answered. Indeed, some may be unanswerable in the short run. Still, it is hoped that the syntheses and ensuing empirical investigation will shed some light on the salient issues and point in the direction of their resolution.














CHAPTER 4
SYNTHESIS

In the previous two chapters discussion focused on two mainstream but distinct schools of thought concerning racial prejudice in America. The first school maintains that racial threats or the perception thereof provide the central mechanism for the generation and use of prejudice. The second school introduces a new, form of abstract prejudice thought to have become widespread in the post civil rights era. In this chapter, a third position is laid out in order to synthesize a set of reasonable theoretical propositions and lay the ground work for testing those propositions.

The (survey research portion of the) literature

discussed in the previous two chapters suffers from a problem common in the social sciences. Namely, the phenomenon being modeled is far more complex than the model.1 In this case, as the proponents of each of these schools have staked out their territories, they have tested their hypotheses against relatively transparent null hypotheses. Additionally, the conversations in each school have been independent of each other. The result of these shortcomings stultifies theory building since research falls into a cycle of replication on




1 Of course, the very nature of modelling requires some simplification, else the exercise would be pointless.

61






62


the one hand or a series of minor internecine squabbles on the other.

Given the broad view of the two schools, one can derive areas of synthesis. Indeed, one expects common ground between the approaches because both schools address prejudice and its causes. one might expect propositions held in common by both schools to be especially robust, theoretically and empirically. Conversely, the greatest opportunity for theory building lies in areas neglected by either or both schools. A number of hypotheses discussed later in the chapter provide a framework for plumbing these interstitial depths.

Despite the existence of any number of common or

uncommon elements between the two approaches, two themes merit serious treatment. The first and most fundamental theme centers on the notion of multi-dimensionality of racial attitudes. Is there merely one kind of attitude, parceled out in varying levels across the nation or are there multiple dimensions of varying intensity and salience? The second (and secondary) issue revolves around defining interests. Do people define their interest narrowly, in terms of the maximization of personal benefits, or are interests more broad, weighed in terms of the maximization of benefits for

family, group, or country? These two issues will be taken up below followed by a brief discussion of the importance contextual factors in understanding attitudes and behavior with a fairly detailed discussion of one such factor, racial composition.






63


Multi-dimensionality


The notion of multi-dimensional racial attitudes dates from the 1950s (e.g. Rokeach 1960; Allport 1954). Many studies identified the various dimensions through the use of intensive personality trait questionnaires. This method resulted in detailed measurements of the respondent's racial attitudes. The studies, however, failed to demonstrate the relevance of the scales to any external conditions or to demonstrate the impact of those attitudes on political behavior or attitudes.

Other studies have excelled in measuring people's

political orientations and evaluations of candidates. For example the American National Election Studies (1952-1992) conducted by the Institute for Social Research at the University of Michigan have consistently measured many fundamental political orientations, behaviors, characteristics and attitudes. Of course, every question a secondary or even a primary researcher might want to ask cannot be asked. Few data sets can tap multiple dimensions of racial attitudes and provide measures of political phenomena against which one may judge the effects of the racial orientations. Hence, most studies cannot take this sort of approach. The proponents of symbolic racism acknowledge the existence of multi-dimensionality (see discussion in Chapter Three), while others are less explicit in their treatment.






64


But the acknowledgment of multi-dimensionality by the proponents of symbolic racism is underdeveloped empirically and theoretically, except to say that the new or symbolic racism replaces the old. one wonders why the old has not yet disappeared or what the different effects of the two sets of prejudicial attitudes might be?

The group conflict orientation held by some of the authors discussed in Chapter Two holds no place for conceptions of multi-dimensionality. From their perspective, material imperatives would outweigh any preferences, psychological predispositions or attitudes. Their view of racial attitudes lying mainly in the material realm makes them ill equipped to explain even some of their own findings. The works by Huckfeldt and Kohfeld (1989) and Key (1949) are less clear about the exact nature of the racial attitudes, but unlike their colleagues they point to racial predispositions resulting in behavior which is contrary to individuals, material interest. Given this lack of agreement it seems reasonable to temper or relax the assumption that actual conflicting interests represents a necessary and sufficient condition for the emergence or existence of racial hostilities.

The two central schools of thought either dismiss the possibility of multi-dimensionality or fail to explain it. Some studies by the proponents of symbolic racism do find






65


multiple dimensions in racial attitudes but fail to explore the significance of multi-dimensionality.2



Attitude A on horizontal axis Attitude B low high

high II

low IV

Figure 4-1. matrix of racial attitudes



There is little reason to expect people's prejudicial attitudes to follow a zero sum pattern, where one new attitude replaces the other. Rather, the attitudes may represent distinct and independent dimensions. A graphical representation of a two dimensional model is presented in figure 4-1. This two by two matrix is not meant to pigeon hole the orientations into four categories. Instead, the two continua fall on the axes. So an individual could fall anywhere in the range of axis A, without restricting their location on axis B.

The literature discussed in the previous two chapters leads one to some propositions concerning the nature of multi-dimensional racial attitudes. This discussion focuses on the theoretical expectations in two dimensions for clarity's sake but the arguments could be extended to more

2 For example, Sears, Hensler and Speer (1979, p. 382n) use factor analysis to identify three racial dimensions: 1) old fashioned racism; 2) opposition to government action; and 3) opposition to racial protest. Still, they combine all the items into a single racial attitude scale.






66


(N) dimensions. Further, the discussion addresses the two dimensions for which measures are available in the data exploited in this study, namely perceived racial threat and symbolic racism. The specific measures and operationalizations are discussed in the next chapter. Specific hypotheses are numbered in the text and summarized in Table 4-1.

First, one would expect two racial dimensions to show some degree of independence, but not be completely uncorrelated (HSYNO1). The similarity in theme should mean that the dimensions share some combination of inputs (what explains the attitude), traits (the geographic distribution of the attitudes or other descriptors), or effects (what the racial attitudes reveal about other political orientations and behaviors). Divergence in some of these is to be expected as evidence of the independence of the dimensions.

For the two dimensions examined here a number of

predictions follow from previous theory. The two dimensions should differ in their inputs. For example, the local racial composition should correlate with the perception of racial threats (HRTo1); this sensitivity has been shown to be more pronounced in the context of the South with its segregationist heritage (e.g. Fossett and Kiecolt l989)(HRTO2). Similarly, an individual's assessment of his or her financial situation of the state of the economy in general should covary with sensitivity to perceived threats (HRTO3). In other words, if one has a dim outlook on the






67


Table 4-1 Derived and Synthetic Hypotheses

Derived Racial Threat Hypotheses
HRTO1: Levels of perceived racial threat will covary with
the percentage black.

HRT02: Levels of perceived racial threat will be higher
among Southerners.

HRTO3:Levels of perceived racial threat will covary with
economic vulnerability.

HRT04:Aggregate (or mean) levels of perceived racial threat
will be highest in the Deep South, less high in the RimSouth and lowest in the non-South.

HRT05: Evaluations of candidates will covary with levels of
perceived racial threat when candidates are identified with
the out group as members or as supporters.

HRT06: Support for safety net policies (exclusive of
racially specific policy positions) will covary with levels
of perceived racial threat.

HRT07: Support for policies which benefits African Americans
in particular will vary with levels of perceived racial
threat.

HRT08: The level of perceived racial threat will covary with
the closeness in income (class proximity) of the racial
groups locally.



Derived Symbolic Racism Hypotheses
HSR01i: Levels of symbolic racism will be uncorrelated with
local racial composition.

HSR02: Levels of symbolic racism will be uncorrelated with
individuals' regional origins or residence.

HSR03: Levels of symbolic racism will be uncorrelated with
economic vulnerability.

HSR04: Aggregate levels of symbolic racism will be
uncorrelated with region.

HSR05: Evaluation of liberal or black candidates will covary
with level of symbolic racism.






68




Table 4-1 (continued)

Derived Symbolic Racism Hypotheses (continued) HSR06: Support for safety net policies will vary with
symbolic racism.

HSR07: Support for policies which benefits African Americans
in particular will vary with levels of symbolic racism.



Synthetic Hypotheses
HSYN01: Perceptions of racial threat and symbolic racism
represent different dimensions of racial attitudes.

HSYN02: The level of perceived racial threat will vary with
the level of income equality between the races.

HSYN03: Symbolic racism will ovary with the percentage
black at low levels, after which the marginal effect of the
racial composition of symbolic racism will diminish. HSYN04: Current level of perceived racial threat and
symbolic racism will ovary with archaic levels of the
percentage black.

HSYN05: Levels of perceived threat will vary with personal
contacts with out group members, as in the work place.



economy (or that one is losing relative to others) then the sense of threat from an external source becomes more salient than it would be under other conditions.

The assumption of abstraction on the part of the proponents of the symbolic racism hypothesis implies independence between symbolic racism and contextual factors like the local racial composition (HSR01) or the region of residence or upbringing (HSR02). Likewise, symbolic racism should not be influenced by the individual's economic outlook because, according to its proponents, symbolic racism is







69


independent of personal life (HSRO3). Indeed, the proponents of symbolic racism are clear that symbolic racism should be completely independent of all environmental factors or personal interests. (Kinder and Sears 1981, p. 416; Kinder and Sears 1985, p. 1143; McConahay and Hough 1976, p. 37; MoConahay 1982, p. 705n; Sears, Hensler and Speer 1979, p. 371)

The two dimensions should differ in their traits. According to the findings discussed in Chapter Two, perceptions of threat should be more pronounced in the South given its tradition of sensitivity to such threats. On the other hand, symbolic racism should not vary regionally, according to its proponents, because of its independence from the individuals' material context.3 According to symbolic racism theory, one would not expect variation among any of the regions including the most likely suspect, the South. So aggregate levels of perceived racial threat should be highest in the South (HRTO4), while aggregate levels of symbolic racism will not vary by region (HSRO4).

In terms of the effects of the two racial dimensions, one might expect a different picture. The perception of


3 Historical conditions in the South and previous research lead one to expect living in the South to have some effect (input) into the make-up of an individual's racial attitudes; this notion is supported by the threat literature, although some scholars see Southern exceptionalism or the uniqueness of Southern culture and politics as fast fading (e.g. Stanley). As a trait, however, geographic distribution must be considered independently form arguments based on the impact of southernness on the racial attitude. This difference is especially salient given the abstractness assumption of symbolic racism.






70


threat should not have a uniform effect on attitudes, issue positions or the evaluation of candidates across the political spectrum. Rather, it would affect only on those issues and evaluations which were related to one of its two fundamental components: race and economic vulnerability. For example, the perception of threat would not be efficacious in predicting the relative evaluations of say, two white candidates in a contest in which the race card was not played by one. It may figure more prominently in situations in which one of the candidates race baits his or her opponent by saying the opponent will give white jobs to blacks. Even then, the effectiveness of the tactic depends on the particular skills of the candidates and other circumstances surrounding that election. One would not expect a measure of threat to predict evaluations without that disposition being made salient. Similarly, the perception of threat might come into play in the evaluation of a black candidate (touching on the other fundamental element) even if economic threats were not particularly salient(HRT05)- Symbolic racism would ovary with support for candidates who appeal to traditional values (HSR05)

Additionally, attitudes related to threat should guide issue orientation when salient to the fundamental elements. So, if one feels economically threatened one might be more supportive of programs to help the unemployed, in general, because this safety net would benefit the threatened individual or his or her reference group (HRT06). On the






71


other hand, one would not expect the threatened individual to support programs which specifically assist groups with whom one perceives an adversarial relationship. So, threatened whites, while supporting some spending programs like aid to the unemployed, would oppose programs specifically assisting blacks, their perceived competition (HRT07).

According to its proponents, symbolic racism should

predict evaluations of candidates and issues. Sears (1986) points out that symbolic racism also explains positions in non-racial issue areas. This, presumably, results from the concept's wedding of conservatism and anti-black affect. To the extent that symbolic racism is a form or expression of prejudice one would expect it to have a more pronounced effect in situations which involved racial objects. In other words, a measure of symbolic racism should explain or predict especially well when the object (or dependent variable) is a black candidate or the issue has racial content(HSR05-07).4


Defining the Nature of Interests


As discussed in the previous chapter, the proponents of symbolic racism test their approach against a single

4 Of course, these expectations are based on the theoretical positions stated by the proponents. The discussion in the previous chapter revealed the symbolic racism theory had not been tested in the national or Southern samples. Had it been, the theory may have been developed somewhat differently. The notion that symbolic predispositions influence current attitudes and behaviors is reasonable enough, but to argue that these symbolic predispositions are completely independent from the context in which individuals find themselves oversteps the bounds of reasonable expectations.






72


alternative hypothesis, namely narrow self interest defined in largely objective terms. Yet recent scholarship argues for a broader conception of interest and identity. For example, a burgeoning social movements literature sees the development of group consciousness (identity with a group beyond the self) as an essential element in explaining movement genesis (McAdam 1982; Zald and McCarthy 1977). veritable cottage industries have sprung up to explain different kinds of consciousness from gender (Rhodebeck 1992) to race (Jackson 1981; Shingles 1981; Miller, Gurin, Gurin and Malanchuk 1981), even to identification with businessmen (Miller, Gurin, Gurin and Malanchuk 1981); these studies found significant links between group identity and political behavior and various forms of political participation. Similarly, much of the extensive group conflict literature (while maintaining that objective material interests between groups are at odds) argues variously that group membership or identity are essential elements of that conflict (e.g. Baker 1975; Blauner 1972; Isaacs 1975; Schemerhorn 1956; Van den Berghe 1981; Wilson 1973). Giles and Evans (1985) found that levels of racial hostility were positively related to levels of white and Anglo-Saxon identity.

Given the weight of theoretical and empirical evidence documenting the importance of group identification and consciousness, one cannot conclude that self-interest represents the sole conduit for the threat mechanism. Indeed, one expects that individuals would be sensitive not






73


only to personal threats (challenges to the individual's job, economic status, security, or political influence) but to similar threats to their families and even their race in general. Further, the perception of a more general threat should predict other political attitudes and behaviors. measures of symbolic racism have not been tested against more general orientations or representations of threat.


Synthesis


The empirical and theoretical controversies discussed

above impels an effort at reconciliation. For example, some of the research discussed in the second chapter argues that white hostility toward blacks results from real economic threats posed by blacks. Others argue that instrumentally marshaled racial hostility prevents a poor white-black coalition from upsetting the status quo.

while neither of these propositions are tested in this study, the mechanism underlying them is. The belief that blacks pose economic, status or power threats, rather than the essential material conditions, may lead individuals to have hostile attitudes or to act on existing prejudicial attitudes. material conditions could make believers out of many, but those conditions and consciousness of them need to be disaggregated.

For example, imagine four individuals working for a firm likely to change the composition of its work force in response to a boycott by a minority group: the first worker's






74


job is threatened because of the changes and he or she knows it; the second one's job is threatened, but he or she thinks (wrongly) it is secure; the third one's job is secure, but he or she thinks (wrongly) that it is threatened; the fourth worker's job is secure and he or she knows it. Despite the difference in the reality of threat posed to the workers, it is evident that there is no reason to expect workers one and three to respond differently to the perceived threat to their jobs. Other things being equal, one would expect similar response among those who have the same perceptions concerning the threats (real and imagined) to their job status.

What expectations are reasonable concerning the two workers who perceive their status as secure? Narrow self interest predicts those workers would not harbor hostility toward the minority group. After all, their interests remain unaffected by the shake-up in the company (at least by the standard of self-interest employed by the proponents of

symbolic racism); they still have their jobs. If, on the other hand, interests are more closely related to group identification, then despite the perceived (rightly or wrongly) security of their positions, they may harbor hostile attitudes. They may harbor those attitudes because the shake-up threatens, not them personally, but people like them, people with whom they identify. or they may feel threatened because the environment has simply become less secure; they could lose their job next.






75


In this chapter, discussion has drawn attention to ways in which political and racial attitudes may be sensitive to and distinct from actual material conditions (as with the worker who incorrectly perceived her job was threatened). People's perceptions of their condition may diverge from the actual circumstances, but they are not divorced from those circumstances. In the above example, the climate of change led the actually unthreatened worker to perceive that his or her job was less secure than it was. The context, in which co-workers with whom he or she worked were likely to lose their jobs, heightened that worker's sensitivity to the perception of threat. This sense of threat occurs even though there may be no actual threat to the worker's job. Behavior is predicated on perceptions that may be inaccurate. And those perceptions, with their attendant measurement errors,, develop within a context. In the next section, discussion focuses on conditions which inform the context of racial attitudes, paying particular attention to the changing interpretations of local racial composition.


Context and the Changing Meaning of Racial Composition


Some of the studies discussed in the second chapter point to different theoretical significances of the local racial composition. To some (e.g. Giles and Evans 1985) the local composition represents a contemporary and actual threat to the lives and livelihood of whites. To others (e.g. Key 1949), racial composition represents a proxy for a labor






76


system with a particular set of class relationships Are these interpretations reasonable in the contemporary context? Is the racial composition a threat in and of itself or is it a measure of something else? Operationally it provides a shorthand measure to describe the relative number of blacks and whites living in a given geographic area. From this, one can not necessarily conclude that composition in and of itself represents some threat. But the local racial composition may be correlated with other salient but unmeasured details of the locality.

The contemporary racial composition is correlated with (see later chapters) racial attitudes in the expected ways but not necessarily for the same reasons that group conflict proponents argued. The first element of the contemporary significance of the local racial compositions may found in its representation as a resource for the mobilization of black interest groups (McAdam 1982, Blalock 1965). McAdam, in particular, details the effects of the urbanization of black America as a resource for the mobilization of interest groups and social movements. In a nutshell, the concentration of blacks in the urban setting allowed for the recognition of common interests, higher incomes than those available on the farm, and the formation of civic groups and


5 Key (1949) wrote before the great migration was complete; racial composition was correlated with the ruralness of Southern counties. Giles' (Giles 1977; Giles and Evans 1985, 1986, Giles and Buckner 1993) work addresses a post-migration America where the percentage black may be more correlated with level of urbanity.






77


churches, which in turn were able to recruit and maintain sophisticated leaders (e.g. educated ministers). The same elements that allowed increased concentrations of blacks in the cities to generate successful social movements may aggravate white hostility to the extent that movement objectives and tactics threaten the economic and political positions of whites.6 In this light, the higher percentage black would elevate the sense of threat among whites. Indeed, because the processes described by McAdam are largely cumulative one would expect a nonlinear relationship between the percentage black and the level of threat. Graphing this relationship would require a quadratic function with an increasing slope (see figure 4-2).

Similarly, the relative ability of blacks to mobilize

other resources may influence the degree to which whites feel threatened. McAdam (1982) and others in the social movements paradigm (particularly the resource mobilization wing) pay particular attention to the ability of nascent movements to generate money as a measure of that movement's strength. In this sense, one might expect the relative wealth of the black community to correlate directly with the level of threat perceived by whites. Blalock (1967) argued that the sense of competition was most heightened among those closest to each other in class position (HRT08). The relative or comparative level of black income vis-a-vis white income should be an

6 For example, Bobo and Gilliam (1990) found that black empowerment in major city governments was linked with white frustration with those governments.





78















y 100













0
0 100





Figure 4-2. Quadratic function with increasing slope






79




effective predictor of threat. Of course, one might also reasonably expect a measure of income equality to correlate inversely with racial hostility (Jacobs 1982). Comparability of income may be an outgrowth of a harmonious racial atmosphere rather than a cause of a toxic one (HSYN02). On the other hand, such arrangements could just as easily promote mutual respect or at least the displacement or loosening of stereotypes (HSYN05)A second way in which the contemporary racial

composition may influence white attitudes lies in increasing the probability of actual conflicts. In a racially monotonic community, one would expect zero cases of racial conflict of any kind. The probability of inter-racial disputes (whether caused by racial hostility or not) could be expected to increase as the percentage of each racial group approached 50% (see figure 4-3).7 This would produce a curve shaped like an inverted U centered on 50%. If symbolic racism is truly abstract, no correlation would be found because of increasing opportunities for inter-racial conflict (HSR01).




7 Conflicts occur between citizens all the time no matter what their race. The increase in transactions between people of different races does not necessarily imply that conflicts resulting from those transactions result in racial hostility. Indeed, other things being equal, the increase in the number of inter-racial transactions may result in a decline in racial hostility because of the increase in the number of positive outcomes from those transactions. However, given the history of racial hostility in America, one would expect (as has been observed) short term increases in hostility.





80















100













0
0 100





Figure 4-3 Parabolic quadratic function






81


A third way in which the current racial composition may accentuate hostility among whites operates through visibility. Potentially threatened whites may be more likely to latch on to racial explanations for their difficulties if they have reminders of the existence, size, and power of the other race. So the more visible the minority population the more likely they would be blamed for problems faced by whites. Again, one might expect a nonlinear relationship between percent black and the level of threat for these reasons. The level of threat would increase with increasing slope as the percentage black increased (like figure 4-2).

The proponents of symbolic racism might argue that

increased visibility of the minority population should have no effect on the level of symbolic racism. However, the visibility argument sketched out above leads one to expect the level of symbolic racism to increase with the percentage black to some threshold level and level off thereafter (HSYN03). If one relaxes the assumption of abstractness, one might expect the symbolic proclivities to increase slightly as the percentage black rose to some level of visibility, making blacks an identifiable segment of the population. After blacks are identified incremental changes should have little effect if the symbolic attitudes are largely abstract (see figure 4-4).





82 100













0
0 100





Figure 4-4. Quadratic function with gradual decay






83




The Archaic significance of Racial Composition


The contemporary significance attached to racial

composition by scholars of group conflict differ dramatically from Key's (1949) reasoning. From Key (also Wright 1977, Mathews and Prothro 1963) one infers that past racial composition affects the current racial climate in localities and states. Introducing this line of reasoning requires the review of some arguments made in more telegraphic form earlier. While focus of this project concerns white racial attitudes in America generally, examining the Southern case proves instructive.

Key (1949) bases his argument on observations of a

region in which a particular economic system reigned with a peculiar set of labor and social institutions; however, King Cotton died. Those institutions no longer dominate the Southern economy. Therefore today's racial composition cannot have the same immediate significance today that the racial composition had when Key (1949) published.Southern Politics. But the archaic racial composition may reveal information about local norms, beliefs and values, or a political culture crystallized in an earlier era. The transformation of the Southern economy and the dramatic demographic shifts make Key's (1949) argument less directly applicable to the current political situation.






84


AS McAdam (1982) points out, the economic system in the South underwent a series of revolutions in the early part of this century. New machinery reduced the need for cheap black labor. New pesticides had a similar effect. Most importantly the collapse of the international cotton markets made the system of cotton dominated monoculture less (or un) profitable by the beginning of World War II. Even as Key (1949) wrote, the basis of his argument was changing (or had changed). Despite the revolution in the Southern economy, changes in the political system and the culture lagged behind by forty years or more. what accounts for this lag? As Tullos (1989) and Scott (1985) document in widely divergent cases, changes in economic systems (in particular agricultural revolutions) may change the class system but not necessarily the relative positions of actors (or families) within the new system. Powerful families may translate or transform their advantages (education, capital and other assets, prestige, and political power) from the previous system into advantages in the new system. A revolution in the economy does not produce an immediate social revolution. This conservation of advantage forestalls the changes one might expect after such a radical transformation as the Southern economy experienced.

The structure of the political system and the entrenched positions of its inhabitants would have delayed immediate changes in the political system. The lack of a grass roots organization in the Republican party made (make) challenges






85


to the hegemonic Democrats costly and difficult. Further, skilled Democratic politicians slowed the expected growth of the Republican party after the one-party system was no longer useful (Lamis 1988). For example, George Wallace continued to win elections into the 1980s, despite his well-earned reputation as a racial extremist. Lamis (1988) credits Wallace's formidable political skills with delaying the election of a Republican governor by more than a decade.

Wallace's success points to an additional element in this explanation of the slow pace of change, beyond the structural factors noted above. This additional element lies in the slow pace of changing values across generations in the South. The values established during the era of King Cotton continued to play a significant role in Southern culture and politics long after that economic system collapsed. This continuation of archaic values may have allowed Wallace to harness fears of racial threats as a source of political capital.

In this sense (or to the extent that this is true) one

finds agreement with the proponents of symbolic racism. That is to say, the process of socialization results in the inculcation of standards and values which may form the basis for later political evaluations. But these values and standards do not operate in a vacuum. So, it is reasonable to expect regional variation in levels of racial hostility, even when proximate explanations are controlled. In other words, the archaic local racial composition should indicate






86


areas with traditions of heightened racial tension, especially in the South. Of course, the contemporary and the archaic racial compositions are not uncorrelated. In this sense, the relationship between the contemporary local racial composition and racial hostilities does not merely reflect the current material context, but also reflects cultural factors which may be unique to the region, state or community.

The patterns of racial antagonism Key (1949) described (using racial composition) continue to affect the current state of race relations in the South and elsewhere. Of course demographic changes and migration and the transformation of the Southern economy have the effect of altering the roots of antagonism in the South. The information revolution and increased access to higher education have muted the transmission of prejudicial values form generation to generation. Still, one expects some degree of continuity within states, counties and local communities. Those raised in communities with traditions of prejudice (or a prejudicial political culture) could be expected to have more prejudicial attitudes than those raised in communities without such traditions because those traditions influence their socialization even if the transmission of values is attenuated because of the factors noted above.

Similarly, the local or state political culture

represents a crucial contextual factor for those who may






87


migrate from places with different traditions as adults. For example, Fossett and Kiecolt's findings reveal that people raised in the South but living in the North were not significantly more likely to harbor prejudicial attitudes than other Northerners (1989, discussed in detail in Chapter Two). Evidently the context in which people find themselves continues to matter after childhood socialization is complete. And, the context dynamically shapes their current values, beliefs and behavior on issues of race. Indeed, "[r]acial socialization does not end with adolescence." (Kinder and Rhodebeck 1982, p. 205). So adults may be influenced by the prevailing political culture and the material conditions in their state and community.

In this sense, the current racial political culture is shaped by two essential elements: 1) traditions, values and beliefs established in the past; and 2) the current material context. Contemporary measures of context may inform research about both of these elements. For example, the contemporary racial composition (and other measurable data) of a community reflects to some degree the visibility of the minority community and other factors discussed in the previous section and may reveal (or contain) information concerning the political culture or the weight of the past. As discussed above, the contemporary racial composition cannot mean the same thing that it did when Key wrote, but the racial composition from that era can (Wright 1977, Mathews and Prothro 1963a,1963b). operationally, data drawn






88


from older censuses can inform research about current racial attitudes by providing a cultural backdrop, pointing to crucial cultural information from the past which may influence contemporary racial attitudes (HSYN04). The influence of the current racial composition on racial attitudes reflects the extent to which those attitudes reflect the sort of social dynamic outlined by social conflict or racial threat theorists (HRT01).

The argument presented above dulls the otherwise sharp distinction between analysis of racial attitudes based on a symbolic politics model and models more firmly grounded in material conditions. It is inaccurate to argue that the basis of racial hostility must be one or the other. material conditions influence the values held by previous, current and future generations because of the methods of transmission from parent, teacher, and supervisor to child, student and employee. These values provide a lens or optic through which current material conditions are viewed. Similarly, current material conditions influence the salience or relevance of that lens in a given situation. Perceptions, beliefs and values cannot be completely divorced from the material world, as the proponents of symbolic racism have claimed.

In the next chapter, discussion focuses on how one might measure and test the propositions laid out in this chapter.














CHAPTER 5
DATA

In the last chapter, discussion centered on the reconciliation of theoretical dilemmas posed by two mainstream schools of thought in the study of whites, racial attitudes. Additionally, hypotheses were derived from theoretical propositions. In this chapter discussion turns to issues concerning the testing of those hypotheses and the measurement of the constituent elements therein. Four topics need coverage in this chapter: 1) the sources of the data used; 2) how the two racial attitudes outlined in previous chapters are measured; 3) how variables that predict those attitudes are measured; and 4) how variables that are predicted by the racial attitudes are measured. Details regarding the measurement of more commonplace variables are provided in the appendix.


Data


The data exploited in this study are drawn from three

sources. First, survey data are taken from the 1986 American National Election Study (ANES). Additionally, two types of census data are used. Contemporary census data are drawn from the 1980 Census of Population and Housing; the specific data used are from the electronic version of STF 3C. That



89






90


summary file will be used to calculate the racial composition for the SMSAs and counties represented in the national sample and to collect other relevant contextual data.1 These data are collected at the SMSA level for respondents residing in SMSAs and at the county level for others. SMSAs, for those who live in them, are thought to be the defining local unit rather than counties or congressional districts. SMSAs tend to have area-wide media markets reinforcing identification with the urban center rather than with constituent counties. County level data are used for respondents not residing in SMSAs. County level data represent the highest level of precision available for merging since census tracts and minor civil divisions have been padded (overwritten with nines) in the 1986 ANES.

Archaic census data are drawn from the ICPSR

reproductions of the U.S. Census of Population and Housing 1% counts from 1900 and 1940 (ICPSR study number 0003). The 100% counts are available in neither electronic nor card format for either the 1900 and the 1940 censuses. The use of the sample counts should make little difference in the project at hand since only minor differences between the sample and 100% counts are likely. Additionally, any variations between the two counts should not produce systematic biases.2

1 The 1990 National Summary File is not yet available, so interpolation for maximum accuracy is not yet feasible.
2 Theoretically both the 1% and the 100% counts are samples, because the survey used to produce the 100% count cannot actually account for everyone. Unfortunately, the 100%




Full Text
62
the one hand or a series of minor internecine squabbles on
the other.
Given the broad view of the two schools, one can derive
areas of synthesis. Indeed, one expects common ground
between the approaches because both schools address prejudice
and its causes. One might expect propositions held in common
by both schools to be especially robust, theoretically and
empirically. Conversely, the greatest opportunity for theory
building lies in areas neglected by either or both schools.
A number of hypotheses discussed later in the chapter provide
a framework for plumbing these interstitial depths.
Despite the existence of any number of common or
uncommon elements between the two approaches, two themes
merit serious treatment. The first and most fundamental
theme centers on the notion of multi-dimensionality of racial
attitudes. Is there merely one kind of attitude, parceled
out in varying levels across the nation or are there multiple
dimensions of varying intensity and salience? The second
(and secondary) issue revolves around defining interests. Do
people define their interest narrowly, in terms of the
maximization of personal benefits, or are interests more
broad, weighed in terms of the maximization of benefits for
family, group, or country? These two issues will be taken up
below followed by a brief discussion of the importance
contextual factors in understanding attitudes and behavior
with a fairly detailed discussion of one such factor, racial
composition.


15
For example, even before Key started the project, dependence
on cheap black labor began decreasing as a result of
mechanization and lessened dependence on mono-culture (McAdam
1982). The civil rights movement has changed the face of
racial politics. Electoral barriers have largely been
erased. Not only have blacks earned the vote, but they have
used it to make government and its policies more
representative. Reapportionment has changed the balance of
power in state governments away from rural bourbons. Key
(1949) himself pointed out the rapid migration of blacks out
of the South.
Despite the changes, Key provides useful insights into
the mechanisms of politics generally. At the very least,
Key's work provides a useful starting place for political
analysis. Often his theses are upheld. For example, Stanley
(1987) endorses Key's estimate that the relaxation of voting
restriction would have a negligible effect on white turnout.
This hypothesis is upheld by many subsequent studies (e.g.
Allport 1954; Blalock 1957; Pettigrew 1957; Wilcox and Roof
1978)
The Contemporary Significance of Race at the Macro Level
Key's analysis of the effect of race on the class
composition of the Democratic party has not fared as well in
the changed environment. The threat (real or imaginary)
posed by black ascendance was used as a mechanism for
maintaining the one party system. This system forced lower


117
respondents' cultural, historical, and contemporary context.
Second, the nature and structure of racial sentiments is more
complex than the proponents of symbolic racism are prepared
to admit. Not only is symbolic racism subject to regional
variation, but it is subject to variations at sub-regional
levels as well. In this sense the abstractness principle of
symbolic racism must be questioned.3
The models predicting the levels of perceived racial
threat are constructed in parallel fashion to the symbolic
racism models just discussed. The results are presented in
Table 6-2.
As was the case with the symbolic racism model, the
first equation shows little of significance. Perceived
racial threat has an inverse relationship with the level of
inter-racial income equality. This runs counter to the
hypothesis derived from Blalock (Hrt08) closeness of class
position promotes racial antagonism. Additionally, the level
of Southern nativity has a very modest positive relationship
with the level of perceived threat. This proves ephemeral;
when other variables are added to the model the apparent
effect of nativity disappears.
3 Note also the consistent effect of personal income of the
symbolic attitudes. While this effects is only significant
at some level between p<.05 and p<.10 it provide evidence
that the symbolic predispositions are not only sensitive to
contextual effects generally but are sensitive to other
aspects of the respondent's personal life. This runs counter
to the theoretical territory staked out by the concept's
proponents.


85
to the hegemonic Democrats costly and difficult. Further,
skilled Democratic politicians slowed the expected growth of
the Republican party after the one-party system was no longer
useful (Lamis 1988). For example, George Wallace continued
to win elections into the 1980s, despite his well-earned
reputation as a racial extremist. Lamis (1988) credits
Wallace's formidable political skills with delaying the
election of a Republican governor by more than a decade.
Wallace's success points to an additional element in
this explanation of the slow pace of change, beyond the
structural factors noted above. This additional element lies
in the slow pace of changing values across generations in the
South. The values established during the era of King Cotton
continued to play a significant role in Southern culture and
politics long after that economic system collapsed. This
continuation of archaic values may have allowed Wallace to
harness fears of racial threats as a source of political
capital.
In this sense (or to the extent that this is true) one
finds agreement with the proponents of symbolic racism. That
is to say, the process of socialization results in the
inculcation of standards and values which may form the basis
for later political evaluations. But these values and
standards do not operate in a vacuum. So, it is reasonable
to expect regional variation in levels of racial hostility,
even when proximate explanations are controlled. In other
words, the archaic local racial composition should indicate


159
threat. In multivariate models, 'Southernness' has the
predicted positive relationship even when other variables
like income and education were controlled; Southerners are
more likely to feel threatened even when other explanations
are controlled (Hrt02 Hrt04)* Additionally, the level of
perceived threat is sensitive to the level of economic
vulnerability as measured by the respondents assessment of
personal finances, even when other variables are controlled
(Hrt03)
Similar support for the hypotheses regarding the
predictive abilities of the respondent's level of perceived
racial threat is found (see Tables 6-3 and 6-4). The level
of perceived racial threat proves to be a significant
predictor of the respondents' evaluations of a black
candidate (Jesse Jackson) and of the respondents' evaluations
of blacks as a group (Hrt05)* Those who felt more threatened
are not less likely to support government assistance programs
in general; indeed, there is a negative relationship between
perceptions of threat and the desire to reduce funding for
programs to help the disadvantaged (Hrt06)- Despite the lack
of resistance to such policies in general, the reverse
relationship is demonstrated with regard to programs
specifically designed to assist blacks. Even when other
variables are controlled, a positive relationship between the
level of perceived racial threat and the likelihood of
respondents favoring funding cuts for programs which help
black is demonstrable (Hrt07)


71
other hand, one would not expect the threatened individual to
support programs which specifically assist groups with whom
one perceives an adversarial relationship. So, threatened
whites, while supporting some spending programs like aid to
the unemployed, would oppose programs specifically assisting
blacks, their perceived competition (Hrt07)*
According to its proponents, symbolic racism should
predict evaluations of candidates and issues. Sears (1986)
points out that symbolic racism also explains positions in
non-racial issue areas. This, presumably, results from the
concept's wedding of conservatism and anti-black affect. To
the extent that symbolic racism is a form or expression of
prejudice one would expect it to have a more pronounced
effect in situations which involved racial objects. In other
words, a measure of symbolic racism should explain or predict
especially well when the object (or dependent variable) is a
black candidate or the issue has racial content(Hsr05-07)4
Defining the Nature of Interests
As discussed in the previous chapter, the proponents of
symbolic racism test their approach against a single
4 Of course, these expectations are based on the theoretical
positions stated by the proponents. The discussion in the
previous chapter revealed the symbolic racism theory had not
been tested in the national or Southern samples. Had it
been, the theory may have been developed somewhat
differently. The notion that symbolic predispositions
influence current attitudes and behaviors is reasonable
enough, but to argue that these symbolic predispositions are
completely independent from the context in which individuals
find themselves oversteps the bounds of reasonable
expectations.


95
implies that the two concepts, racial threat and symbolic
racism, are conceptually and empirically distinct. One could
score high on the symbolic racism scale and low on the racial
threat scale, or any combination of the two. This confirms
hypothesis HsynOI/ listed in Table 4-1.
Each group of items was combined to produce linear
additive scales for each of the orientations. The range of
these scales is sufficient to allow the use of Ordinary Least
Squares (OLS) regression.4
In the next section discussion turns to variables which
are used to explain the racial scales.
Contextual Variables
One of the central criticisms of the symbolic racism
literature levied in the preceding chapters regards its lack
of sensitivity to the context in which political behavior
occurs. With the exception of Kinder and Sears (1981), few
of these studies utilize variables which attempt to tap the
material, geographic or culture context in which the
respondent is situated. With this in mind a number of
contextual variables derived from county and SMSA level
census data will be included in this study. These data from
the censuses of various years are merged with survey data
from the 1986 ANES.
4 The threat scale ranges from 5 to 26 and has 729 valid
cases. The symbolic racism variable ranges from 4 to 20 and
has 809 valid cases. Directionality of the scales was
corrected by taking the inverse of each (dividing each summed
value into one).


40
no longer be a major political force," in American politics
(Kinder and Sears, 1981) is no longer defended by either of
its authors. The proponents of symbolic racism theory are
not of one mind. There are disagreements about many aspects
of the theory. What follows is a portrait of a symbolic
racism that may not match any one scholar's description
exactly but which attempts to capture the mean or essence of
all its proponents; this is meant to provide a basis for a
general critigue of the theory and the generation of
hypotheses. Unlike a straw man, the composite presented
below is likely stronger than any one formulation of the
theory presented to date. This portrait is followed by some
attention to the elements of disagreement and unity within
the symbolic racism camp over central themes. The discussion
in later sections should make the limits of the derivative
statement of symbolic racism apparent and point to specific
places in the literature where disagreements occur. In
subsequent sections, attention is paid to controversies
surround data measurement and methods.
The logic of symbolic racism mirrors the logic of
symbolic politics outlined above. Pre-adult socialization
and adult resocialization result in the creation of a set of
racial predispositions or attitudes which provide an
immutable standard for the evaluation of new issues,
candidates or other stimuli. Current racial attitudes are
not the function of one's situation in life. Rather, they
are abstract. These attitudes or predispositions provide the


34
either gone to men or for which they would have previously
received substandard wages. This represents a good deal for
those who had been discriminated against before, because they
were getting better treatment and wages than they had
previously. White male workers faced increased competition
for positions and wages. This depressed real wages. So,
white men lost the (non)competitive advantage they had
previously held.
Of course, some would argue that working class whites
are right to feel threatened because the inclusion of blacks
in the mainstream work force had a more immediate effect on
those on the lower end of the job market. Neither upper
management nor the owners of capital have much to fear from
the rapid expansion of the labor market after three hundred
years of systematic and costly discrimination. Demonstrating
the actual effects of the desegregation of the work place is
less important for this study than demonstrating whites'
belief that such desegregation poses a threat to them and
that perception of threat influenced behavior.
This sort of supply and demand analysis of the effects
of the relaxation of color and gender-based barriers to
employment on whites makes the Giles and Evans assumption
seem less far-fetched. Of course, the changes described
above could have been a short term shock which later resulted
in a new competitive eguilibrium. This may be an empirical
guestion, but it is one that need not be answered in the
present study.


94
not feel so encumbered by pressures to be politically
correct that they are unwilling to take strong positions.
Principle components analysis of the items used in scale
construction confirms the validity of the racial scales. Two
factors were extracted, together explaining 59.7% of the
variance. The rotated (vari-max) factor loadings are
presented in Table 5-2. The questions addressing the loss of
jobs, promotions, and school admission for whites as a group
or members of the respondents family load on the first
factor. The symbolic racism questions load on the second
factor.3 The loading of these variables on the two factors
Table 5-2. Factor analysis for racial items among whites
with white interviewers
variable
white jobs
white admission
white promotion
family admission
family promotion
factor la factor 2b
57838
.19123
77987
.22617
78021
.29666
80092
.15430
78838
.11088
civil rights push
without welfare
Irish,Italians,etc
try harder
.30130
.15733
.20871
.13206
.50797
.76681
.80920
.81751
Principle components analysis produced two factors. Varimax
rotation converged in three iterations.
a eigenvalue=4.00 explaining 44.5 % of variance
b eigenvalues.37 explaining 15.2 % of variance
Source: ANES 1986
3 Maximum likelihood estimation confirms the results
presented here. Similarly, the use of an oblique rotation,
which does not require the assumption of orthogonality
between factors, also confirms the results presented here.


139
Table 7-2. Perceived Racial Threat Scale Regressed on
Contextual and Individual Level Variables for White
Southerners Only.
entries are unstandardized coefficients from OLS regression
Variables
I
II
III
IV
%black 1940

-0.10
-0.02
-0.22
State %black 1940

-0.01

-0.01
%black 1980
0.39*
0.30
0.41*
0.40
%employed in
agriculture



-0.18
income equality



-5.03
v595 age
-0.03
-0.03
-0.03
-0.02
v602 education
-0.70**
-0.72**
-0.69**
-0.73*
v356 personal finance
0.52
0.51
0.50
0.62
v385 conservatism
0.94**
0.94*
0.93*
0.75*
State %black
1900*Average farm size



-0.0003
%black 19802
-0.01b
-0.01
-0.01b
-0.57
%black 19402

0.003

0.004
%black 1940-%black
1900



-0.09
constant
4.76b
5.00*
4.83b
10.78b
adj R2
0.10
0.08
0.08
0.10
N
130
128
128
128
Source: 1986 ANES; 1900, 1940, 1980 Census.


92
ethica central theoretical pointand the civil rights push
question. This last question may at first seem to address
white perceptions of a threat from black leaders, but closer
examination reveals that the focus or object of the question
is abstract: "the civil rights people," like "those people"
are objects of projected stereotypes. All of the symbolic
racism questions have been used by the proponents of the
concept in their own studies. The perceived racial threat
scale has not previously been used.
Not all cases from the 1986 ANES are utilized for this
analysis. Non-white respondents were excluded, since the
focus of the project is to understand white attitudes. The
1986 ANES was a split-form year. Most of the questions used
to construct the racial scales were limited to form B, so
form A respondents were excluded. Assignment to the forms
was random, so the use of respondents from one form or the
other creates no systematic bias in the sample.
The effects of interviewer race on respondents has been
well documented in cases where black respondents were
interviewed by white interviewers (Anderson, Silver and
Abramson 1988a, 1988b; Schuman and Converse 1971). To
prevent the reverse effect, the few white respondents with
non-white interviewers were excluded to avoid interviewer
bias. Indeed, the average score for white interviewers was
lower (more symbolically prejudiced or threatened) than
scores for respondents with non-white interviewers. This was
more often a problem for threat items than symbolic racism


83
The Archaic Significance of Racial Composition
The contemporary significance attached to racial
composition by scholars of group conflict differ dramatically
from Key's (1949) reasoning. From Key (also Wright 1977,
Mathews and Prothro 1963) one infers that past racial
composition affects the current racial climate in localities
and states. Introducing this line of reasoning requires the
review of some arguments made in more telegraphic form
earlier, while focus of this project concerns white racial
attitudes in America generally, examining the Southern case
proves instructive.
Key (1949) bases his argument on observations of a
region in which a particular economic system reigned with a
peculiar set of labor and social institutions; however, King
Cotton died. Those institutions no longer dominate the
Southern economy. Therefore today's racial composition
cannot have the same immediate significance today that the
racial composition had when Key (1949) published Southern
Politics. But the archaic racial composition may reveal
information about local norms, beliefs and values, or a
political culture crystallized in an earlier era. The
transformation of the Southern economy and the dramatic
demographic shifts make Key's (1949) argument less directly
applicable to the current political situation.


53
Self Interest: The Alternative Hypothesis
The series of symbolic racism studies go to great
lengths to test the various symbolic racism scales against a
single alternative hypotheses. The primary alternative is
represented by some measure of self interest. The proponents
may be justified in their selection of this hypothesis, for
it is compelling. But there has been relatively little
response to criticisms of their operationalization or choice.
For example, the theoretical traditions alluded to in the
previous chapter are given little play in the writings of the
symbolic advocates. Their discussion of the self interest
alternative sometimes draws clear distinctions between its
narrow construction and more general threat hypotheses and
sometimes that distinction is more obscure. The alternative
hypotheses they explore represent a most narrow construction
of 'interest.' On the other hand, this narrow construction
may provide clarity and assure that this particular
alternative is well explored.
One must ask how was this alternative hypothesis
measured. While methodological cheap shots do not provide a
basis for the formulation of a theory, they can help one
critically evaluate the extent to which the evidence
marshaled in support of a theory is supportive. The
proponents of symbolic racism go to great lengths to test


CHAPTER 1
INTRODUCTION
Contemporary studies of white racial prejudice have
tended to concentrate exclusively on one of two factors in
explaining racial attitudes. Proponents of the racial threat
explanation1 look to contemporary demographic and economic
conditions, while proponents of symbolic explanations argue
that values learned early in life explain current levels of
prejudice independent of the individuals current demographic
or economic context. This dissertation makes a case for
augmentation and synthesis of these two schools. It is
argued that individuals may simultaneously hold more than one
kind of racial orientation. Further, racial attitudes are
not only the product of either psychological traits or the
current economic context. Rather, racial attitudes result
from both of these factors and are conditioned by local
historical and cultural conditions.
Recent Trends in the Study of Prejudice
Events in recent years have spelled disaster for the
hard fought gains of the civil rights movement. Supreme
1 Racial threat may be defined as the belief on the part of
whites that black ascendance reduces the chances of white
success in competition for jobs, admissions to schools or
other areas of achievement; it may also represent fears on
the part of whites that blacks pose challenge the security of
whites' persons or property.
1


35
Whether or not local racial composition represents an
actual or imagined threat to whites will not be examined in
this study. The question taken up in this study centers on
the operative cognitive mechanism driving racial politics
among whites. In this sense, one need not answer the
question of whether or not the political, economic or
cultural status of whites is objectively threatened or if
such a specter is merely the result of the political
conjurings of opportunists, entrepreneurs and extremists.
This central difference between scholars in the threat school
is less important than their area of agreement, namely that
the perception of threat can be a potent political force.
The potential reconciliation of these perspectives will be
discussed in more detail in Chapter Four. As the second
major school of political research into the racial attitudes
of whites is discussed in the next chapter, this common
ground will come into even sharper focus and the points of
difference will fade into the background.


CHAPTER 5
DATA
In the last chapter, discussion centered on the
reconciliation of theoretical dilemmas posed by two
mainstream schools of thought in the study of whites' racial
attitudes. Additionally, hypotheses were derived from
theoretical propositions. In this chapter discussion turns
to issues concerning the testing of those hypotheses and the
measurement of the constituent elements therein. Four topics
need coverage in this chapter: 1) the sources of the data
used; 2) how the two racial attitudes outlined in previous
chapters are measured; 3) how variables that predict those
attitudes are measured; and 4) how variables that are
predicted by the racial attitudes are measured. Details
regarding the measurement of more commonplace variables are
provided in the appendix.
Data
The data exploited in this study are drawn from three
sources. First, survey data are taken from the 1986 American
National Election Study (ANES). Additionally, two types of
census data are used. Contemporary census data are drawn
from the 1980 Census of Population and Housing; the specific
data used are from the electronic version of STF 3C. That
89


100
from border South states. More to the point, the politics of
the Black Belt affected whole states, not just the counties
which exploited cheap black labor. States represent more than
mere political units. Rather, states represent areas with
specific and identifiable cultural and political traditions
which provide the context within which ongoing political
processes are played out (Conway 1968). Further, states
provide a backdrop for the socialization process.
Wright (1977) found that the addition of state level
contextual data gave his models significantly more
explanatory power than reduced models. He attributes this to
the effect of reference groups on the formation of attitudes
and vote choices. In other words, the state level data yield
information about the adult socialization process in each
state. His position is consistent with the synthesis laid
out in the previous chapter. In addition to the archaic
county-level census data, state-level data will be drawn from
both the 1940 and the 1900 censuses.
The data available in older censuses are not completely
consistent with that of the current censuses. The percentage
of the local or state population made up of African-Americans
was extracted from both the 1900 and the 1940 censuses. The
data were coded to match as closely as possible contemporary
standards concerning who is black which are based on what
some call the rule of hypo decent (see Fredrikson 1981,
Chapter Three for a discussion of the evolution of this
'rule'). For example, in the 1900 census several different


44
and Sears could declare that America had become essentially
racially egalitarian.
The most provocative assertion of the newness of
symbolic racism is found in Kinder and Sears' 1981 study.
"Since the explicitly segregationist, white supremacist view
has all but disappeared, it can no longer be a political
force .... What has replaced it ..." is symbolic racism
(p. 416). McConahay, too, makes claims that the new racism
in now dominant while old fashioned racism is a thing of the
pre-civil-rights-era past (1982).
While some studies made claims about symbolic racism
replacing other racial attitude, others point to the concept
as simply one form of racial attitude among any number of
racial attitudes. Sears and Kinder (1985) back off from the
replacement hypotheses to a position recognizing the
distinctness of symbolic attitudes as opposed to other forms.
Others point to common elements across different types of
racial attitude. One landmark study which portrayed symbolic
racism as new and emergent in suburban America points to its
similarity to older prejudicial attitudes by saying that
symbolic racism is an "expression of some of the negative
explanation based on the changing language of prejudice.
They argue, for example, that the content of racial
stereotypes has changed over the years, but that the
underlying reasons for the application of prejudicial
stereotypes has not, i.e. bigotry. In this sense, they imply
that the measures evoke less response not because America has
become a land of racial egalitarianism. Rather, the old
questions no longer resonate with the underlying prejudicial
attitudes. In other words, the changing context has rendered
the old measures useless in tapping persistent prejudicial
attitudes.


96
1980 Census
A number of contextual variables from the 1980 census
will be included in the analysis. For example, the local
racial composition is calculated by dividing the black
population by the total population. The economic context is
gauged by a measure of income equality. This variable is
designed to tap the relative economic strength of blacks in a
given locality. The mean black income in an SMSA or county
is divided by the mean white income in the same county.
Some previous studies (e.g. Giles and Evans 1985) have
tried to tap economic strength by using the difference of
mean incomes. The difference of mean incomes has good
features but still may be subject to local variations in the
value of the dollar. For example, in an area where incomes
are generally low, say the Columbus, Georgia SMSA,
differences in mean incomes would be lower than in areas
where incomes are higher, say Los Angeles. So, a difference
in average income of $500 in Columbus, may reflect a great
disparity of economic power, while a difference of $1000 in
Los Angeles might not reflect a significant disparity of
income at all. In this sense, the difference measure is not
interval and may not even be ordinal. The quotient of the
mean black income and the mean white income reduces the
sensitivity of the income measure to local variations in the


103
Indeed, Table 5-3 indicates the data from 1900 and 1940
have virtually identical effects on the level of perceived
racial threat and symbolic racism (see the far right column),
while controlling for the effect of the contemporary racial
composition. Because the effect is so similar and because
the 1940 data are available for more counties, the 1940
county level data will be used in the analysis presented in
the next chapter. Of course, this table has a substantive
interpretation as well. The archaic racial composition has
significant explanatory power for both racial scales even
when the effect of the contemporary racial composition is
controlled. This implies that the historical context of the
community in which ones lives has some explanatory power for
even symbolic racial attitudes. Similarly, the perception of
threat is not merely a function of the current context.
The local political culture influences the perception of
threat as well. Still, the current context does make a
unique contribution toward the explanation of the perception
of threat even when the historical context is controlled.
The next to last row substantiates this contention in the
predicted direction (making the .10 level the appropriate
standard of significance). The findings reported in Table 5-
3 provide support for hypothesis Hsyn04, namely that the
racial attitude scale will covary with archaic levels of
percentage black. Further, the relationship between
contemporary racial attitudes and archaic measures of racial
composition holds even when the contemporary percentage black


This dissertation was submitted to the Graduate Faculty
of the Department of Political Science in the College of
Liberal Arts and Sciences and to the Graduate School and was
accepted as partial fulfillment of the requirements for the
degree of Doctor of Philosophy.
December, 1993
Dean, Graduate School


164
This confirmation of multi-dimensionality has noteworthy
implications for future research and for the evaluation of
past efforts. Scholars interested in measuring prejudicial
attitudes need to include multiple indicators of prejudice
and demonstrate care in how those indicators are combined.
Further, the multi-dimensionality of racial attitudes has
implications for the evaluation of past research. For
example, the failure of previous research to substantiate
racial effects may be the result of insensitivity to the type
of prejudice involved (e.g. Green and Cowden 1992).
The expected relationship between the level of perceived
racial threat and the level of income equality is
substantiated, but as noted above neither theoretical nor
empirical mechanisms are in place to determine causality
(Hsyn02)* The inverse relationship between income equality
and the perception of threat found here may have implications
for group conflict research beyond those noted above. It may
be that competition for jobs represents an area particularly
sensitive to group based prejudice, while competition for
income does not. On the other hand, perhaps income equality
is more a function of employment than wage discrimination.
While these questions follow from the findings of this
dissertation, answering them falls outside the scope of this
research.
The hypothesized relationships between racial
composition, contemporary and archaic, and the racial
attitude scales are largely borne out. The effects of racial


152
dispersion if not the level of symbolic racism in both South
and North.
In the next chapter an effort will be made to reconcile
the findings of the last two chapters with the theoretical
propositions previously enumerated and to point the way for
future research on these and similar issues.


161
research strategies could be carried out in the same
communities.
In the next section, discussion turns to an evaluation
of the hypotheses relating to factors predicting symbolic
racism and an assessment of the concept's explanatory powers.
Derived Symbolic Racism Hypotheses:
While many of the derived hypotheses regarding the
predictors and effects of perceived racial threat are upheld,
the derived hypotheses for symbolic racism fare less well
(see Tables 5-3, 6-1, 7-1, 7-3 and 7-5). Despite the
assertion by the concept's proponents regarding its
abstraction, the measures of symbolic racism employed here
demonstrate sensitivity to contextual effects in general and
both contemporary and archaic local racial composition in
particular (Hsroi) Similarly, the level of symbolic racism
is sensitive to regional factors (Hsr02 and hsr04)*
Southerners are more likely to hold more prejudicial
attitudes than non-Southerners. The evidence presented in
Chapter Six even suggests a moderate relationship between
family income and the level of symbolic racism, despite the
prediction that symbolic racism is unrelated to personal
factors like economic position or income (Hsr03)* All in
all, none of the hypotheses related to the concept's
abstractness are upheld.
Tests of symbolic racism's explanatory power are
presented in Tables 6-3 and 6-4. This racial attitude scale


110
differences (e.g. percentage black 1980 minus 1940, 1940
minus 1900) in the three percentage black time points were
tried in earlier versions of the models presented here.
Neither proved particularly statistically significant, but
the difference between the 1940 and 1900 level has particular
theoretical significance. As an independent variable it
represents the effect of demographic shift before the birth
of the civil rights movement. McAdam illustrates the
importance of the shift in civil rights movement genesis and
longevity. The findings presented here indicate that the
demographic shifts in and of themselves had little direct
effect on white attitudes. Two explanations follow from
this. First, the degree of shift evident in 1940 had not yet
reach a threshold level that was necessary for migration to
have direct effects on white communities. Second, the
demographic shift's primary effect on American race relations
was as an input to movement genesis; therefore no direct
effect should be expected.
The models of the racial scales are presented in stages
to demonstrate more clearly the independent effects of the
types of variables included in the models. First, the racial
scales are regressed on a group of contextual variables.
Then the individual level variables (age, education, income,
personal financial evaluation, and conservatism) are added.
A dummy variable coded one (1) for the respondents living in
the South and zero (0) for all others rounds out the third
model. The final model for each scale includes all these


58
different. Kinder and Sears could have strengthened their
argument by calling attention to this fact.
Data
With only a one exception (McConahay 1982; Louisville),
all the data used in the examinations of symbolic racism are
drawn from non-Southern samples. There are some instances of
national level data set being used to measure general
symbolic politics hypotheses, but none address symbolic
racism specifically.
This leaves a number of city-wide studies conducted in
the north and west and some studies conducted in colleges and
seminaries. The McConahay article of 1982 does feature data
from a Louisville sample collected by the Harris
organization; some consider Louisville a Southern city.
Still, the fact that regional differences in racial tolerance
are often found (see examples in the previous chapter) gives
one cause to expect either that regional differences would
exist in the new racism or that the prevalence of the old
racism would follow some regional pattern. The significance
of the second point depends on the theoretical expectations
concerning the relationship between symbolic and old
fashioned racism: if symbolic racism is a submerged redneck
racism then one would expect either a greater prevalence of
symbolic racism where redneck racism had existed previously
provide an interesting counter example to this trend by not
reporting the constant when it may have been meaningful.


16
class whites into an unnatural coalition with upper class
whites while excluding blacks altogether. After the civil
rights movement, black allegiance to the Democratic party
became almost universal. Now the perception of racial
threats is the basis for another unnatural combination of
classes in not merely the Southern party system but the
national party system (Huckfeldt and Kohfeld 1989).
According to Huckfeldt and Kohfeld's argument, the Democratic
party plays host to middle and lower class blacks and upper
middle and upper class white liberals. The Republican party
combines socially conservative working class whites with
rich, economically conservative whites.
The crux of Huckfeldt and Kohfeld's argument is that the
nature of this unnatural coalition undervalues the
contribution of blacks since their Democratic votes are taken
for granted by both parties. If the parties had to compete
for their votes then, the argument goes, blacks would be
better represented because more elected officials would have
to be more responsive. Further, such diversity could produce
more stable, class based coalitions in the two major parties.
The conclusions drawn from this argument are provocative
and in some way reinforce the importance of the project at
hand. But a keener focus on some of the interior elements of
the argument provide helpful insights. Huckfeldt and Kohfeld
argue that the current coalitions are uncomfortable for a
number of reasons. First, they affirm that the ideological
base of American politics has shifted, that the labor-


81
A third way in which the current racial composition may
accentuate hostility among whites operates through
visibility. Potentially threatened whites may be more likely
to latch on to racial explanations for their difficulties if
they have reminders of the existence, size, and power of the
other race. So the more visible the minority population the
more likely they would be blamed for problems faced by
whites. Again, one might expect a nonlinear relationship
between percent black and the level of threat for these
reasons. The level of threat would increase with increasing
slope as the percentage black increased (like figure 4-2).
The proponents of symbolic racism might argue that
increased visibility of the minority population should have
no effect on the level of symbolic racism. However, the
visibility argument sketched out above leads one to expect
the level of symbolic racism to increase with the percentage
black to some threshold level and level off thereafter
(HSYN03)- If one relaxes the assumption of abstractness, one
might expect the symbolic proclivities to increase slightly
as the percentage black rose to some level of visibility,
making blacks an identifiable segment of the population.
After blacks are identified incremental changes should have
little effect if the symbolic attitudes are largely abstract
(see figure 4-4).


14
tremendous success. In some elections, outcomes had more to
do with the county from which the candidate hailed than his
platform or the black population. Still, as a rule the
analysis is compelling.
This analysis makes much of the differences in the
interests and attitudes between poor whites on one hand and
the planters and industrialists on the other. Indeed, Key
makes much of the claims of the "haves" that race relations
was a problem for the backward rednecks. Despite their
divergent interests and claims, the trigger mechanism of
their racial attitudes was often the same. Both groups
thought they had something to lose from black ascendance.
Planters were concerned about the loss of cheap labor which
threatened their economic well-being and losing control of
the state government through which they maintained their
advantaged position. Key's argument implies that whites
outside the black-belt had less to lose from black ascendance
and even felt less threatened than other whites. Still,
these whites were sensitive to race-baiting. This
sensitivity grew out of the fear of a return to the poverty
and degradation of the Reconstruction era. Key takes great
pains to indicate that these poor whites acted in a way that
was contrary to their class interests because, among other
things, they feared black ascendance.
Some may argue that the usefulness of Key's work has
faded as a result of the remarkable changes that have
occurred in the South and the nation since its publication.


90
summary file will be used to calculate the racial composition
for the SMSAs and counties represented in the national sample
and to collect other relevant contextual data.1 These data
are collected at the SMSA level for respondents residing in
SMSAs and at the county level for others. SMSAs, for those
who live in them, are thought to be the defining local unit
rather than counties or congressional districts. SMSAs tend
to have area-wide media markets reinforcing identification
with the urban center rather than with constituent counties.
County level data are used for respondents not residing in
SMSAs. County level data represent the highest level of
precision available for merging since census tracts and minor
civil divisions have been padded (overwritten with nines) in
the 1986 ANES.
Archaic census data are drawn from the ICPSR
reproductions of the U.S. Census of Population and Housing 1%
counts from 1900 and 1940 (ICPSR study number 0003). The
100% counts are available in neither electronic nor card
format for either the 1900 and the 1940 censuses. The use of
the sample counts should make little difference in the
project at hand since only minor differences between the
sample and 100% counts are likely. Additionally, any
variations between the two counts should not produce
systematic biases.2
1 The 1990 National Summary File is not yet available, so
interpolation for maximum accuracy is not yet feasible.
2 Theoretically both the 1% and the 100% counts are samples,
because the survey used to produce the 100% count cannot
actually account for everyone. Unfortunately, the 100%


8
represent different dimensions of racial attitudes among
whites.
Further, one may argue that current racial attitudes, be
they the result of perceived racial threat or symbolic
racism, are not merely either the result of current
demographic conditions or isolated psychological traits.
Indeed, both may factor into the equation. If the current
material context can influence an individual's racial
attitudes, couldn't the cultural context have an influence as
well?
In Chapter Four, a synthetic theoretical position is
staked out which allows for the simultaneous influence of
personal characteristics, contemporary material context and
local cultural factors. Chapter Five addresses measurement
issues and issues related to multi-dimensional racial
attitudes. National level models predicting the racial
attitudes and their effects are presented in Chapter Six.
Chapter Seven explores regional variations in the sources of
prejudice. A review of the central findings and the
conclusions are found in the final chapter.
Before ending this introductory chapter a cautionary
note concerning the motives of the author is in order.
The purpose of this project lies not in finding more
ways to brand people as racists. Rather, the goal of this
research lies in seeking a better understanding of the
anatomy of prejudicial attitudes. These prejudicial
attitudes may or may not have implications for political life


126
as a leader on the national scene may compel his evaluation
along different lines than blacks in general. Alternatively,
the salient elements of the bases of Southern prejudice may
be contained in the racial scales. In other words, Southern
prejudice is less geographic than cultural. More discussion
of this issue follows in Chapter Seven.
The second equation on Table 6-3 presents the results of
a similar regression analysis with the respondent's
evaluation of blacks as a group (subtracted from their
evaluation of poor people) as the dependent variable. In
this case the results are much less complex. Perceived
threat is a significant predictor of antipathy toward blacks
as a group at the 0.01 level. The symbolic racism scale is
again insignificant and in this equation the parameter
estimate shrinks to 0.04. It would be an exaggeration to say
that perceived racial threat explains most or perhaps even
much of the variation in the dependent variable (note: the
adjusted r^= .10), but it does have a significant effect in
the predicted direction when symbolic racisms contribution
is negligible. The perception of racial threat makes a
significant substantive and statistical contribution to the
make up of anti-black affect.
Like Equation 1, the control variables reveal few
surprising findings. Indeed, of the individually measured
variables only age makes a significant direct contribution to
the model. Like before, it appears that older people are less
disposed to give blacks a favorable evaluation than younger


169
format in addition to other aspects of validity. (Krosnick
1991) Krosnicks findings may also help explain the
differences between the racial threat and symbolic racism
models in the proportion of the variance explained; more of
the variance in the symbolic models may have been explained
because the questions used to construct the scale are of a
superior form.
The implications of this research for the perceived
racial threat school are perhaps less dramatic than those
noted above for symbolic racism. While the attitudes of
threat do show sensitivity to the contemporary racial
context, there is evidence that the history and culture of
the locale also have significant influences on those
attitudes. This represents a refinement or augmentation to
the main body of quantitative literature on the topic. The
history and culture of an area may have implications for
understanding current racial attitudes which cannot be
captured using only contemporary data.
Unlike symbolic racism, the racial threat literature
does make a place for the influence of contextual effects,
but just barely. The contemporary local racial composition
is seen as influencing the level of perceived threat in some
studies or is used as a proxy for threat in others. A place
is made only for the immediate effects of the local racial
composition, theoretically and empirically. While one might
expect the contemporary scene to have important effects, the
present study indicates that the history and culture of the


125
scale is significant in a one-tailed test, the appropriate
standard when the sign of the parameter estimate lies in the
theoretically predicted direction. The effect of perceived
racial threat on Jackson antipathy is not by any means
overwhelming, but some relationship is evident; the more
threatened the respondents feel the less favorable their view
of Jackson. The symbolic racism scale does not have
significant explanatory power in this model.
Other findings among the control variables are
noteworthy but not especially surprising. Older people like
Jackson less. The respondent's level of pessimism in
evaluating the national economy is associated with a more
favorable Jackson rating. Conservatives do not care for
Jackson. Finally, it appears that Jackson is more popular
(or at least less disliked) in areas with higher percentages
of residents employed in agriculture, although this
unpredicted finding is just shy of significance at the 0.05
level.
These findings among the control variables are not
particularly surprising give Jackson's status as a populist
figure, actively reaching out across racial lines to the
downtrodden. A more remarkable finding lies in a variable
which is not significant. White Southerners do not think
significantly less of Jackson than they do of other national
political figures. Perhaps Southerners' propensity toward
prejudicial attitudes does not extend to exceptional
politicians, like Jackson.
In other words, his very success


xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID EBTDPG1G7_UZ8DG7 INGEST_TIME 2014-10-13T21:46:03Z PACKAGE AA00025835_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES


113
variables allows the observation of indirect contextual
effects. For example, once age, education, income, personal
finance, and conservatism are controlled the effect of the
contemporary racial context is revealed; the percentage
black 1980 has a significant effect in the predicted
direction. Contextual effects still have some explanatory
power even when level of education and income are controlled.
This evidence is in opposition to the Fossett and Kiecolt
(1989) assertion that the regional and contextual effects
represent merely local variation in levels of individual
level variables like education.
Adding the South dummy variable to the model clarifies
the nature of the contextual effects further. The Southern
dummy variable seems to pick up the contextual effects that
were associated with cultural variables and allow more of the
effect of the contemporary racial context to be revealed.
The effect of the percentage black in 1980 on symbolic racism
increases from 0.08 in equation II to 0.11 in equation III
(with an accompanying jump in t-values as wellof course
coefficients, t and p are functions of each other.) The
effects of the individual level variables remain essentially
unchanged by the addition of the Southern dummy variable.
The addition of this dummy adds little to the explanatory
power of the model; its main effect lies in the consolidation
of cultural effects and its control allows or reveals an
amplified effect of the contemporary racial context as
measured by the percentage black in 1980.


120
the South's addition to the model. This lack of impact
implies the limited utility of the geographic designation for
its own sake. The importance of the South has less to do
with the set of state names than with identifiable economic
and cultural structures extant within those states. Directly
identifying those structures provides a better method of
capturing whatever it is scholars grope for when referring to
Southern exceptionalism than merely affixing a value of one
to the eleven former confederate states.
The logic of this argument receives even clearer
support when one turns to the results presented in equation
IV of Table 6-2. Here one sees that the addition of the
hybridized variables produces dramatic changes in the model.
These variables include a measure of post-bellum plantation
agriculture and a portrait of southern racial dispersion
circa 1940. These variables more accurately reflect the
salient elements of southerness than the flat geographic
designations. These hybridized variables have similar
effects on the total model for racial threat that they had on
the models of symbolic racism. Namely, more effectively
controlling for the south reveals the erstwhile suppressed
marginal effects of other theoretically significant
variables.
The significant hybridized variables all have negative
parameter estimates. These may be interpreted in similar
fashion to the negative estimate presented above. For
example the interaction term of South and the percentage


79
effective predictor of threat. Of course, one might also
reasonably expect a measure of income equality to correlate
inversely with racial hostility (Jacobs 1982). Comparability
of income may be an outgrowth of a harmonious racial
atmosphere rather than a cause of a toxic one (Hsyn02) On
the other hand, such arrangements could just as easily
promote mutual respect or at least the displacement or
loosening of stereotypes (Hsyn05)*
A second way in which the contemporary racial
composition may influence white attitudes lies in increasing
the probability of actual conflicts. In a racially monotonic
community, one would expect zero cases of racial conflict of
any kind. The probability of inter-racial disputes (whether
caused by racial hostility or not) could be expected to
increase as the percentage of each racial group approached
50% (see figure 4-3).7 This would produce a curve shaped
like an inverted U centered on 50%. If symbolic racism is
truly abstract, no correlation would be found because of
increasing opportunities for inter-racial conflict (Hsroi)*
7 Conflicts occur between citizens all the time no matter
what their race. The increase in transactions between people
of different races does not necessarily imply that conflicts
resulting from those transactions result in racial hostility.
Indeed, other things being equal, the increase in the number
of inter-racial transactions may result in a decline in
racial hostility because of the increase in the number of
positive outcomes from those transactions. However, given
the history of racial hostility in America, one would expect
(as has been observed) short term increases in hostility.


154
In that chapter, threat is defined as a challenge to
political or economic power, or to security of person or
property. Racial threats in this study are defined as the
challenges to whites' political or economic power, or to
their security of person or property posed by blacks. Much
of the racial threat literature uses definitions of racial
threat which are consistent with this definition. In other
words, blacks represent an actual challenge to, for example,
the economic and social standing of whites. However, a
threat need not be actual for it to influence the thoughts
and behavior of an individual. The perception of threat,
whether that perception is accurate or not, may be sufficient
to effect a change of behavior in an individual. For
example, a novice poker player may fold in response to the
raises of a bluffing opponent. The perception of threat may
lead to the adoption of political positions which individuals
might not otherwise take.
Much of the racial threat literature tests the effects
of environmental or contextual factors on white attitudes.
Typically, the local racial composition, or the percentage
black in the respondent's home county or Standard
Metropolitan Statistical Area (SMSA), is used as a predictor
of some prejudicial orientation, often with successful
results. For the most part, the racial threat literature
assumes that the size of the African-American population in
the respondent's locality is an accurate gauge of the racial
threat because it is in those areas that blacks will present


91
Neither the 1940 nor 1900 census provide SMSA level
data. Many of our contemporary SMSAs were not remotely
metropolitan in those years, so many counties now located in
SMSAs were then only counties. Since SMSAs are constructed
of whole counties, one could use the current structure as a
template for aggregating older county level data. Such a
post hoc aggregation is not theoretically warranted. The
theoretical justification detailed above for using SMSA level
data from contemporary censuses cannot apply to data as old
as these for a variety of reasons, including the more
fragmented nature of media markets in earlier eras. Data
drawn from those two years will be at the county level and at
the state level (discussion follows).
Dependent Variables
The choice of items for the racial scales was based
primarily on face validity, the theoretical assumptions
outlined above (see the Appendix for question text) and
consistency with previous studies. The questions scaled to
tap perceptions of racial threat address issues of whites
losing jobs, promotions or school admission as a result of
affirmative action programs or employer choices. The
symbolic racism items are limited to those items which
address issues of black adherence to the Protestant work
sample is not used as a predictor of the population but is
taken as the population. In this sense the 1% sample may
produce a more accurate or at least as good a predictor of
the population because of the care taken in sampling and the
extrapolation of that sample.


64
But the acknowledgment of multi-dimensionality by the
proponents of symbolic racism is underdeveloped empirically
and theoretically, except to say that the new or symbolic
racism replaces the old. One wonders why the old has not yet
disappeared or what the different effects of the two sets of
prejudicial attitudes might be?
The group conflict orientation held by some of the
authors discussed in Chapter Two holds no place for
conceptions of multi-dimensionality. From their perspective,
material imperatives would outweigh any preferences,
psychological predispositions or attitudes. Their view of
racial attitudes lying mainly in the material realm makes
them ill equipped to explain even some of their own findings.
The works by Huckfeldt and Kohfeld (1989) and Key (1949) are
less clear about the exact nature of the racial attitudes,
but unlike their colleagues they point to racial
predispositions resulting in behavior which is contrary to
individuals' material interest. Given this lack of agreement
it seems reasonable to temper or relax the assumption that
actual conflicting interests represents a necessary and
sufficient condition for the emergence or existence of racial
hostilities.
The two central schools of thought either dismiss the
possibility of multi-dimensionality or fail to explain it.
Some studies by the proponents of symbolic racism do find


APPENDIX
Question text from the 1986 American National Election Study
Number for each response in parentheses for whites with white
interviewers.
Racial Threat scale items:
V516 I would like you tell me whether or not you agree or
disagree with the following statement: Affirmative action
programs for blacks have reduced whites' chances for jobs,
promotions and admissions to schools and training
programs.(Do you agree or disagree)
V517 Do you agree strongly or not strongly?
Do you disagree strongly or not strongly?
l)agree strongly(232) 2)agree not strongly(217)
3)disagree not strongly(217) 4)disagree strongly(111)
8)DK(2) 9)NA(0) 0)INAP(957)
Valid percent giving most extreme response: 29.9; 57.8% in 1
plus 2.
V560 what do you think the chances are these days that a
white person won't get admitted to a school while an equally
or less qualified black person gets admitted instead? Is
this very likely, somewhat likely, or not very likely?
l)very likely(230) 3)somewhat likely(343)
5)not very likely(250)
8)DK(44) 9)NA(5) 0)INAP(864)
Valid percent giving most extreme response: 27.9
V563 What do you think the chances are these days that a
white person won't get a job or promotion while an equally or
less qualified black person gets one instead? Is this very
likely, somewhat likely, or not very likely to happen these
days?
1)very likely(224) 3)somewhat likely(410)
3)not very likely(205)
8)DK(2 7) 9)NA(6) 0)INAP(864 )
Valid percent giving most extreme response: 26.7
V569 What do you think the chances are these days that you
or anyone in your family won't get admitted to a school while
an equally or less qualified black person gets admitted
instead? Is it very likely, somewhat likely, or not very
likely?
1)very likely(105) 2)somewhat likely(247)
173


102
Of course, the county level data from 1940 and 1900 are
highly correlated. This correlation makes it impossible to
use both in the regression models presented in the next
chapter. Since they are correlated, one suspects that one
might be as good as the other in explaining current levels of
the racial attitudes discussed above. Examining the partial
correlation coefficients for each year's data on the racial
scale while controlling for the contribution of other years
should give an indication of the veracity of this assumption.
Table 5-3. Black Concentration in 1900, 1940, and 1980 and
the levels of Perceived Racial Threat and Symbolic Racism in
1986 for White Respondents Only.
Controlling for
the % black 1900
RT by
1900
%black
a
SR by
1900
%black
RT by
1940
%black
0.0443
SR by
1940
%black
0.0615*
RT by
1980
%black
0.0588b
SR by
1980
%black
0.0389
Controlling for
the % black 1940
0.0143
0.0101
Controlling for
the % black 1980
0.0942**
0.1372***
0.0913**
0.1445***
0.0480b
0.0163
a Entries are first order partial correlation coefficients.
RT=Racial Threat
SR=Symbolic Racism
b p<0.10
* p<0.05
** p< 0.01
*** pcO.OOl
Source: 1986 ANES, 1980 Census, 1940 Census, and 1900 Census


60
chapter, an alternative approach will be outlined. This
approach is more synthetic than innovative. Indeed,
innovation is difficult to come by in a literature as
"sprawling" as that on racism (Kinder and Rhodebeck 1982).
This new approach attempts to build on the strengths of the
two outlined below. For example, the claims made by
proponents of symbolic politics concerning the enduring
effects of pre-adult socialization to the exclusion of the
material or social context must be tempered. Similarly, the
assumption that 'realistic group conflict' is necessarily
realistic will be relaxed; even if such is the case, an
individual must perceive the conflict as real to act upon it.
Of course, not all questions raised by this discussion will
be answered. Indeed, some may be unanswerable in the short
run. Still, it is hoped that the syntheses and ensuing
empirical investigation will shed some light on the salient
issues and point in the direction of their resolution.


41
source for positions on issues, the evaluation of candidates
and other external stimuli. Ultimately the predispositions
result in behaviors such as voting.
There are two central elements to the racial attitudes.
The first is anti-black affect, a generalized distaste for or
disapproval of blacks. The second essential element is found
in a commitment to traditional values, not limited to but
epitomized by the Protestant work ethic. Thus Kinder and
Sears (1981, also quoted in Sears 1988) list the ingredients
of symbolic racism as
a blend of antiblack affect and the kind of traditional
moral values embodied in the Protestant Ethic ... a
form of resistance to change to the racial status quo
based on moral feelings that blacks violate such
traditional American Values as individualism and self-
reliance, the work ethic, obedience and discipline, (p.
416)
These components work in 'conjunction' together to form
symbolic racial predispositions.
The proponents argue that this symbolic attitude is a
new form of racism, that it is distinct from redneck racism,
"open bigotry of the sort frequently associated with working-
and lower-class (Southern) whites" (McConahay 1982, p. 705),
in either or both of two ways. Redneck racism has either 1)
disappeared or is of dramatically reduced salience in
contemporary society or 2) become old fashioned. That is to
say, it "is no longer fashionable to express these beliefs or
to support these practices openly in the elite circles of our
society." (McConahay 1982, p. 705)


6
that if they just tried harder they could 'succeed on their
own like the Italians, Irish and Jews did.' Kinder and Sears
call this sort of racism symbolic because it is abstract,
based on ideological positions and attitudes rather than
experience. They argue that perceptions of threat and actual
vulnerability are ineffectual predictors of behavior while
symbolic measures excel.
Despite the growing literature on symbolic racism, more
research is needed to refine our understanding. While these
perspectives have drawn some criticism, chiefly from
Sniderman and his co-authors, important elements of the
theory have not been tested. Further, the elements that have
been tested use local samples which might yield results with
little external validity, or validity only if we throw out
the South. Additionally, insufficient effort has been made
to test the symbolic racism theses against racial threat
hypotheses. There are exceptions to be sure. (Green and
Cowden 1992, Kinder and Sears 1981, Sears, Lau, Tyler and
Allen 1980) The theory and measurement of this school of
thought are treated in detail in chapter three.
Synthesis
One need not choose sides in such a debate, for there is
no a priori reason to suspect that the two attitudes are
mutually exclusive. The proponents of symbolic racism theory
may be correct in asserting that an abstract sense of
prejudice is held by many whites, but many may feel


178
Dalmas Taylor (eds.), Eliminating racism: profiles in
controversy. New York: Plenum.
Bobo, L. and F.D. Gilliam 1990 Race and sociopolitical
participation, and black empowerment. American
Political Science Review. 84:377-93.
Boyd, L. H. Jr., and G.R. Iverson 1979 Contextual analysis:
concepts and statistical techniques. Belmont,
California: Wadsworth.
Button, James 1989 Blacks and social change: the impact of
the civil rights movement in southern communities
Princeton: Princeton University Press.
Conway, Mary Margaret 1968 The white backlash re-examined:
Wallace and the 1964 primaries. Social Science
Quarterly 49:710-719.
Edelman, Murray 1971 Politics as symbolic action: Mass
arousal and quiescence. Chicago: Markum Press.
Fossett, Mark A. and K. Jill Kiecolt 1989 The relative size
of minority populations and white racial attitudes
Social Science Quarterly. 70(4)820-835.
Fredrickson, George M. 1981 White Supremacy: a comparative
study of American and South American history. New York:
Oxford University Press.
Giles, Michael W. 1977 Percent black and racial hostility:
an old assumption reexamined. Social Science Quarterly
58:412-417.
Giles, Michael W. and Melanie Buckner 1993 David Duke and
Black Threat: An old hypothesis revisited Journal of
Politics 55:702-713.
Giles, Michael W. and Arthur S. Evans 1985 External threat,
perceived threat, and group identity. Social Science
Quarterly 66(1) 50-66.
Giles, Michael W. and Arthur S. Evans 1986 The power
approach to intergroup hostility. Journal of Conflict
Resolution. 30(3): 469-486.
Goertz, Gary 1992 Contextual theories and indicators in
world politics. World Politics 17(4):285-303.
Green, Donald P. and Jonathan A. Cowden 1992 Who protests:
self-interest and white opposition to busing. Journal
of Politics 54 (May) .-471-496.


31
the Bubbas in Babylon, like that of a fish out of water, is
different than it would be 'Down Home.'
Using the same data (1972 ANES) exploited by Giles and
Evans (1985), Fossett and Kiecolt (1989) re-estimate the
effects of several contextual and individual predictor
variables on their shared measure of perceived threat. They
find that local racial composition influences their measure
of threat in the expected direction for all samples and
subsamples. Again, their specification differs from that of
Giles and Evans (1985) by the inclusion of the community size
control. Community size is negatively related to perceived
threat in all samples, so urbanness appears to have an
ameliorating effect on prejudicial attitudes. What they do
not find is a significant relationship between either a South
dummy variable or southern origins and their measure of
perceived threat. This runs counter to what one would
expect. Indeed, no variable predicting their measure of
threat has a significant regional difference.
Fossett and Kiecolt (1989) go further and examine the
effect of the threat scale on support for racial integration.
Here they find, as expected, that the threat scale has a
significant inverse relationship with support for
integration. Even with the threat scale in the model, the
percentage black has the expected (inverse) effect in all
samples. But in these models there are important regional
variations. First the threat scale has a more pronounced
effect in the Southern model. Second the Southern origins


180
Congressional voting. American Journal of Political
Science. 23:495-527.
Kinder, Donald R. and D. Roderick Kiewiet 1981 Sociotropic
politics. British Journal of Political Science 11:129-
161.
Kinder, Donald R. and L.A. Rhodebeck 1982 Continuities in
support for racial equality, 1972-1976. Public Opinion
Quarterly 46:195-215.
Kinder, Donald R. and David 0. Sears 1981 Prejudice and
politics: symbolic racism versus racial threats to the
good life. Journal of Personality and Social
Psychology. 40:414-431.
Kinder, Donald R. and David 0. Sears 1985 White opposition
to busing: on conceptualizing and operationalizing group
conflict. Journal of Personality and Social Psychology.
48:1141-1147.
King, Gary 1986 How not to lie with statistics: Avoiding
common mistakes in quantitative political science.
American Journal of Political Science 30:666-687.
Kousser, J. Morgan 1974 The shaping of southern politics.
New Haven: Yale.
Krosnick, Jon A. 1991 The stability of political
preferences: Comparisons of symbolic and nonsymbolic
attitudes. American Journal of Political Science
35:547-76.
Lamis, Alexander P. 1988 The two-party south. New York:
Oxford University Press.
Lieberson, Stanley 1980 Pieces of the pie: blacks and
white immigrants since 1880. Berkeley: University of
California Press.
Matthews, Donald R. and James W. Prothro 1963a Social and
economic factors and negro registration in the South.
American Political Science Review 57:24-44.
Matthews, Donald R. and James W. Prothro 1963b Political
factors and negro voter registration in the South.
American Political Science Review 57:355-367.
McAdam, Doug 1982 Political process and the development of
the black insurgency, 1930-1970. Chicago: University of
Chicago Press.
McConahay, John B. 1982 Is it the buses or the blacks:
Self-interest versus racial attitudes as correlates of


119
The second equation in Table 6-2 adds the block of
individual level variables, as above. And like the symbolic
racism models, education and conservatism show consistent
effects in predicting this racial attitude. Note, however,
that the ameliorative effect of education is reduced by half
in the racial threat model from the level extant in the
symbolic racism model. Conservatism aggravates the racial
scale in the prejudicial direction. Note also that the
respondent's pessimism regarding personal finances is
directly related (although with limited magnitude) to the
level of perceived threat; the more precarious the
prospective financial evaluation, the more threatened the
respondent feels. The inverse relationship between the level
of income equality and the level of perceived threat is
actually amplified by the addition of the individual level
variables. In other words, when factors closely associated
with individual economic vulnerability (such as age,
education, and actual income) are controlled the impact of
relative income levels between the races is more accurately
represented. After this initial amplification, the impact of
income equality remains constant throughout the other
equations.
The addition of the South dummy variable has almost no
impact on the model (see equation III). The dummy variable
has almost no direct effect; its beta is virtually
indistinguishable from zero. Additionally, none of the other
parameter estimates demonstrate any changes as a result of


47
or a street corner. The residue of anti-black affect reduces
the level of sympathy resulting in the assignment of blame
for the outcast group. So the perceived failings of the out
group to live up to expectations and the attendant, but
abstract, disdain for that out group result in the symbolic
racial attitude.
Despite criticism over the combination of analytically
and (potentially) empirically distinct elements (note
especially the frontal assault by Sniderman and Tetlock 1986a
and a less successful effort by Bobo 1983), the proponents
stand by the 'conjunction' premise.
Abstraction vs. Self Interest
The concept of symbolic racism was introduced in Sears
and Kinder's 1971 study of the 1969 Mayoral election in Los
Angeles. This election pitted a conservative white
candidate, Yorty, against a liberal black candidate, Bradley.
The concept developed inductively from their analysis of
survey questions asked during the election (Kinder 1986,
McConahay 1976). From the responses to the questions the
authors inferred that a new form of racism had taken root.
This new racism consisted of "abstract moral assertions about
blacks' behavior as a group . ."(Sears and McConahay
1973)
This new racism was thought to be abstract, not based in
the individual's current material conditions, and hence
symbolic. If not based in material conditions, where does


43
values had replaced the old racism (Kinder and Sears 1981).
Lately, both Kinder (1986) and Sears (1988) have backed off
from that position and now argue that the old fashioned
racism still exists and is different from symbolic racism and
that both have anti-black affect in common. All proponents
defend the distinctiveness of symbolic racism from previously
recognized forms of prejudice.
The new racism was distinguished from two other
recognized forms of racism: 1) generalized (in)egalitarianism
or 'redneck racism' which fostered "legalized or formalized
racial discrimination, [and] inherent racial inferiority"
(Sears and McConahay 1973, 138) among other things; and 2)
personal racial threats, characterized by threats to an
individual's job, housing, or safety.
Kinder and Sears (1971, 1981) dismissed the importance
of redneck or old-fashioned racism and were joined in this
dismissal by associates McConahay and Hough (1976) and
various constellations of co-authors (examples follow). They
point to declining levels of support in surveys for
segregationist positions as evidence of the declining
significance of old fashioned prejudice3. So by 1981, Kinder
3 See Schuman, Steeh and Bobo (1985) for an extensive
discussion of the declining levels of support for segregation
over time. In that study, they indicate a period of decline
in prejudicial attitudes throughout the civil rights era
before a new equilibrium (at relatively low, but not
inconsequential levels) is reached. Few respondents to
segregationist survey questions indicate the most prejudicial
positions. But as Sniderman and Tetlock point out (1986a,
also Sniderman, Tetlock, Piazza, and Kendrick 1991) it is not
evident why few choose the prejudicial responses to those
questions. Sniderman and Tetlock (1986a) pose one


5
authors share a common perspective based on a few simple
premises. First, the changes wrought in America in the 1950s
and 1960s have changed the nature of racial attitudes held by
most Americans. It is no longer socially acceptable to hold
openly racist positions, especially positions based on
segregation or white supremacy. So, old fashioned racism,
defined as the belief in segregationist values, cannot be
advocated openly.
The second premise maintains that racism has a new form
of expression. Whites now use their adherence to
traditional, conservative values to legitimize anti-black
positions. This allows them to explain prejudicial feelings
and beliefs in terms of acceptable ideological stances. The
use of these values is independent of the individual's
demographic or economic context. In other words, one's class
or location and proximity to blacks have no effect on one's
racial attitudes.
This orientation is commonly termed 'symbolic racism,' a
combination of anti-black affect and traditional values which
(according to its proponents) has replaced the old racism
rooted in segregationist values (Kinder and Sears 1981).
According to Kinder and Sears, symbolic racism refers to a
sense of disdain for those who fail to adhere to the
Protestant work ethic and thus seek handouts, lack thrift or
punctuality, fail to repress sexual urges or delay
gratification to work for distant goals. A symbolic racist
is likely to blame the victims of past injustice by saying


106
dependent variables has been discussed. Similar treatments
of the contextual and endogenous variables were provided.
Further, evidence was brought to bear on two hypotheses
outlined in Chapter Four (HsynOI and Hsyn04)- The two racial
scale were found to occupy different dimensions, as
predicted, and the significance of archaic racial composition
was confirmed.
In the next chapter discussion turns to address modeling
procedures and hypothesis testing.


88
from older censuses can inform research about current racial
attitudes by providing a cultural backdrop, pointing to
crucial cultural information from the past which may
influence contemporary racial attitudes (Hsyn04) The
influence of the current racial composition on racial
attitudes reflects the extent to which those attitudes
reflect the sort of social dynamic outlined by social
conflict or racial threat theorists (HrtOI)*
The argument presented above dulls the otherwise sharp
distinction between analysis of racial attitudes based on a
symbolic politics model and models more firmly grounded in
material conditions. It is inaccurate to argue that the
basis of racial hostility must be one or the other. Material
conditions influence the values held by previous, current and
future generations because of the methods of transmission
from parent, teacher, and supervisor to child, student and
employee. These values provide a lens or optic through which
current material conditions are viewed. Similarly, current
material conditions influence the salience or relevance of
that lens in a given situation. Perceptions, beliefs and
values cannot be completely divorced from the material world,
as the proponents of symbolic racism have claimed.
In the next chapter, discussion focuses on how one might
measure and test the propositions laid out in this chapter.


149
Table 7-6. Standard Deviation of Perceived Racial Threat
Regressed on Contextual Variables in Each County.
entries are unstandardized coefficients from OLS regression
white Southern respondents only
Variables I
Mean Racial Threat 0.82***
II
0.79***
III
0.95***
%black 1940
State %black 1940
%black 1980
%employed in
agriculture
income equality
0.000005 0.001
-0.0004
-0.0003
0.001
0.05b
constant
adj
N
-0.03*
0.58
24
-0.03
0.48
23
-0.06
0.53
23
white Northern respondents only
Variables
I
II
Mean Racial Threat
0.69***
0.81***
%black 1940

-0.0005*
State %black 1940


%black 1980


%employed in
agriculture


income equality
__ _
III
0.83***
0.00005
-0.001
-0.0002
0.001
-0.01
constant
adj R^
N
-0.02*** -0.03*** -0.02b
0.57 0.65 0.64
73 73 65
Source 1986 ANES; 1900, 1940 and 1980 Census


CHAPTER 4
SYNTHESIS
in the previous two chapters discussion focused on two
mainstream but distinct schools of thought concerning racial
prejudice in America. The first school maintains that racial
threats or the perception thereof provide the central
mechanism for the generation and use of prejudice. The
second school introduces a 'new' form of abstract prejudice
thought to have become widespread in the post civil rights
era. In this chapter, a third position is laid out in order
to synthesize a set of reasonable theoretical propositions
and lay the ground work for testing those propositions.
The (survey research portion of the) literature
discussed in the previous two chapters suffers from a problem
common in the social sciences. Namely, the phenomenon being
modeled is far more complex than the model.1 In this case,
as the proponents of each of these schools have staked out
their territories, they have tested their hypotheses against
relatively transparent null hypotheses. Additionally, the
conversations in each school have been independent of each
other. The result of these shortcomings stultifies theory
building since research falls into a cycle of replication on
1 Of course, the very nature of modelling requires some
simplification, else the exercise would be pointless.
61


112
Table 6-1. Symbolic Racism Regressed on Contextual and
Individual Level Variables for White Respondents Only.
entries are unstandardized coefficients from OLS regression
Variables
I
II
III
IV
%black 1940
0.05b
0.03
0.01
0.13*
State %black40
0.000001b
0.000002b
-0.0000003
-0.0000002
%black 1980
0.005
0.08*
0.11**
0.13b
%employed in agriculture
0.25b
0.08
0.09
0.02
nativity
-0.01
-0.03b
-0.02
-0.03b
income equality
-1.48
-2.36
-2.16
-3.27*
v595 age

0.02b
0.02b
0.02
v602 education

-0.99***
-0.99***
-1.00***
v733 family income

-0.07b
-0.07b
-0.07b
v356 personal finance

0.06
0.05
0.05
v385 conservatism

0.63***
0.62***
0.61***
south


1.55b
3.40*
%black 1900* average farm
size 1900



-0.0001
%black 19802



-0.002
south*%black40



-0.15**
%black 1940-%black 1900



0.05
constant
adj R2
N
11.51***
0.07
758
14.03***
0.23
558
13.54***
0.23
558
14.99***
0.24
540
Source: 1986 ANES; 1900,
b p-value < 0.10
* p-value < 0.05
** p-value < 0.01
*** p-value < 0.001
1940 and 1980 Census


23
percentage of registered whites voting for Duke and the
percentage of eligible whites voting for Duke. Further, the
direction of some of the significant controls lends support
to the notion that the mechanism generating Duke's support is
related to perception of threat. For example, median income
is negatively correlated with Duke support, indicating that
the most economically vulnerable localities were more likely
to support Duke. The percentage of whites unemployed was
significant and positive in the equation predicting the
percentage of eligible whites voting for Duke. This implies
that Duke's appeal was strong and had the effect of
mobilizing voters who were particularly vulnerable and most
likely to feel threatened by black ascendance. Urbanization
and inmigration had the expected ameliorative effects.
While both Hertzog (1991) and Giles and Buckner (1993)
demonstrate support for the relationship between threat and
candidate support in the aggregate and Giles and Buckner
(1993) find some support that implies that threat is one of
the mechanisms driving this relationship, both studies suffer
from limitations imposed by their data. There is only so
much one can learn from even the creative use of aggregate
level data. Individual level data are required to establish
the nature of the mechanism driving the behavior recognized
by these scholars.


ACKNOWLEDGMENTS
I wish to express my gratitude to some of those who
helped make this dissertation possible. A complete catalogue
of all who have inspired, assisted, cajoled, or otherwise
pushed me into action is not possible. Let me try to thank a
few of you.
I am indebted to the members of my doctoral committee
for their helpful comments at various stages of this project.
To Peggy Conway, I must express particular gratitude for her
encouragement, criticism, inspiration, and administrative
heavy lifting.
I owe special gratitude to my family. My parents have
inspired me through their dedication to service. My wife,
Frannie, has provided unwavering support and timely
distractions, both of which were essential for the completion
of this project.
iii


Copyright
by
Randolph C
1993
Horn


CHAPTER 8
CONCLUSIONS: FARTHER ALONG
This chapter reviews the theoretical bases of this
dissertation and the central empirical findings. First, the
central theoretical positions in the racial threat literature
detailed in Chapter Two are discussed, followed by a brief
review of the essential tenets of the symbolic racism school
of thought presented in Chapter Three. After this, a
synopsis of the synthetic theoretical positions developed in
Chapter Four is laid out. The discussion of these
theoretical developments is followed by a review and
assessment of the hypotheses generated from those theories
and their tests. The chapter closes with the presentation of
the central conclusions and ramifications of this research.
Racial Threat
This dissertation assesses the theoretical strengths of
two major schools of thought in the political science
literature on white prejudice: the racial threat literature
and the symbolic racism literature. This section features a
brief review of the central theoretical tenets of the racial
threat literature. A number of representative works in this
field are reviewed in detail in Chapter Two.
153


54
this (self-interest) alternative hypothesis. Unfortunately,
the extensive tests provided leave lingering questions.
First, a couple of general methodological issues should
be addressed. In a number of studies, the proponents use a
number of tests to gauge the relative explanatory power of
self-interest and symbolic racism (Sears, Hensler and Speer
1979; Sears, Lau, Tyler and Allen 1980; Sears and Citrin
1982; Kinder and Sears 1981; McConahay 1982). Some of the
tests consistently present difficulties in interpretation.
While there is no reason to suspect an intent to obfuscate,
one finds a consistent pattern of modeling procedures which
could underestimate the effects of alternative hypotheses.
These three tests are discussed below. The first two types
of test involve problems associated with the comparison of a
scale to a collection of individual variables and the third
involves questions concerning the proper comparison of
populations.
First, a symbolic racism scale is compared with a number
of measures for self interest. For example, in Kinder and
Sears 1981 study, the effect of the symbolic racism scale is
compared with that of a handful of self interest indicators
in a regression equation predicting support for one of the
mayoral candidates in the 1973 election in Los Angeles. The
scale has a consistently significant effect in the predicted
direction while the other measures tend not to have
significant effects. Of course, comparing a scale to a
collection of, presumably, correlated measures presents some


38
partisan identification. These enduring predispositions
determine political attitudes, like issue positions and
candidate evaluations, and behaviors, like voting or
participation in demonstrations.
The impact of the symbolic predispositions attitudes or
behavior occurs to the exclusion of an individual's current
personal or economic situation. In this perspective, the
material context has little influence on the individual.
Indeed, the effects of actions based on self-interest (or
what people think their self-interest is) is largely
relegated to the private sphere.
Of course, this caricature of this theoretical
traditions may seem a bit far-fetched. A less exaggerated
example from the economic voting literature may provide a
more favorable reading. Some have thought that voters in
America engaged in pocket-book voting, making their electoral
choice in favor of the candidate most likely to benefit them
financially (Bloom and Price 1975; Tufte 1978) or closest to
them on the content of issues (e. g. Page 1977; Riker and
Ordeshook 1973). Students of symbolic politics have argued
that this is not the case (Kinder and Kiewiet 1979; 1981).
Rather than making their decision based on their individual
interest, voters cast their ballots based on how they think
the country as a whole is doing, sometimes called
'sociotropic voting'. Thus the choice revolves around which
candidate is best for the country, not around which candidate
is best for the individual. Indeed, Kinder and Kiewiet found


174
4)somewhat unlikely(234) 5)very unlikely(223)
6)NA(12) 8)DK(44) 9)NA(7) 0)INAP(864)
Valid percent giving most extreme response: 12.8; 42.9% in 1
plus 2.
V583 What are the chances these days that you or anyone in
your family won't get a job or promotion while an equally or
less qualified black employee receives one instead? Is this
very likely, somewhat likely, somewhat unlikely, or very
unlikely
l)very likely(104) 2)somewhat likely(240)
4)somewhat unlikely(233) 5)very unlikely(245)
8)DK(37) 9)NA(5) 0)INAP(864)
Valid percent giving most extreme response: 12.5; 41.4% in 1
plus 2.
Symbolic Racism scale items:
V559 Some people say that the civil rights people have
pushed too fast. Others say feel that they haven't pushed
fast enough. How about you: do you think that civil rights
leaders are trying to push too fast, are going too slowly, or
are they moving at about the right speed?
l)too fast(241) 3)about right(523) 5)
too slowly(72)
8)DK(20) 9)NA(16) 0)INAP(864)
Valid percent giving most extreme response: 28.8
V565 Now I am going to read several statements, as I did
before. After each one, I would like you to tell me whether
you agree strongly, agree somewhat, or disagree strongly with
the statement. As before, you can just give me the number of
your choice from the booklet, the first statement is...
Most blacks who receive money from welfare programs could get
along without it if they tried?
l)agree strongly(215) 2)agree
somewhat(298)
3)neither agree nor disagree(121)
4 jdisagree somewhat(159) 5)disagree strongly(54)
8)DK(17) 9)NA(8) 0)INAP(864)
Valid percent giving most extreme response: 25.4; 60.6% in 1
plus 2.
V568 Irish, Italians, Jewish and many other minorities
overcame prejudice and worked their way up. Blacks should do
the same without any special favors?
1)agree strongly(280) 2)agree somewhat(287)
3)neither agree nor disagree(105)
4)disagree somewhat(143) 5)disagree strongly(39)
8)DK(10) 9)NA(8) 0)INAP(864)


160
The only racial threat hypothesis not upheld involved
the relationship between the parity of income between the
races and the level of perceived racial threat (Hrt08)* It
was predicted that the level of perceived racial threat would
increase as incomes approached parity because increased
levels of competition would spur the perception that blacks
posed an economic threat to whites. This contention is not
supported in the multivariate models presented in Table 6-2.
Further, the significant coefficients indicate that the
reverse is true. There are two reasonable explanations for
this unexpected finding. First, income parity between the
races may represent a measure of interracial good will;
incomes are close because of the absence of prejudice. On
the other hand, such interracial goodwill may be the effect
of the similarity in income levels between the races, because
common income levels give whites and blacks more in common
and racial animosities are reduced. Neither the theory nor
data discussed in this study can provide a conclusive
explanation for this puzzle. Still, one may imagine research
strategies for addressing this issue. For example,
quantitative longitudinal studies linking the level of income
parity and the number and intensity of racially motivated
incidents in a number of carefully selected communities may
shed light on the dilemma. Similarly, qualitative cultural
and historical analysis of race relations in a few well
selected cases would also prove useful. Better yet, both


168
while some count the consistency with which the symbolic
racism scale explains a variety of issue positions (see Table
6-4, for example) as a strength, it may well point to the
concepts greatest weakness. Why should one expect a racial
attitude to explain a respondent's spending priorities for
space research? Why, if the scale really taps something
racial, does the scale not have remarkably more explanatory
power when the dependent variables have racial content? The
sensitivity of the racial scale to racial contextual effects
implies that there is some racial component to it, but one
does not get the same impression by viewing its effects.
Indeed, the consistent resistance to any spending may provide
an endorsement of the position taken by the primary
detractors of symbolic racism (e.g. Sniderman et al. 1985),
namely, that the concept represents conservatism, not a new
racial orientation. Evidence is not found that could
disprove this contention.
Again, the sensitivity to racial contextual effects
provides some evidence that there may be some racial content
in the concept, but the uniformity of effects raise questions
about the concepts validity. The efficacy of the measure in
explaining spending preferences may be a function of the
facility with which the constituent variables tap fiscal
conservatism. Further, the design of the questions
themselves may contribute to explaining the uniformity of
effects. The unfolding response format used to measure the
symbolic attitudes may be more stable by the force of the


144
prejudicial attitude, while conservatism is positively
associated with the same. Again, the effects are consistent
across all equation. It should be noted, however, that the
explanatory power of these variables is somewhat diminished
in this table. In equation one the quadratic term for the
contemporary racial composition, but not the untransformed
term, is significant indicating a dramatic relationship
between the level of perceived threat and the local racial
composition. This effect (again) vanishes when the measures
of archaic racial composition are included in the model
estimated in equation II. As above, however, the measures of
past racial composition fall short of the accept level of
statistical significance. The state percentage black for
1940 is significant at the p<0.10 level.
Equation III provides a different model. In this
equation, the archaic quadratic and state level terms are
dropped out and the other contextual variables are included.
In this model, the local percentage black for 1940
demonstrates a significant positive relationship with the
level of perceived racial threat. The other variables change
little. From this one gathers that the current level of
perceived racial threat is sensitive to cultural patterns
established in the middle of this century. This equation
also reveals an unexpected finding. The interaction term
between the percentage black at the state level in 1900 and
the average farm size in the state that year has a
significant negative relationship with the level of perceived


170
region should be included in theoretical models and can be
incorporated into empirical models.
Despite the technical limitations confronted in this
study, the evidence points clearly to the importance of
culture and historical context in conditioning and shaping
contemporary attitudes. While this study may have been
innovative in its use of the available contextual data, one
should not view its importance as merely technical. Rather,
the study provides empirical confirmation of the importance
of political culture in theoretical constructions.
Of course, there are limitations to the carrying
strength of this pronouncement. America is changing rapidly.
The communication revolution is changing our culture in as of
yet unforeseen ways. The American public is more transient
than ever; the acorn falls further from the tree than it did
in the past. Major demographic shifts are occurring: the
growth of the sunbelt, the growth of the aged as a proportion
of the population, and the increasing importance and numbers
of people of color. All these factors have changed and are
making America more diverse. These changes have the
potential to decrease the importance of local historical
factors. For the time being, however, those historical and
cultural factors have some force and can contribute to our
understanding of the nature of politics in America and
elsewhere.
Finally, even though mass attitudes not policy analysis
comprise the focus of the present study, this research has


25
the four studies find support for the threat hypothesis
(Giles 1977; Giles and Evans 1986; and Fossett and Kiecolt
1989). Giles and Evans's 1985 analysis marks the only
exception.
The Giles (1977) study attempts to analyze the effect of
racial composition on racial intolerance or hostility. Using
data from the 1972 ANES, Giles constructed a racial attitude
score for each respondent. The six questions used to
construct the scale covered a wide range of topics related to
race including the federal government's role in guaranteeing
equal opportunity, the federal government's role in
overseeing school desegregation, the role of "the government"
in assuring equal access to public accommodations, the pace
of the civil rights push, the trade off between whites' right
to maintain segregated neighborhoods and blacks' right to
live where they wish, and whether the respondent favors
desegregation, strict segregation or something in between.
He found that the expected relationship held, but only for
respondents in the South. The regional differences may have
been attributable, in part, to the heightened sensitivity of
Southerners at the time to federally mandated desegregation
programs; mandated desegregation of northern schools, for
example, did not come until later. Half the items in the
scale address the role of the government in mandating civil
rights reforms. Of course, even this heightened sensitivity
on the part of Southerners would follow from their greater
sense of threat from the government and ascendant blacks.


127
people. Of the contextual variables, only one has a
statistically significant effect in this model. The Southern
dummy variable is significant. Southerners are more likely
to rate blacks less favorably than the poor than are non-
Southerners. This is striking given the lack of such an
effect in Equation 1. The implication of this difference is
remarkable. Southerners may be more disposed to hold
prejudicial attitudes about blacks as a minority group, but
this generalized prejudice does not carry over to their
analysis of Jackson (and presumably other leaders) once other
factors are taken into account, namely their economic
vulnerability and relatedly their perception of racial
threat.
What are the effects of the racial attitudes scale in
predicting policy positions, namely federal spending
priorities? Respondents were asked whether federal spending
should be increased, decreased or kept the same in a range of
policy areas (see the appendix for question text). Since the
use of OLS regression on trichotomous variables is fraught
with liabilities a different strategy was adopted for the
models presented in Table 6-4. A single response category
was selected and coded one (1) while the other two response
categories where coded zero (0). In all cases, the response
'decrease federal funding* held the plurality of responses.
Coding this category one (1) maximizes the possible
variability in the resulting two categories. Recoding the
responses in this manner allows for the use of logistic


28
measure perceived threat. It was later adopted by Fossett
and Kiecolt (1989). Fossett and Kiecolt (1989) reanalyze the
same data used by Giles and Evans (1985), but their findings
differ. While Giles and Evans (1985) find that racial
composition is not generally linked to their threat scale,
the more recent study finds the expected relationship between
local racial composition and the threat scale, and not only
for Southern respondents. Fossett and Kiecolt (1989) found
the percent black to predict their measure of threat in the
total sample and in both the Southern and non-Southern
subsamples. The difference in their findings results from
differences in model specification, discussed later in this
chapter.
Despite its apparent usefulness, the scale is not
without problems. First, one might question the face
validity of the items. One could think that the civil rights
people are pushing too fast because of a strategic
orientation; perhaps a sympathetic respondent believes that
more economic power should be amassed before continued
political maneuvers proceed. Similarly, one may think that
any civil rights push is too fast regardless of one's
vulnerability to threats from ascendant blacks. A second and
related problem lies how the items have been perceived and
utilized by other scholars; this issue will be taken up in
the next chapter.
Fossett and Kiecolt (1989) also look for regional
differences in respondent's support for integration, but are


132
assist blacks even though the relationship is reversed when
the questions involve more general categories of
disadvantaged groups. The sense of vulnerability possessed
by those who perceived racial threats translates into the
adoption of racially prejudicial attitudes.
The same cannot be said of the symbolic attitudes. The
level of symbolic racism is not associated in any significant
way with evaluations of blacks leaders or blacks in general,
once the perception of threat is controlled. Further, the
relationships between symbolic racism and federal spending
priorities do not support the notion that there is a racially
prejudiced component to the concept. Symbolic racism is not
more associated with the desire to cut funding for programs
that target blacks than with a more general desire to cut any
program that helps (or targets) disadvantaged groups.
Further, the fourth equation indicates the positive
relationship between the level of symbolic racism and the
desire to cut funding for space research. Even conservatism
is associated with support for space research.
By all appearances symbolic racism seems not to hold
real racially prejudicial content; it has no effect once
perceived racial threat is controlled. Rather, symbolic
racism or more properly the questions used to measure it
appears to be a more accurate gauge of a particular component
of conservatism than anything inherently anti-black. In
particular, the questions measure the belief in self-reliance
well. However, respondents who claim that 'blacks could get


7
threatened as well. Indeed, the two orientations may exist
independently of each other in the same individual. These
constructs may reflect different dimensions of contemporary
racial attitudes. Individuals may evince high levels of
symbolic prejudice, or feel racially threatened, or some
combination of the two.
Similarly, one need not buy the entire theoretical
package of either camp. For example, one might easily
disagree with racial threat theorists who maintain that the
cause of racial conflict is the incompatibility of the
interests of whites and blacks. A political economist could
probably demonstrate that black ascendance poses little
threat to most whites. What is important from the behavioral
perspective is the perception of conflict or threat. Still,
the material context can influence the perception of threat.
For example, Green and Cowden's (1992) finding that
respondents with children who would be bused were more likely
to protest the busing plan in Boston implies that the
perceptions of threat may influence behavior dramatically,
even if it does not influence one's tendency to articulate
symbolically racist responses to survey questions. From this
literature, one may gather that the perception of threat and
symbolic racism may each have an independent force in
influencing different kinds of positions on issues and
evaluative behaviors. Rather than trying to supplant one
perspective with another, the argument presented emphasizes
recognizing the possibility that the two perspectives


135
interest. The models discussed thus far address the
prediction of the central tendencies of the racial attitudes.
Later in this chapter, models are constructed to assess the
effects of the contextual variables on the dispersion or
variance in the two racial attitudes.
Southern Models of Racial Attitudes
The analysis for the Southern sample is addressed first.
The models predicting the level of symbolic racism are
presented in Table 7-1. The results reflect some of the
essential findings reported in the previous chapter. For
example, equation I reveals that the individual level
variables of education and conservatism each have significant
explanatory power. Higher levels of education are associated
with decreases in the level of symbolic racism. Higher
levels of conservatism are associated with higher levels of
symbolic racism. These effects change little from the
splitting of the sample or the addition of other variables in
equations II and III.
The first equation indicates that the contemporary
percentage black has a positive relationship with the level
of symbolic racism. The negative value in the quadratic term
for that variable indicates that the relationship is
curvilinear.1 The relationship between the contemporary
1 The local maximum may be calculated by applying the formula
-2a/b, where a is the parameter estimate (not reported) for
the percentage black and b is the parameter estimate (not
reported) for the quadratic term. (Giles and Evans 1986)


138
between the percentage black at 1940 and 1980 masks the
independent effects of each.
A similar estimation strategy for Southerners'
perceptions of racial threat produced the results presented
in Table 7-2. The results are similar to those presented in
Table 7-1. Education demonstrates a consistently
ameliorative effect on the level of perceived threat across
the four equations. Similarly, conservatism consistently
aggravates the level of perceived racial threat across all
four equations. The contemporary racial composition has an
almost identical effect in predicting the level of perceived
threat vis-a-vis its effect in predicting the level of
symbolic racism. As before the relationship is curvilinear
with a local maximum of 75%. The central effect of the local
racial composition is conveyed by the non-quadratic term;
there is a significant positive relationship between the
contemporary percentage black and the level of perceived
racial threat. The marginal decline in the effect occurring
after the percentage black reaches 75% is insufficient to
vitiate the basic relationship; higher levels of perceived
racial threat are associated with higher percentages black in
the respondent's locality.
The addition of control variables and the archaic
measures of racial context do little to improve the
predictive power of the model presented in equation I. As in
the previous table, the effect of the contemporary racial
composition falls slightly out of the range of accepted


98
Nativity is associated with the influx of people and the
influx of newer and more diverse ideas. This measure is
calculated by dividing the number of residents born in the
state by the total population for that area. The result is
the proportion native to that area. One expects areas with
more in-migration to be more racially progressive.
1900 and 1940 Censuses
In addition to the variables drawn from the 1980 census
other census data are exploited. One expects the weight of
the past to continue to affect contemporary racial attitudes.
Demographic data drawn from previous eras are expected to
provide a gauge of the existence of such legacies (see
previous chapter for a more detailed discussion of this
topic.) Within the catalogue of extant censuses, two years
are of particular theoretical import. The 1900 census
represents the first comprehensive accounting of the nation's
population after the overthrow of the regime of the
reconstruction era in the South. In this sense, data from
this year represents a baseline of data for the Jim Crow era.
Furthermore, these data predate changes accompanying the
demise of the cotton economy.
The 1940 census provides the most recent counting of the
nation's population before the commencement of large scale
protest and other civil rights activities garnered extensive
national coverage. This argument follows Wrights (1977)
contention that the desegregation of the military after World


104
is controlled. This hypothesis and others will be explored
more thoroughly in the following chapters.
Endogenous variables
While the primary focus of this project concerns the
theory and measurement of racial orientations, it is useful
to demonstrate that these racial orientations influence
broader political concerns. That is to say, it is important
to demonstrate that racial orientations influence the
individuals' political lives and through their actions the
greater community. As such, it seems appropriate to attempt
to measure the effects of racial orientations on policy
positions and other political evaluation. The usefulness of
perceived threat and symbolic racism as explanatory variables
is gauged on two types of dependent variables taken from the
1986 ANES (question text in the Appendix): Policy positions
and candidate evaluations.
For the sake of comparison, symbolic racism and
perceived racial threat will be used to predict a variety of
issue positions such as federal spending on minorities, the
unemployed, food stamps, and space research. All of these
are available in the 1986 ANES.
Similarly, since 1986 was not a Presidential election
year, the 1986 ANES does not include candidate evaluations
for presidential candidates. It does contain feeling
thermometers for national political figures. It is expected
that racial orientations will have differential effects on


26
Unfortunately, he was not able to tap perceived threat
directly. Still, his findings support the notion that the
perception of a threat is an intervening step between the
context, individual attributes and political attitudes and
behaviors. In this case, the articulation of intolerance
provides an additional clue of the existence of perceived
threat at the individual level.
The Giles and Evans study of 1985 uses data from the
1972 ANES and marks a departure from the earlier findings in
two ways. First, they find what is on first appearances a
viable measure of threat. In that study, respondents were
asked if they thought the civil rights people were pushing
too fast, too slow, or at the right pace. Additionally, they
were asked if black people had too much, too little or the
right amount of influence (See the Appendix for actual text).
These two items were combined to make an additive scale of
threat. So, respondents who thought the movement was going
too fast and thought blacks had too much influence scored
highest on the threat scale.
The second way in which this study marks a departure
from previous ones lies in its findings. Giles and Evans
(1985) argue that there is no significant relationship
between local racial composition (percentage black) and
perceived threat. Additionally, this non-relationship holds
up even when regional differences are accounted for. This
would seems to represent a chink in the armor of the
proponents of the threat hypothesis. But appearances can be


CHAPTER 6
NATIONAL MODELS OF PREJUDICE AND ITS EFFECTS
In the last chapter discussion focused on two related
issues: the data used in this project and its appropriateness
for the same. Special attention was paid to identifying
sources of bias. Additionally two theoretically distinct
racial dimensions were identified. This chapter builds on
the last by modeling the sources of those racial attitude
scales using multivariate procedures. Similarly, the effect
of the racial scales on other political attitudes and issue
positions is gauged through multivariate procedures.
Procedures
The last chapter discussed the construction of two
racial attitude scales. The ranges of these scales allow the
use of Ordinary Least Squares (OLS) regression.1 OLS allows
measurement of the independent effects of salient variables
while controlling for the effects of other salient variables.
Another advantage of OLS is its ease of interpretation and
familiarity to many readers. The scales are modeled in
similar fashion:
1 The threat scale ranges from 5 to 26 and has 729 valid
cases. The symbolic racism variable ranges from 4 to 20 and
has 809 valid cases. Directionality of the scales was
corrected by taking the inverse of each (dividing each summed
value into one and multiplying by 100).
107


148
Northerners (at the bottom). For both Northerners and
Southerners, the mean level of symbolic racism is associated
with increases in the amount of variation within each county.
For each region, one additional variable sheds light on the
construction and dispersion of racial attitudes. In the
South, the percentage black in 1940 is negatively associated
with the standard deviation of the symbolic racism scale. In
other words, the percentage black from 1940 is associated
with greater unity of opinion within the South on this racial
scale, as Key might have predicted. The standard deviation
of symbolic racism in the North also shows sensitivity to a
contextual variable. The higher the contemporary percentage
black the greater the variation in the level of symbolic
racism in the non-Southern counties. In other words,
respondents in areas with higher percentages of blacks imply
a greater diversity of opinion on racial issues.
Similar analysis using the standard deviation in the
level of perceived racial threat is presented in Table 7-6.
For both Northerners and Southerners alike, the standard
deviation in perceived racial threat increases with the mean
level of perceived racial threat. That is to say, that as
the average level of perceived racial threat increases, the
range of opinion also increases. For Southerners one other
contextual variable makes a significant impact of the
dispersion of the perception of racial threat. Equation III
reveals a positive relationship between the level of income
equality and the dispersion of perceived threat. The closer


63
Multi-dimensionality
The notion of multi-dimensional racial attitudes dates
from the 1950s (e.g. Rokeach 1960; Allport 1954). Many
studies identified the various dimensions through the use of
intensive personality trait questionnaires. This method
resulted in detailed measurements of the respondent's racial
attitudes. The studies, however, failed to demonstrate the
relevance of the scales to any external conditions or to
demonstrate the impact of those attitudes on political
behavior or attitudes.
Other studies have excelled in measuring people's
political orientations and evaluations of candidates. For
example the American National Election Studies (1952-1992)
conducted by the Institute for Social Research at the
University of Michigan have consistently measured many
fundamental political orientations, behaviors,
characteristics and attitudes. Of course, every question a
secondary or even a primary researcher might want to ask
cannot be asked. Few data sets can tap multiple dimensions
of racial attitudes and provide measures of political
phenomena against which one may judge the effects of the
racial orientations. Hence, most studies cannot take this
sort of approach. The proponents of symbolic racism
acknowledge the existence of multi-dimensionality (see
discussion in Chapter Three), while others are less explicit
in their treatment.


163
are blacks than it does for other objects. Despite the
satisfaction of those hypotheses concerning federal spending
priorities, one is left with lingering doubt about the
usefulness of the concept in explaining racial attitudes.
Indeed, the findings presented in Table 6-4 imply that the
measure of symbolic racism could as easily represent a
measure of fiscal conservatism.
In the next section the synthetic hypotheses will be
evaluated.
Synthetic Hypotheses:
While the hypotheses presented in the two previous
sections were derived from purist readings of the literature
in the respective schools of thought, the hypotheses
discussed in this sections result from critical reading of
both literatures and in that sense are synthetic.
For example, the notion of multi-dimensionality does not
fit comfortably into the racial threat literature and only
fits in the symbolic racism literature to the extent that
symbolic racism is thought to replace old-fashioned racism.
Table 5-2 demonstrates that two racial attitudes need not
reflect the same attitudinal dimension (HsynOI)* To the
extent that the two racial scales are actually measuring
racial attitudes, those attitudes enjoy some independence
from each other. Individuals may have variable levels of
each attitudes. One may be disposed toward symbolic
prejudice and yet not feel threatened or visa versa.


American National Election Studies (ANES) to assess the
influence of racial composition and other contextual
variables on perceptions of racial threat and symbolic racial
attitudes. These orientations, in turn, are used to predict
issue positions and evaluations of candidates.
The analysis are conducted on two levels. First, the
general hypotheses of interest are tested at the national
level. Factor analysis confirms that the two racial
attitudes occupy different dimensions. However, both
perceptions of threat and symbolic racism are influenced by
contextual measures like the local percentage black. The
racial measures differ in their ability to predict issue
positions or evaluations of groups or candidates; perception
of threat is a better predictor of candidate and group
evaluations, while symbolic racism is a consistent predictor
of federal spending preferences. A second level of analysis
applies the same techniques to compare the South with other
regions; the same factors which explain differences in racial
attitudes between the regions do not explain marginal
differences within the South. The dissertation demonstrates
empirically the importance of local history and political
culture in explaining contemporary political and racial
orientation.


13
pursuit of their goals. The solidarity of elites within the
Democratic party stifled class-based or populist efforts to
address the agenda of the have-nots, be they white or black.
The one-party south and the techniques of control used by its
elites proved effective in preventing the development of
class based alliances between blacks and poor whites.
In addition to this basic set of relationships, Key
stresses the importance of the local racial composition.
Recall Key's statement that, "[wjhatever phase of the
southern political process one seeks to understand, sooner or
later the trail of inquiry leads to the Negro." (p. 5) The
desire on the part of whites to control blacks is greatest,
so the argument goes, in the those areas in which blacks make
up a larger proportion of the population, namely the black-
belt. Planters sought to maintain the system which provided
cheap labor for their estates. Whites outside the black-belt
were, hence, less concerned about the maintenance of white
supremacy. So, it was in the black belt that the subjugation
of blacks and the enforcement of a unified front on the race
question by white elites was most effective. In some states,
Key uses maps to demonstrate the strong correlation between
those counties with more than 50% black and strong support
for particular candidates and high levels of turnout. The
political impact of the local racial composition was potent.
While this description bears some resemblance to the
broad view of Key's canvas, there were of course variations.
Some populists, like "Big Jim" Folsom in Alabama, enjoyed


27
deceiving. The Giles and Evans (1985) finding should not be
accepted uncritically. Later efforts, discussed below,
revealed shortcomings in their modeling procedures. Still,
they served the academy by stimulating discussion of this
issue.
Additionally, Giles and Evans (1985) established a
relationship between the respondent's racial attitudes and
his or her level of identification with white people as a
group; the more hostile the respondents were, the greater
their attachment to whites. In other words, the sense of
racial polarization was evident not only in terms of people's
positions on about the civil rights people, or black people,
but also in terms of how they saw themselves. Similarly,
those who identified British ancestry, as opposed to those
who identified no specific ethnic heritage, felt warmer
toward white people. So, identification with whites as a
reference groups was enhanced by both the Giles and Evans
(1985) racial threat scale and by identification with some
other group, in this case the British. One caveat about
these findings is in order. Giles and Evans (1985) did not
include controls for region in the model estimating
attachment to whites. This may present a significant source
of bias if a greater proportion of Southerners trace their
ancestry to Britain than do Northerners, a not altogether
remote possibility.
The scale developed by Giles and Evans (1985) represents
the most serious attempt by political scientists to date to


131
that aid blacks.4 Evidently the feelings of economic
vulnerability held by those with higher levels of perceived
racial threat make them more favorably disposed to approve of
programs that help those who, like themselves, may need a leg
up. This apparent benevolence to programs designed to help
the poor does not extend to programs designed to help blacks
specifically. In this way, the racial component of their
vulnerability comes into play. Those with higher levels of
perceived racial threat are more likely to support cuts in
programs that help blacks than others. Also in this model
the contemporary racial context is associated with support
for cuts in funding for programs that aid blacks. The local
racial context has an effect not captured by the measured
components of racial prejudice.
The models presented in Tables 6-3 and 6-4 endorse the
validity of the perceived racial threat concept. The level
of perceived racial threat is related to anti-black
evaluations of candidates and blacks. This anti-black affect
is further evidenced by the relationship between federal
spending priorities and the level of perceived threat. The
level of perceived threat is associated with the anti-black
position on questions involving the support of programs that
4 The parameter estimate for perceived racial threat (0.03)
has a p-value of about 0.11, just over the 0.10 limit for a
one-tailed test. While statements made about this estimate
must be more tentative than statements made about estimates
with lower p-values, ignoring the estimate because it is
0.01 over the limit is tantamount to throwing the baby out
with the bath water.


55
difficulties. Assuming the measures for self-interest were
valid, they might not appear as significant while appearing
in an equation with other measures of the same concept; the
explained variance may have been spread thinly over the
multiple indicator so thinly that no individual measure had a
statistically significant effect. In other words, each of
the parameter estimates for the measures of self interest
report the effect of that measure of self interest while
controlling for the effects of other measures of the same
concept. Finding even a small effect may have theoretically
significant ramifications (see for example, King 1986),
namely that self-interest makes some contribution.
A second kind of test is used to compare the predictive
abilities of the two concepts. For example, in their table
3, Kinder and Sears (1981) estimate a regression equation
with the symbolic racism scale as the sole independent
variable. They note the adjusted r2, the percentage of
variance in the dependent variable explained by all the
independent variables adjusted for the total number of
variables in the model. Next they add the several measure of
self-interest to the model and note the adjusted r2 for the
total model including the symbolic racism scale and the
measures of self-interest. By subtracting the first r2 from
the second they can gauge the additional amount variance
explained by the measures of self-interest. But herein lies
the rub. In the calculation the adjusted r2, symbolic racism
is treated like a single variable while the collection of


BIOGRAPHICAL SKETCH
Randolph C. Horn was born in Opelika, Alabama and raised
in Birmingham, Alabama. He graduated from the University of
the South in Sewannee, Tennessee in 1987 with a Bachelor of
Arts Degree in Third World studies and political science. He
received his Master of Arts degree from the University of
Florida in 1990. He has managed to keep his sense of humor
despite extensive graduate study.
184


45
feelings underlying old fashioned or red-neck racism, but it
differs from them in its other psychological roots and in
many of its specific forms of expression." (McConahay and
Hough 1976, p. 24) A similar claim is made in a later study
by McConahay (1982, p. 705) A 1979 study's factor analysis
found three racial dimensions, but since all were correlated
with anti-busing attitudes they were treated as one dimension
(Sears, Hensler and Speer 1979, p. 382n).
The original claims of newness have been softened
somewhat. The proponents do maintain that this expression of
racial attitudes is a post civil rights era phenomenon, that
it is distinct from other racial attitudes, and that it is
more important.
Combination or Multi-dimensionality
While most proponents of symbolic racism agree that
traditional values and anti-black affect are ingredients of
symbolic racism, there is some dissension as to the exact
nature of the relationships between the concepts. Kinder
recently claimed that symbolic racism is the 'conjunction' of
these values and racial prejudice. (1986) The precise
relationship between the two elements has never been
specified. McConahay and Hough (1976) stress this facet of
symbolic racism by calling attention to a number of
constituent elements. For McConahay and Hough (1976),
symbolic racism has blacks as a group as its object, a moral
tone (p. 36-7), addresses issues surrounding the violation of


167
abstract, somehow distinct from the personal lives of people
or the contexts in which they live. Both the contemporary
and the archaic racial context influence the symbolic
attitudes. Additionally the attitude was subject to
significant regional variation, although the theory expects
none due to the principle of abstractness.
In the face of this evidence, it seems reasonable to
temper the claims of abstractness. Such tempering would
bring the notion of symbolic racism more in line with
existing theory and research in symbolic politics. Indeed,
other research by the major proponents of symbolic racism in
the area of symbolic politics seems to accommodate context
and other non-psychological constructs such as peer
influences (adult socialization) quite comfortably (see for
example, Kinder and Rhodebeck 1982). Against this backdrop,
the tenacity with which the proponents of symbolic racism
cleave to the notion of abstractness seems ill-placed. Could
it be that whites are sensitive to the events and history in
their locale and to the influences of friends and co-worker
and yet still remain unsympathetic to the plight of the
blacks they imagine living on welfare in the inner city?
This renovated construction of the principle of abstraction
yields a better fit with the available evidence.
The evidence presented in this frontal assault may seem
more damaging than it is. After all, the slight theoretical
adjustment noted above deflects the lion's share of that
assault. A more troubling issue lies in the gray areas.


52
Sears (1988)6, contains none of the contested measures and is
drawn from a wide range of studies. One must conclude that
these items have some measure of face validity for Sears,
Kinder, McConahay and the other advocates of the concept.
Upon review of Sears' list of the currently accepted measures
for symbolic racism, one observes that the questions have one
fundamental thing in common. They all have blacks as their
objects. This marks a point of theoretical consistency. If
the concept of symbolic racism has at its core the notion
that 'blacks violate traditional American values' then it
follows that survey questions meant to tap that core should
have blacks as their focus. Beyond this, some of the
questions in this list clearly indicate specific values, in
particular self-reliance, or getting ahead without government
assistance. Other questions not on this list show remarkable
face validity in terms of their apparent ability to capture
both antiblack affect and values such as self-reliance. The
"blacks could get ahead like the Jews, Irish and Italians"
question is a particularly good example of this (see the
appendix for the text of questions used in this study). These
and other questions will be discussed at greater length in
the chapter on methods.
6 Kinder (1986, pp. 156-7) provides a more exhaustive list.
The shorter list showcased here is adequate to give the
flavor of the measures.


142
measure of contemporary racial composition does not have a
significant effect in predicting the level of symbolic
racism, but the quadratic term does. This indicates that the
relationship between the local racial composition and the
level of symbolic racism is dramatic; as the percentage black
increases so too does its marginal effect on the level of
symbolic racism.
When the measures of archaic racial composition are
added the apparent effect of the contemporary racial
composition vanishes, but the effect of the archaic measures
fall short of significance (p<0.11 for the quadratic term,
and p<0.22 for the untransformed term--not shown). From this
we may gather that the archaic racial composition has the
bulk of the explanatory power. In analyses not presented
here, it is apparent that the effect of the archaic racial
composition is suppressed unless the contemporary racial
composition is controlled. In other words, if we control for
the current racial context, the percentage black in 1940
becomes a near significant predictor of the current level of
symbolic racism. Unfortunately, due to the dwindling number
of cases, this analysis must be presented with some
hesitancy.
Similar analysis of the models predicting the level of
perceived racial threat among non-Southern respondents is
presented in Table 7-4. Again, the same relationships
between education and conservatism and the racial attitude
are evident. Education is negatively associated with the


19
does not indicate that race must always have an effect.
Indeed, by pointing to a specific contest in which race
baiting was particularly effective, one is pushed to
recognize that race baiting is less effective in other
contests or was not employed at all.
Aggregate Analyses
Huckfeldt and Kohfeld (1989) identify the macro level
relationship and pose potential solutions for the
disequilibrium afflicting the party system. They do not, nor
do they claim to, nail down the operative mechanism driving
the white flight from the Democratic coalition. Some recent
papers have added support to this line of inquiry.
Two analyses of recent elections shed some light on the
aggregate effects of local racial composition in electoral
contests. First, Hertzog (1991) made creative use of county
level returns in his analysis of African-American Douglas
Wilder's 1989 election as governor of Virginia. At first it
seems counter intuitive that the first elected black governor
would come from a Southern state, especially the Southern
state with the lowest percentage of black population of all
Southern states. Hertzog's (1991) analysis, influenced by
Key (1949), argues that this low percentage black actually
facilitated Wilder's success. He presents compelling
evidence to substantiate an inverse relationship between the
percentage of African-Americans and the proportion of the
white vote falling in Wilder's column.


140
significance levels when other correlated variable are added
to the model (see equation II). The simple removal of the
state percentage black for 1940 reduces colinearity
sufficiently to push the contemporary racial context back
toward significance. The marginal effects of the hybridized
variable are less pronounced in these models than in the
national models presented in the previous chapter. As above,
the explanation for this probably lies in the fact that the
division of the sample by region removes the stark relief
presented by the dramatic cultural differences between those
regions.
Northern Models of Racial Attitudes
Parallel analysis for the non-Southern cases (called
Northern in tables) is presented in Tables 7-3 and 7-4. The
equations modeling symbolic racism are presented in Table 7-
3. Here the relationship between education and the level of
symbolic racism holds similar magnitude and identical
direction to that of the Southern sample. Education
consistently and significantly ameliorates the level of
prejudicial attitude across all three equations. Similarly,
conservatism aggravates the level of symbolic racism. Higher
levels of conservatism are associated with higher levels of
symbolic racism across all equations. Age also displays a
moderately positive relationship with the level of symbolic
racism (see equation I) but this relationship diminishes with
the addition of contextual variables. The untransformed


50
symbolic politics hypotheses which happened to include racial
measures. For example Sniderman and Tetlock (1986) criticize
Sears, Hensler, and Speer (1979) and Sears, Lau, Tyler and
Allen (1980) for combining measures of old fashioned and
symbolic racism in these studies. Kinder 1986 and Sears 19885
respond that those studies do not measure symbolic racism but
symbolic politics generally. Rather than dwelling on
contested measures, what follows is a discussion of the
current concepts and measures most widely accepted by the
proponents of the theory, setting up a rationale for the
measures used in the present study.
While symbolic racism is the conjunction of antiblack
affect and traditional values, Sears argues that it manifests
itself in two ways: "(a) antagonism toward blacks' "pushing
too hard' and moving too fast, especially (though not
exclusively) through the use of violence and (b) resentment
toward special favors for blacks . ." (Sears 1988, p.
56) These manifestations, more or less, reflect the
constituent theoretical parts. For example, a belief in
self-reliance combined with antiblack affect may make one
more disposed to latch onto stereotypes concerning welfare.
Table 3.1 below list examples of the operationalization of
these concepts in a series of symbolic racism studies
conducted between 1976 and 1985. This list, compiled by
5 Sears (1988) claims that these two studies did not measure
symbolic racism on page 64 after crediting those same
studies as examples of the concept's efficacy on page 59.


RACIAL COMPOSITION AND RACIAL THREAT
THE ANATOMY OF CONTEMPORARY RACISM
BY
RANDOLPH CLAIBORNE HORN
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSPHY
UNIVERSITY OF FLORIDA
1993

Copyright
by
Randolph C
1993
Horn

ACKNOWLEDGMENTS
I wish to express my gratitude to some of those who
helped make this dissertation possible. A complete catalogue
of all who have inspired, assisted, cajoled, or otherwise
pushed me into action is not possible. Let me try to thank a
few of you.
I am indebted to the members of my doctoral committee
for their helpful comments at various stages of this project.
To Peggy Conway, I must express particular gratitude for her
encouragement, criticism, inspiration, and administrative
heavy lifting.
I owe special gratitude to my family. My parents have
inspired me through their dedication to service. My wife,
Frannie, has provided unwavering support and timely
distractions, both of which were essential for the completion
of this project.
iii

TABLE OF CONTENTS
ACKNOWLEDGMENTS iii
LIST OF TABLES vi
LIST OF FIGURES viii
ABSTRACT ix
CHAPTER 1 1
INTRODUCTION 1
Recent Trends in the Study of Prejudice 1
Perceptions of Threat 2
Symbolic Racism 4
Synthesis 6
CHAPTER 2 10
PERCEPTIONS OF RACIAL THREAT: THEORY AND MEASUREMENT 10
Foundations 11
The Contemporary Significance of Race at the Macro
Level 15
Aggregate Analyses 19
Individual Analyses 24
Reconciliation 32
CHAPTER 3 36
SYMBOLIC RACISM: THEORY AND MEASUREMENT 36
Symbolic Politics in a Nutshell 37
Theoretical Underpinnings of Symbolic Racism 39
Newness or Distinctiveness from Old Fashioned Racism 42
Combination or Multi-dimensionality 45
Abstraction vs. Self Interest 47
Methods and Measurement 49
Self Interest: The Alternative Hypothesis 53
Data 58
CHAPTER 4 61
SYNTHESIS 61
Multi-dimensionality 63
Defining the Nature of Interests 71
Synthesis
IV

Context and the Changing Meaning of Racial
Composition 75
The Archaic Significance of Racial Composition 83
CHAPTER 5 89
DATA 89
Data 89
Dependent Variables 91
Contextual Variables 95
1980 Census 96
1900 and 1940 Censuses 98
Endogenous variables 104
CHAPTER 6 107
NATIONAL MODELS OF PREJUDICE AND ITS EFFECTS 107
Procedures 107
Models Explaining the Origins of Racial Attitudes Ill
Racial Attitudes and Other Political Orientations 121
CHAPTER 7 134
REGIONAL ANALYSIS 134
Southern Models of Racial Attitudes 135
Northern Models of Racial Attitudes 140
Regional Differences in Attitudinal Unity 146
CHAPTER 8 152
CONCLUSIONS: FARTHER ALONG 152
Racial Threat 152
Symbolic Racism 154
The Synthetic Theoretical Position 155
Assessing the Derived Racial Threat Hypotheses 157
Derived Symbolic Racism Hypotheses: 160
Synthetic Hypotheses 162
Conclusions 165
APPENDIX 172
Racial Threat scale items: 172
Symbolic Racism scale items: 173
Coding of other items: 174
Data Derived from the Census 175
BIOGRAPHICAL SKETCH 184
v

LIST OF TABLES
Table 5-1. T-test for race of interviewer bias on
whites 93
Table 5-2. Factor analysis for racial items among
whites with white interviewers 94
Table 5-3. Black Concentration in 1900, 1940, and 1980
and the levels of Perceived Racial Threat and Symbolic
Racism in 1986 for White Respondents Only 102
Table 6-1. Symbolic Racism Regressed on Contextual and
Individual Level Variables for White Respondents
Only 112
Table 6-2. Perceived Racial Threat Scale Regressed on
Contextual and Individual Level Variables for Whites
Only 118
Table 6-3. Regression of Evaluation of Candidate and
Group Objects on Racial Attitudes for White
Respondents Only 123
Table 6-4. Logistic Regression of Decreased Federal
Spending on Racial Attitudes for White Respondents
Only 129
Table 7-1. Symbolic Racism Regressed on Contextual and
Individual Level Variables for White Southern
Respondents Only 136
Table 7-2. Perceived Racial Threat Scale Regressed on
Contextual and Individual Level Variables for White
Southerners Only 139
Table 7-3. Symbolic Racism Regressed on Contextual and
Individual Level Variables for White Northern
Respondents Only 141
Table 7-4. Perceived Racial Threat Scale Regressed on
Contextual and Individual Level Variables for White
Northerners Only 143
Table 7-5. Standard deviation of Symbolic Racism
Regressed on Contextual Variables in Each County 147
vi

Table 7-6. Standard Deviation of Perceived Racial
Threat Regressed on Contextual Variables in Each
County
149
Vll

LIST OF FIGURES
Figure 4-1. Matrix of racial attitudes 65
Figure 4-2. Quadratic function with increasing slope 78
Figure 4-3. Parabolic quadratic function 80
Figure 4-4. Quadratic function with gradual decay 82
viii

Abstract of Dissertation Presented to the Graduate
School of the University of Florida in Partial Fulfillment of
the Requirements for the Degree of Doctor of Philosophy
RACIAL COMPOSITION AND RACIAL THREAT:
THE ANATOMY OF CONTEMPORARY RACISM
By
Randolph Claiborne Horn
December, 1993
Chairman: M. Margaret Conway
Major Department: Political Science
While the use of thinly veiled racial issues in recent
elections serves as a reminder of the salience of race in
American politics, scholars disagree about the causes and
effects of white racial attitudes. The proponents of
symbolic racism and group conflict approaches have engaged in
a debate over which single approach best explains racism.
This dissertation demonstrates that symbolic attitudes and
perceptions of racial threat occupy different dimensions in
the construction of white racial attitudes. Racial attitudes
do not exist in a vacuum; the components and the effects of
those attitudes are tempered by historical factors and the
contemporary racial context. Further, each dimension has a
differential impact on political behavior.
This project combines composition data from the 1900,
1940, and 1980 censuses with survey data taken from the 1986
ix

American National Election Studies (ANES) to assess the
influence of racial composition and other contextual
variables on perceptions of racial threat and symbolic racial
attitudes. These orientations, in turn, are used to predict
issue positions and evaluations of candidates.
The analysis are conducted on two levels. First, the
general hypotheses of interest are tested at the national
level. Factor analysis confirms that the two racial
attitudes occupy different dimensions. However, both
perceptions of threat and symbolic racism are influenced by
contextual measures like the local percentage black. The
racial measures differ in their ability to predict issue
positions or evaluations of groups or candidates; perception
of threat is a better predictor of candidate and group
evaluations, while symbolic racism is a consistent predictor
of federal spending preferences. A second level of analysis
applies the same techniques to compare the South with other
regions; the same factors which explain differences in racial
attitudes between the regions do not explain marginal
differences within the South. The dissertation demonstrates
empirically the importance of local history and political
culture in explaining contemporary political and racial
orientation.

CHAPTER 1
INTRODUCTION
Contemporary studies of white racial prejudice have
tended to concentrate exclusively on one of two factors in
explaining racial attitudes. Proponents of the racial threat
explanation1 look to contemporary demographic and economic
conditions, while proponents of symbolic explanations argue
that values learned early in life explain current levels of
prejudice independent of the individuals current demographic
or economic context. This dissertation makes a case for
augmentation and synthesis of these two schools. It is
argued that individuals may simultaneously hold more than one
kind of racial orientation. Further, racial attitudes are
not only the product of either psychological traits or the
current economic context. Rather, racial attitudes result
from both of these factors and are conditioned by local
historical and cultural conditions.
Recent Trends in the Study of Prejudice
Events in recent years have spelled disaster for the
hard fought gains of the civil rights movement. Supreme
1 Racial threat may be defined as the belief on the part of
whites that black ascendance reduces the chances of white
success in competition for jobs, admissions to schools or
other areas of achievement; it may also represent fears on
the part of whites that blacks pose challenge the security of
whites' persons or property.
1

2
Court decisions have made it easier to discriminate and have
whittled away at affirmative action programs. Extremist
groups, such as those led by Tom Metzger and manned by skin
heads, seem on the rise. Indeed, the number of hate groups
nation-wide increased 27% between 1990 and 1991. (Southern
Poverty Law Center 1992) The qualified electoral success of
David Duke in Louisiana seems to scare everyone, except of
course his supporters.
While these events point to an increasingly engaged
white backlash, other events seem to offer contradictory
signals. Douglas Wilder was elected Governor of Virginiaa
feat which required substantial white support. Throughout
America the number of African-American elected officials has
been increasing steadily since 1970, largely through the
efforts of black candidates and voters. (Joint Center for
Political Studies 1988)
Perceptions of Threat
What explains these seemingly contradictory trends? One
starting point is the white mind. Both Wilder and Duke have
been supported by white voters. This support may spring from
the same pattern of racial orientation. The notion of racial
threat is exemplified by whites who feel that their jobs,
admission to schools, or lifestyles are in jeopardy because
of African-American advances. More generally, one may define
the perception of racial threat as the belief on the part of
whites that blacks:

3
1) as a class or as individuals may detract from whites'
economic well being by providing competition for
jobs that would otherwise go to whites;
2) present challenges to physical well being (i.e. that
criminal activity engaged in by blacks will endanger
their persons or property); or that
3) present a serious challenge to whites' political
standing (e.g. that black officials or officials
elected by blacks will wrest control of the ship of
state from whites and use the state apparatus to the
disadvantage of whites).
Perhaps low levels of perceived racial threat allowed some
Virginia whites to vote for Wilder. Similarly, high levels
of perceived racial threat fueled support for Duke in
Louisiana. Studies by Hertzog (1991) and Giles and Buckner
(1993) support this notion by finding a relationship between
local racial composition and support for candidates at the
district level. These studies echo the findings of earlier
scholars (Key 1949, Blalock 1957, Allport 1954, Pettigrew
1959, and Giles 1977).
While both Hertzog (1991) and Giles and Buckner (1993)
claim the relationship between racial composition and support
for these candidates is a function of racial threat, they are
unable to substantiate this because they lack individual-
level data; they do not demonstrate that whites actually feel
threatened. While from a behavioral standpoint it is only
necessary to demonstrate that individuals perceive a threat
(or feel threatened), the position adopted by Giles and
Buckner and others posits that the material interests of
whites actually are threatened; the necessity of this
contention is assessed in the next chapter. Using

4
individual-level data to construct a racial threat scale
would facilitate the study of the relationship between
perception of racial threat and political behavior. A
number of other studies support the salience of white
perceptions of racial threat and the effects of local racial
composition. (Fossett and Kiecolt 1989, Giles and Evans
1985, 1986)
The difficulty encountered by those who hold theoretical
stock in the racial threat concept is understandable.
Relatively few of the national surveys contain items that
would allow measurement of such a notion. Still, the 1986
American National Election Study (ANES) contains an extensive
battery of racially oriented questions. Many of these items
may be used to measure perceptions of racial threat. (See
the Appendix for actual text and marginals). The theoretical
ground staked out by the proponents of the racial threat
framework may be measured, surveyed or tested with individual
level survey data which is widely available.
Symbolic Racism
Despite a number of papers indicating the salience of
perceptions of racial threat and local racial composition in
influencing white attitudes and voting behavior, the lion's
share of research on white racial attitudes has situated
itself in the symbolic racism paradigm (Kinder and Rhodebeck
1982, Kinder and Sears 1981, 1985, McConahay 1982, McConahay
and Hough 1976, Sears, Hensler, and Speer 1979). These

5
authors share a common perspective based on a few simple
premises. First, the changes wrought in America in the 1950s
and 1960s have changed the nature of racial attitudes held by
most Americans. It is no longer socially acceptable to hold
openly racist positions, especially positions based on
segregation or white supremacy. So, old fashioned racism,
defined as the belief in segregationist values, cannot be
advocated openly.
The second premise maintains that racism has a new form
of expression. Whites now use their adherence to
traditional, conservative values to legitimize anti-black
positions. This allows them to explain prejudicial feelings
and beliefs in terms of acceptable ideological stances. The
use of these values is independent of the individual's
demographic or economic context. In other words, one's class
or location and proximity to blacks have no effect on one's
racial attitudes.
This orientation is commonly termed 'symbolic racism,' a
combination of anti-black affect and traditional values which
(according to its proponents) has replaced the old racism
rooted in segregationist values (Kinder and Sears 1981).
According to Kinder and Sears, symbolic racism refers to a
sense of disdain for those who fail to adhere to the
Protestant work ethic and thus seek handouts, lack thrift or
punctuality, fail to repress sexual urges or delay
gratification to work for distant goals. A symbolic racist
is likely to blame the victims of past injustice by saying

6
that if they just tried harder they could 'succeed on their
own like the Italians, Irish and Jews did.' Kinder and Sears
call this sort of racism symbolic because it is abstract,
based on ideological positions and attitudes rather than
experience. They argue that perceptions of threat and actual
vulnerability are ineffectual predictors of behavior while
symbolic measures excel.
Despite the growing literature on symbolic racism, more
research is needed to refine our understanding. While these
perspectives have drawn some criticism, chiefly from
Sniderman and his co-authors, important elements of the
theory have not been tested. Further, the elements that have
been tested use local samples which might yield results with
little external validity, or validity only if we throw out
the South. Additionally, insufficient effort has been made
to test the symbolic racism theses against racial threat
hypotheses. There are exceptions to be sure. (Green and
Cowden 1992, Kinder and Sears 1981, Sears, Lau, Tyler and
Allen 1980) The theory and measurement of this school of
thought are treated in detail in chapter three.
Synthesis
One need not choose sides in such a debate, for there is
no a priori reason to suspect that the two attitudes are
mutually exclusive. The proponents of symbolic racism theory
may be correct in asserting that an abstract sense of
prejudice is held by many whites, but many may feel

7
threatened as well. Indeed, the two orientations may exist
independently of each other in the same individual. These
constructs may reflect different dimensions of contemporary
racial attitudes. Individuals may evince high levels of
symbolic prejudice, or feel racially threatened, or some
combination of the two.
Similarly, one need not buy the entire theoretical
package of either camp. For example, one might easily
disagree with racial threat theorists who maintain that the
cause of racial conflict is the incompatibility of the
interests of whites and blacks. A political economist could
probably demonstrate that black ascendance poses little
threat to most whites. What is important from the behavioral
perspective is the perception of conflict or threat. Still,
the material context can influence the perception of threat.
For example, Green and Cowden's (1992) finding that
respondents with children who would be bused were more likely
to protest the busing plan in Boston implies that the
perceptions of threat may influence behavior dramatically,
even if it does not influence one's tendency to articulate
symbolically racist responses to survey questions. From this
literature, one may gather that the perception of threat and
symbolic racism may each have an independent force in
influencing different kinds of positions on issues and
evaluative behaviors. Rather than trying to supplant one
perspective with another, the argument presented emphasizes
recognizing the possibility that the two perspectives

8
represent different dimensions of racial attitudes among
whites.
Further, one may argue that current racial attitudes, be
they the result of perceived racial threat or symbolic
racism, are not merely either the result of current
demographic conditions or isolated psychological traits.
Indeed, both may factor into the equation. If the current
material context can influence an individual's racial
attitudes, couldn't the cultural context have an influence as
well?
In Chapter Four, a synthetic theoretical position is
staked out which allows for the simultaneous influence of
personal characteristics, contemporary material context and
local cultural factors. Chapter Five addresses measurement
issues and issues related to multi-dimensional racial
attitudes. National level models predicting the racial
attitudes and their effects are presented in Chapter Six.
Chapter Seven explores regional variations in the sources of
prejudice. A review of the central findings and the
conclusions are found in the final chapter.
Before ending this introductory chapter a cautionary
note concerning the motives of the author is in order.
The purpose of this project lies not in finding more
ways to brand people as racists. Rather, the goal of this
research lies in seeking a better understanding of the
anatomy of prejudicial attitudes. These prejudicial
attitudes may or may not have implications for political life

9
in America; likely as not, they do. It is instructive to
recall R. K. Merton's conception of the relationship between
prejudice and discrimination (1957). In a nutshell, Merton
pointed out that the two concepts represented different
dimensions; imagine a contingency table with prejudice on one
axis and discrimination on the other. It is possible, Merton
hypothesized, that one can be highly prejudiced and not
discriminate, never acting on that prejudice. Similarly,
nonprejudiced individuals may engage in discriminatory
activities.
At times provocative expressions are used. For example,
the expression (not originated by this author) symbolic
racism carries some unnecessary shock value. The expression
is really a misnomer, for the concept refers to a particular
theory of prejudice and need not carry the pejorative
implications of racism. Branding people as racists may have
few positive results, like kicking the dog. Nonetheless,
this term is used in this research because it is used by
other scholars in the field.
This study provides neither apology for nor an
admonition of prejudice per se. Rather, the author hopes to
provide insights into the causes and effects of racial
prejudice. It is hoped that the development of a more
sophisticated understanding of prejudice may contribute to
the amelioration of social conditions which have ill effects
for whites and blacks alike; if we are to defeat the forces
of ignorance and prejudice, we must first understand them.

CHAPTER 2
PERCEPTIONS OF RACIAL THREAT: THEORY AND MEASUREMENT
The claims of some scholars that control of the symbols
of the race debate represents the central struggle of
American politics (Omi and Winant, 1986) may be somewhat over
stated. Still, race seems to represent a fundamental
cleavage in American society. Even thirty years after the
heyday of the civil rights movement, the United states
continues to be a highly segregated society. This social and
geographic segregation has implications for the political
sphere as well. Race or race baiting continues to be used as
a trump card in elections throughout the nation. To be sure,
the race card is not always effective, but in many contests
it continues to be decisive.
The role of race in the literature of political science
is not unified by a common theoretical orientation, unit of
analysis, or methodology. Two important albeit divergent
streams are easily identifiable. The first tends to focus on
the effects of race at some level of aggregation. Scholars
in this stream may examine the effect of race or the local
racial composition on party systems, class differences or
tendencies within an electorate in particular contexts or
environments. The second stream addresses the psychology of
racial attitudes in individuals, apart from any specific
10

11
context. This second stream will be taken up in Chapter
Three.
In this chapter, two examples of the first school will
be examined in some detail. The first is a classic with a
regional focus, and the second an important contemporary
effort which explores the significance of race for electoral
and party politics nationally. This will be followed by a
close examination of the current work in this school which
uses aggregate and individual level data. While there are
some notable debates among the proponents of this general
area of scholarship, the similarities deserve more emphasis
than the points of disagreement.
Foundations
V. 0. Key's (1949) analysis in Southern Politics is
typical of the first and older line of argument. Key argued
that understanding the role of race was essential to
understanding Southern politics (p. 5). The means through
which blacks were socially and politically subjugated
thwarted the political articulation of class differences
among whites. In particular, the one-party system common
throughout the region prevented the empowerment of both
blacks and poor whites through the political arena. Blacks
were excluded from the system altogether. Extra-legal
sanctions against those who sought change were severe.
Blacks who braved those sanctions faced a maze of legal

12
obstacles to participation, including poll taxes and literacy
tests.
Poor whites were stymied by the one party system in two
ways. First, the barriers designed to exclude blacks from
participation if evenly applied would have also excluded many
whites. Key dismisses the effects of legal barriers to white
participation. This view is upheld by Harold Stanley's
analysis of participation patterns since the revocation of
those barriers (Stanley 1987; for contrary views see Kousser
1974; and Rusk and Stucker 1978). More important, according
to Key, were other factors: low education, a lock by elites
on real power, the punishment of dissenters, the obfuscation
of factions caused by the one party system and race baiting.
Indeed,
[t]he presence of the black provides a ready instrument
for the destruction of tendencies toward class division
among the whites. The specter of Reconstruction, of
Negro government can still be used to quell incipient
rebellion by discontented whites. (Key 1949, p. 655)
Key paints a picture of the South in which Democratic
single-party rule operated to the benefit of upper class
whites, planters and industrialists. Cheap black labor was
exploited and blacks had little opportunity to participate in
the political arena. Poor whites (rednecks, peckerwoods and
peckerheads) enjoyed a similarly degraded position. They
farmed inferior lands and had little hope for meaningful
participation. Poor whites who did join the fray met with
stiff retribution or faced near impossible odds in the

13
pursuit of their goals. The solidarity of elites within the
Democratic party stifled class-based or populist efforts to
address the agenda of the have-nots, be they white or black.
The one-party south and the techniques of control used by its
elites proved effective in preventing the development of
class based alliances between blacks and poor whites.
In addition to this basic set of relationships, Key
stresses the importance of the local racial composition.
Recall Key's statement that, "[wjhatever phase of the
southern political process one seeks to understand, sooner or
later the trail of inquiry leads to the Negro." (p. 5) The
desire on the part of whites to control blacks is greatest,
so the argument goes, in the those areas in which blacks make
up a larger proportion of the population, namely the black-
belt. Planters sought to maintain the system which provided
cheap labor for their estates. Whites outside the black-belt
were, hence, less concerned about the maintenance of white
supremacy. So, it was in the black belt that the subjugation
of blacks and the enforcement of a unified front on the race
question by white elites was most effective. In some states,
Key uses maps to demonstrate the strong correlation between
those counties with more than 50% black and strong support
for particular candidates and high levels of turnout. The
political impact of the local racial composition was potent.
While this description bears some resemblance to the
broad view of Key's canvas, there were of course variations.
Some populists, like "Big Jim" Folsom in Alabama, enjoyed

14
tremendous success. In some elections, outcomes had more to
do with the county from which the candidate hailed than his
platform or the black population. Still, as a rule the
analysis is compelling.
This analysis makes much of the differences in the
interests and attitudes between poor whites on one hand and
the planters and industrialists on the other. Indeed, Key
makes much of the claims of the "haves" that race relations
was a problem for the backward rednecks. Despite their
divergent interests and claims, the trigger mechanism of
their racial attitudes was often the same. Both groups
thought they had something to lose from black ascendance.
Planters were concerned about the loss of cheap labor which
threatened their economic well-being and losing control of
the state government through which they maintained their
advantaged position. Key's argument implies that whites
outside the black-belt had less to lose from black ascendance
and even felt less threatened than other whites. Still,
these whites were sensitive to race-baiting. This
sensitivity grew out of the fear of a return to the poverty
and degradation of the Reconstruction era. Key takes great
pains to indicate that these poor whites acted in a way that
was contrary to their class interests because, among other
things, they feared black ascendance.
Some may argue that the usefulness of Key's work has
faded as a result of the remarkable changes that have
occurred in the South and the nation since its publication.

15
For example, even before Key started the project, dependence
on cheap black labor began decreasing as a result of
mechanization and lessened dependence on mono-culture (McAdam
1982). The civil rights movement has changed the face of
racial politics. Electoral barriers have largely been
erased. Not only have blacks earned the vote, but they have
used it to make government and its policies more
representative. Reapportionment has changed the balance of
power in state governments away from rural bourbons. Key
(1949) himself pointed out the rapid migration of blacks out
of the South.
Despite the changes, Key provides useful insights into
the mechanisms of politics generally. At the very least,
Key's work provides a useful starting place for political
analysis. Often his theses are upheld. For example, Stanley
(1987) endorses Key's estimate that the relaxation of voting
restriction would have a negligible effect on white turnout.
This hypothesis is upheld by many subsequent studies (e.g.
Allport 1954; Blalock 1957; Pettigrew 1957; Wilcox and Roof
1978)
The Contemporary Significance of Race at the Macro Level
Key's analysis of the effect of race on the class
composition of the Democratic party has not fared as well in
the changed environment. The threat (real or imaginary)
posed by black ascendance was used as a mechanism for
maintaining the one party system. This system forced lower

16
class whites into an unnatural coalition with upper class
whites while excluding blacks altogether. After the civil
rights movement, black allegiance to the Democratic party
became almost universal. Now the perception of racial
threats is the basis for another unnatural combination of
classes in not merely the Southern party system but the
national party system (Huckfeldt and Kohfeld 1989).
According to Huckfeldt and Kohfeld's argument, the Democratic
party plays host to middle and lower class blacks and upper
middle and upper class white liberals. The Republican party
combines socially conservative working class whites with
rich, economically conservative whites.
The crux of Huckfeldt and Kohfeld's argument is that the
nature of this unnatural coalition undervalues the
contribution of blacks since their Democratic votes are taken
for granted by both parties. If the parties had to compete
for their votes then, the argument goes, blacks would be
better represented because more elected officials would have
to be more responsive. Further, such diversity could produce
more stable, class based coalitions in the two major parties.
The conclusions drawn from this argument are provocative
and in some way reinforce the importance of the project at
hand. But a keener focus on some of the interior elements of
the argument provide helpful insights. Huckfeldt and Kohfeld
argue that the current coalitions are uncomfortable for a
number of reasons. First, they affirm that the ideological
base of American politics has shifted, that the labor-

17
management dichotomy no longer represents the sole axis of
ideological placement. New left issues such as environmental
concerns and social change issues (e.g. feminism, gay rights,
etc.) don't fit on the old continuum and represent a
different dimension.1 So, the Democratic coalition of the
New Deal era is complicated by the addition of a new
ideological dimension, and old-style labor liberals find they
have less and less in common with new-left-lifestyle
liberals.
The influx of black voters into the Democratic party
after the passage of the Voting Rights Act of 1965 has
further complicated the task of coalition building in the
party system. In particular, working class whites have
become less supportive of the party. When one confronts this
issue in terms of electoral coalitions, it becomes clear that
the inclusion of blacks displaces working class whites
disproportionately. In other words, the larger the
proportion blacks voters comprise of the Democratic vote, the
smaller the proportion of whites in that same Democratic
electorate. This would seems merely tautological but for the
fact that the reduction in the white contribution (in terms
of Democratic votes) comes disproportionately from the
1 It should be noted that while Huckfeldt and Kohfeld
identify the salience of the changing landscape of American
ideological orientations, they distance themselves from those
who would couch those changes in terms of the development of
"post-material values." Indeed, they demonstrate many of the
shortcoming of post-materialist theory.

18
working class. Indeed, there is a strong inverse
relationship.
So, according to Huckfeldt and Kohfeld (1989), middle
class and Working class blacks make strange Democratic
bedfellows with new left upper class whites and some residue
of the working class whites who do not convert or demobilize.
The Republican coalition is similarly uncomfortable with its
combination of upper class fiscal conservatives and working
class whites. The authors also provide anecdotal data of the
splintering effects of race in particular contests. Again
the assumption concerning the "unnaturalness" of the
coalitions lies in the class differences between the various
groups.
Theoretically, working class whites and working class
blacks have compatible material interests. Similarly,
according to Huckfeldt and Kohfeld (1989), middle class
whites and blacks are presumed to have the same material
interests. Still, working class whites tend to flee the
Democratic column in inverse proportions to black
contributions to that column, even though it is in their
material interest to pursue like policies and candidates.
They flee because rightly or wrongly they feel as though
their interests are incompatible. They flee because they
feel that black politicians or politicians whose agenda gains
obvious black supports pose some threat if elected.
Despite the potent effect race can have on electoral
coalitions, the argument put forth by Huckfeldt and Kohfeld

19
does not indicate that race must always have an effect.
Indeed, by pointing to a specific contest in which race
baiting was particularly effective, one is pushed to
recognize that race baiting is less effective in other
contests or was not employed at all.
Aggregate Analyses
Huckfeldt and Kohfeld (1989) identify the macro level
relationship and pose potential solutions for the
disequilibrium afflicting the party system. They do not, nor
do they claim to, nail down the operative mechanism driving
the white flight from the Democratic coalition. Some recent
papers have added support to this line of inquiry.
Two analyses of recent elections shed some light on the
aggregate effects of local racial composition in electoral
contests. First, Hertzog (1991) made creative use of county
level returns in his analysis of African-American Douglas
Wilder's 1989 election as governor of Virginia. At first it
seems counter intuitive that the first elected black governor
would come from a Southern state, especially the Southern
state with the lowest percentage of black population of all
Southern states. Hertzog's (1991) analysis, influenced by
Key (1949), argues that this low percentage black actually
facilitated Wilder's success. He presents compelling
evidence to substantiate an inverse relationship between the
percentage of African-Americans and the proportion of the
white vote falling in Wilder's column.

20
Of course, many other factors contributed to Wilders
success (Jones and Clemons 1993). Wilder, the former
Lieutenant Governor, had risen through a series of
influential Virginia offices, so there were no doubts about
his qualifications. He was endorsed by many prominent
Democratic leaders in the state. Wilder's reputation and
campaign statements placed him clearly as a "social moderate
and a fiscal conservative" (Jones and Clemons 1993, p. 140).
Wilder's pro-choice stance allowed him to benefit from the
urgency lent the abortion issue by the U.S. Supreme Court's
1989 Webster decision. Wilder was perceived as tough on
crime and most thought his opponent responsible for negative
campaigning. Jones and Clemons (1993) claim that Wilder
intentionally distanced himself from the black community and
from racial issues, running what they term a 'deracialized'
campaign.
The traditionally perceived relationship between racial
composition and racial attitudes and white political behavior
is born out by Hertzog's (1991) analysis. Of course, the
percent black does not automatically determine victory in
electoral contest. Much more is at play. In the Wilder
case, one sees a highly qualified moderate Democrat running a
professional campaign, endorsed by important pols, on the
popular side of a divisive issue, and intentionally de
emphasizing race. Wilder's performance was optimal. Still
the pattern of white support remains. Hertzog (1991) argues
that those whites living in areas with fewer blacks felt less

21
threatened and could more easily cast their votes for the
candidate they preferred for some reason other than race.
Those who felt threatened had more difficulty in casting a
vote for Wilder, other things being equal.
Of course, Hertzog (1991) is not able to substantiate
the intervening variable, or mechanism, of threat without
individual level data. It is entirely possible that the
relationship he observes, albeit a strong one, is spurious.
It is possible, but unlikely, that any other unmeasured
demographic or social characteristic would correlate well
enough with racial composition to produce such a pronounced
effect. Indeed, Wilder's extremely narrow victory, despite
wider margins in opinion polls, implies the importance of
latent prejudices of which individuals were loath to speak.
Other scholars bring a similar sort of analysis to bear
on a most different case. Giles and Buckner (1993) examined
the pattern of support for David Duke in the 1990 Louisiana
race for a seat in the U.S. Senate. Duke, widely perceived a
racist, ran as a Republican and found himself in a run-off
with Bennett Johnston, a Democrat. Duke, former leader of
the Ku Klux Klan, has been characterized by some as a
purveyor of thinly veiled racist jingoism. Less benevolent
observers call him an overt racist. Duke tends to deny that
he is a racist, saying he has changed since his days with the
Klan. Still he does consider himself an advocate for white
people's rights; he even founded an organization called the
National Association for the Advancement of White People

22
(NAAWP). There is little doubt about the racial content of
Duke's appeal.
Giles and Buckner's (1993) findings are consistent with
those of Hertzog. They find a strong relationship between
local racial composition and the level of white support for
Duke2. Hertzog's estimates of white support for Wilder were
complicated by that candidate's significant black support.
Giles and Buckner assume that Duke garnered a negligible
number of black votes. This reasonable assumption allows
them to more fully examine the central relationship and to
introduce important controls in a multivariate analysis.
Despite the fact that they have only aggregate level
data, Giles and Buckner (1993) include a number of controls
by including census data at the SMSA and parish level. These
controls provide a convenient opportunity to assess the
current impact of theoretically important variables from
earlier work. Included as controls are two measures of
social status, median white income and the percentage whites
with a high school education. Additionally, the percent of
whites unemployed, the percent living in urban areas, the
percent of white inmigration, and the percentage of the
population coming of age after 1961 are also included.
Even with all these controls, their measure of racial
composition proves efficacious in predicting both the
2 Operationally, Giles and Buckner use the percentage of all
registered voters who are listed as black. They argue that
this better represents the threat to white political hegemony
than a population based figure. Of course, the two figures
are so highly correlated that it makes no difference in the
analysis.

23
percentage of registered whites voting for Duke and the
percentage of eligible whites voting for Duke. Further, the
direction of some of the significant controls lends support
to the notion that the mechanism generating Duke's support is
related to perception of threat. For example, median income
is negatively correlated with Duke support, indicating that
the most economically vulnerable localities were more likely
to support Duke. The percentage of whites unemployed was
significant and positive in the equation predicting the
percentage of eligible whites voting for Duke. This implies
that Duke's appeal was strong and had the effect of
mobilizing voters who were particularly vulnerable and most
likely to feel threatened by black ascendance. Urbanization
and inmigration had the expected ameliorative effects.
While both Hertzog (1991) and Giles and Buckner (1993)
demonstrate support for the relationship between threat and
candidate support in the aggregate and Giles and Buckner
(1993) find some support that implies that threat is one of
the mechanisms driving this relationship, both studies suffer
from limitations imposed by their data. There is only so
much one can learn from even the creative use of aggregate
level data. Individual level data are required to establish
the nature of the mechanism driving the behavior recognized
by these scholars.

24
Individual Analyses
Despite the well-developed literature linking the size
of minority populations and various measure of racial
inequality, there are only a few studies using survey data
which seek substantiate the connection between local racial
composition and the attitudes and behaviors of individuals.
Really, only a handful of recent studies have systematically
confronted the assumptions emanating from the aggregate
literature discussed above (Fossett and Kiecolt 1989; Giles
and Evans 1985, 1986; and Giles 1977). All of these studies
performed secondary analysis of nationally sampled data
collected by the two main academic research centers, the
General Social Survey (GSS) conducted by National Opinion
Research Center (NORC) and the American National Election
Studies (ANES) by the Institute for Social Research (ISR).
Despite the obvious attraction of testing hypotheses with a
nationally drawn sample, the use of such data is not without
costs. Finding valid measures for the theoretical concepts
of interest usually proves difficult and is often impossible.
The creative operationalizations used by Giles and Evans and
Fossett and Kiecolt represent serious, if not altogether
successful, attempts to overcome the limitations imposed by
the data. Though none satisfactorily address the
intervening threat mechanism, they do merit discussion. A
brief discussion of these recent studies follows. Three of

25
the four studies find support for the threat hypothesis
(Giles 1977; Giles and Evans 1986; and Fossett and Kiecolt
1989). Giles and Evans's 1985 analysis marks the only
exception.
The Giles (1977) study attempts to analyze the effect of
racial composition on racial intolerance or hostility. Using
data from the 1972 ANES, Giles constructed a racial attitude
score for each respondent. The six questions used to
construct the scale covered a wide range of topics related to
race including the federal government's role in guaranteeing
equal opportunity, the federal government's role in
overseeing school desegregation, the role of "the government"
in assuring equal access to public accommodations, the pace
of the civil rights push, the trade off between whites' right
to maintain segregated neighborhoods and blacks' right to
live where they wish, and whether the respondent favors
desegregation, strict segregation or something in between.
He found that the expected relationship held, but only for
respondents in the South. The regional differences may have
been attributable, in part, to the heightened sensitivity of
Southerners at the time to federally mandated desegregation
programs; mandated desegregation of northern schools, for
example, did not come until later. Half the items in the
scale address the role of the government in mandating civil
rights reforms. Of course, even this heightened sensitivity
on the part of Southerners would follow from their greater
sense of threat from the government and ascendant blacks.

26
Unfortunately, he was not able to tap perceived threat
directly. Still, his findings support the notion that the
perception of a threat is an intervening step between the
context, individual attributes and political attitudes and
behaviors. In this case, the articulation of intolerance
provides an additional clue of the existence of perceived
threat at the individual level.
The Giles and Evans study of 1985 uses data from the
1972 ANES and marks a departure from the earlier findings in
two ways. First, they find what is on first appearances a
viable measure of threat. In that study, respondents were
asked if they thought the civil rights people were pushing
too fast, too slow, or at the right pace. Additionally, they
were asked if black people had too much, too little or the
right amount of influence (See the Appendix for actual text).
These two items were combined to make an additive scale of
threat. So, respondents who thought the movement was going
too fast and thought blacks had too much influence scored
highest on the threat scale.
The second way in which this study marks a departure
from previous ones lies in its findings. Giles and Evans
(1985) argue that there is no significant relationship
between local racial composition (percentage black) and
perceived threat. Additionally, this non-relationship holds
up even when regional differences are accounted for. This
would seems to represent a chink in the armor of the
proponents of the threat hypothesis. But appearances can be

27
deceiving. The Giles and Evans (1985) finding should not be
accepted uncritically. Later efforts, discussed below,
revealed shortcomings in their modeling procedures. Still,
they served the academy by stimulating discussion of this
issue.
Additionally, Giles and Evans (1985) established a
relationship between the respondent's racial attitudes and
his or her level of identification with white people as a
group; the more hostile the respondents were, the greater
their attachment to whites. In other words, the sense of
racial polarization was evident not only in terms of people's
positions on about the civil rights people, or black people,
but also in terms of how they saw themselves. Similarly,
those who identified British ancestry, as opposed to those
who identified no specific ethnic heritage, felt warmer
toward white people. So, identification with whites as a
reference groups was enhanced by both the Giles and Evans
(1985) racial threat scale and by identification with some
other group, in this case the British. One caveat about
these findings is in order. Giles and Evans (1985) did not
include controls for region in the model estimating
attachment to whites. This may present a significant source
of bias if a greater proportion of Southerners trace their
ancestry to Britain than do Northerners, a not altogether
remote possibility.
The scale developed by Giles and Evans (1985) represents
the most serious attempt by political scientists to date to

28
measure perceived threat. It was later adopted by Fossett
and Kiecolt (1989). Fossett and Kiecolt (1989) reanalyze the
same data used by Giles and Evans (1985), but their findings
differ. While Giles and Evans (1985) find that racial
composition is not generally linked to their threat scale,
the more recent study finds the expected relationship between
local racial composition and the threat scale, and not only
for Southern respondents. Fossett and Kiecolt (1989) found
the percent black to predict their measure of threat in the
total sample and in both the Southern and non-Southern
subsamples. The difference in their findings results from
differences in model specification, discussed later in this
chapter.
Despite its apparent usefulness, the scale is not
without problems. First, one might question the face
validity of the items. One could think that the civil rights
people are pushing too fast because of a strategic
orientation; perhaps a sympathetic respondent believes that
more economic power should be amassed before continued
political maneuvers proceed. Similarly, one may think that
any civil rights push is too fast regardless of one's
vulnerability to threats from ascendant blacks. A second and
related problem lies how the items have been perceived and
utilized by other scholars; this issue will be taken up in
the next chapter.
Fossett and Kiecolt (1989) also look for regional
differences in respondent's support for integration, but are

29
unable to support Giles' (1977) Southern exceptionalism
explanation. Fossett and Kiecolt's (1989) results differ
because of a slight difference in the way they specified
their models. Using data from the 1976 and 1977 GSS, they
find that support for integration was inversely related to
the percentage black in all regions. They included a measure
of the respondent's community size, because 'urbanness'
should produce greater tolerance (Anderson 1962; Jackman
1978; Thomlinson 1969). Indeed, their data support the
ameliorating effect of 'urbanness.' Moreover, the inclusion
of urbanness in the models reveals the relationship between
racial composition and perceived threat. Presumably,
respondents from urban areas are so much more tolerant in
general than those residing in rural areas that the predicted
linear effect of racial composition is obscured unless
community size is controlled.
Fossett and Kiecolt's (1989) rebuff of Giles' (1977)
lies in their finding that the direction of the relationship
is the same for both southern and non-southern sub samples.
This they take as support for "Lieberson's (1980) hypothesis
that regional differences in white racial attitudes result in
part from regional differences in the distribution of certain
contextual variable." (Fossett and Kiecolt 1989, p.829) This
may be true in part, but their own data indicate that even
with several control variables a South dummy variable remains
statistically significant in the prejudicial direction.
Similarly, the effects of percentage black and a measure of

30
the respondent's Southern origins are significantly different
in models estimated for Southern and non-Southern subsamples.
So, in addition to "differences in the regional distribution
of certain contextual variables," Southernness make a
significant and unique contribution. The regional difference
is one of degree not of kind, but it remains a significant
difference.
As might be expected, the various measures of
Southernness have their effect in the more prejudicial
direction. This points in the direction of another
unrealized finding from the data presented by Fossett and
Kiecolt (1989). The effect of Southern origins is
insignificant in the non-southern sample. This
insignificance may be the result of a paucity of cases.
Alternatively, the insignificance of this variable may be
like the dog that did not bark of Sherlock Holmes fame. The
fact that Southern transplants in the North are not
significantly more prejudicial that their neighbors may point
to the importance of context in structuring political
attitudes. This interpretation is supported by [or
supports] contextual arguments made by Wright (1977) and
others, namely that the most pronounced and important
contextual effects are mediated through one's primary group
processes. In other words, "contextual characteristics
operate by influencing the types of people with whom one is
likely to interact." (Wright 1977, p.505) The behavior of

31
the Bubbas in Babylon, like that of a fish out of water, is
different than it would be 'Down Home.'
Using the same data (1972 ANES) exploited by Giles and
Evans (1985), Fossett and Kiecolt (1989) re-estimate the
effects of several contextual and individual predictor
variables on their shared measure of perceived threat. They
find that local racial composition influences their measure
of threat in the expected direction for all samples and
subsamples. Again, their specification differs from that of
Giles and Evans (1985) by the inclusion of the community size
control. Community size is negatively related to perceived
threat in all samples, so urbanness appears to have an
ameliorating effect on prejudicial attitudes. What they do
not find is a significant relationship between either a South
dummy variable or southern origins and their measure of
perceived threat. This runs counter to what one would
expect. Indeed, no variable predicting their measure of
threat has a significant regional difference.
Fossett and Kiecolt (1989) go further and examine the
effect of the threat scale on support for racial integration.
Here they find, as expected, that the threat scale has a
significant inverse relationship with support for
integration. Even with the threat scale in the model, the
percentage black has the expected (inverse) effect in all
samples. But in these models there are important regional
variations. First the threat scale has a more pronounced
effect in the Southern model. Second the Southern origins

32
variable, as above, is significant in the prejudicial
direction only in the Southern sample; Southerners in exile
are not less likely to support integration that their
neighbors. Further, the difference of these two variables
alone accounts for a 13.4% increase in the amount of variance
explained in the Southern sample over the Northern sample.
These difference reinforce the importance of context in
allowing both access to socialized values and in allowing the
use of those values in determining policy preferences.
Reconciliation
This discussion of the interpretation of regional
contextual variable points to another question deserving
further discussion. Giles and Evans (1985, 1986) treat
racial composition as a measure of external threat. This
interpretation is consistent with some of the sociological
literature on 'ethnic conflict' or competitive ethnicity (see
Giles and Evans 1985, 1986 for a review of this literature
and citations). In the 1985 article they make the
distinction between external threat and perceived threat.
Perceived threat was measured by asking the respondents
questions while the percent black represented the sole
measure of external threat. Giles and Evans (1985) may be
correct in making this assumption. In this way, they
distinguish themselves from Marxian and modernization
theorists who characterize race as an ascriptive identifier
and of diminishing salience in developed societies.

33
The discussion of the role of race in politics by Key
(1949) and Huckfeldt and Kohfeld (1989) implies that race is
not primordial, that it does not represent an unresolvable
cleavage. For Key (1949) the fundamental struggle of
Southern politics was class based. Race was used, in part,
as an instrument to thwart insurgent actions by the have-
nots. Key (1949) characterizes the net effect of racial
politics as anti-democratic because the system prevented the
recognition of objective interests and action based on those
interests. In this sense race is an unnatural basis for the
formation of group cleavages. Rather, tenant farmers and
yeoman farmers had common interests regardless of race. Race
baiting prevented the recognition of these common interests
and other forms of repression discouraged participation not
only of blacks, but of poor whites as well. Huckfeldt and
Kohfeld's analysis is similar. Race represents an unnatural
cleavage that prevents the formation of class based political
parties.
While Giles and Evans (1985,1986) do not test the
validity of their assumption that whites actually are
threatened by local blacks or cite empirical studies which
support it, there may be reason to think that there is
something to the notion. For example, the social change
movements of the 1960s and 1970s created an environment in
which racial and gender discrimination was made more
difficult. As a result, women and blacks entered the job
market and began taking some of the jobs that would have

34
either gone to men or for which they would have previously
received substandard wages. This represents a good deal for
those who had been discriminated against before, because they
were getting better treatment and wages than they had
previously. White male workers faced increased competition
for positions and wages. This depressed real wages. So,
white men lost the (non)competitive advantage they had
previously held.
Of course, some would argue that working class whites
are right to feel threatened because the inclusion of blacks
in the mainstream work force had a more immediate effect on
those on the lower end of the job market. Neither upper
management nor the owners of capital have much to fear from
the rapid expansion of the labor market after three hundred
years of systematic and costly discrimination. Demonstrating
the actual effects of the desegregation of the work place is
less important for this study than demonstrating whites'
belief that such desegregation poses a threat to them and
that perception of threat influenced behavior.
This sort of supply and demand analysis of the effects
of the relaxation of color and gender-based barriers to
employment on whites makes the Giles and Evans assumption
seem less far-fetched. Of course, the changes described
above could have been a short term shock which later resulted
in a new competitive eguilibrium. This may be an empirical
guestion, but it is one that need not be answered in the
present study.

35
Whether or not local racial composition represents an
actual or imagined threat to whites will not be examined in
this study. The question taken up in this study centers on
the operative cognitive mechanism driving racial politics
among whites. In this sense, one need not answer the
question of whether or not the political, economic or
cultural status of whites is objectively threatened or if
such a specter is merely the result of the political
conjurings of opportunists, entrepreneurs and extremists.
This central difference between scholars in the threat school
is less important than their area of agreement, namely that
the perception of threat can be a potent political force.
The potential reconciliation of these perspectives will be
discussed in more detail in Chapter Four. As the second
major school of political research into the racial attitudes
of whites is discussed in the next chapter, this common
ground will come into even sharper focus and the points of
difference will fade into the background.

CHAPTER 3
SYMBOLIC RACISM: THEORY AND MEASUREMENT
The role of race in politics has stimulated much
controversy in political science. In the last chapter,
discussion focused on approaches which point to the
perception of threat as the operative mechanism of white
racial antagonism. While scholars disagree as to the verity
of the threat, many see its perception as instrumental in
explaining, in part, larger political phenomena. This school
of thought finds its recent roots in the literature of
Southern politics and related changes to the party system
since the civil rights era. Its methods are largely
aggregative, although a handful of individual level studies
exist. Some of these were discussed.
In this chapter, discussion centers on different
explanations of the role of race in politics. Symbolic
racism1 represents perhaps the dominant paradigm in the
political science literature on white racial attitudes. This
1 "Symbolic racism" refers to a school of thought in the
literature on American racial prejudice. Other names for
this school have been bowed, new racism and modern racism.
These alternatives have all the flaws and none of the
strengths of the original term. Symbolic racism represents
the most universally recognized moniker for this particular
theoretical tradition. Many have attacked the expression on
a number of grounds. The use of the term in this study is
not meant as an endorsement of its content or its
appropriateness as a label for that content. Rather, its use
is adopted, for the purpose of clarity only, to refer to this
school of thought.
36

37
school of thought grows out of the more general orientation
of symbolic politics. Since this theoretical tradition is
psychological, its method is exclusively individually
oriented. Although a few experimental designs have been
conducted, most work in this area has exploited survey data.
The chapter opens with a brief discussion of symbolic
politics and the role of racial attitudes in that theory of
politics. This is followed by a description of the
development of the essential elements of the theory and its
landmark studies. The chapter closes with an account of the
major critiques of the approach and its methods.
Symbolic Politics in a Nutshell
The symbolic racism orientation grows out of the
symbolic politics tradition2. A thumbnail sketch of this
orientation will provide the foundation for the later
treatment of symbolic racism. The symbolic tradition argues
that political behavior is driven by cognitive and affective
processes. Pre-adult socialization and later resocialization
creates a number of affective orientations or symbolic
predispositions which provide an enduring standard for the
evaluation of new stimuli. Among these affective
orientations are numbered racial attitudes, ideology and
2 The brand of symbolic politics advocated by Sears (see
Sears 1988 for a summary), in particular, shares some common
ground with Edelman's (1971) articulation, but features one
dramatic point of departure. While Edelman left room for the
influence of self interest, Sears and his colleagues dismiss
such influences.

38
partisan identification. These enduring predispositions
determine political attitudes, like issue positions and
candidate evaluations, and behaviors, like voting or
participation in demonstrations.
The impact of the symbolic predispositions attitudes or
behavior occurs to the exclusion of an individual's current
personal or economic situation. In this perspective, the
material context has little influence on the individual.
Indeed, the effects of actions based on self-interest (or
what people think their self-interest is) is largely
relegated to the private sphere.
Of course, this caricature of this theoretical
traditions may seem a bit far-fetched. A less exaggerated
example from the economic voting literature may provide a
more favorable reading. Some have thought that voters in
America engaged in pocket-book voting, making their electoral
choice in favor of the candidate most likely to benefit them
financially (Bloom and Price 1975; Tufte 1978) or closest to
them on the content of issues (e. g. Page 1977; Riker and
Ordeshook 1973). Students of symbolic politics have argued
that this is not the case (Kinder and Kiewiet 1979; 1981).
Rather than making their decision based on their individual
interest, voters cast their ballots based on how they think
the country as a whole is doing, sometimes called
'sociotropic voting'. Thus the choice revolves around which
candidate is best for the country, not around which candidate
is best for the individual. Indeed, Kinder and Kiewiet found

39
support for this proposition in their 1981 study, even when
measures of satisfaction with income and satisfaction with
current financial status were controlled.
Of course, this literature is not without critics, but
neither is it without fans. In this case discussed above,
the symbolic politics approach contributes to our
understanding of the calculus of the vote choice. At the
same time, its usefulness in understanding politics in
general may be reasonably questioned. The approach
constitutes a black box of sorts; the symbolic
predispositions go in one side and the vote comes out the
other. One might imagine that the context in which this
process occurs might have some influence on it. How does the
voter evaluate what is 'good for the country' economically
vis-a-vis what is 'good for the country' militarily or
racially or educationally? Do voters project their
preferences as the most benevolent path for the country?
Despite these and other obvious criticisms, one can see the
possibility that an individual might suppress his or her
economic self-interest in making the vote choice.
Theoretical Underpinnings of Symbolic Racism
The theory of symbolic racism was introduced about 20
years ago (Sears and Kinder 1971). It has not remained
static since that time but has evolved. In particular, the
claims made by its proponents early on have been tempered.
For example, The assertion that traditional prejudice, "can

40
no longer be a major political force," in American politics
(Kinder and Sears, 1981) is no longer defended by either of
its authors. The proponents of symbolic racism theory are
not of one mind. There are disagreements about many aspects
of the theory. What follows is a portrait of a symbolic
racism that may not match any one scholar's description
exactly but which attempts to capture the mean or essence of
all its proponents; this is meant to provide a basis for a
general critigue of the theory and the generation of
hypotheses. Unlike a straw man, the composite presented
below is likely stronger than any one formulation of the
theory presented to date. This portrait is followed by some
attention to the elements of disagreement and unity within
the symbolic racism camp over central themes. The discussion
in later sections should make the limits of the derivative
statement of symbolic racism apparent and point to specific
places in the literature where disagreements occur. In
subsequent sections, attention is paid to controversies
surround data measurement and methods.
The logic of symbolic racism mirrors the logic of
symbolic politics outlined above. Pre-adult socialization
and adult resocialization result in the creation of a set of
racial predispositions or attitudes which provide an
immutable standard for the evaluation of new issues,
candidates or other stimuli. Current racial attitudes are
not the function of one's situation in life. Rather, they
are abstract. These attitudes or predispositions provide the

41
source for positions on issues, the evaluation of candidates
and other external stimuli. Ultimately the predispositions
result in behaviors such as voting.
There are two central elements to the racial attitudes.
The first is anti-black affect, a generalized distaste for or
disapproval of blacks. The second essential element is found
in a commitment to traditional values, not limited to but
epitomized by the Protestant work ethic. Thus Kinder and
Sears (1981, also quoted in Sears 1988) list the ingredients
of symbolic racism as
a blend of antiblack affect and the kind of traditional
moral values embodied in the Protestant Ethic ... a
form of resistance to change to the racial status quo
based on moral feelings that blacks violate such
traditional American Values as individualism and self-
reliance, the work ethic, obedience and discipline, (p.
416)
These components work in 'conjunction' together to form
symbolic racial predispositions.
The proponents argue that this symbolic attitude is a
new form of racism, that it is distinct from redneck racism,
"open bigotry of the sort frequently associated with working-
and lower-class (Southern) whites" (McConahay 1982, p. 705),
in either or both of two ways. Redneck racism has either 1)
disappeared or is of dramatically reduced salience in
contemporary society or 2) become old fashioned. That is to
say, it "is no longer fashionable to express these beliefs or
to support these practices openly in the elite circles of our
society." (McConahay 1982, p. 705)

42
Finally, like other symbolic predispositions in the
symbolic politics paradigm, symbolic racial attitudes were
thought to be completely abstract. While older forms of
racial attitudes had been thought to be related to the
context within which the individual found him or herself,
this new set of attitudes was thought to have no relationship
to the individual's context or material interests.
In sum, the essence of symbolic racism, as argued by its
proponents, is characterized by three points. First, it is a
new racial attitude, distinct from previously examined
prejudicial attitudes. Second, it is comprised of two
central elements, antiblack affect and traditional values.
Third, it is wholly abstract, completely unrelated to an
individual's context or material interests. These three
characteristics seem to represent the best and most widely
accepted 'definition' of the concept. The proponents of
symbolic racism have stood by these elements in the face of
some of their more trenchant critics (see the Sniderman and
Tetlock (1986a) exchange with Kinder 1986, or the chapters by
Sears and Bobo in Katz and Taylor, 1988). These exchanges
provide a more detailed history of the concepts development
and weaknesses than space permits here. A more detailed
discussion of the three central elements follows.
Newness or Distinctiveness from Old Fashioned Racism
Early articulations of symbolic racism maintained that
the new combination of anti-black affect and traditional

43
values had replaced the old racism (Kinder and Sears 1981).
Lately, both Kinder (1986) and Sears (1988) have backed off
from that position and now argue that the old fashioned
racism still exists and is different from symbolic racism and
that both have anti-black affect in common. All proponents
defend the distinctiveness of symbolic racism from previously
recognized forms of prejudice.
The new racism was distinguished from two other
recognized forms of racism: 1) generalized (in)egalitarianism
or 'redneck racism' which fostered "legalized or formalized
racial discrimination, [and] inherent racial inferiority"
(Sears and McConahay 1973, 138) among other things; and 2)
personal racial threats, characterized by threats to an
individual's job, housing, or safety.
Kinder and Sears (1971, 1981) dismissed the importance
of redneck or old-fashioned racism and were joined in this
dismissal by associates McConahay and Hough (1976) and
various constellations of co-authors (examples follow). They
point to declining levels of support in surveys for
segregationist positions as evidence of the declining
significance of old fashioned prejudice3. So by 1981, Kinder
3 See Schuman, Steeh and Bobo (1985) for an extensive
discussion of the declining levels of support for segregation
over time. In that study, they indicate a period of decline
in prejudicial attitudes throughout the civil rights era
before a new equilibrium (at relatively low, but not
inconsequential levels) is reached. Few respondents to
segregationist survey questions indicate the most prejudicial
positions. But as Sniderman and Tetlock point out (1986a,
also Sniderman, Tetlock, Piazza, and Kendrick 1991) it is not
evident why few choose the prejudicial responses to those
questions. Sniderman and Tetlock (1986a) pose one

44
and Sears could declare that America had become essentially
racially egalitarian.
The most provocative assertion of the newness of
symbolic racism is found in Kinder and Sears' 1981 study.
"Since the explicitly segregationist, white supremacist view
has all but disappeared, it can no longer be a political
force .... What has replaced it ..." is symbolic racism
(p. 416). McConahay, too, makes claims that the new racism
in now dominant while old fashioned racism is a thing of the
pre-civil-rights-era past (1982).
While some studies made claims about symbolic racism
replacing other racial attitude, others point to the concept
as simply one form of racial attitude among any number of
racial attitudes. Sears and Kinder (1985) back off from the
replacement hypotheses to a position recognizing the
distinctness of symbolic attitudes as opposed to other forms.
Others point to common elements across different types of
racial attitude. One landmark study which portrayed symbolic
racism as new and emergent in suburban America points to its
similarity to older prejudicial attitudes by saying that
symbolic racism is an "expression of some of the negative
explanation based on the changing language of prejudice.
They argue, for example, that the content of racial
stereotypes has changed over the years, but that the
underlying reasons for the application of prejudicial
stereotypes has not, i.e. bigotry. In this sense, they imply
that the measures evoke less response not because America has
become a land of racial egalitarianism. Rather, the old
questions no longer resonate with the underlying prejudicial
attitudes. In other words, the changing context has rendered
the old measures useless in tapping persistent prejudicial
attitudes.

45
feelings underlying old fashioned or red-neck racism, but it
differs from them in its other psychological roots and in
many of its specific forms of expression." (McConahay and
Hough 1976, p. 24) A similar claim is made in a later study
by McConahay (1982, p. 705) A 1979 study's factor analysis
found three racial dimensions, but since all were correlated
with anti-busing attitudes they were treated as one dimension
(Sears, Hensler and Speer 1979, p. 382n).
The original claims of newness have been softened
somewhat. The proponents do maintain that this expression of
racial attitudes is a post civil rights era phenomenon, that
it is distinct from other racial attitudes, and that it is
more important.
Combination or Multi-dimensionality
While most proponents of symbolic racism agree that
traditional values and anti-black affect are ingredients of
symbolic racism, there is some dissension as to the exact
nature of the relationships between the concepts. Kinder
recently claimed that symbolic racism is the 'conjunction' of
these values and racial prejudice. (1986) The precise
relationship between the two elements has never been
specified. McConahay and Hough (1976) stress this facet of
symbolic racism by calling attention to a number of
constituent elements. For McConahay and Hough (1976),
symbolic racism has blacks as a group as its object, a moral
tone (p. 36-7), addresses issues surrounding the violation of

46
the Protestant ethic (p. 41; also McConahay 1982), and has as
it "most important psychological factor . political and
economic conservatism . ." (p. 37)
A similar approach is taken by Sears, Hensler and
Speer's (1979) articulation of symbolic politics' "anti
busing sentiment as primarily derived from long-standing
political predispositions, specifically racial prejudice and
political conservatism, which in turn can be traced to
political socialization in preadult life." (p. 372). Kinder
and Sears (1981) point to symbolic racism's "blend of
antiblack affect and the kind of traditional American moral
values embodied in the Protestant ethic." (p. 416; also
endorsed in Kinder and Sears 1985; Kinder 1986; Sears 1988)
The formulation by Sears and Citrin is similar, "the
conjunction of two separate underlying dimensions: a
specifically anti-black attitude, and conservative value
priorities, which stress individual effort and responsibility
rather than government activism." (1982, p. 168)
Nowhere in the literature on this topic is there a clear
articulation of how the elements fit together, why they
should not be treated as distinct dependent or independent
variables or how they are different from garden-variety
prejudice and conservatism. One gathers that the
relationship is mutually reinforcing. A dedication to the
values embodied in the work ethic reduces sympathy with the
unemployed, underemployed, or otherwise less successful folk
one might hear aboutperhaps poor blacks on the television

47
or a street corner. The residue of anti-black affect reduces
the level of sympathy resulting in the assignment of blame
for the outcast group. So the perceived failings of the out
group to live up to expectations and the attendant, but
abstract, disdain for that out group result in the symbolic
racial attitude.
Despite criticism over the combination of analytically
and (potentially) empirically distinct elements (note
especially the frontal assault by Sniderman and Tetlock 1986a
and a less successful effort by Bobo 1983), the proponents
stand by the 'conjunction' premise.
Abstraction vs. Self Interest
The concept of symbolic racism was introduced in Sears
and Kinder's 1971 study of the 1969 Mayoral election in Los
Angeles. This election pitted a conservative white
candidate, Yorty, against a liberal black candidate, Bradley.
The concept developed inductively from their analysis of
survey questions asked during the election (Kinder 1986,
McConahay 1976). From the responses to the questions the
authors inferred that a new form of racism had taken root.
This new racism consisted of "abstract moral assertions about
blacks' behavior as a group . ."(Sears and McConahay
1973)
This new racism was thought to be abstract, not based in
the individual's current material conditions, and hence
symbolic. If not based in material conditions, where does

48
the attitude have its roots? Like other predisposition in
the symbolic politics paradigm, symbolic racism has its roots
in pre-adult and adult socialization. Let us leave aside for
the moment issues arising from the effects of material
conditions on socialization.
All the studies of symbolic racism draw the distinction
between the abstract, or value driven, nature of symbolic
politics and the material context in which the individual
finds him- or herself4. This abstractness principle is
generally presented in juxtaposition to the notion of threats
to personal lives (Kinder and Sears 1981, p. 416, Sears and
Citrin 1982) or the effects of personal experiences
(McConahay and Hough 1976, p. 37; McConahay 1982, p. 705n;
Sears, Lau, Tyler and Allen 1980, p. 671; Sears and Lau 1983,
p. 224) The articulation found in Sears, Hensler and Speer
(1979) is particularly behavioral in its emphasis on the
stimulus-response mechanism:
Whether or not the issue has some tangible consequence
for the adult voter's personal life is irrelevant.
One's relevant personal 'stake' in the issue is an
emotional, symbolic one; it triggers long-held, habitual
responses. (1979, p. 371)
4 The position articulated in a more general study of
symbolic attitudes is inconsistent with the statement. In
Kinder and Rhodebeck's analysis of the 1972-76 ANES panel
study (1982) they state, "These stability coefficients . .
imply both that racial beliefs are comparably durable . .
and that they are far from fixed. Racial socialization does
not end with adolescence." (p. 205)

49
Two statements by Kinder and Sears (1985) reinforce this
point.
Resistance to change in the racial status quo would
seem, there-fore, to have little to do with the direct
tangible threats blacks might pose to personal life, and
a great deal to do, instead, with prejudice and values,
(p. 1141)
We say that personal life is usually compartmentalized
away from political life; that any real or imagined
threats posed by blacks to whites' private lives seldom
spill over into antiblack political positions, (p. 1143)
Given the evidence of regional variation presented in
the previous chapter, it is difficult to accept uncritically
the assertion that racial attitudes are divorced from
context. Still, the proponents of symbolic racism (speaking
for themselves) make their position abundantly clear.
Methods and Measurement
Given its ad hoc development, symbolic racism did not
start out with a set of theoretically derived measures.
Rather the concept was deduced from measures meant to tap
general racial attitudes in the Los Angeles studies (Sears
and Kinder 1970, 1971; Kinder and Sears 1981). Since that
time, efforts have been made to develop better measures.
These efforts have met with varying degrees of success.
While critics of symbolic racism point to shortcomings in the
measures used in a number of studies employing national level
data, Kinder (1986) and Sears (1988) argue that these studies
were not designed to test symbolic racism per se. Rather,
they argue that the studies were examinations of general

50
symbolic politics hypotheses which happened to include racial
measures. For example Sniderman and Tetlock (1986) criticize
Sears, Hensler, and Speer (1979) and Sears, Lau, Tyler and
Allen (1980) for combining measures of old fashioned and
symbolic racism in these studies. Kinder 1986 and Sears 19885
respond that those studies do not measure symbolic racism but
symbolic politics generally. Rather than dwelling on
contested measures, what follows is a discussion of the
current concepts and measures most widely accepted by the
proponents of the theory, setting up a rationale for the
measures used in the present study.
While symbolic racism is the conjunction of antiblack
affect and traditional values, Sears argues that it manifests
itself in two ways: "(a) antagonism toward blacks' "pushing
too hard' and moving too fast, especially (though not
exclusively) through the use of violence and (b) resentment
toward special favors for blacks . ." (Sears 1988, p.
56) These manifestations, more or less, reflect the
constituent theoretical parts. For example, a belief in
self-reliance combined with antiblack affect may make one
more disposed to latch onto stereotypes concerning welfare.
Table 3.1 below list examples of the operationalization of
these concepts in a series of symbolic racism studies
conducted between 1976 and 1985. This list, compiled by
5 Sears (1988) claims that these two studies did not measure
symbolic racism on page 64 after crediting those same
studies as examples of the concept's efficacy on page 59.

51
Table 3.1 Sears's two dimensions of symbolic racism measures
Antagonism Toward Blacks' Demands
Blacks are getting too demanding in their push for equal
rights. (Agree)b,c
Blacks shouldn't push themselves where they are not
wanted. (Agree)a
Some say that the civil rights people have been trying
to push too fast. Other feel they haven't pushed fast
enough. (Trying to push to fast)d
It is easy to understand the anger of black people in
America. (Disagree)b,c
Resentment About Special Favors for Blacks
Over the past few years, the government and the news
media have shown more respect to blacks than they deserve.
(Agree)b
Over the past few years, blacks have gotten more
economically than they deserve. (Agree)b
The government should not make any special effort to
help blacks and other racial minorities because they should
help themselves. (Agree)e
Do you think blacks who receive money from welfare
programs could get along without it if they tried, or do you
think they really need the help? (Could get along)a
Do you think the Los Angeles city officials pay more,
less, or the same attention to a request or complaint from a
black person as from a white person? (More)a
a Kinder and Sears (1981).
b McConahay (1982)
c McConahay and Hough (1976)
d Sears and Allen (1984)
e Sears and Citrin (1982)
Source: Sears (1988, p. 57)

52
Sears (1988)6, contains none of the contested measures and is
drawn from a wide range of studies. One must conclude that
these items have some measure of face validity for Sears,
Kinder, McConahay and the other advocates of the concept.
Upon review of Sears' list of the currently accepted measures
for symbolic racism, one observes that the questions have one
fundamental thing in common. They all have blacks as their
objects. This marks a point of theoretical consistency. If
the concept of symbolic racism has at its core the notion
that 'blacks violate traditional American values' then it
follows that survey questions meant to tap that core should
have blacks as their focus. Beyond this, some of the
questions in this list clearly indicate specific values, in
particular self-reliance, or getting ahead without government
assistance. Other questions not on this list show remarkable
face validity in terms of their apparent ability to capture
both antiblack affect and values such as self-reliance. The
"blacks could get ahead like the Jews, Irish and Italians"
question is a particularly good example of this (see the
appendix for the text of questions used in this study). These
and other questions will be discussed at greater length in
the chapter on methods.
6 Kinder (1986, pp. 156-7) provides a more exhaustive list.
The shorter list showcased here is adequate to give the
flavor of the measures.

53
Self Interest: The Alternative Hypothesis
The series of symbolic racism studies go to great
lengths to test the various symbolic racism scales against a
single alternative hypotheses. The primary alternative is
represented by some measure of self interest. The proponents
may be justified in their selection of this hypothesis, for
it is compelling. But there has been relatively little
response to criticisms of their operationalization or choice.
For example, the theoretical traditions alluded to in the
previous chapter are given little play in the writings of the
symbolic advocates. Their discussion of the self interest
alternative sometimes draws clear distinctions between its
narrow construction and more general threat hypotheses and
sometimes that distinction is more obscure. The alternative
hypotheses they explore represent a most narrow construction
of 'interest.' On the other hand, this narrow construction
may provide clarity and assure that this particular
alternative is well explored.
One must ask how was this alternative hypothesis
measured. While methodological cheap shots do not provide a
basis for the formulation of a theory, they can help one
critically evaluate the extent to which the evidence
marshaled in support of a theory is supportive. The
proponents of symbolic racism go to great lengths to test

54
this (self-interest) alternative hypothesis. Unfortunately,
the extensive tests provided leave lingering questions.
First, a couple of general methodological issues should
be addressed. In a number of studies, the proponents use a
number of tests to gauge the relative explanatory power of
self-interest and symbolic racism (Sears, Hensler and Speer
1979; Sears, Lau, Tyler and Allen 1980; Sears and Citrin
1982; Kinder and Sears 1981; McConahay 1982). Some of the
tests consistently present difficulties in interpretation.
While there is no reason to suspect an intent to obfuscate,
one finds a consistent pattern of modeling procedures which
could underestimate the effects of alternative hypotheses.
These three tests are discussed below. The first two types
of test involve problems associated with the comparison of a
scale to a collection of individual variables and the third
involves questions concerning the proper comparison of
populations.
First, a symbolic racism scale is compared with a number
of measures for self interest. For example, in Kinder and
Sears 1981 study, the effect of the symbolic racism scale is
compared with that of a handful of self interest indicators
in a regression equation predicting support for one of the
mayoral candidates in the 1973 election in Los Angeles. The
scale has a consistently significant effect in the predicted
direction while the other measures tend not to have
significant effects. Of course, comparing a scale to a
collection of, presumably, correlated measures presents some

55
difficulties. Assuming the measures for self-interest were
valid, they might not appear as significant while appearing
in an equation with other measures of the same concept; the
explained variance may have been spread thinly over the
multiple indicator so thinly that no individual measure had a
statistically significant effect. In other words, each of
the parameter estimates for the measures of self interest
report the effect of that measure of self interest while
controlling for the effects of other measures of the same
concept. Finding even a small effect may have theoretically
significant ramifications (see for example, King 1986),
namely that self-interest makes some contribution.
A second kind of test is used to compare the predictive
abilities of the two concepts. For example, in their table
3, Kinder and Sears (1981) estimate a regression equation
with the symbolic racism scale as the sole independent
variable. They note the adjusted r2, the percentage of
variance in the dependent variable explained by all the
independent variables adjusted for the total number of
variables in the model. Next they add the several measure of
self-interest to the model and note the adjusted r2 for the
total model including the symbolic racism scale and the
measures of self-interest. By subtracting the first r2 from
the second they can gauge the additional amount variance
explained by the measures of self-interest. But herein lies
the rub. In the calculation the adjusted r2, symbolic racism
is treated like a single variable while the collection of

56
self-interest measures, also measuring a single concept, are
treated as multiple unrelated measures. This comparison
presents a bias against the unsealed items, a bias that is
amplified by colinearity among the unsealed items.
A third test breaks the sample into smaller groupings
based on the responses to one of the measures designed to tap
self-interest. The complete model is then re-estimated
(without the measure used to divide the sample) for each sub
group of the total sample. As above, the symbolic racism
scale is found to have significant predictive power while the
measures of self-interest fail. This interpretation is
problematic. First, the division into the groups represents
just another way of controlling for the effect of the
variable used as the divisive criterion. Such a division is
reasonable under a number of circumstances, for example if
nonlinear effects are expected for different levels of the
criterion variable or if the criterion variable was shown to
empirically or has a theoretically significant interaction
with another variable in the model. The presentation by
Kinder and Sears (1981) is not much of an improvement over
having the criterion variable as another independent
variable.
A second problem associated with this test is that it
rests on a comparison of the marginal effects of the
independent variables across 'populations' or sub-samples
rather than on the different absolute levels within each sub
sample. In this sense, as above, one expects the marginal

57
effects to be the same as when the criterion variable was
controlled statistically in an earlier equation. If there
were differences in the absolute levels of the dependent
variable, in this case the evaluation of a mayoral
candidates, how would one know? The marginal effects of the
independent variables should not be very different from the
equations where the criterion was included as a control, but
one would find differences in the constant of the equation.
Technically, the constant is the value of the dependent
variable when all the independent variables are zero. In a
sense, the constant gives a measure of the dependent variable
before the contributions of the independent variables are
included. In this light, the differences between the
subsamples could reside in the constant. For example, those
living in the area where busing is occurring may be so much
more prejudicial than those in other areas that the marginal
effects of the other independent variables is negligible; in
such a situation, comparing constants would expose or
disconfirm this possibility. Kinder and Sears (1981) do not
report the constant, so the comparison across sub-samples is
incomplete.7 Of course, the constants may not have been
7 In general, the omission of the constant is not an
egregious transgression. Generally, when one or more of the
independent variables lacks a meaningful zero point (i.e.
zero falls outside the accepted range for the independent
variable or the models in general) the constant may be
omitted. For example, if year is a predictor of automobile
cost, the value of the constant (the cost of an automobile in
the year zero) is of little interest. The trend in many
journals is to report the constant even when it is
meaningless, for no apparent reason. Here Kinder and Sears

58
different. Kinder and Sears could have strengthened their
argument by calling attention to this fact.
Data
With only a one exception (McConahay 1982; Louisville),
all the data used in the examinations of symbolic racism are
drawn from non-Southern samples. There are some instances of
national level data set being used to measure general
symbolic politics hypotheses, but none address symbolic
racism specifically.
This leaves a number of city-wide studies conducted in
the north and west and some studies conducted in colleges and
seminaries. The McConahay article of 1982 does feature data
from a Louisville sample collected by the Harris
organization; some consider Louisville a Southern city.
Still, the fact that regional differences in racial tolerance
are often found (see examples in the previous chapter) gives
one cause to expect either that regional differences would
exist in the new racism or that the prevalence of the old
racism would follow some regional pattern. The significance
of the second point depends on the theoretical expectations
concerning the relationship between symbolic and old
fashioned racism: if symbolic racism is a submerged redneck
racism then one would expect either a greater prevalence of
symbolic racism where redneck racism had existed previously
provide an interesting counter example to this trend by not
reporting the constant when it may have been meaningful.

59
or a decreased level of symbolic racism where redneck racism
prevails because people there would not need to veil their
attitudes. This would be especially useful if tests were
provided comparing the two forms of racism.
Of course, given the framework of symbolic politics,
there is no expected relationship between context and
symbolic predispositions. In this sense, the theoretical
problems caused by the loci of the various studies only apply
to the alternative hypotheses. Still, the use of national
level data or comparisons of absolute level of symbolic
racism across regions might provide a useful test of the
degree to which the new concept is actually symbolic. If it
were truly abstract then there should be no differences
across regions. Previous studies have not tested this. A
fuller elaboration of this hypotheses is outlined in the next
chapter.
The framework of symbolic racism rests on three pillars:
distinctiveness from known forms of prejudice; the
combination of antiblack affect and traditional values; and
the divorce of symbolic attitudes from context and self
interest. The concept has been tested against a narrowly
constructed self-interest alternative hypothesis. The data
used for those tests have been drawn from municipal and
state-wide samples, primarily collected outside the South.
In this chapter and the one before it, discussion has
focused on two major theoretical traditions in the study of
white racial attitudes in political science. In the next

60
chapter, an alternative approach will be outlined. This
approach is more synthetic than innovative. Indeed,
innovation is difficult to come by in a literature as
"sprawling" as that on racism (Kinder and Rhodebeck 1982).
This new approach attempts to build on the strengths of the
two outlined below. For example, the claims made by
proponents of symbolic politics concerning the enduring
effects of pre-adult socialization to the exclusion of the
material or social context must be tempered. Similarly, the
assumption that 'realistic group conflict' is necessarily
realistic will be relaxed; even if such is the case, an
individual must perceive the conflict as real to act upon it.
Of course, not all questions raised by this discussion will
be answered. Indeed, some may be unanswerable in the short
run. Still, it is hoped that the syntheses and ensuing
empirical investigation will shed some light on the salient
issues and point in the direction of their resolution.

CHAPTER 4
SYNTHESIS
in the previous two chapters discussion focused on two
mainstream but distinct schools of thought concerning racial
prejudice in America. The first school maintains that racial
threats or the perception thereof provide the central
mechanism for the generation and use of prejudice. The
second school introduces a 'new' form of abstract prejudice
thought to have become widespread in the post civil rights
era. In this chapter, a third position is laid out in order
to synthesize a set of reasonable theoretical propositions
and lay the ground work for testing those propositions.
The (survey research portion of the) literature
discussed in the previous two chapters suffers from a problem
common in the social sciences. Namely, the phenomenon being
modeled is far more complex than the model.1 In this case,
as the proponents of each of these schools have staked out
their territories, they have tested their hypotheses against
relatively transparent null hypotheses. Additionally, the
conversations in each school have been independent of each
other. The result of these shortcomings stultifies theory
building since research falls into a cycle of replication on
1 Of course, the very nature of modelling requires some
simplification, else the exercise would be pointless.
61

62
the one hand or a series of minor internecine squabbles on
the other.
Given the broad view of the two schools, one can derive
areas of synthesis. Indeed, one expects common ground
between the approaches because both schools address prejudice
and its causes. One might expect propositions held in common
by both schools to be especially robust, theoretically and
empirically. Conversely, the greatest opportunity for theory
building lies in areas neglected by either or both schools.
A number of hypotheses discussed later in the chapter provide
a framework for plumbing these interstitial depths.
Despite the existence of any number of common or
uncommon elements between the two approaches, two themes
merit serious treatment. The first and most fundamental
theme centers on the notion of multi-dimensionality of racial
attitudes. Is there merely one kind of attitude, parceled
out in varying levels across the nation or are there multiple
dimensions of varying intensity and salience? The second
(and secondary) issue revolves around defining interests. Do
people define their interest narrowly, in terms of the
maximization of personal benefits, or are interests more
broad, weighed in terms of the maximization of benefits for
family, group, or country? These two issues will be taken up
below followed by a brief discussion of the importance
contextual factors in understanding attitudes and behavior
with a fairly detailed discussion of one such factor, racial
composition.

63
Multi-dimensionality
The notion of multi-dimensional racial attitudes dates
from the 1950s (e.g. Rokeach 1960; Allport 1954). Many
studies identified the various dimensions through the use of
intensive personality trait questionnaires. This method
resulted in detailed measurements of the respondent's racial
attitudes. The studies, however, failed to demonstrate the
relevance of the scales to any external conditions or to
demonstrate the impact of those attitudes on political
behavior or attitudes.
Other studies have excelled in measuring people's
political orientations and evaluations of candidates. For
example the American National Election Studies (1952-1992)
conducted by the Institute for Social Research at the
University of Michigan have consistently measured many
fundamental political orientations, behaviors,
characteristics and attitudes. Of course, every question a
secondary or even a primary researcher might want to ask
cannot be asked. Few data sets can tap multiple dimensions
of racial attitudes and provide measures of political
phenomena against which one may judge the effects of the
racial orientations. Hence, most studies cannot take this
sort of approach. The proponents of symbolic racism
acknowledge the existence of multi-dimensionality (see
discussion in Chapter Three), while others are less explicit
in their treatment.

64
But the acknowledgment of multi-dimensionality by the
proponents of symbolic racism is underdeveloped empirically
and theoretically, except to say that the new or symbolic
racism replaces the old. One wonders why the old has not yet
disappeared or what the different effects of the two sets of
prejudicial attitudes might be?
The group conflict orientation held by some of the
authors discussed in Chapter Two holds no place for
conceptions of multi-dimensionality. From their perspective,
material imperatives would outweigh any preferences,
psychological predispositions or attitudes. Their view of
racial attitudes lying mainly in the material realm makes
them ill equipped to explain even some of their own findings.
The works by Huckfeldt and Kohfeld (1989) and Key (1949) are
less clear about the exact nature of the racial attitudes,
but unlike their colleagues they point to racial
predispositions resulting in behavior which is contrary to
individuals' material interest. Given this lack of agreement
it seems reasonable to temper or relax the assumption that
actual conflicting interests represents a necessary and
sufficient condition for the emergence or existence of racial
hostilities.
The two central schools of thought either dismiss the
possibility of multi-dimensionality or fail to explain it.
Some studies by the proponents of symbolic racism do find

65
multiple dimensions
in racial attitudes but
fail to explore
the significance of
multi-dimensionality.2
Attitude A on
horizontal axis
Attitude B
low
high
high
I
II
low
III
IV
Figure 4-1. matrix of racial attitudes
There is little reason to expect people's prejudicial
attitudes to follow a zero sum pattern, where one new
attitude replaces the other. Rather, the attitudes may
represent distinct and independent dimensions. A graphical
representation of a two dimensional model is presented in
figure 4-1. This two by two matrix is not meant to pigeon
hole the orientations into four categories. Instead, the two
continua fall on the axes. So an individual could fall
anywhere in the range of axis A, without restricting their
location on axis B.
The literature discussed in the previous two chapters
leads one to some propositions concerning the nature of
multi-dimensional racial attitudes. This discussion focuses
on the theoretical expectations in two dimensions for
clarity's sake but the arguments could be extended to more
2 For example, Sears, Hensler and Speer (1979, p. 382n) use
factor analysis to identify three racial dimensions: 1) old
fashioned racism; 2) opposition to government action; and 3)
opposition to racial protest. Still, they combine all the
items into a single racial attitude scale.

66
(N) dimensions. Further, the discussion addresses the two
dimensions for which measures are available in the data
exploited in this study, namely perceived racial threat and
symbolic racism. The specific measures and
operationalizations are discussed in the next chapter.
Specific hypotheses are numbered in the text and summarized
in Table 4-1.
First, one would expect two racial dimensions to show
some degree of independence, but not be completely
uncorrelated (HsynOI)- The similarity in theme should mean
that the dimensions share some combination of inputs (what
explains the attitude), traits (the geographic distribution
of the attitudes or other descriptors), or effects (what the
racial attitudes reveal about other political orientations
and behaviors). Divergence in some of these is to be
expected as evidence of the independence of the dimensions.
For the two dimensions examined here a number of
predictions follow from previous theory. The two dimensions
should differ in their inputs. For example, the local racial
composition should correlate with the perception of racial
threats (HrtoD? this sensitivity has been shown to be more
pronounced in the context of the South with its
segregationist heritage (e.g. Fossett and Kiecolt
1989)(Hrt02) Similarly, an individual's assessment of his
or her financial situation of the state of the economy in
general should covary with sensitivity to perceived threats
(Hrt03) In other words, if one has a dim outlook on the

67
Table 4-1 Derived and Synthetic Hypotheses
Derived Racial Threat Hypotheses
Hrt01: Levels of perceived racial threat will covary with
the percentage black.
Hrt02: Levels of perceived racial threat will be higher
among Southerners.
HRT03:Levels of perceived racial threat will covary with
economic vulnerability.
Hrt045Aggregate (or mean) levels of perceived racial threat
will be highest in the Deep South, less high in the Rim-
South and lowest in the non-South.
Hrt05! Evaluations of candidates will covary with levels of
perceived racial threat when candidates are identified with
the out group as members or as supporters.
Hrt06: Support for safety net policies (exclusive of
racially specific policy positions) will covary with levels
of perceived racial threat.
hRT07: Support for policies which benefits African Americans
in particular will vary with levels of perceived racial
threat.
hRT08! The level of perceived racial threat will covary with
the closeness in income (class proximity) of the racial
groups locally.
Derived Symbolic Racism Hypotheses
HsroI: Levels of symbolic racism will be uncorrelated with
local racial composition.
Hsr02: Levels of symbolic racism will be uncorrelated with
individuals' regional origins or residence.
HSR03: Levels of symbolic racism will be uncorrelated with
economic vulnerability.
HSR04: Aggregate levels of symbolic racism will be
uncorrelated with region.
Hsr05: Evaluation of liberal or black candidates will covary
with level of symbolic racism.

68
Table 4-1 (continued)
Derived Symbolic Racism Hypotheses (continued)
Hsr06: Support for safety net policies will vary with
symbolic racism.
hSR07: Support for policies which benefits African Americans
in particular will vary with levels of symbolic racism.
Synthetic Hypotheses
HsYN01: Perceptions of racial threat and symbolic racism
represent different dimensions of racial attitudes.
HSYN02! The level of perceived racial threat will vary with
the level of income equality between the races.
HSYN03: Symbolic racism will covary with the percentage
black at low levels, after which the marginal effect of the
racial composition of symbolic racism will diminish.
HSYN04: Current level of perceived racial threat and
symbolic racism will covary with archaic levels of the
percentage black.
hSYN05: Levels of perceived threat will vary with personal
contacts with out group members, as in the work place.
economy (or that one is losing relative to others) then the
sense of threat from an external source becomes more salient
than it would be under other conditions.
The assumption of abstraction on the part of the
proponents of the symbolic racism hypothesis implies
independence between symbolic racism and contextual factors
like the local racial composition (Hsroi) or the region of
residence or upbringing (Hsr02)* Likewise, symbolic racism
should not be influenced by the individual's economic outlook
because, according to its proponents, symbolic racism is

69
independent of personal life (Hsr03)* Indeed, the proponents
of symbolic racism are clear that symbolic racism should be
completely independent of all environmental factors or
personal interests. (Kinder and Sears 1981, p. 416; Kinder
and Sears 1985, p. 1143; McConahay and Hough 1976, p. 37;
McConahay 1982, p. 705n; Sears, Hensler and Speer 1979, p.
371)
The two dimensions should differ in their traits.
According to the findings discussed in Chapter Two,
perceptions of threat should be more pronounced in the South
given its tradition of sensitivity to such threats. On the
other hand, symbolic racism should not vary regionally,
according to its proponents, because of its independence from
the individuals' material context.3 According to symbolic
racism theory, one would not expect variation among any of
the regions including the most likely suspect, the South. So
aggregate levels of perceived racial threat should be highest
in the South (Hrt04)^ while aggregate levels of symbolic
racism will not vary by region (Hsr04)*
In terms of the effects of the two racial dimensions,
one might expect a different picture. The perception of
3 Historical conditions in the South and previous research
lead one to expect living in the South to have some effect
(input) into the make-up of an individual's racial attitudes;
this notion is supported by the threat literature, although
some scholars see Southern exceptionalism or the uniqueness
of Southern culture and politics as fast fading (e.g.
Stanley). As a trait, however, geographic distribution must
be considered independently form arguments based on the
impact of southernness on the racial attitude. This
difference is especially salient given the abstractness
assumption of symbolic racism.

70
threat should not have a uniform effect on attitudes, issue
positions or the evaluation of candidates across the
political spectrum. Rather, it would affect only on those
issues and evaluations which were related to one of its two
fundamental components: race and economic vulnerability. For
example, the perception of threat would not be efficacious in
predicting the relative evaluations of say, two white
candidates in a contest in which the race card was not played
by one. It may figure more prominently in situations in
which one of the candidates race baits his or her opponent by
saying the opponent will give white jobs to blacks. Even
then, the effectiveness of the tactic depends on the
particular skills of the candidates and other circumstances
surrounding that election. One would not expect a measure of
threat to predict evaluations without that disposition being
made salient. Similarly, the perception of threat might come
into play in the evaluation of a black candidate (touching on
the other fundamental element) even if economic threats were
not particularly salient(Hrt05) Symbolic racism would
covary with support for candidates who appeal to traditional
values (Hsr05)
Additionally, attitudes related to threat should guide
issue orientation when salient to the fundamental elements.
So, if one feels economically threatened one might be more
supportive of programs to help the unemployed, in general,
because this safety net would benefit the threatened
individual or his or her reference group (Hri>06)* On the

71
other hand, one would not expect the threatened individual to
support programs which specifically assist groups with whom
one perceives an adversarial relationship. So, threatened
whites, while supporting some spending programs like aid to
the unemployed, would oppose programs specifically assisting
blacks, their perceived competition (Hrt07)*
According to its proponents, symbolic racism should
predict evaluations of candidates and issues. Sears (1986)
points out that symbolic racism also explains positions in
non-racial issue areas. This, presumably, results from the
concept's wedding of conservatism and anti-black affect. To
the extent that symbolic racism is a form or expression of
prejudice one would expect it to have a more pronounced
effect in situations which involved racial objects. In other
words, a measure of symbolic racism should explain or predict
especially well when the object (or dependent variable) is a
black candidate or the issue has racial content(Hsr05-07)4
Defining the Nature of Interests
As discussed in the previous chapter, the proponents of
symbolic racism test their approach against a single
4 Of course, these expectations are based on the theoretical
positions stated by the proponents. The discussion in the
previous chapter revealed the symbolic racism theory had not
been tested in the national or Southern samples. Had it
been, the theory may have been developed somewhat
differently. The notion that symbolic predispositions
influence current attitudes and behaviors is reasonable
enough, but to argue that these symbolic predispositions are
completely independent from the context in which individuals
find themselves oversteps the bounds of reasonable
expectations.

72
alternative hypothesis, namely narrow self interest defined
in largely objective terms. Yet recent scholarship argues
for a broader conception of interest and identity. For
example, a burgeoning social movements literature sees the
development of group consciousness (identity with a group
beyond the self) as an essential element in explaining
movement genesis (McAdam 1982; Zald and McCarthy 1977).
Veritable cottage industries have sprung up to explain
different kinds of consciousness from gender (Rhodebeck 1992)
to race (Jackson 1981; Shingles 1981; Miller, Gurin, Gurin
and Malanchuk 1981), even to identification with businessmen
(Miller, Gurin, Gurin and Malanchuk 1981); these studies
found significant links between group identity and political
behavior and various forms of political participation.
Similarly, much of the extensive group conflict literature
(while maintaining that objective material interests between
groups are at odds) argues variously that group membership or
identity are essential elements of that conflict (e.g. Baker
1975; Blauner 1972; Isaacs 1975; Schemerhorn 1956; Van den
Berghe 1981; Wilson 1973). Giles and Evans (1985) found that
levels of racial hostility were positively related to levels
of white and Anglo-Saxon identity.
Given the weight of theoretical and empirical evidence
documenting the importance of group identification and
consciousness, one cannot conclude that self-interest
represents the sole conduit for the threat mechanism.
Indeed, one expects that individuals would be sensitive not

73
only to personal threats (challenges to the individual's job,
economic status, security, or political influence) but to
similar threats to their families and even their race in
general. Further, the perception of a more general threat
should predict other political attitudes and behaviors.
Measures of symbolic racism have not been tested against more
general orientations or representations of threat.
Synthesis
The empirical and theoretical controversies discussed
above impels an effort at reconciliation. For example, some
of the research discussed in the second chapter argues that
white hostility toward blacks results from real economic
threats posed by blacks. Others argue that instrumentally
marshaled racial hostility prevents a poor white-black
coalition from upsetting the status quo.
While neither of these propositions are tested in this
study, the mechanism underlying them is. The belief that
blacks pose economic, status or power threats, rather than
the essential material conditions, may lead individuals to
have hostile attitudes or to act on existing prejudicial
attitudes. Material conditions could make believers out of
many, but those conditions and consciousness of them need to
be disaggregated.
For example, imagine four individuals working for a firm
likely to change the composition of its work force in
response to a boycott by a minority group: the first worker's

74
job is threatened because of the changes and he or she knows
it; the second one's job is threatened, but he or she thinks
(wrongly) it is secure; the third one's job is secure, but he
or she thinks (wrongly) that it is threatened; the fourth
worker's job is secure and he or she knows it. Despite the
difference in the reality of threat posed to the workers, it
is evident that there is no reason to expect workers one and
three to respond differently to the perceived threat to their
jobs. Other things being equal, one would expect similar
response among those who have the same perceptions concerning
the threats (real and imagined) to their job status.
What expectations are reasonable concerning the two
workers who perceive their status as secure? Narrow self
interest predicts those workers would not harbor hostility
toward the minority group. After all, their interests remain
unaffected by the shake-up in the company (at least by the
standard of self-interest employed by the proponents of
symbolic racism); they still have their jobs. If, on the
other hand, interests are more closely related to group
identification, then despite the perceived (rightly or
wrongly) security of their positions, they may harbor hostile
attitudes. They may harbor those attitudes because the
shake-up threatens, not them personally, but people like
them, people with whom they identify. Or they may feel
threatened because the environment has simply become less
secure; they could lose their job next.

75
In this chapter, discussion has drawn attention to ways
in which political and racial attitudes may be sensitive to
and distinct from actual material conditions (as with the
worker who incorrectly perceived her job was threatened).
People's perceptions of their condition may diverge from the
actual circumstances, but they are not divorced from those
circumstances. In the above example, the climate of change
led the actually unthreatened worker to perceive that his or
her job was less secure than it was. The context, in which
co-workers with whom he or she worked were likely to lose
their jobs, heightened that worker's sensitivity to the
perception of threat. This sense of threat occurs even
though there may be no actual threat to the worker's job.
Behavior is predicated on perceptions that may be inaccurate.
And those perceptions, with their attendant 'measurement
errors,' develop within a context. In the next section,
discussion focuses on conditions which inform the context of
racial attitudes, paying particular attention to the changing
interpretations of local racial composition.
Context and the Changing Meaning of Racial Composition
Some of the studies discussed in the second chapter
point to different theoretical significances of the local
racial composition. To some (e.g. Giles and Evans 1985) the
local composition represents a contemporary and actual threat
to the lives and livelihood of whites. To others (e.g. Key
1949), racial composition represents a proxy for a labor

76
system with a particular set of class relationships.5 Are
these interpretations reasonable in the contemporary context?
Is the racial composition a threat in and of itself or is it
a measure of something else? Operationally it provides a
shorthand measure to describe the relative number of blacks
and whites living in a given geographic area. From this, one
can not necessarily conclude that composition in and of
itself represents some threat. But the local racial
composition may be correlated with other salient but
unmeasured details of the locality.
The contemporary racial composition is correlated with
(see later chapters) racial attitudes in the expected ways
but not necessarily for the same reasons that group conflict
proponents argued. The first element of the contemporary
significance of the local racial compositions may found in
its representation as a resource for the mobilization of
black interest groups (McAdam 1982, Blalock 1965). McAdam,
in particular, details the effects of the urbanization of
black America as a resource for the mobilization of interest
groups and social movements. In a nutshell, the
concentration of blacks in the urban setting allowed for the
recognition of common interests, higher incomes than those
available on the farm, and the formation of civic groups and
5 Key (1949) wrote before the great migration was complete;
racial composition was correlated with the ruralness of
Southern counties. Giles' (Giles 1977; Giles and Evans 1985,
1986, Giles and Buckner 1993) work addresses a post-migration
America where the percentage black may be more correlated
with level of urbanity.

77
churches, which in turn were able to recruit and maintain
sophisticated leaders (e.g. educated ministers). The same
elements that allowed increased concentrations of blacks in
the cities to generate successful social movements may
aggravate white hostility to the extent that movement
objectives and tactics threaten the economic and political
positions of whites.^ in this light, the higher percentage
black would elevate the sense of threat among whites.
Indeed, because the processes described by McAdam are largely
cumulative one would expect a nonlinear relationship between
the percentage black and the level of threat. Graphing this
relationship would require a quadratic function with an
increasing slope (see figure 4-2).
Similarly, the relative ability of blacks to mobilize
other resources may influence the degree to which whites feel
threatened. McAdam (1982) and others in the social movements
paradigm (particularly the resource mobilization wing) pay
particular attention to the ability of nascent movements to
generate money as a measure of that movement's strength. In
this sense, one might expect the relative wealth of the black
community to correlate directly with the level of threat
perceived by whites. Blalock (1967) argued that the sense of
competition was most heightened among those closest to each
other in class position (Hrt08)* The relative or comparative
level of black income vis-a-vis white income should be an
6 For example, Bobo and Gilliam (1990) found that black
empowerment in major city governments was linked with white
frustration with those governments.

78
Figure 4-2. Quadratic function with increasing slope

79
effective predictor of threat. Of course, one might also
reasonably expect a measure of income equality to correlate
inversely with racial hostility (Jacobs 1982). Comparability
of income may be an outgrowth of a harmonious racial
atmosphere rather than a cause of a toxic one (Hsyn02) On
the other hand, such arrangements could just as easily
promote mutual respect or at least the displacement or
loosening of stereotypes (Hsyn05)*
A second way in which the contemporary racial
composition may influence white attitudes lies in increasing
the probability of actual conflicts. In a racially monotonic
community, one would expect zero cases of racial conflict of
any kind. The probability of inter-racial disputes (whether
caused by racial hostility or not) could be expected to
increase as the percentage of each racial group approached
50% (see figure 4-3).7 This would produce a curve shaped
like an inverted U centered on 50%. If symbolic racism is
truly abstract, no correlation would be found because of
increasing opportunities for inter-racial conflict (Hsroi)*
7 Conflicts occur between citizens all the time no matter
what their race. The increase in transactions between people
of different races does not necessarily imply that conflicts
resulting from those transactions result in racial hostility.
Indeed, other things being equal, the increase in the number
of inter-racial transactions may result in a decline in
racial hostility because of the increase in the number of
positive outcomes from those transactions. However, given
the history of racial hostility in America, one would expect
(as has been observed) short term increases in hostility.

80
Figure 4-3 Parabolic quadratic function

81
A third way in which the current racial composition may
accentuate hostility among whites operates through
visibility. Potentially threatened whites may be more likely
to latch on to racial explanations for their difficulties if
they have reminders of the existence, size, and power of the
other race. So the more visible the minority population the
more likely they would be blamed for problems faced by
whites. Again, one might expect a nonlinear relationship
between percent black and the level of threat for these
reasons. The level of threat would increase with increasing
slope as the percentage black increased (like figure 4-2).
The proponents of symbolic racism might argue that
increased visibility of the minority population should have
no effect on the level of symbolic racism. However, the
visibility argument sketched out above leads one to expect
the level of symbolic racism to increase with the percentage
black to some threshold level and level off thereafter
(HSYN03)- If one relaxes the assumption of abstractness, one
might expect the symbolic proclivities to increase slightly
as the percentage black rose to some level of visibility,
making blacks an identifiable segment of the population.
After blacks are identified incremental changes should have
little effect if the symbolic attitudes are largely abstract
(see figure 4-4).

82
Figure 4-4. Quadratic function with gradual decay

83
The Archaic Significance of Racial Composition
The contemporary significance attached to racial
composition by scholars of group conflict differ dramatically
from Key's (1949) reasoning. From Key (also Wright 1977,
Mathews and Prothro 1963) one infers that past racial
composition affects the current racial climate in localities
and states. Introducing this line of reasoning requires the
review of some arguments made in more telegraphic form
earlier, while focus of this project concerns white racial
attitudes in America generally, examining the Southern case
proves instructive.
Key (1949) bases his argument on observations of a
region in which a particular economic system reigned with a
peculiar set of labor and social institutions; however, King
Cotton died. Those institutions no longer dominate the
Southern economy. Therefore today's racial composition
cannot have the same immediate significance today that the
racial composition had when Key (1949) published Southern
Politics. But the archaic racial composition may reveal
information about local norms, beliefs and values, or a
political culture crystallized in an earlier era. The
transformation of the Southern economy and the dramatic
demographic shifts make Key's (1949) argument less directly
applicable to the current political situation.

84
As McAdam (1982) points out, the economic system in the
South underwent a series of revolutions in the early part of
this century. New machinery reduced the need for cheap black
labor. New pesticides had a similar effect. Most
importantly the collapse of the international cotton markets
made the system of cotton dominated monoculture less (or un)
profitable by the beginning of World War II. Even as Key
(1949) wrote, the basis of his argument was changing (or had
changed). Despite the revolution in the Southern economy,
changes in the political system and the culture lagged behind
by forty years or more. What accounts for this lag? As
Tullos (1989) and Scott (1985) document in widely divergent
cases, changes in economic systems (in particular
agricultural revolutions) may change the class system but not
necessarily the relative positions of actors (or families)
within the new system. Powerful families may translate or
transform their advantages (education, capital and other
assets, prestige, and political power) from the previous
system into advantages in the new system. A revolution in
the economy does not produce an immediate social revolution.
This conservation of advantage forestalls the changes one
might expect after such a radical transformation as the
Southern economy experienced.
The structure of the political system and the entrenched
positions of its inhabitants would have delayed immediate
changes in the political system. The lack of a grass roots
organization in the Republican party made (make) challenges

85
to the hegemonic Democrats costly and difficult. Further,
skilled Democratic politicians slowed the expected growth of
the Republican party after the one-party system was no longer
useful (Lamis 1988). For example, George Wallace continued
to win elections into the 1980s, despite his well-earned
reputation as a racial extremist. Lamis (1988) credits
Wallace's formidable political skills with delaying the
election of a Republican governor by more than a decade.
Wallace's success points to an additional element in
this explanation of the slow pace of change, beyond the
structural factors noted above. This additional element lies
in the slow pace of changing values across generations in the
South. The values established during the era of King Cotton
continued to play a significant role in Southern culture and
politics long after that economic system collapsed. This
continuation of archaic values may have allowed Wallace to
harness fears of racial threats as a source of political
capital.
In this sense (or to the extent that this is true) one
finds agreement with the proponents of symbolic racism. That
is to say, the process of socialization results in the
inculcation of standards and values which may form the basis
for later political evaluations. But these values and
standards do not operate in a vacuum. So, it is reasonable
to expect regional variation in levels of racial hostility,
even when proximate explanations are controlled. In other
words, the archaic local racial composition should indicate

86
areas with traditions of heightened racial tension,
especially in the South. Of course, the contemporary and the
archaic racial compositions are not uncorrelated. In this
sense, the relationship between the contemporary local racial
composition and racial hostilities does not merely reflect
the current material context, but also reflects cultural
factors which may be unique to the region, state or
community.
The patterns of racial antagonism Key (1949) described
(using racial composition) continue to affect the current
state of race relations in the South and elsewhere. Of
course demographic changes and migration and the
transformation of the Southern economy have the effect of
altering the roots of antagonism in the South. The
information revolution and increased access to higher
education have muted the transmission of prejudicial values
form generation to generation. Still, one expects some
degree of continuity within states, counties and local
communities. Those raised in communities with traditions of
prejudice (or a prejudicial political culture) could be
expected to have more prejudicial attitudes than those raised
in communities without such traditions because those
traditions influence their socialization even if the
transmission of values is attenuated because of the factors
noted above.
Similarly, the local or state political culture
represents a crucial contextual factor for those who may

87
migrate from places with different traditions as adults. For
example, Fossett and Kiecolt's findings reveal that people
raised in the South but living in the North were not
significantly more likely to harbor prejudicial attitudes
than other Northerners (1989, discussed in detail in Chapter
Two). Evidently the context in which people find themselves
continues to matter after childhood socialization is
complete. And, the context dynamically shapes their current
values, beliefs and behavior on issues of race. Indeed,
[r]acial socialization does not end with adolescence."
(Kinder and Rhodebeck 1982, p. 205). So adults may be
influenced by the prevailing political culture and the
material conditions in their state and community.
In this sense, the current racial political culture is
shaped by two essential elements: 1) traditions, values and
beliefs established in the past; and 2) the current material
context. Contemporary measures of context may inform
research about both of these elements. For example, the
contemporary racial composition (and other measurable data)
of a community reflects to some degree the visibility of the
minority community and other factors discussed in the
previous section and may reveal (or contain) information
concerning the political culture or the weight of the past.
As discussed above, the contemporary racial composition
cannot mean the same thing that it did when Key wrote, but
the racial composition from that era can (Wright 1977,
Mathews and Prothro 1963a,1963b). Operationally, data drawn

88
from older censuses can inform research about current racial
attitudes by providing a cultural backdrop, pointing to
crucial cultural information from the past which may
influence contemporary racial attitudes (Hsyn04) The
influence of the current racial composition on racial
attitudes reflects the extent to which those attitudes
reflect the sort of social dynamic outlined by social
conflict or racial threat theorists (HrtOI)*
The argument presented above dulls the otherwise sharp
distinction between analysis of racial attitudes based on a
symbolic politics model and models more firmly grounded in
material conditions. It is inaccurate to argue that the
basis of racial hostility must be one or the other. Material
conditions influence the values held by previous, current and
future generations because of the methods of transmission
from parent, teacher, and supervisor to child, student and
employee. These values provide a lens or optic through which
current material conditions are viewed. Similarly, current
material conditions influence the salience or relevance of
that lens in a given situation. Perceptions, beliefs and
values cannot be completely divorced from the material world,
as the proponents of symbolic racism have claimed.
In the next chapter, discussion focuses on how one might
measure and test the propositions laid out in this chapter.

CHAPTER 5
DATA
In the last chapter, discussion centered on the
reconciliation of theoretical dilemmas posed by two
mainstream schools of thought in the study of whites' racial
attitudes. Additionally, hypotheses were derived from
theoretical propositions. In this chapter discussion turns
to issues concerning the testing of those hypotheses and the
measurement of the constituent elements therein. Four topics
need coverage in this chapter: 1) the sources of the data
used; 2) how the two racial attitudes outlined in previous
chapters are measured; 3) how variables that predict those
attitudes are measured; and 4) how variables that are
predicted by the racial attitudes are measured. Details
regarding the measurement of more commonplace variables are
provided in the appendix.
Data
The data exploited in this study are drawn from three
sources. First, survey data are taken from the 1986 American
National Election Study (ANES). Additionally, two types of
census data are used. Contemporary census data are drawn
from the 1980 Census of Population and Housing; the specific
data used are from the electronic version of STF 3C. That
89

90
summary file will be used to calculate the racial composition
for the SMSAs and counties represented in the national sample
and to collect other relevant contextual data.1 These data
are collected at the SMSA level for respondents residing in
SMSAs and at the county level for others. SMSAs, for those
who live in them, are thought to be the defining local unit
rather than counties or congressional districts. SMSAs tend
to have area-wide media markets reinforcing identification
with the urban center rather than with constituent counties.
County level data are used for respondents not residing in
SMSAs. County level data represent the highest level of
precision available for merging since census tracts and minor
civil divisions have been padded (overwritten with nines) in
the 1986 ANES.
Archaic census data are drawn from the ICPSR
reproductions of the U.S. Census of Population and Housing 1%
counts from 1900 and 1940 (ICPSR study number 0003). The
100% counts are available in neither electronic nor card
format for either the 1900 and the 1940 censuses. The use of
the sample counts should make little difference in the
project at hand since only minor differences between the
sample and 100% counts are likely. Additionally, any
variations between the two counts should not produce
systematic biases.2
1 The 1990 National Summary File is not yet available, so
interpolation for maximum accuracy is not yet feasible.
2 Theoretically both the 1% and the 100% counts are samples,
because the survey used to produce the 100% count cannot
actually account for everyone. Unfortunately, the 100%

91
Neither the 1940 nor 1900 census provide SMSA level
data. Many of our contemporary SMSAs were not remotely
metropolitan in those years, so many counties now located in
SMSAs were then only counties. Since SMSAs are constructed
of whole counties, one could use the current structure as a
template for aggregating older county level data. Such a
post hoc aggregation is not theoretically warranted. The
theoretical justification detailed above for using SMSA level
data from contemporary censuses cannot apply to data as old
as these for a variety of reasons, including the more
fragmented nature of media markets in earlier eras. Data
drawn from those two years will be at the county level and at
the state level (discussion follows).
Dependent Variables
The choice of items for the racial scales was based
primarily on face validity, the theoretical assumptions
outlined above (see the Appendix for question text) and
consistency with previous studies. The questions scaled to
tap perceptions of racial threat address issues of whites
losing jobs, promotions or school admission as a result of
affirmative action programs or employer choices. The
symbolic racism items are limited to those items which
address issues of black adherence to the Protestant work
sample is not used as a predictor of the population but is
taken as the population. In this sense the 1% sample may
produce a more accurate or at least as good a predictor of
the population because of the care taken in sampling and the
extrapolation of that sample.

92
ethica central theoretical pointand the civil rights push
question. This last question may at first seem to address
white perceptions of a threat from black leaders, but closer
examination reveals that the focus or object of the question
is abstract: "the civil rights people," like "those people"
are objects of projected stereotypes. All of the symbolic
racism questions have been used by the proponents of the
concept in their own studies. The perceived racial threat
scale has not previously been used.
Not all cases from the 1986 ANES are utilized for this
analysis. Non-white respondents were excluded, since the
focus of the project is to understand white attitudes. The
1986 ANES was a split-form year. Most of the questions used
to construct the racial scales were limited to form B, so
form A respondents were excluded. Assignment to the forms
was random, so the use of respondents from one form or the
other creates no systematic bias in the sample.
The effects of interviewer race on respondents has been
well documented in cases where black respondents were
interviewed by white interviewers (Anderson, Silver and
Abramson 1988a, 1988b; Schuman and Converse 1971). To
prevent the reverse effect, the few white respondents with
non-white interviewers were excluded to avoid interviewer
bias. Indeed, the average score for white interviewers was
lower (more symbolically prejudiced or threatened) than
scores for respondents with non-white interviewers. This was
more often a problem for threat items than symbolic racism

93
items. T-tests for some of these items are displayed in
Table 5-1. Note that three of the items yielded significant
differences by the interviewer groups. Accordingly,
respondents with non-white interviewers were excluded from
the analysis.
The questions used to construct these racial scales are
highly charged. One may wonder if respondents would be
comfortable revealing their positions on these touchy issues.
The frequencies for these items are reported in the Appendix.
There we see that between 12.5% and 32.8% felt comfortable
enough to choose the most extreme positions. Typically 41.4%
to 66.4% selected one of the two most extreme responses.
These figures imply that the respondents do
Table 5-1. T-test for race of interviewer bias on whites
Civil rights push
white interviewer
non-white interviewer
mean
2.60
3.06
pooled t-
value
-2.27
P
.02
separate
t-value
-2.24
P
.03
white admission
white interviewer
non-white interviewer
3.05
3.61
-2.05
.04
-2.04
.05
family admission
white interviewer
non-white
interviewer
3.32
3.83
-1.88
.06
-2.03
.05
white promotion
white interviewer
non-white interviewer
2.95
3.35
-1.58
.11
-1.43
.16
family promotion
white interviewer
non-white interviewer
3.36
3.67
-1.11
.27
-1.10
.28
Source: 1986 ANES

94
not feel so encumbered by pressures to be politically
correct that they are unwilling to take strong positions.
Principle components analysis of the items used in scale
construction confirms the validity of the racial scales. Two
factors were extracted, together explaining 59.7% of the
variance. The rotated (vari-max) factor loadings are
presented in Table 5-2. The questions addressing the loss of
jobs, promotions, and school admission for whites as a group
or members of the respondents family load on the first
factor. The symbolic racism questions load on the second
factor.3 The loading of these variables on the two factors
Table 5-2. Factor analysis for racial items among whites
with white interviewers
variable
white jobs
white admission
white promotion
family admission
family promotion
factor la factor 2b
57838
.19123
77987
.22617
78021
.29666
80092
.15430
78838
.11088
civil rights push
without welfare
Irish,Italians,etc
try harder
.30130
.15733
.20871
.13206
.50797
.76681
.80920
.81751
Principle components analysis produced two factors. Varimax
rotation converged in three iterations.
a eigenvalue=4.00 explaining 44.5 % of variance
b eigenvalues.37 explaining 15.2 % of variance
Source: ANES 1986
3 Maximum likelihood estimation confirms the results
presented here. Similarly, the use of an oblique rotation,
which does not require the assumption of orthogonality
between factors, also confirms the results presented here.

95
implies that the two concepts, racial threat and symbolic
racism, are conceptually and empirically distinct. One could
score high on the symbolic racism scale and low on the racial
threat scale, or any combination of the two. This confirms
hypothesis HsynOI/ listed in Table 4-1.
Each group of items was combined to produce linear
additive scales for each of the orientations. The range of
these scales is sufficient to allow the use of Ordinary Least
Squares (OLS) regression.4
In the next section discussion turns to variables which
are used to explain the racial scales.
Contextual Variables
One of the central criticisms of the symbolic racism
literature levied in the preceding chapters regards its lack
of sensitivity to the context in which political behavior
occurs. With the exception of Kinder and Sears (1981), few
of these studies utilize variables which attempt to tap the
material, geographic or culture context in which the
respondent is situated. With this in mind a number of
contextual variables derived from county and SMSA level
census data will be included in this study. These data from
the censuses of various years are merged with survey data
from the 1986 ANES.
4 The threat scale ranges from 5 to 26 and has 729 valid
cases. The symbolic racism variable ranges from 4 to 20 and
has 809 valid cases. Directionality of the scales was
corrected by taking the inverse of each (dividing each summed
value into one).

96
1980 Census
A number of contextual variables from the 1980 census
will be included in the analysis. For example, the local
racial composition is calculated by dividing the black
population by the total population. The economic context is
gauged by a measure of income equality. This variable is
designed to tap the relative economic strength of blacks in a
given locality. The mean black income in an SMSA or county
is divided by the mean white income in the same county.
Some previous studies (e.g. Giles and Evans 1985) have
tried to tap economic strength by using the difference of
mean incomes. The difference of mean incomes has good
features but still may be subject to local variations in the
value of the dollar. For example, in an area where incomes
are generally low, say the Columbus, Georgia SMSA,
differences in mean incomes would be lower than in areas
where incomes are higher, say Los Angeles. So, a difference
in average income of $500 in Columbus, may reflect a great
disparity of economic power, while a difference of $1000 in
Los Angeles might not reflect a significant disparity of
income at all. In this sense, the difference measure is not
interval and may not even be ordinal. The quotient of the
mean black income and the mean white income reduces the
sensitivity of the income measure to local variations in the

97
value of the dollar. The lower the quotient the greater the
income disparity in favor of whites.5
Modernization is expected to have an impact or racial
attitudes. One expects to find a variations in context among
different types of communities. Button (1989) has
demonstrated that the pattern of race relations in southern
communities often varies along a New South/Old South
continuum, with New South communities reflecting more
progressive patterns. It is not possible to replicate
Button's qualitative approach when analyzing national
surveys, but it is possible to construct proxy variables
which quantitatively tap the national differences
corresponding to the New South/Old South continuum, like
urbanity.
But measuring urbanity presents thorny issues. For
example, the suburbanization of America may render measures
based on population density moot. People living in sprawling
and elite suburban communities may have little in common with
farmers or small town folk in areas with identical population
density or belt codes. To address problems like this, two
variables which measure theoretically distinct elements of
modernity and urbanization have been designed: the percentage
of the labor force employed in agriculture and nativity.
5 To expand the example, if the mean incomes in Columbus
were $11,000 and $11,500 the difference is $500 and the
quotient is .956. If the mean incomes in Los Angeles were
$33,000 and $34,000, the difference would be $1,000 and the
quotient .971.

98
Nativity is associated with the influx of people and the
influx of newer and more diverse ideas. This measure is
calculated by dividing the number of residents born in the
state by the total population for that area. The result is
the proportion native to that area. One expects areas with
more in-migration to be more racially progressive.
1900 and 1940 Censuses
In addition to the variables drawn from the 1980 census
other census data are exploited. One expects the weight of
the past to continue to affect contemporary racial attitudes.
Demographic data drawn from previous eras are expected to
provide a gauge of the existence of such legacies (see
previous chapter for a more detailed discussion of this
topic.) Within the catalogue of extant censuses, two years
are of particular theoretical import. The 1900 census
represents the first comprehensive accounting of the nation's
population after the overthrow of the regime of the
reconstruction era in the South. In this sense, data from
this year represents a baseline of data for the Jim Crow era.
Furthermore, these data predate changes accompanying the
demise of the cotton economy.
The 1940 census provides the most recent counting of the
nation's population before the commencement of large scale
protest and other civil rights activities garnered extensive
national coverage. This argument follows Wrights (1977)
contention that the desegregation of the military after World

99
War II represented the first act of the nation's civil rights
drama. After that point, the values of individuals were more
subject to influence from either side in the ensuing
struggle. Data from 1940 should provide a snapshot of the
immediate post cotton era but precede the dawning of the
national civil rights movement.
Key (1949) implies that the states have unique political
attributes. In this way the state represents more than an
(often) artificial line around a large territory. Rather,
states develop unique cultures of their own. This has much
to do with the hand states are dealt in terms of natural
resources and the like. But the states' culture also follows
from the importance of states as political units. In this
sense, state politics influence not only the political
culture in the state but even culture in general. Surely the
converse is true as well, that each state's politics are
influenced by the exigencies of the (political) culture.
Still, Key argues many common traits in the South influence
the region's and the states' political development; it is in
this sense that Key spoke of universal effects like the
'friends and neighbors vote' and regional effects like the
cotton economy within the context of state politics.
So, dirt farmers from North Alabama and North
Mississippi have a lot in common, but the fact that they hail
from different states is important. Similarly, dirt farmers
from those two Deep South states (i.e. states entrenched in
the cotton economy) have less in common with dirt farmers

100
from border South states. More to the point, the politics of
the Black Belt affected whole states, not just the counties
which exploited cheap black labor. States represent more than
mere political units. Rather, states represent areas with
specific and identifiable cultural and political traditions
which provide the context within which ongoing political
processes are played out (Conway 1968). Further, states
provide a backdrop for the socialization process.
Wright (1977) found that the addition of state level
contextual data gave his models significantly more
explanatory power than reduced models. He attributes this to
the effect of reference groups on the formation of attitudes
and vote choices. In other words, the state level data yield
information about the adult socialization process in each
state. His position is consistent with the synthesis laid
out in the previous chapter. In addition to the archaic
county-level census data, state-level data will be drawn from
both the 1940 and the 1900 censuses.
The data available in older censuses are not completely
consistent with that of the current censuses. The percentage
of the local or state population made up of African-Americans
was extracted from both the 1900 and the 1940 censuses. The
data were coded to match as closely as possible contemporary
standards concerning who is black which are based on what
some call the rule of hypo decent (see Fredrikson 1981,
Chapter Three for a discussion of the evolution of this
'rule'). For example, in the 1900 census several different

101
categories were provided for people of racially mixed
ancestry each of which had a technical definition (e.g.
mulatto, griffe, quadroon, octoroon) based on the
distribution of the mixture. By today's (culturally
dependent) standards and those used universally in the
research discussed in the previous chapters, all those
categories are subgroups of the general category black.
The 1900 census has a category called 'Colored' which
refers to people of color from places other than Sub-Saharan
Africa. These people might be thought of as Asian, Semitic,
Mestizos, Latin Americans, Filipinos or any of a number of
(typically) nationality-based titles not valued at the time.
The same category name in the 1940 census refers to people
with any (visible) African ancestors and as such is
consistent with contemporary standards of racial distinction.
The percentage black is calculated at both the county and
state level.
An additional variable is drawn from the 1900 census to
gauge in an additional way the effect of the cotton economy
on racial attitudes in the South in particular. The average
farm size at the county and state level should give
information about the prevalence of large farms in those
units. When combined (as an interaction term) with the local
racial composition the result should provide a measure of
plantation-economy-ness or latifundization. This will prove
particularly important for the discussion of the Southern
region in Chapter Seven.

102
Of course, the county level data from 1940 and 1900 are
highly correlated. This correlation makes it impossible to
use both in the regression models presented in the next
chapter. Since they are correlated, one suspects that one
might be as good as the other in explaining current levels of
the racial attitudes discussed above. Examining the partial
correlation coefficients for each year's data on the racial
scale while controlling for the contribution of other years
should give an indication of the veracity of this assumption.
Table 5-3. Black Concentration in 1900, 1940, and 1980 and
the levels of Perceived Racial Threat and Symbolic Racism in
1986 for White Respondents Only.
Controlling for
the % black 1900
RT by
1900
%black
a
SR by
1900
%black
RT by
1940
%black
0.0443
SR by
1940
%black
0.0615*
RT by
1980
%black
0.0588b
SR by
1980
%black
0.0389
Controlling for
the % black 1940
0.0143
0.0101
Controlling for
the % black 1980
0.0942**
0.1372***
0.0913**
0.1445***
0.0480b
0.0163
a Entries are first order partial correlation coefficients.
RT=Racial Threat
SR=Symbolic Racism
b p<0.10
* p<0.05
** p< 0.01
*** pcO.OOl
Source: 1986 ANES, 1980 Census, 1940 Census, and 1900 Census

103
Indeed, Table 5-3 indicates the data from 1900 and 1940
have virtually identical effects on the level of perceived
racial threat and symbolic racism (see the far right column),
while controlling for the effect of the contemporary racial
composition. Because the effect is so similar and because
the 1940 data are available for more counties, the 1940
county level data will be used in the analysis presented in
the next chapter. Of course, this table has a substantive
interpretation as well. The archaic racial composition has
significant explanatory power for both racial scales even
when the effect of the contemporary racial composition is
controlled. This implies that the historical context of the
community in which ones lives has some explanatory power for
even symbolic racial attitudes. Similarly, the perception of
threat is not merely a function of the current context.
The local political culture influences the perception of
threat as well. Still, the current context does make a
unique contribution toward the explanation of the perception
of threat even when the historical context is controlled.
The next to last row substantiates this contention in the
predicted direction (making the .10 level the appropriate
standard of significance). The findings reported in Table 5-
3 provide support for hypothesis Hsyn04, namely that the
racial attitude scale will covary with archaic levels of
percentage black. Further, the relationship between
contemporary racial attitudes and archaic measures of racial
composition holds even when the contemporary percentage black

104
is controlled. This hypothesis and others will be explored
more thoroughly in the following chapters.
Endogenous variables
While the primary focus of this project concerns the
theory and measurement of racial orientations, it is useful
to demonstrate that these racial orientations influence
broader political concerns. That is to say, it is important
to demonstrate that racial orientations influence the
individuals' political lives and through their actions the
greater community. As such, it seems appropriate to attempt
to measure the effects of racial orientations on policy
positions and other political evaluation. The usefulness of
perceived threat and symbolic racism as explanatory variables
is gauged on two types of dependent variables taken from the
1986 ANES (question text in the Appendix): Policy positions
and candidate evaluations.
For the sake of comparison, symbolic racism and
perceived racial threat will be used to predict a variety of
issue positions such as federal spending on minorities, the
unemployed, food stamps, and space research. All of these
are available in the 1986 ANES.
Similarly, since 1986 was not a Presidential election
year, the 1986 ANES does not include candidate evaluations
for presidential candidates. It does contain feeling
thermometers for national political figures. It is expected
that racial orientations will have differential effects on

105
the respondent's evaluations of national political figures.
How people rate such figures is important because such
ratings have been directly linked to vote choices (e.g. Page
and Jones 1979). For this variable, the Jesse Jackson
feeling thermometer is subtracted from the average rating of
Mario Cuomo and Robert Dole. This is intended to tap the
respondents rating of Jackson in light of the respondent's
rating of political figures in general. Cuomo and Dole are
selected because they represent prominent leaders in each of
the two major parties. This measure might be improved by
using more leaders (objects) to improve the accuracy of the
mean from which the Jackson rating is subtracted. However,
increasing the number of figures used in the calculation
increases the number of missing cases. Further, Gary Hart
represented the only Democrat rated besides Cuomo and
Jackson.
Group evaluations follow a similar method. For this
variable, the feeling thermometer rating of blacks is
subtracted from the rating of poor people. While poor people
do not represent an especially popular group, neither do they
represent an especially threatening group.
Multivariate tests of the effects of the racial scales
on these endogenous variables are discussed in greater detail
in the following chapter.
This chapter has focused on issues concerning the data
used in the present study. The data sources have been
outlined. The construction of scales used to measure the

106
dependent variables has been discussed. Similar treatments
of the contextual and endogenous variables were provided.
Further, evidence was brought to bear on two hypotheses
outlined in Chapter Four (HsynOI and Hsyn04)- The two racial
scale were found to occupy different dimensions, as
predicted, and the significance of archaic racial composition
was confirmed.
In the next chapter discussion turns to address modeling
procedures and hypothesis testing.

CHAPTER 6
NATIONAL MODELS OF PREJUDICE AND ITS EFFECTS
In the last chapter discussion focused on two related
issues: the data used in this project and its appropriateness
for the same. Special attention was paid to identifying
sources of bias. Additionally two theoretically distinct
racial dimensions were identified. This chapter builds on
the last by modeling the sources of those racial attitude
scales using multivariate procedures. Similarly, the effect
of the racial scales on other political attitudes and issue
positions is gauged through multivariate procedures.
Procedures
The last chapter discussed the construction of two
racial attitude scales. The ranges of these scales allow the
use of Ordinary Least Squares (OLS) regression.1 OLS allows
measurement of the independent effects of salient variables
while controlling for the effects of other salient variables.
Another advantage of OLS is its ease of interpretation and
familiarity to many readers. The scales are modeled in
similar fashion:
1 The threat scale ranges from 5 to 26 and has 729 valid
cases. The symbolic racism variable ranges from 4 to 20 and
has 809 valid cases. Directionality of the scales was
corrected by taking the inverse of each (dividing each summed
value into one and multiplying by 100).
107

108
y=a+piXl . +P4X4+P6Z1 . +piOZ5+rxiZi . +rxjZj+e
where:
y = either the racial threat or symbolic racism scale
Xl=Age
X2=Education
X3=Income
X4=evaluation of the national economic performance
X5=evaluation of personal economic performance
X6=Ideology
Zl=South
Z2ij=Percent black 1980,1940,1900
Z3=Percent employed in agriculture
Z4=Income equality quotient
Z5=Nativity
XiZi to XjZj = all possible combinations of Xs and Zs.
Normally one would check all the two-way interactions as
a matter of course. When interaction terms are significant,
special estimation procedures are required. Typically one
would re-estimate the model for different levels of one of
the constituent variables. Most often, one finds minor
differences in slope on the other variables and only
occasionally opposite signs. Here, one has other reasons for
the inclusion of interaction terms. For instance,
interaction terms can be an effective way of modeling
contextual variables. Boyd and Iverson (1979) provide an

109
extensive treatment of this technique (see also Iverson 1991,
Goertz 1992).
In the research reported here, most of the two-way
interaction terms are theoretically and statistically
insignificant. One interaction term has particular
theoretical significance and is included in the models
estimated and presented in this chapter, the product of a
South dummy variable and the percentage black 1940. As noted
above the appropriate treatment for statistically significant
interaction terms involves re-estimation of the model at
fixed levels of one of the constituent variables. This
treatment is carried out and discussed in Chapter Seven.
In addition to the tests for interaction effects three
other variables are included in the analysis of the origins
of the racial orientations. The first is a measure of the
degree to which the state economies were dominated by
plantation style agriculture in 1900. This variable is
constructed by taking the product (multiplying) the average
farm size at the state level in 1900 by the percentage black
at the state level for the same year. The second variable,
as discussed in Chapter Four, is the square of the percentage
black in 1980.
The third additional variable is the amount of change in
the percentage black at the county level over 40 year periods
between 1900 and 1940 and 1940 and 1980. For example, this
is computed by subtracting the percentage black in 1900 from
the percentage black in 1940. Both of the possible

110
differences (e.g. percentage black 1980 minus 1940, 1940
minus 1900) in the three percentage black time points were
tried in earlier versions of the models presented here.
Neither proved particularly statistically significant, but
the difference between the 1940 and 1900 level has particular
theoretical significance. As an independent variable it
represents the effect of demographic shift before the birth
of the civil rights movement. McAdam illustrates the
importance of the shift in civil rights movement genesis and
longevity. The findings presented here indicate that the
demographic shifts in and of themselves had little direct
effect on white attitudes. Two explanations follow from
this. First, the degree of shift evident in 1940 had not yet
reach a threshold level that was necessary for migration to
have direct effects on white communities. Second, the
demographic shift's primary effect on American race relations
was as an input to movement genesis; therefore no direct
effect should be expected.
The models of the racial scales are presented in stages
to demonstrate more clearly the independent effects of the
types of variables included in the models. First, the racial
scales are regressed on a group of contextual variables.
Then the individual level variables (age, education, income,
personal financial evaluation, and conservatism) are added.
A dummy variable coded one (1) for the respondents living in
the South and zero (0) for all others rounds out the third
model. The final model for each scale includes all these

Ill
variable plus the two interaction terms, the square of the
percentage black 1980 and the difference variable discussed
above.
Models Explaining the Origins of Racial Attitudes
The models for symbolic racism are presented in
Table 6-1. Equation I presents the contextual variables.
Three things are noteworthy in this equation. First, the
percentage black at both the state and county level for 1940
are positively related to the symbolic racism scale. This
seems to indicate that the historical context has some
effect, albeit a weak one, on the level of symbolic racial
attitude despite the theoretical claims of abstraction.
Similarly, the significance of the agricultural employment
variable has some influence on the symbolic scale, indicating
some sensitivity to the current context (although this may be
an indirect effect, see below). Finally, note that the
contextual variables alone explain relatively little of the
variance, about 7%. These relationships change somewhat when
the individual level variables are added to the equation.
Equation II in Table 6-1 presents the next model
with both contextual effects and individual level variables.
Note that education and conservatism have dramatic direct
effects on the symbolic racism scale; increased levels of
education reduces the level of symbolic racism, while
conservatism is associated with higher levels of the
prejudicial attitude. The inclusion of the individual level

112
Table 6-1. Symbolic Racism Regressed on Contextual and
Individual Level Variables for White Respondents Only.
entries are unstandardized coefficients from OLS regression
Variables
I
II
III
IV
%black 1940
0.05b
0.03
0.01
0.13*
State %black40
0.000001b
0.000002b
-0.0000003
-0.0000002
%black 1980
0.005
0.08*
0.11**
0.13b
%employed in agriculture
0.25b
0.08
0.09
0.02
nativity
-0.01
-0.03b
-0.02
-0.03b
income equality
-1.48
-2.36
-2.16
-3.27*
v595 age

0.02b
0.02b
0.02
v602 education

-0.99***
-0.99***
-1.00***
v733 family income

-0.07b
-0.07b
-0.07b
v356 personal finance

0.06
0.05
0.05
v385 conservatism

0.63***
0.62***
0.61***
south


1.55b
3.40*
%black 1900* average farm
size 1900



-0.0001
%black 19802



-0.002
south*%black40



-0.15**
%black 1940-%black 1900



0.05
constant
adj R2
N
11.51***
0.07
758
14.03***
0.23
558
13.54***
0.23
558
14.99***
0.24
540
Source: 1986 ANES; 1900,
b p-value < 0.10
* p-value < 0.05
** p-value < 0.01
*** p-value < 0.001
1940 and 1980 Census

113
variables allows the observation of indirect contextual
effects. For example, once age, education, income, personal
finance, and conservatism are controlled the effect of the
contemporary racial context is revealed; the percentage
black 1980 has a significant effect in the predicted
direction. Contextual effects still have some explanatory
power even when level of education and income are controlled.
This evidence is in opposition to the Fossett and Kiecolt
(1989) assertion that the regional and contextual effects
represent merely local variation in levels of individual
level variables like education.
Adding the South dummy variable to the model clarifies
the nature of the contextual effects further. The Southern
dummy variable seems to pick up the contextual effects that
were associated with cultural variables and allow more of the
effect of the contemporary racial context to be revealed.
The effect of the percentage black in 1980 on symbolic racism
increases from 0.08 in equation II to 0.11 in equation III
(with an accompanying jump in t-values as wellof course
coefficients, t and p are functions of each other.) The
effects of the individual level variables remain essentially
unchanged by the addition of the Southern dummy variable.
The addition of this dummy adds little to the explanatory
power of the model; its main effect lies in the consolidation
of cultural effects and its control allows or reveals an
amplified effect of the contemporary racial context as
measured by the percentage black in 1980.

114
While the ability of the southern dummy variable to
capture or suppress the effects of the cultural and
historical proxy variables in the models may seem dramatic
enough, it pales in comparison to the effects on the model
when the hybridized variables are added. Of these variables
one in particular is of note in the models explaining
symbolic racism. The interaction term of the south dummy and
the percentage black 1940 not only has a significant inverse
effect but it reveals effects suppressed by the southern
dummy variable alone. Again this adaptation of the model has
little influence on the effects of the individual level
explanatory variables. Education continues to be associated
with the amelioration of the prejudicial attitude while
conservatism remains associated with its aggravation, and
both of these effects are of consistent magnitudes across all
models.
The interpretation of models with complex interaction
terms pushes scholars into hazardous territory, mainly
because it is impossible to interpret the marginal effects of
the constituent elements of the interaction terms.2 In this
case, however, the task may be less perilous because the
interaction term contains a dummy variable. The term
demonstrates the direct effect of racial composition in the
south, circa 1940, on symbolically prejudicial attitudes.
Remarkably the relationship is inverse. On its face, one
2 Re-estimation of less complex models upholds the
substantive interpretation presented here.

115
gathers that as the levels of percentage black (1940)
increase the levels of symbolic racism decrease. Re
estimation of the total models for south and non-south in the
next chapter indicates that in general southerners have more
prejudicial attitudes than non southerners. The significance
of the negative parameter in this case lies in understanding
the marginal effect of the percentage black (1940) within the
southern states. Namely, the interaction term cancels out
the effect of the percentage black in 1940 in the model,
controlling for all the other variables in the model. In
other words, even though the South has higher levels of the
both prejudice and percentage black 1940, the marginal effect
of the local percentage black in 1940 in the South is nil.
The variable has explanatory power for the nation as a whole,
but is less useful in explaining variation within the South
alone. Other factors explain the variation in levels of
prejudicial attitudes within the South. The analysis
presented in the next chapter sheds some light on these
issues.
Taken together the block of hybrid variables should
represent a more effective control of Southernness than the
south dummy variable. After all, the essence of Key's
arguments about southern political life rests on the economic
and political factors, not on the precise latitude and
longitude on the globe. Put another way, living below the
Mason Dixon line has more to do with the (central tendencies
of the) people and the economy than with the fact that a spot

116
in the South lies one inch or one hundred miles below that
line.
While some of the hybrid variables are not statistically
significant in and of themselves, their inclusion in these
models provides a more sophisticated and accurate control of
the factors which determine 'Southernness'. Having taken
Southernness into account the interpretation of the first
block of contextual variables becomes more clear. For
example, once the South is fully controlled the percentage
black circa 1940 is directly and strongly related to the
level of symbolic racism. The prima face interpretation of
this rests on the notion that those areas in the north which
had larger black populations in 1940 demonized the new black
migrants.
Other important changes occur in the model with the
addition of more effective southern control variables. For
example, the effect of income equality is revealed as
significant. Where the incomes of African Americans and
whites are closer to parity, the levels of symbolic prejudice
are lower.
The interpretation of these models reveals some striking
findings. First, even though the symbolic attitudes are more
closely tied to individual level characteristics, they are
sensitive to contextual effects. This is evident despite the
repeated insistence of the theory's proponents that their
symbolic nature is completely abstract (see chapter 3 for
examples). Symbolic racism is clearly related to the

117
respondents' cultural, historical, and contemporary context.
Second, the nature and structure of racial sentiments is more
complex than the proponents of symbolic racism are prepared
to admit. Not only is symbolic racism subject to regional
variation, but it is subject to variations at sub-regional
levels as well. In this sense the abstractness principle of
symbolic racism must be questioned.3
The models predicting the levels of perceived racial
threat are constructed in parallel fashion to the symbolic
racism models just discussed. The results are presented in
Table 6-2.
As was the case with the symbolic racism model, the
first equation shows little of significance. Perceived
racial threat has an inverse relationship with the level of
inter-racial income equality. This runs counter to the
hypothesis derived from Blalock (Hrt08) closeness of class
position promotes racial antagonism. Additionally, the level
of Southern nativity has a very modest positive relationship
with the level of perceived threat. This proves ephemeral;
when other variables are added to the model the apparent
effect of nativity disappears.
3 Note also the consistent effect of personal income of the
symbolic attitudes. While this effects is only significant
at some level between p<.05 and p<.10 it provide evidence
that the symbolic predispositions are not only sensitive to
contextual effects generally but are sensitive to other
aspects of the respondent's personal life. This runs counter
to the theoretical territory staked out by the concept's
proponents.

118
Table 6-2. Perceived Racial Threat Scale Regressed on
Contextual and Individual Level Variables for Whites Only,
entries are unstandardized coefficients from OLS regression
Variables
I
II
hi
IV
%black 1940
0.04
0.04
0.04
0.18***
State %black 1940
0.0000003
0.0000005
0.0000005
0.000003
%black 1980
0.04
0.04
0.04
0.13b
%employed in agriculture
-0.10
-0.22
-0.22
-0.09
nativity
0.02b
0.004
0.004
-0.02
income equality
-2.79*
-3.81**
-3.80**
-3.64*
v595 age

0.001
0.001
0.00004
v602 education

-0.39***
-0.39***
-0.38***
v733 family income

CM
O
O
1
-0.02
-0.02
v356 personal finance

0.29b
0.29b
0.36*
v385 conservatism

0.49***
0.49***
0.45**
south


0.04
2.44b
%black 1900* average farm
size 1900



-0.0006*
%black 19802



-0.005b
south*%black40



-0.19***
%black 1940-%black 1900



0.07
constant
7.46***
7.87***
7.86***
8.20***
adj R2
0.05
0.11
0.11
0.14
N
682
517
517
502
Source: 1986 ANES; 1900,1940 and
1980 Censuses
b p-value < 0.10
* p-value < 0.05
** p-value < 0.01
*** p-value < 0.001

119
The second equation in Table 6-2 adds the block of
individual level variables, as above. And like the symbolic
racism models, education and conservatism show consistent
effects in predicting this racial attitude. Note, however,
that the ameliorative effect of education is reduced by half
in the racial threat model from the level extant in the
symbolic racism model. Conservatism aggravates the racial
scale in the prejudicial direction. Note also that the
respondent's pessimism regarding personal finances is
directly related (although with limited magnitude) to the
level of perceived threat; the more precarious the
prospective financial evaluation, the more threatened the
respondent feels. The inverse relationship between the level
of income equality and the level of perceived threat is
actually amplified by the addition of the individual level
variables. In other words, when factors closely associated
with individual economic vulnerability (such as age,
education, and actual income) are controlled the impact of
relative income levels between the races is more accurately
represented. After this initial amplification, the impact of
income equality remains constant throughout the other
equations.
The addition of the South dummy variable has almost no
impact on the model (see equation III). The dummy variable
has almost no direct effect; its beta is virtually
indistinguishable from zero. Additionally, none of the other
parameter estimates demonstrate any changes as a result of

120
the South's addition to the model. This lack of impact
implies the limited utility of the geographic designation for
its own sake. The importance of the South has less to do
with the set of state names than with identifiable economic
and cultural structures extant within those states. Directly
identifying those structures provides a better method of
capturing whatever it is scholars grope for when referring to
Southern exceptionalism than merely affixing a value of one
to the eleven former confederate states.
The logic of this argument receives even clearer
support when one turns to the results presented in equation
IV of Table 6-2. Here one sees that the addition of the
hybridized variables produces dramatic changes in the model.
These variables include a measure of post-bellum plantation
agriculture and a portrait of southern racial dispersion
circa 1940. These variables more accurately reflect the
salient elements of southerness than the flat geographic
designations. These hybridized variables have similar
effects on the total model for racial threat that they had on
the models of symbolic racism. Namely, more effectively
controlling for the south reveals the erstwhile suppressed
marginal effects of other theoretically significant
variables.
The significant hybridized variables all have negative
parameter estimates. These may be interpreted in similar
fashion to the negative estimate presented above. For
example the interaction term of South and the percentage

121
black in 1940 indicates not that the level of perceived
racial threat in the south is lower than in the North;
analysis presented in the next chapter reveals that precisely
the opposite is true. Rather, its marginal effect within the
South is zero, because the interaction term cancels out the
effect percentage black in 1940, controlling for all the
other variables in the model. The local racial composition
in 1940 explains some of the variation in contemporary racial
prejudice nationally, but is not able to explain marginal
variations within the South alone. Re-estimation of these
models by region in the next chapter permits a more accurate
interpretation.
After controlling for the South, the marginal effects of
variables already in the model change. The effect of the
racial composition 1940 increases dramatically as does the
impact of that same variable at the state level. This seems
a potent endorsement of Wright's (1977) finding that archaic
racial composition influences political attitudes and
behavior today. Additionally, the impact of the contemporary
racial context comes into play in the predicted direction.
The introduction of more accurate control mechanisms for
'region' allow the revelation of the expected theoretical
relationships.
Racial Attitudes and Other Political Orientations
Having demonstrated differences in the origins of the
two racial attitudes, discussion now turns to evaluation of

122
the effects of the two attitudes on other politically
relevant variables. The focus on the racial attitudes shifts
from explaining the origins of the attitudes themselves to
exploring how the racial attitudes explain and predict other
aspects of political behavior. Two types of questions are
addressed. First, how do racial attitudes influence the
evaluation of candidates and groups? Second, how do racial
attitudes influence the respondent's policy preferences, in
this case federal spending preferences? Discussion now turns
to gauging the effect of the racial attitudes on evaluations
of candidates and groups.
The design of dependent variables presents some problems
for this portion of the study. First the evaluations of
candidates is problematic. Ideally, one would use the
evaluation of a widely known and well publicized presidential
candidate. Since the data were collected during an off year
election, no evaluations of presidential candidates are
available. Other political figures with national reputations
are used, even though they were not actually campaigning at
the time. Second, the split form survey forces amendment of
standard procedures for evaluation of feeling thermometers.
These two factors work together to reduce the response rates
on the feeling thermometer evaluations. Normally, one takes
the difference between the rating of the group or person of
interest and the average of all the others. This technique
is problematic when respondents rate few candidates and
groups. Here a more conservative approach is taken (see the

123
Table 6-3. Regression of Evaluation of Candidate and Group
Objects on Racial Attitudes for White Respondents Only.
Entries are unstandardized coefficients from OLS regression.
Jackson
Group
Antipathy
Antipathy
Variable
Racial Attitude
Equation 1
Equation 2
Perceived threat
0.57a b
0.64**
Symbolic racism
0.39
0.04
Rs Characteristics
Age
0.40***
0.22**
Education
-1.83
-0.79
Family Income
0.42
-0.29
National Economic
evaluation
-3.70*
1.12
Personal economic
evaluation
-1.92
0.62
Conservatism
2.56*
-0.54
Context
Percentage Black
-29.42
7.61
Percentage employed
in Agriculture
-294.94b
97.73
Income equality
-12.68
11.90
Nativity
-17.21
9.93
South
3.12
6.06*
Constant
43.79*
-4.01
N
254
493
Adjusted
.15
.10
F/p
4.39/.0001
5.00./.0001
a OLS parameter estimates
b pco.10 p<0.05 ** p<0.01
Source: 1986 ANES, 1980 Census.
pcO.OOl

124
appendix for details). The measure of candidate evaluation
takes the difference between the only nationally prominent
black politician mentioned in the survey and the average of
two other politicians, one Democratic and one Republican.
Similarly, the measure for group evaluation takes the
difference in feeling thermometer rating between blacks and
the poor. The independent variables for both models are the
same. They include the two racial scales, a group of
individual level variables, and measures of the current
context. The results of the models are found in Table 6-3.
Note that the parameter estimates presented are
unstandardized, since the dependent variables in both models
have the same metric.
Table 6-3 presents models of the respondents'
evaluations of a prominent black leader, namely Jesse
Jackson, and of blacks as a group. The variables are coded
such that the dependent variable represents a measure of the
respondents disdain for the attitude object. Accordingly,
Equation 1 is labeled Jackson Antipathy; higher levels of the
dependent variable indicate that respondents rate Jackson
lower than other national political figures. Similarly,
Equation 2 is labeled Group Antipathy: higher levels of this
variable indicate that blacks as a group were rated lower
than the poor.
Equation 1 presents the results of the model predicting
Jackson Antipathy. While neither of the racial scales is
significant at the (two-tailed) 0.05 level, the racial threat

125
scale is significant in a one-tailed test, the appropriate
standard when the sign of the parameter estimate lies in the
theoretically predicted direction. The effect of perceived
racial threat on Jackson antipathy is not by any means
overwhelming, but some relationship is evident; the more
threatened the respondents feel the less favorable their view
of Jackson. The symbolic racism scale does not have
significant explanatory power in this model.
Other findings among the control variables are
noteworthy but not especially surprising. Older people like
Jackson less. The respondent's level of pessimism in
evaluating the national economy is associated with a more
favorable Jackson rating. Conservatives do not care for
Jackson. Finally, it appears that Jackson is more popular
(or at least less disliked) in areas with higher percentages
of residents employed in agriculture, although this
unpredicted finding is just shy of significance at the 0.05
level.
These findings among the control variables are not
particularly surprising give Jackson's status as a populist
figure, actively reaching out across racial lines to the
downtrodden. A more remarkable finding lies in a variable
which is not significant. White Southerners do not think
significantly less of Jackson than they do of other national
political figures. Perhaps Southerners' propensity toward
prejudicial attitudes does not extend to exceptional
politicians, like Jackson.
In other words, his very success

126
as a leader on the national scene may compel his evaluation
along different lines than blacks in general. Alternatively,
the salient elements of the bases of Southern prejudice may
be contained in the racial scales. In other words, Southern
prejudice is less geographic than cultural. More discussion
of this issue follows in Chapter Seven.
The second equation on Table 6-3 presents the results of
a similar regression analysis with the respondent's
evaluation of blacks as a group (subtracted from their
evaluation of poor people) as the dependent variable. In
this case the results are much less complex. Perceived
threat is a significant predictor of antipathy toward blacks
as a group at the 0.01 level. The symbolic racism scale is
again insignificant and in this equation the parameter
estimate shrinks to 0.04. It would be an exaggeration to say
that perceived racial threat explains most or perhaps even
much of the variation in the dependent variable (note: the
adjusted r^= .10), but it does have a significant effect in
the predicted direction when symbolic racisms contribution
is negligible. The perception of racial threat makes a
significant substantive and statistical contribution to the
make up of anti-black affect.
Like Equation 1, the control variables reveal few
surprising findings. Indeed, of the individually measured
variables only age makes a significant direct contribution to
the model. Like before, it appears that older people are less
disposed to give blacks a favorable evaluation than younger

127
people. Of the contextual variables, only one has a
statistically significant effect in this model. The Southern
dummy variable is significant. Southerners are more likely
to rate blacks less favorably than the poor than are non-
Southerners. This is striking given the lack of such an
effect in Equation 1. The implication of this difference is
remarkable. Southerners may be more disposed to hold
prejudicial attitudes about blacks as a minority group, but
this generalized prejudice does not carry over to their
analysis of Jackson (and presumably other leaders) once other
factors are taken into account, namely their economic
vulnerability and relatedly their perception of racial
threat.
What are the effects of the racial attitudes scale in
predicting policy positions, namely federal spending
priorities? Respondents were asked whether federal spending
should be increased, decreased or kept the same in a range of
policy areas (see the appendix for question text). Since the
use of OLS regression on trichotomous variables is fraught
with liabilities a different strategy was adopted for the
models presented in Table 6-4. A single response category
was selected and coded one (1) while the other two response
categories where coded zero (0). In all cases, the response
'decrease federal funding* held the plurality of responses.
Coding this category one (1) maximizes the possible
variability in the resulting two categories. Recoding the
responses in this manner allows for the use of logistic

128
regression. Logistic regression allows for multivariate
modeling, in the same manner as OLS regression, but for
models with dichotomous dependent variable. The only
essential caveat is that the standard per unit
interpretations given OLS parameter estimates should be
avoided since logistic parameter estimates are functions of
changes in the odds of a particular dependent response
category, not units change. The signs and P-values are
interpreted the same way.
Table 6-4 present the results of four logistic
regression equations estimated using federal funding
questions as dependent variables. The first equation
demonstrates the predictive power of the racial scales in
explaining the desire of the respondent to decrease funding
for food stamps. Note that the two scales have opposite
signs in this model. Those who rate higher on the symbolic
racism scale are more likely to support decreases in funding
for food stamp programs. The pattern for perceived racial
threat is the opposite. The more threatened the individual
the more likely that individual will resist decreases to food
stamp funding. (Discussion follows below.) Of the control
variables, only one has a significant effect. As might be
expected those with higher family incomes are less supportive
of food stamp programs that those earning less. Perhaps the
relatively better off believe they have little to gain by
supporting programs that help the poor.

129
Table 6-4. Logistic Regression of Decreased Federal Spending
on Racial Attitudes for White Respondents Only.
Variable
Racial Attitude
Threat
Food Stamps
-0.04 013
Aid
Unemployed
-0.05b
Aid Blacks
0.03
Space
Research
0.01
Symbolic racism
0.25***
0.18***
0.22***
0.08*
R's Characteristics
Age
0.0002
-0.02*
-0.0009
-0.003
Education
0.04
-0.002
-0.01
-0.25**
Family Income
0.08***
0.02
-0.004
0.01
National Economic
-0.10
-0.32*
-0.03
-0.02
evaluation
Personal economic
-0.06
-0.07)
-0.05
0.18b
evaluation
Conservatism
0.26
0.30*
0.36
-0.25*
Context
Percentage Black
0.30
1.38
2.67b
-0.97
Percentage employed
10.90
17.24b
9.91
12.28
in Agriculture
Income equality
0.37
0.36
-0.54
0.59
Nativity
-0.74
-1.16
-0.13
1.18
South
0.18
-0.38
0.23
-0.45
Constant
-0.67
-0.14
-0.24
0.08
N
497
519
502
518
df=13
model Chi-square/p
88.17/0.0001
42.08/0.0001
90.55/0.0001
40.27/0.0001
a parameter estimates from logistic regression indicating
support for decreased federal spending
* p<0.05 ** p<0.01 *** pcO.OOl
b p<0.05 in a one-tailed test
Sources 1986 ANES, 1980 Census.

130
The second equation modeled in this table has as its
dependent variable support for cutting funding for aid to the
unemployed. Like the previous equation, the racial scales
split on this issue. Those with higher levels of threat are
more likely to resist spending cuts for unemployment
benefits, while those with higher ratings on the symbolic
racism scale are more likely to support cutting funds for aid
to the unemployed. (Discussion of this phenomenon follows
below.) The control variables hold few surprises.
Conservatives support cutting spending. Those with
pessimistic outlooks on the national economy and the older
Americans resist spending cuts for the unemployed. Older
Americans might remember hard times, while the most
pessimistic might be more likely to envision the usefulness
of such programs than those with more optimistic outlooks.
Equation three has the respondents' attitudes on funding
for programs that aid blacks as the dependent variable. In
this equation the split between the two racial attitudes
evaporates. As before, those with high symbolic racism
scores are more likely to desire funding cuts in this area
than those with lower scores on that scale. But in this
model the effect of perceived racial threat is reversed. In
the previous two equations respondents with higher racial
threat scores were more resistant to cutting funding for
programs that assisted the unemployed or provided food stamps
for the poor. In this equation, those with higher levels of
perceived threat are more like to support cuts in programs

131
that aid blacks.4 Evidently the feelings of economic
vulnerability held by those with higher levels of perceived
racial threat make them more favorably disposed to approve of
programs that help those who, like themselves, may need a leg
up. This apparent benevolence to programs designed to help
the poor does not extend to programs designed to help blacks
specifically. In this way, the racial component of their
vulnerability comes into play. Those with higher levels of
perceived racial threat are more likely to support cuts in
programs that help blacks than others. Also in this model
the contemporary racial context is associated with support
for cuts in funding for programs that aid blacks. The local
racial context has an effect not captured by the measured
components of racial prejudice.
The models presented in Tables 6-3 and 6-4 endorse the
validity of the perceived racial threat concept. The level
of perceived racial threat is related to anti-black
evaluations of candidates and blacks. This anti-black affect
is further evidenced by the relationship between federal
spending priorities and the level of perceived threat. The
level of perceived threat is associated with the anti-black
position on questions involving the support of programs that
4 The parameter estimate for perceived racial threat (0.03)
has a p-value of about 0.11, just over the 0.10 limit for a
one-tailed test. While statements made about this estimate
must be more tentative than statements made about estimates
with lower p-values, ignoring the estimate because it is
0.01 over the limit is tantamount to throwing the baby out
with the bath water.

132
assist blacks even though the relationship is reversed when
the questions involve more general categories of
disadvantaged groups. The sense of vulnerability possessed
by those who perceived racial threats translates into the
adoption of racially prejudicial attitudes.
The same cannot be said of the symbolic attitudes. The
level of symbolic racism is not associated in any significant
way with evaluations of blacks leaders or blacks in general,
once the perception of threat is controlled. Further, the
relationships between symbolic racism and federal spending
priorities do not support the notion that there is a racially
prejudiced component to the concept. Symbolic racism is not
more associated with the desire to cut funding for programs
that target blacks than with a more general desire to cut any
program that helps (or targets) disadvantaged groups.
Further, the fourth equation indicates the positive
relationship between the level of symbolic racism and the
desire to cut funding for space research. Even conservatism
is associated with support for space research.
By all appearances symbolic racism seems not to hold
real racially prejudicial content; it has no effect once
perceived racial threat is controlled. Rather, symbolic
racism or more properly the questions used to measure it
appears to be a more accurate gauge of a particular component
of conservatism than anything inherently anti-black. In
particular, the questions measure the belief in self-reliance
well. However, respondents who claim that 'blacks could get

133
along without welfare if they tried harder, or that 'blacks
could get ahead like the Jews, Italians, or Irish' might make
similar statements about anyone on welfare or any immigrant
group. One recalls the quip about Archie Bunker, "Hes not a
bigot, he hates everybody the same."
In the next chapter, discussion focuses on regional
difference in the salience, construction and distribution of
racial attitudes.

CHAPTER 7
REGIONAL ANALYSIS
In the previous chapter, the analysis revealed
differences in the causes and effects of the two racial
scales, symbolic racism and perceived racial threat. In this
chapter, the analysis focuses more sharply on the causes of
the two racial attitudes. In particular, the discussion
attempts to draw out issues related to the regional effects
apparent in the models presented in the last chapter.
Tables 6-1 and 6-2 indicate significant regional
differences in the explanation of racial attitudes. In this
chapter, those models are re-estimated after splitting the
sample into two groups, North and South. The division
follows the same criteria for the coding of the Southern
dummy variable detailed in the appendix. A caveat should be
issued at this point. The division of the sample into two
regional subgroups markedly reduces the number of cases to
the point of straining the statistical models. Indeed, even
the models in the previous chapter pushed the limits
somewhat. Given this restraint the analysis presented in
this chapter does not exactly mirror that of the previous
chapter. For example, some variables are dropped from the
analysis. Like the last chapter, the models are presented in
stages to reveal the effects that the addition of control and
contextual variables have on the primary variables of
134

135
interest. The models discussed thus far address the
prediction of the central tendencies of the racial attitudes.
Later in this chapter, models are constructed to assess the
effects of the contextual variables on the dispersion or
variance in the two racial attitudes.
Southern Models of Racial Attitudes
The analysis for the Southern sample is addressed first.
The models predicting the level of symbolic racism are
presented in Table 7-1. The results reflect some of the
essential findings reported in the previous chapter. For
example, equation I reveals that the individual level
variables of education and conservatism each have significant
explanatory power. Higher levels of education are associated
with decreases in the level of symbolic racism. Higher
levels of conservatism are associated with higher levels of
symbolic racism. These effects change little from the
splitting of the sample or the addition of other variables in
equations II and III.
The first equation indicates that the contemporary
percentage black has a positive relationship with the level
of symbolic racism. The negative value in the quadratic term
for that variable indicates that the relationship is
curvilinear.1 The relationship between the contemporary
1 The local maximum may be calculated by applying the formula
-2a/b, where a is the parameter estimate (not reported) for
the percentage black and b is the parameter estimate (not
reported) for the quadratic term. (Giles and Evans 1986)

136
Table 7-1. Symbolic Racism Regressed on Contextual and
Individual Level Variables for White Southern Respondents
Only.
entries are unstandardized coefficients from OLS regression
Variables
i
II
III
%black 1940

-0.25
-0.16
State %black 1940

-0.04
-0.11
%black 1980
0.45*
0.71*
0.49
%employed in
agriculture


0.38
income equality


-5.63
v595 age
-0.0008
0.003
0.01
v602 education
1.35***
-1.27***
-1.22**
v733 family income



v356 personal finance



v385 conservatism
0.93**
0.91*
0.74*
state %black
1900*average farm size


0.0002
%black 19802
-0.01*
-0.02*
-0.01
%black 19402

0.004
0.001
%black 1940-%black
1900


-0.09
constant
11.51***
10.61***
11.13*
adj R2
0.17
0.15
0.16
N
146
144
144
Source: 1986 ANES;1900,1940,1980 Census

137
percentage black and the level of symbolic racism is positive
until the local maximum is reached at 72%, where the slope
reverses. In other words, the marginal effect of increases
in the percentage black on the level of symbolic racism
decreases as that percentage increases. This reduction in
the marginal effect should not distract the reader from the
essential finding of a positive relationship between the
contemporary racial composition and the level of symbolic
racism.
Equations II and III in Table 7-1 include additional
variables. The relationships discussed above demonstrate
little change with the addition of these variables, none of
which have significant independent effects. The effect of
the contemporary percentage black and its quadratic term in
equation III are just shy of significance at the 0.05 level.
The removal (not shown) of the percentage black 1940 for the
state or county push those variables back into the
significant range. The marginal effects of the archaic
racial composition and the other hybridized variables are
indistinguishable from zero in this reduced sample. Change
from the national models (presented in Table 6-1) may result
from the strain on the models from the reduced number of
cases. Alternatively, the significance of those variables in
the national models may reflected the dramatic cultural
differences between the regions. Once the ultimate regional
control, the division of the sample, is exercised the effects
become markedly less dramatic. Further, the correlation

138
between the percentage black at 1940 and 1980 masks the
independent effects of each.
A similar estimation strategy for Southerners'
perceptions of racial threat produced the results presented
in Table 7-2. The results are similar to those presented in
Table 7-1. Education demonstrates a consistently
ameliorative effect on the level of perceived threat across
the four equations. Similarly, conservatism consistently
aggravates the level of perceived racial threat across all
four equations. The contemporary racial composition has an
almost identical effect in predicting the level of perceived
threat vis-a-vis its effect in predicting the level of
symbolic racism. As before the relationship is curvilinear
with a local maximum of 75%. The central effect of the local
racial composition is conveyed by the non-quadratic term;
there is a significant positive relationship between the
contemporary percentage black and the level of perceived
racial threat. The marginal decline in the effect occurring
after the percentage black reaches 75% is insufficient to
vitiate the basic relationship; higher levels of perceived
racial threat are associated with higher percentages black in
the respondent's locality.
The addition of control variables and the archaic
measures of racial context do little to improve the
predictive power of the model presented in equation I. As in
the previous table, the effect of the contemporary racial
composition falls slightly out of the range of accepted

139
Table 7-2. Perceived Racial Threat Scale Regressed on
Contextual and Individual Level Variables for White
Southerners Only.
entries are unstandardized coefficients from OLS regression
Variables
I
II
III
IV
%black 1940

-0.10
-0.02
-0.22
State %black 1940

-0.01

-0.01
%black 1980
0.39*
0.30
0.41*
0.40
%employed in
agriculture



-0.18
income equality



-5.03
v595 age
-0.03
-0.03
-0.03
-0.02
v602 education
-0.70**
-0.72**
-0.69**
-0.73*
v356 personal finance
0.52
0.51
0.50
0.62
v385 conservatism
0.94**
0.94*
0.93*
0.75*
State %black
1900*Average farm size



-0.0003
%black 19802
-0.01b
-0.01
-0.01b
-0.57
%black 19402

0.003

0.004
%black 1940-%black
1900



-0.09
constant
4.76b
5.00*
4.83b
10.78b
adj R2
0.10
0.08
0.08
0.10
N
130
128
128
128
Source: 1986 ANES; 1900, 1940, 1980 Census.

140
significance levels when other correlated variable are added
to the model (see equation II). The simple removal of the
state percentage black for 1940 reduces colinearity
sufficiently to push the contemporary racial context back
toward significance. The marginal effects of the hybridized
variable are less pronounced in these models than in the
national models presented in the previous chapter. As above,
the explanation for this probably lies in the fact that the
division of the sample by region removes the stark relief
presented by the dramatic cultural differences between those
regions.
Northern Models of Racial Attitudes
Parallel analysis for the non-Southern cases (called
Northern in tables) is presented in Tables 7-3 and 7-4. The
equations modeling symbolic racism are presented in Table 7-
3. Here the relationship between education and the level of
symbolic racism holds similar magnitude and identical
direction to that of the Southern sample. Education
consistently and significantly ameliorates the level of
prejudicial attitude across all three equations. Similarly,
conservatism aggravates the level of symbolic racism. Higher
levels of conservatism are associated with higher levels of
symbolic racism across all equations. Age also displays a
moderately positive relationship with the level of symbolic
racism (see equation I) but this relationship diminishes with
the addition of contextual variables. The untransformed

141
Table 7-3. Symbolic Racism Regressed on Contextual and
Individual Level Variables for White Northern Respondents
Only.
entries are unstandardized coefficients from OLS regression
Variables
I
II
III
%black 1940

-0.04
-0.03
State %black 1940

-0.13
-0.16
%black 1980
-0.03
0.06
0.04
%employed in
agriculture


0.02
income equality


0.23
v595 age
0.02*
0.02b
0.02
v602 education
-0.98***
-0.97***
-1.01**
v385 conservatism
0.58***
0.60*
0.16**
State %black
1900*Average farm size


-0.0005
%black 19802
0.004*
-0.0006
0.002
%black 19402

0.003
0.003
%black 1940-%black
1900


0.09
constant
11.09***
11.23***
11.33**
adj R2
0.22
0.22
0.22
N
482
448
430
Source: 1986 ANES; 1900, 1940 and 1980 Census

142
measure of contemporary racial composition does not have a
significant effect in predicting the level of symbolic
racism, but the quadratic term does. This indicates that the
relationship between the local racial composition and the
level of symbolic racism is dramatic; as the percentage black
increases so too does its marginal effect on the level of
symbolic racism.
When the measures of archaic racial composition are
added the apparent effect of the contemporary racial
composition vanishes, but the effect of the archaic measures
fall short of significance (p<0.11 for the quadratic term,
and p<0.22 for the untransformed term--not shown). From this
we may gather that the archaic racial composition has the
bulk of the explanatory power. In analyses not presented
here, it is apparent that the effect of the archaic racial
composition is suppressed unless the contemporary racial
composition is controlled. In other words, if we control for
the current racial context, the percentage black in 1940
becomes a near significant predictor of the current level of
symbolic racism. Unfortunately, due to the dwindling number
of cases, this analysis must be presented with some
hesitancy.
Similar analysis of the models predicting the level of
perceived racial threat among non-Southern respondents is
presented in Table 7-4. Again, the same relationships
between education and conservatism and the racial attitude
are evident. Education is negatively associated with the

143
Table 7-4. Perceived Racial Threat Scale Regressed on
Contextual and Individual Level Variables for White
Northerners Only.
entries are unstandardized coefficients from OLS regression
Variables
i
ii
hi
IV
%black 1940

0.16
0.14*
0.23*
State %black 1940

-0.13b

-0.01
%black 1980
-0.02
-0.07
-0.02
-0.06
%employed in
agriculture


-0.10
-0.07
income equality


-2.84
-2.43
v595 age
0.003
0.004
0.005
0.004
v602 education
-0.31**
-0.32**
-0.26*
-0.27*
v356 personal finance
0.20
0.17
0.26
0.23
v385 conservatism
0.39**
0.41*
0.39**
0.39**
state %black
1900*average farm size


-0.002***
-0.002***
%black 19802
0.004*
0.006
0.001
0.004
%black 19402

-0.003

-0.003
%black 1940-%black
1900


-0.08
-0.08
constant
6.14***
6.29*
7.51***
7.47***
adj R2
0.08
0.08
0.10
0.11
N
445
413
398
398
Source: 1986 ANES;
1900, 1940
and 1980
Census.

144
prejudicial attitude, while conservatism is positively
associated with the same. Again, the effects are consistent
across all equation. It should be noted, however, that the
explanatory power of these variables is somewhat diminished
in this table. In equation one the quadratic term for the
contemporary racial composition, but not the untransformed
term, is significant indicating a dramatic relationship
between the level of perceived threat and the local racial
composition. This effect (again) vanishes when the measures
of archaic racial composition are included in the model
estimated in equation II. As above, however, the measures of
past racial composition fall short of the accept level of
statistical significance. The state percentage black for
1940 is significant at the p<0.10 level.
Equation III provides a different model. In this
equation, the archaic quadratic and state level terms are
dropped out and the other contextual variables are included.
In this model, the local percentage black for 1940
demonstrates a significant positive relationship with the
level of perceived racial threat. The other variables change
little. From this one gathers that the current level of
perceived racial threat is sensitive to cultural patterns
established in the middle of this century. This equation
also reveals an unexpected finding. The interaction term
between the percentage black at the state level in 1900 and
the average farm size in the state that year has a
significant negative relationship with the level of perceived

145
racial threat. Neither of the individual terms demonstrates
significance when placed in similar models without the
partner. The variable was designed to measure the level of
plantation-style agriculture in the South and was expected to
be positively correlated with level of prejudicial attitude
primarily in the South. A tentative explanation might
include the notion that the big agricultural states at the
turn of the century are not the same states which attracted
black migrants in the middle of the century. The competition
for factory and other jobs in the urban and industrial areas
produced a predisposition to perceptions of threat. The
questions raised by this finding cannot be answered in the
present study, but are worthy of further research.
A further adjustment to the model is presented in
equation IV on Table 7-4. In this equation, two racial
composition variables from 1940 which were excluded from the
previous model are included, the state percentage black and
the quadratic term for the county level percentage black.
The addition of these statistically insignificant variables
has little effect on the other independent variables in the
model, save one. The standardized regression coefficient for
the percentage black 1940 enjoys a substantial increase in
its explanatory power. The findings reported in this and the
previous paragraph point to the importance of historic and
contextual factors in understanding contemporary racial
attitudes.

146
Regional Differences in Attltudinal Unity
The discussion thus far has focused on the effects of
independent variables on an individual's level of prejudicial
racial attitudes. In this section discussion addresses the
effects of contextual variables on the dispersion of racial
attitudes within a given area. Recall that the
interpretation of Key's (1949) work given in the second
chapter implies a connection between the percentage black and
the dispersion of racial attitudes in the South. Namely,
because of the perceived threat posed to whites by blacks,
higher percentages black will be associated with greater
unity of attitudes.
This hypothesis is assessed in this section by using the
standard deviation within each county of the two racial
scales as the dependent variables regressed on contextual
variables used in the preceding analysis. Again, some
caveats are in order. Some counties may have only a few
respondents from which to calculate the standard deviation.
(The Ns in the tables refer to the number of counties used in
the analysis.) Because of this the results presented in the
next two tables only point the way toward future research on
this topic rather than providing verifiable proof of the
issues at hand.
Table 7-5 presents the results from models with the
standard deviation of symbolic racism as the dependent
variable for Southerners (at the top of the page) and

147
Table 7-5. Standard deviation of Symbolic Racism Regressed
on Contextual Variables in Each County
entries are unstandardized coefficients from OLS regression
white Southern respondents only
Variables
I
II
Mean Symbolic Racism
0.48**
0.42*
%black 1940

-0.001b
State %black 1940

0.0004
%black 1980

0.001
%employed in
agriculture

-0.003
income equality

-0.002
constant
-0.01
-0.004
adj R2
0.30
0.18
N
27
26
white Northern respondents only
Variables
I
II
Mean Symbolic Racism
0.45***
0.44**
%black 1940

-0.0005
State %black 1940

-0.001
%black 1980

0.001*
%employed in
agriculture

0.001
income equality

0.004
constant
-0.01
-0.02
ad j R2
0.31
0.35
N
78
70
Source: 1986 ANES; 1900, 1940 and 1980 Census

148
Northerners (at the bottom). For both Northerners and
Southerners, the mean level of symbolic racism is associated
with increases in the amount of variation within each county.
For each region, one additional variable sheds light on the
construction and dispersion of racial attitudes. In the
South, the percentage black in 1940 is negatively associated
with the standard deviation of the symbolic racism scale. In
other words, the percentage black from 1940 is associated
with greater unity of opinion within the South on this racial
scale, as Key might have predicted. The standard deviation
of symbolic racism in the North also shows sensitivity to a
contextual variable. The higher the contemporary percentage
black the greater the variation in the level of symbolic
racism in the non-Southern counties. In other words,
respondents in areas with higher percentages of blacks imply
a greater diversity of opinion on racial issues.
Similar analysis using the standard deviation in the
level of perceived racial threat is presented in Table 7-6.
For both Northerners and Southerners alike, the standard
deviation in perceived racial threat increases with the mean
level of perceived racial threat. That is to say, that as
the average level of perceived racial threat increases, the
range of opinion also increases. For Southerners one other
contextual variable makes a significant impact of the
dispersion of the perception of racial threat. Equation III
reveals a positive relationship between the level of income
equality and the dispersion of perceived threat. The closer

149
Table 7-6. Standard Deviation of Perceived Racial Threat
Regressed on Contextual Variables in Each County.
entries are unstandardized coefficients from OLS regression
white Southern respondents only
Variables I
Mean Racial Threat 0.82***
II
0.79***
III
0.95***
%black 1940
State %black 1940
%black 1980
%employed in
agriculture
income equality
0.000005 0.001
-0.0004
-0.0003
0.001
0.05b
constant
adj
N
-0.03*
0.58
24
-0.03
0.48
23
-0.06
0.53
23
white Northern respondents only
Variables
I
II
Mean Racial Threat
0.69***
0.81***
%black 1940

-0.0005*
State %black 1940


%black 1980


%employed in
agriculture


income equality
__ _
III
0.83***
0.00005
-0.001
-0.0002
0.001
-0.01
constant
adj R^
N
-0.02*** -0.03*** -0.02b
0.57 0.65 0.64
73 73 65
Source 1986 ANES; 1900, 1940 and 1980 Census

150
the average income for the races is to parity, the greater
the diversity or range of level of perceived racial threat.
While income equality may not tell much about the absolute
levels of threat, it may reveal something about areas in
which a wider range of attitudes exists. The implication is
that where income equality is high and hence competition more
stiff two countervailing conditions may co-exist. Some
whites may develop closer associations with those blacks with
whom they share common interests and means. Others may
experience heightened sensitivity due to competition for
scarce resources, like jobs or housing.
The effects of the independent variables other than the
mean level of perceived threat on the model for Northern
respondents is muted. Equation II indicates that the
percentage black from 1940 has a significant negative effect
on the standard deviation of perceived racial threat. Those
areas with larger black population in the middle of the
century have greater unity of opinion currently on this
racial scale. The relationship appears to be fleeting,
however. When other variables are added to the model, the
effect becomes insignificant.
The findings presented in this chapter have sought to
shed light on regional differences in the racial attitudes.
The results indicate that the measures that are efficacious
in demonstrating the differences between regions are not as
useful in demonstrating the differences between individuals
or counties within the regions. Since the non-South is more

151
diverse than the South, it is reasonable to expect the more
diversity from the models of the non-South. In this sense,
the variable which measured the level of 'Southerness' in the
last chapter retain some explanatory power for states which
are like the South. The relatively more homogenous nature of
the Southern region makes it stand out in rather stark relief
from the rest of the country, but at the same time makes it
more difficult to identify the causes and explanations of
Southern peculiarities within an exclusively Southern sample.
For example, the effects of leadership in Southern
communities on the attitudes and behavior of residents is
evinced by the uneven pattern of desegregation in the South
during the 1960s. Still, localized effects like this, while
important, go unmeasured in this research. The measured
markers of the South are demonstrative in the national model,
but lackluster once the sample is broken up by region.
One thing is relatively apparent. Understanding the
roots of both racial scales requires an appreciation for both
the present and the historical context. This seems true even
for understanding symbolic racism which showed some
sensitivity to contextual effects in both the South and the
North, despite the ardent assertions by its proponents that
the racial orientation is context-free. Not only do the
individual level regression models imply such a connection
but the analysis of contextual effects on the standard
deviations of the racial scales implies the same. The racial
context at the county level seemed to influence the

152
dispersion if not the level of symbolic racism in both South
and North.
In the next chapter an effort will be made to reconcile
the findings of the last two chapters with the theoretical
propositions previously enumerated and to point the way for
future research on these and similar issues.

CHAPTER 8
CONCLUSIONS: FARTHER ALONG
This chapter reviews the theoretical bases of this
dissertation and the central empirical findings. First, the
central theoretical positions in the racial threat literature
detailed in Chapter Two are discussed, followed by a brief
review of the essential tenets of the symbolic racism school
of thought presented in Chapter Three. After this, a
synopsis of the synthetic theoretical positions developed in
Chapter Four is laid out. The discussion of these
theoretical developments is followed by a review and
assessment of the hypotheses generated from those theories
and their tests. The chapter closes with the presentation of
the central conclusions and ramifications of this research.
Racial Threat
This dissertation assesses the theoretical strengths of
two major schools of thought in the political science
literature on white prejudice: the racial threat literature
and the symbolic racism literature. This section features a
brief review of the central theoretical tenets of the racial
threat literature. A number of representative works in this
field are reviewed in detail in Chapter Two.
153

154
In that chapter, threat is defined as a challenge to
political or economic power, or to security of person or
property. Racial threats in this study are defined as the
challenges to whites' political or economic power, or to
their security of person or property posed by blacks. Much
of the racial threat literature uses definitions of racial
threat which are consistent with this definition. In other
words, blacks represent an actual challenge to, for example,
the economic and social standing of whites. However, a
threat need not be actual for it to influence the thoughts
and behavior of an individual. The perception of threat,
whether that perception is accurate or not, may be sufficient
to effect a change of behavior in an individual. For
example, a novice poker player may fold in response to the
raises of a bluffing opponent. The perception of threat may
lead to the adoption of political positions which individuals
might not otherwise take.
Much of the racial threat literature tests the effects
of environmental or contextual factors on white attitudes.
Typically, the local racial composition, or the percentage
black in the respondent's home county or Standard
Metropolitan Statistical Area (SMSA), is used as a predictor
of some prejudicial orientation, often with successful
results. For the most part, the racial threat literature
assumes that the size of the African-American population in
the respondent's locality is an accurate gauge of the racial
threat because it is in those areas that blacks will present

155
the stiffest competition for jobs and the benefits of modern
society. Additionally the research on the effects of racial
threat has found regional differences in the level and effect
of threats and prejudice; respondents living in the South
tend to be more prejudicial. While the intuitive reaction to
this finding lies in appeals to the influence of the South's
confederate and segregationist past, racial threat scholars
has argued that the effect is an indicator of the South's
traditionally lower levels of education or other individually
measurable variables.
In the next section a brief review of the second major
school of white prejudice, or symbolic racism, is presented.
Symbolic Racism
The notion of symbolic racism represents an extension of
the symbolic politics paradigm. The theory and measurement
of symbolic racism is discussed in detail in Chapter Three.
In general, proponents of the paradigm argue that cognitive
and affective processes drive political behavior, rather than
material conditions or an individual's political and social
environment. Affective orientations or symbolic
predispositions are set during pre-adult socialization.
These predispositions are then used as benchmark for the
evaluation of later stimuli and form the basis of political
attitudes and behaviors. The proponents of symbolic racism
argue that racial attitudes follow this same pattern. So,

156
whites' current evaluations of blacks are based on the
symbolic predispositions galvanized in youth.
More to the point, the proponents of symbolic racism
argue that contemporary prejudicial attitudes result from a
combination of anti-black affect, a generalized distaste for
blacks, and traditional values exemplified by the Protestant
work ethic. Whites may be disposed to dislike blacks
generally, but the whites' belief that blacks fail to live up
to the standard set by the work ethic aggravates the disdain
dramatically. Further, this violation of these cherished
values provides a socially acceptable avenue for the
articulation of prejudice; whites are free to vent their
prejudicial attitudes as long as they couch their disdain in
terms of blacks' failure to 'work hard'.
A final ingredient of symbolic racism is found in the
assumption of abstraction. The proponents of the concept
argue that this form of prejudice is independent of an
individual's personal life, self interest, or economic,
political or social context. So, environmental factors found
significant in explaining prejudice in the racial threat
literature are thought not to explain symbolic racism.
The Synthetic Theoretical Position
Chapter Four presented a synthetic position which
accounts for some of the incompatibilities between the two
schools of thought discussed above and includes elements
missing from both. The novelty of the synthetic position

157
lies in two central areas. First, people have the ability to
use more than one form of racial attitude. Symbolic racism
and the perception of racial threat are represented in
different dimensions. In measurement models, people may
posses varying levels of each orientation. Indeed,
individuals may access more racial orientations than this
study could measure.
The second main innovation of the synthetic theoretical
positions lies in the attention paid to the effects of
cultural and historical traits on racial attitudes. People
do not live in a vacuum, so their attitudes, racial and
otherwise, are subject to influence by the local political
culture. The history of a locality influences the ways in
which current residents see themselves and the community. In
particular, past patterns of race relations or past economic
structures, such as plantation agriculture, are thought to
have particular explanatory value in addition to being
measurable.
The measurement of the historical and cultural
differences between regions relies on data from past censuses
as predictors of current racial attitudes. Data are drawn
from the 1900 and 1940 censuses to represent the effects of
past economic and demographic influences on current
attitudes. Demographic and economic patterns established in
an earlier age influence today's political culture.
Similarly, data drawn from the 1980 census provides
information about the current local context.

158
The analysis presented in Chapters Five, Six and Seven
provides evidence with which one may assess the theoretical
propositions laid out in Chapter Four. In the next three
sections, those hypotheses will be assessed in light of the
preceding analysis. Additionally, general conclusions about
the research will be drawn as well as a brief discussion of
the prospects for future research in this field. The
hypotheses were divided into three categories: some were
direct derivations of the theoretical positions laid out by
the proponents of the racial threat school, some were
similarly derived from the symbolic racism school, and some
were synthetic reconciliations of the two schools. Those
hypotheses will be discussed in the following three sections
of this chapter. The hypotheses derived from the racial
threat literature will be taken up first.
Assessing the Derived Racial Threat Hypotheses
Many of the hypotheses derived from the racial threat
literature are upheld or supported by the analysis presented
in the preceding chapters (see Tables 5-3, 6-2, 7-2, 7-4 and
7-6). The hypotheses involving the predictors of perceived
racial threat are especially robust. The percentage black
1980 as well as the archaic percentage black (1940)
demonstrate positive and significant associations with the
level of perceived racial threat, even when numerous other
variables were controlled (Hrtoi)- Similarly, respondents
living in the South have higher levels of perceived racial

159
threat. In multivariate models, 'Southernness' has the
predicted positive relationship even when other variables
like income and education were controlled; Southerners are
more likely to feel threatened even when other explanations
are controlled (Hrt02 Hrt04)* Additionally, the level of
perceived threat is sensitive to the level of economic
vulnerability as measured by the respondents assessment of
personal finances, even when other variables are controlled
(Hrt03)
Similar support for the hypotheses regarding the
predictive abilities of the respondent's level of perceived
racial threat is found (see Tables 6-3 and 6-4). The level
of perceived racial threat proves to be a significant
predictor of the respondents' evaluations of a black
candidate (Jesse Jackson) and of the respondents' evaluations
of blacks as a group (Hrt05)* Those who felt more threatened
are not less likely to support government assistance programs
in general; indeed, there is a negative relationship between
perceptions of threat and the desire to reduce funding for
programs to help the disadvantaged (Hrt06)- Despite the lack
of resistance to such policies in general, the reverse
relationship is demonstrated with regard to programs
specifically designed to assist blacks. Even when other
variables are controlled, a positive relationship between the
level of perceived racial threat and the likelihood of
respondents favoring funding cuts for programs which help
black is demonstrable (Hrt07)

160
The only racial threat hypothesis not upheld involved
the relationship between the parity of income between the
races and the level of perceived racial threat (Hrt08)* It
was predicted that the level of perceived racial threat would
increase as incomes approached parity because increased
levels of competition would spur the perception that blacks
posed an economic threat to whites. This contention is not
supported in the multivariate models presented in Table 6-2.
Further, the significant coefficients indicate that the
reverse is true. There are two reasonable explanations for
this unexpected finding. First, income parity between the
races may represent a measure of interracial good will;
incomes are close because of the absence of prejudice. On
the other hand, such interracial goodwill may be the effect
of the similarity in income levels between the races, because
common income levels give whites and blacks more in common
and racial animosities are reduced. Neither the theory nor
data discussed in this study can provide a conclusive
explanation for this puzzle. Still, one may imagine research
strategies for addressing this issue. For example,
quantitative longitudinal studies linking the level of income
parity and the number and intensity of racially motivated
incidents in a number of carefully selected communities may
shed light on the dilemma. Similarly, qualitative cultural
and historical analysis of race relations in a few well
selected cases would also prove useful. Better yet, both

161
research strategies could be carried out in the same
communities.
In the next section, discussion turns to an evaluation
of the hypotheses relating to factors predicting symbolic
racism and an assessment of the concept's explanatory powers.
Derived Symbolic Racism Hypotheses:
While many of the derived hypotheses regarding the
predictors and effects of perceived racial threat are upheld,
the derived hypotheses for symbolic racism fare less well
(see Tables 5-3, 6-1, 7-1, 7-3 and 7-5). Despite the
assertion by the concept's proponents regarding its
abstraction, the measures of symbolic racism employed here
demonstrate sensitivity to contextual effects in general and
both contemporary and archaic local racial composition in
particular (Hsroi) Similarly, the level of symbolic racism
is sensitive to regional factors (Hsr02 and hsr04)*
Southerners are more likely to hold more prejudicial
attitudes than non-Southerners. The evidence presented in
Chapter Six even suggests a moderate relationship between
family income and the level of symbolic racism, despite the
prediction that symbolic racism is unrelated to personal
factors like economic position or income (Hsr03)* All in
all, none of the hypotheses related to the concept's
abstractness are upheld.
Tests of symbolic racism's explanatory power are
presented in Tables 6-3 and 6-4. This racial attitude scale

162
demonstrates no significant explanatory power in predicting
the respondents' evaluation of either a national political
figure (Jesse Jackson) or blacks as a group (Hsr05),
controlling for other variables. The scale does a better
job, however, in predicting the respondents' federal spending
priorities. The level of symbolic racism is positively
related to the likelihood respondents would prefer spending
cuts in all issue areas presented in Table 6-4. This finding
is consistent with the positions laid out in Chapter Four.
Because of adherence to traditional values, symbolic racism
was expected to demonstrate an inverse relationship with
support for safety net policies (Hsr06) A similar inverse
relationship was expected between symbolic racism and
programs which benefit blacks in particular (Hsr07)*
However, the consistency of the effects for all forms of
federal spending raises questions about the usefulness of the
concept. Why, for example, would the racial attitude predict
the respondents' desire to cut funding for the space program?
Similarly, why, given the racial content of the concept, does
the symbolic racism scale not have more explanatory power in
predicting attitudes concerning programs that help blacks
specifically than it does the less racially specific program
of food stamps?
The pattern implied in Tables 6-3 and 6-4 indicates that
the measures, if not the concept, of symbolic racism are
flawed. The symbolic racism scale does not have more
explanatory power when the objects of evaluations or policies

163
are blacks than it does for other objects. Despite the
satisfaction of those hypotheses concerning federal spending
priorities, one is left with lingering doubt about the
usefulness of the concept in explaining racial attitudes.
Indeed, the findings presented in Table 6-4 imply that the
measure of symbolic racism could as easily represent a
measure of fiscal conservatism.
In the next section the synthetic hypotheses will be
evaluated.
Synthetic Hypotheses:
While the hypotheses presented in the two previous
sections were derived from purist readings of the literature
in the respective schools of thought, the hypotheses
discussed in this sections result from critical reading of
both literatures and in that sense are synthetic.
For example, the notion of multi-dimensionality does not
fit comfortably into the racial threat literature and only
fits in the symbolic racism literature to the extent that
symbolic racism is thought to replace old-fashioned racism.
Table 5-2 demonstrates that two racial attitudes need not
reflect the same attitudinal dimension (HsynOI)* To the
extent that the two racial scales are actually measuring
racial attitudes, those attitudes enjoy some independence
from each other. Individuals may have variable levels of
each attitudes. One may be disposed toward symbolic
prejudice and yet not feel threatened or visa versa.

164
This confirmation of multi-dimensionality has noteworthy
implications for future research and for the evaluation of
past efforts. Scholars interested in measuring prejudicial
attitudes need to include multiple indicators of prejudice
and demonstrate care in how those indicators are combined.
Further, the multi-dimensionality of racial attitudes has
implications for the evaluation of past research. For
example, the failure of previous research to substantiate
racial effects may be the result of insensitivity to the type
of prejudice involved (e.g. Green and Cowden 1992).
The expected relationship between the level of perceived
racial threat and the level of income equality is
substantiated, but as noted above neither theoretical nor
empirical mechanisms are in place to determine causality
(Hsyn02)* The inverse relationship between income equality
and the perception of threat found here may have implications
for group conflict research beyond those noted above. It may
be that competition for jobs represents an area particularly
sensitive to group based prejudice, while competition for
income does not. On the other hand, perhaps income equality
is more a function of employment than wage discrimination.
While these questions follow from the findings of this
dissertation, answering them falls outside the scope of this
research.
The hypothesized relationships between racial
composition, contemporary and archaic, and the racial
attitude scales are largely borne out. The effects of racial

165
composition on racial scales tend to be curvilinear. The use
of quadratic terms indicates that the local racial
composition has its greatest impact on the level of symbolic
racism at lower levels (of percentage black); as the
percentage black increases, its marginal effect on the level
of symbolic racism (or slope) decreases (Hsyn03)* Levels of
symbolic prejudice increase in response to the local racial
composition to a point. This implies that the tendency to
harbor symbolic prejudice is related to exposure to blacks or
information about blacks within the community. This exposure
may increase the opportunities for the attitudes to be used
and, albeit erroneously, reinforced. But this cycle of use
and reinforcement only affects prejudicial attitudes to a
point, after which saturation sets in, preventing escalation
of the prejudicial attitude.
Similarly, the current levels of perceived racial threat
vary with the archaic racial composition (Hsyn04)* The
contemporary racial context also makes an independent
contribution to the models explaining perceived racial
threat. This relationship is also curvilinear, with
decreases in the marginal effects of the racial context
occurring as the percentage black increased beyond a certain
point. But this point is generally outside the realistic
range; the effect of racial composition of the perception of
racial threat begins to decline at about 120% black in most
models. Though quadratic estimation improves fit, the actual
effect of racial composition for most models is close to

166
linear. This implies that whites in areas with significant
or even majority black population are more likely to resort
to or harbor perceptions of racially based threat. Racial
animosities are likely to be aggravated when jobs are scarce.
Discussion in the next section assesses the value and
substantive significance of these findings and raises issues
for future research in this area.
Conclusions
The study just completed suffers from any number of
problems afflicting empirical social science research in
general. For example, the theoretical and measurement models
alike are oversimplified and are not able to capture the full
complexities of life's rich pageant or the depths of its
evils. Similarly, even the simplified models suffer from
statistical ailments. The low number of cases strains
analysis at times. The colinearity between contextual
variables obscures relationships. Despite these limitations
and the concomitant caveats, a number of conclusions may be
reasonably drawn.
The matter of symbolic racism is addressed first.
Evidence from this study indicates that some of the claims of
symbolic racism's proponents are untenable. The concept is
measured using widely accepted indicators, yet its behavior
is remarkably inconsistent with the claims of its authors.
Most notably, little support is found for the notion that
symbolic racism, or the scale used to measure it, is actually

167
abstract, somehow distinct from the personal lives of people
or the contexts in which they live. Both the contemporary
and the archaic racial context influence the symbolic
attitudes. Additionally the attitude was subject to
significant regional variation, although the theory expects
none due to the principle of abstractness.
In the face of this evidence, it seems reasonable to
temper the claims of abstractness. Such tempering would
bring the notion of symbolic racism more in line with
existing theory and research in symbolic politics. Indeed,
other research by the major proponents of symbolic racism in
the area of symbolic politics seems to accommodate context
and other non-psychological constructs such as peer
influences (adult socialization) quite comfortably (see for
example, Kinder and Rhodebeck 1982). Against this backdrop,
the tenacity with which the proponents of symbolic racism
cleave to the notion of abstractness seems ill-placed. Could
it be that whites are sensitive to the events and history in
their locale and to the influences of friends and co-worker
and yet still remain unsympathetic to the plight of the
blacks they imagine living on welfare in the inner city?
This renovated construction of the principle of abstraction
yields a better fit with the available evidence.
The evidence presented in this frontal assault may seem
more damaging than it is. After all, the slight theoretical
adjustment noted above deflects the lion's share of that
assault. A more troubling issue lies in the gray areas.

168
while some count the consistency with which the symbolic
racism scale explains a variety of issue positions (see Table
6-4, for example) as a strength, it may well point to the
concepts greatest weakness. Why should one expect a racial
attitude to explain a respondent's spending priorities for
space research? Why, if the scale really taps something
racial, does the scale not have remarkably more explanatory
power when the dependent variables have racial content? The
sensitivity of the racial scale to racial contextual effects
implies that there is some racial component to it, but one
does not get the same impression by viewing its effects.
Indeed, the consistent resistance to any spending may provide
an endorsement of the position taken by the primary
detractors of symbolic racism (e.g. Sniderman et al. 1985),
namely, that the concept represents conservatism, not a new
racial orientation. Evidence is not found that could
disprove this contention.
Again, the sensitivity to racial contextual effects
provides some evidence that there may be some racial content
in the concept, but the uniformity of effects raise questions
about the concepts validity. The efficacy of the measure in
explaining spending preferences may be a function of the
facility with which the constituent variables tap fiscal
conservatism. Further, the design of the questions
themselves may contribute to explaining the uniformity of
effects. The unfolding response format used to measure the
symbolic attitudes may be more stable by the force of the

169
format in addition to other aspects of validity. (Krosnick
1991) Krosnicks findings may also help explain the
differences between the racial threat and symbolic racism
models in the proportion of the variance explained; more of
the variance in the symbolic models may have been explained
because the questions used to construct the scale are of a
superior form.
The implications of this research for the perceived
racial threat school are perhaps less dramatic than those
noted above for symbolic racism. While the attitudes of
threat do show sensitivity to the contemporary racial
context, there is evidence that the history and culture of
the locale also have significant influences on those
attitudes. This represents a refinement or augmentation to
the main body of quantitative literature on the topic. The
history and culture of an area may have implications for
understanding current racial attitudes which cannot be
captured using only contemporary data.
Unlike symbolic racism, the racial threat literature
does make a place for the influence of contextual effects,
but just barely. The contemporary local racial composition
is seen as influencing the level of perceived threat in some
studies or is used as a proxy for threat in others. A place
is made only for the immediate effects of the local racial
composition, theoretically and empirically. While one might
expect the contemporary scene to have important effects, the
present study indicates that the history and culture of the

170
region should be included in theoretical models and can be
incorporated into empirical models.
Despite the technical limitations confronted in this
study, the evidence points clearly to the importance of
culture and historical context in conditioning and shaping
contemporary attitudes. While this study may have been
innovative in its use of the available contextual data, one
should not view its importance as merely technical. Rather,
the study provides empirical confirmation of the importance
of political culture in theoretical constructions.
Of course, there are limitations to the carrying
strength of this pronouncement. America is changing rapidly.
The communication revolution is changing our culture in as of
yet unforeseen ways. The American public is more transient
than ever; the acorn falls further from the tree than it did
in the past. Major demographic shifts are occurring: the
growth of the sunbelt, the growth of the aged as a proportion
of the population, and the increasing importance and numbers
of people of color. All these factors have changed and are
making America more diverse. These changes have the
potential to decrease the importance of local historical
factors. For the time being, however, those historical and
cultural factors have some force and can contribute to our
understanding of the nature of politics in America and
elsewhere.
Finally, even though mass attitudes not policy analysis
comprise the focus of the present study, this research has

171
policy implications. The first and most obvious lies in the
consistent effect of education on the level of prejudicial
attitudes in all models. Policy makers wishing to blunt the
force of prejudice should turn their attention to the
schools. This study show a clear relationship between the
length of education and the amelioration of prejudicial
attitudes. Finding ways to keep youth of America in school
would seem to reduce the likelihood of their adopting
extremist positions. The content of education probably has
some effect on the propensity of individuals to adopt less
prejudicial attitudes, but the findings of this study do not
speak directly to that point.
Calling for educational improvements represents a
particularly un-controversial policy recommendation for any
study read largely by educators. The second central
recommendation flowing from this research may be more
controversial. The willingness of 'threatened whites' to
support programs which aid the disadvantaged may point the
way to the formation of political coalitions which may elect
officials able to influence the factors causing prejudice.
In other words, poor whites and poor blacks face many of the
same challenges. Black and white candidates may enjoy
success in some areas by appealing to these common challenges
and building multi-racial electoral and governing coalitions.
The study presents a number of questions for future
research. As noted above, unraveling the relationship
between income parity and prejudicial attitudes presents a

172
scholarly puzzle yet to be solved. At this writing the
ultimate fate of minority access districts is unknown. There
is merit in understanding the effects of contextual factors
on the electoral coalitions established in those and other
districts with significant minority populations. Similarly,
the approach taken in this study may be useful in exploring
other areas of mass political behavior. For example, one can
easily see how an area's economic history and context may
influence a variety of political orientations: ideological
dispositions, economic policy preferences, or receptivity to
taxation to name a few. Finally, the findings of this study
imply that future research should continue to find innovative
ways to asses the effects of historical and cultural factors,
whether that research take quantitative form or not.

APPENDIX
Question text from the 1986 American National Election Study
Number for each response in parentheses for whites with white
interviewers.
Racial Threat scale items:
V516 I would like you tell me whether or not you agree or
disagree with the following statement: Affirmative action
programs for blacks have reduced whites' chances for jobs,
promotions and admissions to schools and training
programs.(Do you agree or disagree)
V517 Do you agree strongly or not strongly?
Do you disagree strongly or not strongly?
l)agree strongly(232) 2)agree not strongly(217)
3)disagree not strongly(217) 4)disagree strongly(111)
8)DK(2) 9)NA(0) 0)INAP(957)
Valid percent giving most extreme response: 29.9; 57.8% in 1
plus 2.
V560 what do you think the chances are these days that a
white person won't get admitted to a school while an equally
or less qualified black person gets admitted instead? Is
this very likely, somewhat likely, or not very likely?
l)very likely(230) 3)somewhat likely(343)
5)not very likely(250)
8)DK(44) 9)NA(5) 0)INAP(864)
Valid percent giving most extreme response: 27.9
V563 What do you think the chances are these days that a
white person won't get a job or promotion while an equally or
less qualified black person gets one instead? Is this very
likely, somewhat likely, or not very likely to happen these
days?
1)very likely(224) 3)somewhat likely(410)
3)not very likely(205)
8)DK(2 7) 9)NA(6) 0)INAP(864 )
Valid percent giving most extreme response: 26.7
V569 What do you think the chances are these days that you
or anyone in your family won't get admitted to a school while
an equally or less qualified black person gets admitted
instead? Is it very likely, somewhat likely, or not very
likely?
1)very likely(105) 2)somewhat likely(247)
173

174
4)somewhat unlikely(234) 5)very unlikely(223)
6)NA(12) 8)DK(44) 9)NA(7) 0)INAP(864)
Valid percent giving most extreme response: 12.8; 42.9% in 1
plus 2.
V583 What are the chances these days that you or anyone in
your family won't get a job or promotion while an equally or
less qualified black employee receives one instead? Is this
very likely, somewhat likely, somewhat unlikely, or very
unlikely
l)very likely(104) 2)somewhat likely(240)
4)somewhat unlikely(233) 5)very unlikely(245)
8)DK(37) 9)NA(5) 0)INAP(864)
Valid percent giving most extreme response: 12.5; 41.4% in 1
plus 2.
Symbolic Racism scale items:
V559 Some people say that the civil rights people have
pushed too fast. Others say feel that they haven't pushed
fast enough. How about you: do you think that civil rights
leaders are trying to push too fast, are going too slowly, or
are they moving at about the right speed?
l)too fast(241) 3)about right(523) 5)
too slowly(72)
8)DK(20) 9)NA(16) 0)INAP(864)
Valid percent giving most extreme response: 28.8
V565 Now I am going to read several statements, as I did
before. After each one, I would like you to tell me whether
you agree strongly, agree somewhat, or disagree strongly with
the statement. As before, you can just give me the number of
your choice from the booklet, the first statement is...
Most blacks who receive money from welfare programs could get
along without it if they tried?
l)agree strongly(215) 2)agree
somewhat(298)
3)neither agree nor disagree(121)
4 jdisagree somewhat(159) 5)disagree strongly(54)
8)DK(17) 9)NA(8) 0)INAP(864)
Valid percent giving most extreme response: 25.4; 60.6% in 1
plus 2.
V568 Irish, Italians, Jewish and many other minorities
overcame prejudice and worked their way up. Blacks should do
the same without any special favors?
1)agree strongly(280) 2)agree somewhat(287)
3)neither agree nor disagree(105)
4)disagree somewhat(143) 5)disagree strongly(39)
8)DK(10) 9)NA(8) 0)INAP(864)

175
Valid percent giving most extreme response: 32.8; 66.4% in 1
plus 2.
V579 It's really a matter of some people not trying hard
enough; if blacks would only try harder they could be just as
well off as whites.
l)agree strongly(190) 2)agree
somewhat(317)
3)neither agree nor disagree(115)
4)disagree somewhat(161) 5)disagree strongly(71)
8)DK(9) 9)NA(9) 0)INAP(864)
Valid percent giving most extreme response: 22.2; 59.4% in 1
plus 2.
Coding of other items:
V356 Rs economic situation--We are interested in how people
are getting along these days. Would you say that you (and
your family living here) are better or worse off financially
than you were a year ago.
Is that much better/worse off or somewhat better/worse off.
l)Much better 5)Much worse.
V373 R's perception of National economyHow about the
economy in the country as a whole? Would you say that over
the past year the nation's economy has gotten better, stayed
about the same, or gotten worse?
Would you say much better/worse or somewhat better/worse?
l)Much better 5)Much worse.
V385 R's self placement on seven point ideological scale.
1) Extremely liberal 7)Extremely conservative.
V595 R's Age Actual age.
17) seventeen years 99) 99 years or older
V602 Summary of R's education.
1) 8 grades or less 2) 9-11 grades completed
3) H.S. Diploma/GED
4) More than 12 years 5) AA degree 6) BA level
7) Advanced degrees
V733 R's family income for 1985.
1) less than $2,999 22) $75,000 and over
South. l)Alabama, Arkansas, Florida, Georgia, Louisiana,
Mississippi, North Carolina, South Carolina, Tennessee,
Texas, Virginia.
0) all others
Leadership Evaluation constructed with Feeling Thermometers
(FT)

176
Mario Cuomo=CFT, Robert Dole=DFT, Jesse Jackson=JFT
((CFT+DFT)/2)-JFT=Leadership evaluation
Group evaluation constructed with FTs
Poor People FT-Blacks FT=Group score
Federal spending questions-
If you had a say in making up the federal budget this year,
for which of the following programs would you like to see
spending increased and for which would you like to see
spending decreased. Should federal spending on be
increased, decreased or kept about the same?
1) Decreased 0) all others
Data Derived from the Census
Percentage black the black population divided by the total
population for each SMSA (where available) or county.
Percentage employed in agriculture the number employed in
agriculture, forestry and fishing divided by the total
population for each SMSA (where available) or county.
Nativity the number of residents born in the state they now
reside in divided by the total population for each SMSA
(where available) or county.
Income quotient the income earned by blacks divided by the
income earned by whites for each SMSA (where available) or
county.

LIST OF REFERENCES
Anderson, Barbara A., Brian D. Silver and Paul R. Abramson
1988a The effects of the race of the interviewer on
measures of electoral participation by blacks in the SRC
national election studies. Public Opinion Quarterly
52:53-83.
Anderson, Barbara A., Brian D. Silver and Paul R. Abramson
1988b The effects of the race of the interviewer on
race-related attitudes of black respondents in SRC/CPS
National Election Studies. Public Opinion Quarterly
52:289-324.
Anderson, Neis 1962 The urban way of life. International
Journal of Comparative Sociology. 3:175-188.
Allport, J.D. 1954 The nature of prejudice. Reading, MA:
Addison-Wesley.
Baker, Donald G. 1975 Race, power and white siege cultures.
Social Dynamics. 1(2):143-157 .
Blalock, Hubert M., jr. 1957 Percent non-white and
discrimination in the South. American Sociological
Review 21:583-5883.
Blalock, Hubert M., jr. 1965 Toward a theory of minority
group relations. New York: Wiley.
Blalock, Hubert M., jr. 1967 Causal inferences, closed
populations and measures of association. American
Political Science Review 61:130-136.
Blauner, Bob 1972 Racial oppression in America. New York
Harper and Row.
Bloom, H.S. and H.D. Price 1975 Voter response to short run
economic conditions: the asymmetric effect of prosperity
and recession. American Political Science Review
69:1240-1254.
Bobo, Larry 1983 Whites' opposition to busing: symbolic
racism or realistic group conflict? Journal of
Personality and Social Psychology 45:1196-1210.
Bobo, Larry 1988 Group conflict, prejudice, and the paradox
of contemporary racial attitudes. In Phylis A. Katz and
177

178
Dalmas Taylor (eds.), Eliminating racism: profiles in
controversy. New York: Plenum.
Bobo, L. and F.D. Gilliam 1990 Race and sociopolitical
participation, and black empowerment. American
Political Science Review. 84:377-93.
Boyd, L. H. Jr., and G.R. Iverson 1979 Contextual analysis:
concepts and statistical techniques. Belmont,
California: Wadsworth.
Button, James 1989 Blacks and social change: the impact of
the civil rights movement in southern communities
Princeton: Princeton University Press.
Conway, Mary Margaret 1968 The white backlash re-examined:
Wallace and the 1964 primaries. Social Science
Quarterly 49:710-719.
Edelman, Murray 1971 Politics as symbolic action: Mass
arousal and quiescence. Chicago: Markum Press.
Fossett, Mark A. and K. Jill Kiecolt 1989 The relative size
of minority populations and white racial attitudes
Social Science Quarterly. 70(4)820-835.
Fredrickson, George M. 1981 White Supremacy: a comparative
study of American and South American history. New York:
Oxford University Press.
Giles, Michael W. 1977 Percent black and racial hostility:
an old assumption reexamined. Social Science Quarterly
58:412-417.
Giles, Michael W. and Melanie Buckner 1993 David Duke and
Black Threat: An old hypothesis revisited Journal of
Politics 55:702-713.
Giles, Michael W. and Arthur S. Evans 1985 External threat,
perceived threat, and group identity. Social Science
Quarterly 66(1) 50-66.
Giles, Michael W. and Arthur S. Evans 1986 The power
approach to intergroup hostility. Journal of Conflict
Resolution. 30(3): 469-486.
Goertz, Gary 1992 Contextual theories and indicators in
world politics. World Politics 17(4):285-303.
Green, Donald P. and Jonathan A. Cowden 1992 Who protests:
self-interest and white opposition to busing. Journal
of Politics 54 (May) .-471-496.

179
Hertzog, Mark W. 1991 White flight in the voting booth:
The racial composition of localities and partisan voting
in Virginia in the 1980s. Paper presented to the
Southern Political Science Association November 7-9,
Tampa, Florida.
Huckfeldt, Robert and Carol Weitzel Kohfeld 1989 Race and
the decline of class in American politics Chicago:
University of Chicago Press.
Isaacs, H. 1975 Idols of the tribe. New York: Harper &
Row.
Iverson, Gudmund R. 1991 Contextual analysis. Newbury
Park, California: Sage.
Jackman, Mary R. 1978 General and applied tolerance: does
education Increase Commitment to racial Integration
American Journal of Political Science 22:302-324.
Jackson, Byran 0. 1981 The effects of racial group
consciousness on political mobilization in American
cities. Western Political Quarterly 34:631-646.
Jacobs, Davids 1982 Competition, scale and political
explanations for inequality: an integrated study of
sectoral explanations at the aggregate level. American
Sociological review. 47:600-14.
Joint Center for Political Studies 1988 Black elected
officials: a national roster Washington D.C.: Joint
Center for Political Studies Press.
Jones, Charles E. and Michael L. Clemons 1993 A model of
crossover voting: An assessment of the Wilder victory in
Georgia Persons (ed.) Dilemmas of black politics:
issues of leadership and strategy. New York: Harper
Collins.
Katz, Phyllis A. and Dalmas A. Taylor (eds.) 1988
Eliminating racism: profiles in controversy New York:
Plenum Press.
Key, V.O. 1949 Southern politics in state and nation. New
York: Vintage.
Kinder, Donald R. 1986 The continuing American dilemma:
white resistance to racial change 40 years after Myrdal.
Journal of Social Issues. 41:151-171.
Kinder, Donald R. and D. Roderick Kiewiet 1979 Economic
discontent and political behavior: the role of personal
grievances and collective economic judgments in

180
Congressional voting. American Journal of Political
Science. 23:495-527.
Kinder, Donald R. and D. Roderick Kiewiet 1981 Sociotropic
politics. British Journal of Political Science 11:129-
161.
Kinder, Donald R. and L.A. Rhodebeck 1982 Continuities in
support for racial equality, 1972-1976. Public Opinion
Quarterly 46:195-215.
Kinder, Donald R. and David 0. Sears 1981 Prejudice and
politics: symbolic racism versus racial threats to the
good life. Journal of Personality and Social
Psychology. 40:414-431.
Kinder, Donald R. and David 0. Sears 1985 White opposition
to busing: on conceptualizing and operationalizing group
conflict. Journal of Personality and Social Psychology.
48:1141-1147.
King, Gary 1986 How not to lie with statistics: Avoiding
common mistakes in quantitative political science.
American Journal of Political Science 30:666-687.
Kousser, J. Morgan 1974 The shaping of southern politics.
New Haven: Yale.
Krosnick, Jon A. 1991 The stability of political
preferences: Comparisons of symbolic and nonsymbolic
attitudes. American Journal of Political Science
35:547-76.
Lamis, Alexander P. 1988 The two-party south. New York:
Oxford University Press.
Lieberson, Stanley 1980 Pieces of the pie: blacks and
white immigrants since 1880. Berkeley: University of
California Press.
Matthews, Donald R. and James W. Prothro 1963a Social and
economic factors and negro registration in the South.
American Political Science Review 57:24-44.
Matthews, Donald R. and James W. Prothro 1963b Political
factors and negro voter registration in the South.
American Political Science Review 57:355-367.
McAdam, Doug 1982 Political process and the development of
the black insurgency, 1930-1970. Chicago: University of
Chicago Press.
McConahay, John B. 1982 Is it the buses or the blacks:
Self-interest versus racial attitudes as correlates of

181
anti busing attitudes in Louisville. Journal of
Politics 44:692-720.
McConahay, John B. and J. C. Hough, jr. 1976 Symbolic
racism. Journal of Social Issues. 31:2345.[32(2):23
45]
Merton, R.K. 1957 Social theory and Social Structure. New
York: Free Press.
Miller, Arthur H., Patricia Gurin, Gerald Gurin and Okasana
Malanchuk 1981 Group consciousness and political
participation. American Journal of Political Science
25(August):494-511.
Omi, M. and Howard Winant 1986 Racial formation in the
United States. New York: Routledge and Kegan Paul.
Page, Benjamin 1977 Elections and social choice: the state
of the evidence. American Journal of Political Science.
21:639-68.
Page, Benjamin and Calvin Jones 1979 Reciprocal effects of
policy preferences, party loyalties and the vote.
American Political Science Review 73:1071-90.
Pettigrew, Thomas F. 1959 Regional variations in Anti Negro
prejudice. Journal of Abnormal and Social Psychology.
59:28-36.
Rhodebeck, L.A. 1992 The politicization of gender
consciousness: a pooled cross-sectional analysis. Paper
presented to the annual meeting of the American
Political Science Association, Chicago, IL.
Riker, William D. and Peter C. Ordeshook 1973 An
introduction to positive political theory. Engelwood
Cliffs, N.J.: Prentice Hall.
Rokeach, Milton 1960 The open and closed mind. New York:
Basic Books.
Rusk, Jerrold G. and John J. Stucker 1978 The effect of the
Southern system of election laws on voting
participation: A reply to V.O. Key, Jr. in Joel H.
Silbey, Allan G. Bogue and William H. Flanagan (eds.)
The history of American electoral behavior. Princeton:
Princeton University Press.
Schermerhorn, R.A. 1956 Power as a primary concept in the
study of minorities. Social Forces. 35:53-56.

182
Schuman, Howard and Jean Converse 1971 Effects of black and
white interviewers on black response in 1968. Public
Opinion Quarterly 35:44-68.
Schuman, Howard, Charlotte Steeh and Lawrence Bobo 1985
Racial attitudes in America: Trends and interpretions.
Cambridge, MA:Harvard University Press.
Scott, James C. 1985 Weapons of the weak: everyday forms of
peasant resistance. New Haven : Yale University Press.
Sears, David 0. 1988 Symbolic racism In P. Katz and D
Taylor (eds.) Eliminating racism, pp. 53-81. New York
Plenum Press.
Sears, David 0. and H.M. Allen, jr. 1984 The trajectory of
local desegregation controversies and whites' opposition
to busing. In N. Miller and M. Brewer (eds.) Groups in
contact: the psychology of desegregation. New York:
Academic Press.
Sears, David 0. and J. Citrin 1982 Tax revolt: Something
for nothing in California. Cambridge: Harvard
University Press.
Sears, David 0., C.P. Hensler, and L.K. Speer 1979 White's
opposition to "busing": self-interest or symbolic
politics? American Political Science Review. 73:369-
384.
Sears, David 0. and D.R. Kinder 1970 The good life, 'white
racism,' and the Los Angeles voter. Paper delivered to
the annual meeting of the Western Psychological
Association, Los Angeles.
Sears, David 0. and D.R. Kinder 1971 Racial tensions and
voting in Los Angeles. In w.z. Hirsh (ed.), Los Angeles:
Viability and Prospects for metropolitan leadership, pp.
51-89. New York Praeger.
Sears, David 0. and D.R. Kinder 1985 White's opposition to
busing: On conceptualizing and operationalizing group
conflict. Journal of Personality and Social Psychology.
48:1141-1147.
Sears, David 0., R.R. Lau 1983 Inducing apparently self-
interested political preferences American Journal of
Political Science 27:223-252.
Sears, David 0., R.R. Lau, T.R. Tyler, and H.M. Allen, jr.
1980 Self interest vs. symbolic politics in policy
attitudes and presidential voting. American Political
Science Review 74:670-684.

183
Shingles, Richard D. 1981 Black consciousness and political
participation: the missing link. American Political
Science Review 75:76-91.
Sniderman, Paul, and Philip E. Tetlock 1986a Symbolic
racism: Problems of motive attribution in political
analysis. Journal of Social Issues 41:129-150.
Sniderman, Paul, and Philip E. Tetlock 1986b Reflections on
American racism. Journal of Social Issues 41:173-187.
Sniderman, Paul, Thomas Piazza, Philip E. Tetlock and Ann
Kendrick 1991 The New Racism. American Journal of
Political Science 35(3):423-47.
Southern Poverty Law Center 1991 Klanwatch Reports Record
Rise in Hate Groups SPLC Report 21(2):1-3.
Stanley, Harold W. 1987 Voter mobilization and the politics
of race: the south and universal suffrage. New York:
Praeger.
Thomlinson, Ralph 1969 Urban Structure. New York: Random
House.
Tufte, Edward R. 1978 Political control of the economy.
Princeton, N.J.: Princeton University Press.
Tullos, Allen 1989 Habits of industry: white culture and
the transformation of the Carolina piedmont. Chapel
Hill: University of North Carolina Press.
Van den Berghe, P.L. 1967 Race and racism: A comparative
perspective. New York: Wiley.
Van den Berghe, P.L. 1981 The ethnic phenomenon. New
York:: Elsevier North-Holland.
Wilcox, Jerry and W Clark Roof 1978 Percent black and
Black-white status inequality: Southern versus non-
Southern patterns. Social Science quarterly 59:421-
434 .
Wilson, w. j. 1973 Power, racism and privilege. New York:
Free Press.
Wright, Gerald C. 1977 Contextual models of electoral
behavior: the southern Wallace vote. American Political
Science Review. 71:497-508.
Zald, Mayer N. and John D. McCarthy 1977 Resource
mobilization and social movements: a partial theory.
American Journal of Sociology 82(6)1212-1241.

BIOGRAPHICAL SKETCH
Randolph C. Horn was born in Opelika, Alabama and raised
in Birmingham, Alabama. He graduated from the University of
the South in Sewannee, Tennessee in 1987 with a Bachelor of
Arts Degree in Third World studies and political science. He
received his Master of Arts degree from the University of
Florida in 1990. He has managed to keep his sense of humor
despite extensive graduate study.
184

I certify that I have read this study and that in my
opinion it conforms to acceptable standards of scholarly
presentation and is fully adequate, in scope and quality, as
a dissertation for the degree of Doctor of Philosophy.
{S- U fcA (cr-rvli H
M.Margaret ''Conway, Chairing
Professor of Political Science
I certify that I have read this study and that in my
opinion it conforms to acceptable standards of scholarly
presentation and is fully adequate, in scope and quality, as
a dissertation for the degree of Doctor of Philosophy.
Leonard Beegh
Professor of
I certify that I have read this study and that in my
opinion it conforms to acceptable standards of scholarly
presentation and is fully adequate, in scope and quality, as
a dissertation for the degree of Doctor o ^Philosophy.
James Button
Professor of Political Science
I certify that I have read this study and that in my
opinion it conforms to acceptable standards of scholarly
presentation and is fully adequate, in scope and quality, as
a dissertation for the degree of Doctor of. Philosophy.
A V CL r
V'^--\V-V AW/t vt\
WayneY.. Francis
Professor of Political Science
I certify that I have read this study and that in my
opinion it conforms to acceptable standards of scholarly
presentation and is fully adequate, in scope and quality,
a dissertation for the degree of Doctor of Philosophy.
Michael D. Martin
Associate Professor of
Political Science
as

This dissertation was submitted to the Graduate Faculty
of the Department of Political Science in the College of
Liberal Arts and Sciences and to the Graduate School and was
accepted as partial fulfillment of the requirements for the
degree of Doctor of Philosophy.
December, 1993
Dean, Graduate School



67
Table 4-1 Derived and Synthetic Hypotheses
Derived Racial Threat Hypotheses
Hrt01: Levels of perceived racial threat will covary with
the percentage black.
Hrt02: Levels of perceived racial threat will be higher
among Southerners.
HRT03:Levels of perceived racial threat will covary with
economic vulnerability.
Hrt045Aggregate (or mean) levels of perceived racial threat
will be highest in the Deep South, less high in the Rim-
South and lowest in the non-South.
Hrt05! Evaluations of candidates will covary with levels of
perceived racial threat when candidates are identified with
the out group as members or as supporters.
Hrt06: Support for safety net policies (exclusive of
racially specific policy positions) will covary with levels
of perceived racial threat.
hRT07: Support for policies which benefits African Americans
in particular will vary with levels of perceived racial
threat.
hRT08! The level of perceived racial threat will covary with
the closeness in income (class proximity) of the racial
groups locally.
Derived Symbolic Racism Hypotheses
HsroI: Levels of symbolic racism will be uncorrelated with
local racial composition.
Hsr02: Levels of symbolic racism will be uncorrelated with
individuals' regional origins or residence.
HSR03: Levels of symbolic racism will be uncorrelated with
economic vulnerability.
HSR04: Aggregate levels of symbolic racism will be
uncorrelated with region.
Hsr05: Evaluation of liberal or black candidates will covary
with level of symbolic racism.


TABLE OF CONTENTS
ACKNOWLEDGMENTS iii
LIST OF TABLES vi
LIST OF FIGURES viii
ABSTRACT ix
CHAPTER 1 1
INTRODUCTION 1
Recent Trends in the Study of Prejudice 1
Perceptions of Threat 2
Symbolic Racism 4
Synthesis 6
CHAPTER 2 10
PERCEPTIONS OF RACIAL THREAT: THEORY AND MEASUREMENT 10
Foundations 11
The Contemporary Significance of Race at the Macro
Level 15
Aggregate Analyses 19
Individual Analyses 24
Reconciliation 32
CHAPTER 3 36
SYMBOLIC RACISM: THEORY AND MEASUREMENT 36
Symbolic Politics in a Nutshell 37
Theoretical Underpinnings of Symbolic Racism 39
Newness or Distinctiveness from Old Fashioned Racism 42
Combination or Multi-dimensionality 45
Abstraction vs. Self Interest 47
Methods and Measurement 49
Self Interest: The Alternative Hypothesis 53
Data 58
CHAPTER 4 61
SYNTHESIS 61
Multi-dimensionality 63
Defining the Nature of Interests 71
Synthesis
IV


56
self-interest measures, also measuring a single concept, are
treated as multiple unrelated measures. This comparison
presents a bias against the unsealed items, a bias that is
amplified by colinearity among the unsealed items.
A third test breaks the sample into smaller groupings
based on the responses to one of the measures designed to tap
self-interest. The complete model is then re-estimated
(without the measure used to divide the sample) for each sub
group of the total sample. As above, the symbolic racism
scale is found to have significant predictive power while the
measures of self-interest fail. This interpretation is
problematic. First, the division into the groups represents
just another way of controlling for the effect of the
variable used as the divisive criterion. Such a division is
reasonable under a number of circumstances, for example if
nonlinear effects are expected for different levels of the
criterion variable or if the criterion variable was shown to
empirically or has a theoretically significant interaction
with another variable in the model. The presentation by
Kinder and Sears (1981) is not much of an improvement over
having the criterion variable as another independent
variable.
A second problem associated with this test is that it
rests on a comparison of the marginal effects of the
independent variables across 'populations' or sub-samples
rather than on the different absolute levels within each sub
sample. In this sense, as above, one expects the marginal


20
Of course, many other factors contributed to Wilders
success (Jones and Clemons 1993). Wilder, the former
Lieutenant Governor, had risen through a series of
influential Virginia offices, so there were no doubts about
his qualifications. He was endorsed by many prominent
Democratic leaders in the state. Wilder's reputation and
campaign statements placed him clearly as a "social moderate
and a fiscal conservative" (Jones and Clemons 1993, p. 140).
Wilder's pro-choice stance allowed him to benefit from the
urgency lent the abortion issue by the U.S. Supreme Court's
1989 Webster decision. Wilder was perceived as tough on
crime and most thought his opponent responsible for negative
campaigning. Jones and Clemons (1993) claim that Wilder
intentionally distanced himself from the black community and
from racial issues, running what they term a 'deracialized'
campaign.
The traditionally perceived relationship between racial
composition and racial attitudes and white political behavior
is born out by Hertzog's (1991) analysis. Of course, the
percent black does not automatically determine victory in
electoral contest. Much more is at play. In the Wilder
case, one sees a highly qualified moderate Democrat running a
professional campaign, endorsed by important pols, on the
popular side of a divisive issue, and intentionally de
emphasizing race. Wilder's performance was optimal. Still
the pattern of white support remains. Hertzog (1991) argues
that those whites living in areas with fewer blacks felt less


158
The analysis presented in Chapters Five, Six and Seven
provides evidence with which one may assess the theoretical
propositions laid out in Chapter Four. In the next three
sections, those hypotheses will be assessed in light of the
preceding analysis. Additionally, general conclusions about
the research will be drawn as well as a brief discussion of
the prospects for future research in this field. The
hypotheses were divided into three categories: some were
direct derivations of the theoretical positions laid out by
the proponents of the racial threat school, some were
similarly derived from the symbolic racism school, and some
were synthetic reconciliations of the two schools. Those
hypotheses will be discussed in the following three sections
of this chapter. The hypotheses derived from the racial
threat literature will be taken up first.
Assessing the Derived Racial Threat Hypotheses
Many of the hypotheses derived from the racial threat
literature are upheld or supported by the analysis presented
in the preceding chapters (see Tables 5-3, 6-2, 7-2, 7-4 and
7-6). The hypotheses involving the predictors of perceived
racial threat are especially robust. The percentage black
1980 as well as the archaic percentage black (1940)
demonstrate positive and significant associations with the
level of perceived racial threat, even when numerous other
variables were controlled (Hrtoi)- Similarly, respondents
living in the South have higher levels of perceived racial


146
Regional Differences in Attltudinal Unity
The discussion thus far has focused on the effects of
independent variables on an individual's level of prejudicial
racial attitudes. In this section discussion addresses the
effects of contextual variables on the dispersion of racial
attitudes within a given area. Recall that the
interpretation of Key's (1949) work given in the second
chapter implies a connection between the percentage black and
the dispersion of racial attitudes in the South. Namely,
because of the perceived threat posed to whites by blacks,
higher percentages black will be associated with greater
unity of attitudes.
This hypothesis is assessed in this section by using the
standard deviation within each county of the two racial
scales as the dependent variables regressed on contextual
variables used in the preceding analysis. Again, some
caveats are in order. Some counties may have only a few
respondents from which to calculate the standard deviation.
(The Ns in the tables refer to the number of counties used in
the analysis.) Because of this the results presented in the
next two tables only point the way toward future research on
this topic rather than providing verifiable proof of the
issues at hand.
Table 7-5 presents the results from models with the
standard deviation of symbolic racism as the dependent
variable for Southerners (at the top of the page) and


87
migrate from places with different traditions as adults. For
example, Fossett and Kiecolt's findings reveal that people
raised in the South but living in the North were not
significantly more likely to harbor prejudicial attitudes
than other Northerners (1989, discussed in detail in Chapter
Two). Evidently the context in which people find themselves
continues to matter after childhood socialization is
complete. And, the context dynamically shapes their current
values, beliefs and behavior on issues of race. Indeed,
[r]acial socialization does not end with adolescence."
(Kinder and Rhodebeck 1982, p. 205). So adults may be
influenced by the prevailing political culture and the
material conditions in their state and community.
In this sense, the current racial political culture is
shaped by two essential elements: 1) traditions, values and
beliefs established in the past; and 2) the current material
context. Contemporary measures of context may inform
research about both of these elements. For example, the
contemporary racial composition (and other measurable data)
of a community reflects to some degree the visibility of the
minority community and other factors discussed in the
previous section and may reveal (or contain) information
concerning the political culture or the weight of the past.
As discussed above, the contemporary racial composition
cannot mean the same thing that it did when Key wrote, but
the racial composition from that era can (Wright 1977,
Mathews and Prothro 1963a,1963b). Operationally, data drawn


Ill
variable plus the two interaction terms, the square of the
percentage black 1980 and the difference variable discussed
above.
Models Explaining the Origins of Racial Attitudes
The models for symbolic racism are presented in
Table 6-1. Equation I presents the contextual variables.
Three things are noteworthy in this equation. First, the
percentage black at both the state and county level for 1940
are positively related to the symbolic racism scale. This
seems to indicate that the historical context has some
effect, albeit a weak one, on the level of symbolic racial
attitude despite the theoretical claims of abstraction.
Similarly, the significance of the agricultural employment
variable has some influence on the symbolic scale, indicating
some sensitivity to the current context (although this may be
an indirect effect, see below). Finally, note that the
contextual variables alone explain relatively little of the
variance, about 7%. These relationships change somewhat when
the individual level variables are added to the equation.
Equation II in Table 6-1 presents the next model
with both contextual effects and individual level variables.
Note that education and conservatism have dramatic direct
effects on the symbolic racism scale; increased levels of
education reduces the level of symbolic racism, while
conservatism is associated with higher levels of the
prejudicial attitude. The inclusion of the individual level


136
Table 7-1. Symbolic Racism Regressed on Contextual and
Individual Level Variables for White Southern Respondents
Only.
entries are unstandardized coefficients from OLS regression
Variables
i
II
III
%black 1940

-0.25
-0.16
State %black 1940

-0.04
-0.11
%black 1980
0.45*
0.71*
0.49
%employed in
agriculture


0.38
income equality


-5.63
v595 age
-0.0008
0.003
0.01
v602 education
1.35***
-1.27***
-1.22**
v733 family income



v356 personal finance



v385 conservatism
0.93**
0.91*
0.74*
state %black
1900*average farm size


0.0002
%black 19802
-0.01*
-0.02*
-0.01
%black 19402

0.004
0.001
%black 1940-%black
1900


-0.09
constant
11.51***
10.61***
11.13*
adj R2
0.17
0.15
0.16
N
146
144
144
Source: 1986 ANES;1900,1940,1980 Census


105
the respondent's evaluations of national political figures.
How people rate such figures is important because such
ratings have been directly linked to vote choices (e.g. Page
and Jones 1979). For this variable, the Jesse Jackson
feeling thermometer is subtracted from the average rating of
Mario Cuomo and Robert Dole. This is intended to tap the
respondents rating of Jackson in light of the respondent's
rating of political figures in general. Cuomo and Dole are
selected because they represent prominent leaders in each of
the two major parties. This measure might be improved by
using more leaders (objects) to improve the accuracy of the
mean from which the Jackson rating is subtracted. However,
increasing the number of figures used in the calculation
increases the number of missing cases. Further, Gary Hart
represented the only Democrat rated besides Cuomo and
Jackson.
Group evaluations follow a similar method. For this
variable, the feeling thermometer rating of blacks is
subtracted from the rating of poor people. While poor people
do not represent an especially popular group, neither do they
represent an especially threatening group.
Multivariate tests of the effects of the racial scales
on these endogenous variables are discussed in greater detail
in the following chapter.
This chapter has focused on issues concerning the data
used in the present study. The data sources have been
outlined. The construction of scales used to measure the


Abstract of Dissertation Presented to the Graduate
School of the University of Florida in Partial Fulfillment of
the Requirements for the Degree of Doctor of Philosophy
RACIAL COMPOSITION AND RACIAL THREAT:
THE ANATOMY OF CONTEMPORARY RACISM
By
Randolph Claiborne Horn
December, 1993
Chairman: M. Margaret Conway
Major Department: Political Science
While the use of thinly veiled racial issues in recent
elections serves as a reminder of the salience of race in
American politics, scholars disagree about the causes and
effects of white racial attitudes. The proponents of
symbolic racism and group conflict approaches have engaged in
a debate over which single approach best explains racism.
This dissertation demonstrates that symbolic attitudes and
perceptions of racial threat occupy different dimensions in
the construction of white racial attitudes. Racial attitudes
do not exist in a vacuum; the components and the effects of
those attitudes are tempered by historical factors and the
contemporary racial context. Further, each dimension has a
differential impact on political behavior.
This project combines composition data from the 1900,
1940, and 1980 censuses with survey data taken from the 1986
ix


128
regression. Logistic regression allows for multivariate
modeling, in the same manner as OLS regression, but for
models with dichotomous dependent variable. The only
essential caveat is that the standard per unit
interpretations given OLS parameter estimates should be
avoided since logistic parameter estimates are functions of
changes in the odds of a particular dependent response
category, not units change. The signs and P-values are
interpreted the same way.
Table 6-4 present the results of four logistic
regression equations estimated using federal funding
questions as dependent variables. The first equation
demonstrates the predictive power of the racial scales in
explaining the desire of the respondent to decrease funding
for food stamps. Note that the two scales have opposite
signs in this model. Those who rate higher on the symbolic
racism scale are more likely to support decreases in funding
for food stamp programs. The pattern for perceived racial
threat is the opposite. The more threatened the individual
the more likely that individual will resist decreases to food
stamp funding. (Discussion follows below.) Of the control
variables, only one has a significant effect. As might be
expected those with higher family incomes are less supportive
of food stamp programs that those earning less. Perhaps the
relatively better off believe they have little to gain by
supporting programs that help the poor.


141
Table 7-3. Symbolic Racism Regressed on Contextual and
Individual Level Variables for White Northern Respondents
Only.
entries are unstandardized coefficients from OLS regression
Variables
I
II
III
%black 1940

-0.04
-0.03
State %black 1940

-0.13
-0.16
%black 1980
-0.03
0.06
0.04
%employed in
agriculture


0.02
income equality


0.23
v595 age
0.02*
0.02b
0.02
v602 education
-0.98***
-0.97***
-1.01**
v385 conservatism
0.58***
0.60*
0.16**
State %black
1900*Average farm size


-0.0005
%black 19802
0.004*
-0.0006
0.002
%black 19402

0.003
0.003
%black 1940-%black
1900


0.09
constant
11.09***
11.23***
11.33**
adj R2
0.22
0.22
0.22
N
482
448
430
Source: 1986 ANES; 1900, 1940 and 1980 Census


39
support for this proposition in their 1981 study, even when
measures of satisfaction with income and satisfaction with
current financial status were controlled.
Of course, this literature is not without critics, but
neither is it without fans. In this case discussed above,
the symbolic politics approach contributes to our
understanding of the calculus of the vote choice. At the
same time, its usefulness in understanding politics in
general may be reasonably questioned. The approach
constitutes a black box of sorts; the symbolic
predispositions go in one side and the vote comes out the
other. One might imagine that the context in which this
process occurs might have some influence on it. How does the
voter evaluate what is 'good for the country' economically
vis-a-vis what is 'good for the country' militarily or
racially or educationally? Do voters project their
preferences as the most benevolent path for the country?
Despite these and other obvious criticisms, one can see the
possibility that an individual might suppress his or her
economic self-interest in making the vote choice.
Theoretical Underpinnings of Symbolic Racism
The theory of symbolic racism was introduced about 20
years ago (Sears and Kinder 1971). It has not remained
static since that time but has evolved. In particular, the
claims made by its proponents early on have been tempered.
For example, The assertion that traditional prejudice, "can


183
Shingles, Richard D. 1981 Black consciousness and political
participation: the missing link. American Political
Science Review 75:76-91.
Sniderman, Paul, and Philip E. Tetlock 1986a Symbolic
racism: Problems of motive attribution in political
analysis. Journal of Social Issues 41:129-150.
Sniderman, Paul, and Philip E. Tetlock 1986b Reflections on
American racism. Journal of Social Issues 41:173-187.
Sniderman, Paul, Thomas Piazza, Philip E. Tetlock and Ann
Kendrick 1991 The New Racism. American Journal of
Political Science 35(3):423-47.
Southern Poverty Law Center 1991 Klanwatch Reports Record
Rise in Hate Groups SPLC Report 21(2):1-3.
Stanley, Harold W. 1987 Voter mobilization and the politics
of race: the south and universal suffrage. New York:
Praeger.
Thomlinson, Ralph 1969 Urban Structure. New York: Random
House.
Tufte, Edward R. 1978 Political control of the economy.
Princeton, N.J.: Princeton University Press.
Tullos, Allen 1989 Habits of industry: white culture and
the transformation of the Carolina piedmont. Chapel
Hill: University of North Carolina Press.
Van den Berghe, P.L. 1967 Race and racism: A comparative
perspective. New York: Wiley.
Van den Berghe, P.L. 1981 The ethnic phenomenon. New
York:: Elsevier North-Holland.
Wilcox, Jerry and W Clark Roof 1978 Percent black and
Black-white status inequality: Southern versus non-
Southern patterns. Social Science quarterly 59:421-
434 .
Wilson, w. j. 1973 Power, racism and privilege. New York:
Free Press.
Wright, Gerald C. 1977 Contextual models of electoral
behavior: the southern Wallace vote. American Political
Science Review. 71:497-508.
Zald, Mayer N. and John D. McCarthy 1977 Resource
mobilization and social movements: a partial theory.
American Journal of Sociology 82(6)1212-1241.


66
(N) dimensions. Further, the discussion addresses the two
dimensions for which measures are available in the data
exploited in this study, namely perceived racial threat and
symbolic racism. The specific measures and
operationalizations are discussed in the next chapter.
Specific hypotheses are numbered in the text and summarized
in Table 4-1.
First, one would expect two racial dimensions to show
some degree of independence, but not be completely
uncorrelated (HsynOI)- The similarity in theme should mean
that the dimensions share some combination of inputs (what
explains the attitude), traits (the geographic distribution
of the attitudes or other descriptors), or effects (what the
racial attitudes reveal about other political orientations
and behaviors). Divergence in some of these is to be
expected as evidence of the independence of the dimensions.
For the two dimensions examined here a number of
predictions follow from previous theory. The two dimensions
should differ in their inputs. For example, the local racial
composition should correlate with the perception of racial
threats (HrtoD? this sensitivity has been shown to be more
pronounced in the context of the South with its
segregationist heritage (e.g. Fossett and Kiecolt
1989)(Hrt02) Similarly, an individual's assessment of his
or her financial situation of the state of the economy in
general should covary with sensitivity to perceived threats
(Hrt03) In other words, if one has a dim outlook on the


CHAPTER 2
PERCEPTIONS OF RACIAL THREAT: THEORY AND MEASUREMENT
The claims of some scholars that control of the symbols
of the race debate represents the central struggle of
American politics (Omi and Winant, 1986) may be somewhat over
stated. Still, race seems to represent a fundamental
cleavage in American society. Even thirty years after the
heyday of the civil rights movement, the United states
continues to be a highly segregated society. This social and
geographic segregation has implications for the political
sphere as well. Race or race baiting continues to be used as
a trump card in elections throughout the nation. To be sure,
the race card is not always effective, but in many contests
it continues to be decisive.
The role of race in the literature of political science
is not unified by a common theoretical orientation, unit of
analysis, or methodology. Two important albeit divergent
streams are easily identifiable. The first tends to focus on
the effects of race at some level of aggregation. Scholars
in this stream may examine the effect of race or the local
racial composition on party systems, class differences or
tendencies within an electorate in particular contexts or
environments. The second stream addresses the psychology of
racial attitudes in individuals, apart from any specific
10


RACIAL COMPOSITION AND RACIAL THREAT
THE ANATOMY OF CONTEMPORARY RACISM
BY
RANDOLPH CLAIBORNE HORN
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSPHY
UNIVERSITY OF FLORIDA
1993


CHAPTER 7
REGIONAL ANALYSIS
In the previous chapter, the analysis revealed
differences in the causes and effects of the two racial
scales, symbolic racism and perceived racial threat. In this
chapter, the analysis focuses more sharply on the causes of
the two racial attitudes. In particular, the discussion
attempts to draw out issues related to the regional effects
apparent in the models presented in the last chapter.
Tables 6-1 and 6-2 indicate significant regional
differences in the explanation of racial attitudes. In this
chapter, those models are re-estimated after splitting the
sample into two groups, North and South. The division
follows the same criteria for the coding of the Southern
dummy variable detailed in the appendix. A caveat should be
issued at this point. The division of the sample into two
regional subgroups markedly reduces the number of cases to
the point of straining the statistical models. Indeed, even
the models in the previous chapter pushed the limits
somewhat. Given this restraint the analysis presented in
this chapter does not exactly mirror that of the previous
chapter. For example, some variables are dropped from the
analysis. Like the last chapter, the models are presented in
stages to reveal the effects that the addition of control and
contextual variables have on the primary variables of
134


143
Table 7-4. Perceived Racial Threat Scale Regressed on
Contextual and Individual Level Variables for White
Northerners Only.
entries are unstandardized coefficients from OLS regression
Variables
i
ii
hi
IV
%black 1940

0.16
0.14*
0.23*
State %black 1940

-0.13b

-0.01
%black 1980
-0.02
-0.07
-0.02
-0.06
%employed in
agriculture


-0.10
-0.07
income equality


-2.84
-2.43
v595 age
0.003
0.004
0.005
0.004
v602 education
-0.31**
-0.32**
-0.26*
-0.27*
v356 personal finance
0.20
0.17
0.26
0.23
v385 conservatism
0.39**
0.41*
0.39**
0.39**
state %black
1900*average farm size


-0.002***
-0.002***
%black 19802
0.004*
0.006
0.001
0.004
%black 19402

-0.003

-0.003
%black 1940-%black
1900


-0.08
-0.08
constant
6.14***
6.29*
7.51***
7.47***
adj R2
0.08
0.08
0.10
0.11
N
445
413
398
398
Source: 1986 ANES;
1900, 1940
and 1980
Census.


175
Valid percent giving most extreme response: 32.8; 66.4% in 1
plus 2.
V579 It's really a matter of some people not trying hard
enough; if blacks would only try harder they could be just as
well off as whites.
l)agree strongly(190) 2)agree
somewhat(317)
3)neither agree nor disagree(115)
4)disagree somewhat(161) 5)disagree strongly(71)
8)DK(9) 9)NA(9) 0)INAP(864)
Valid percent giving most extreme response: 22.2; 59.4% in 1
plus 2.
Coding of other items:
V356 Rs economic situation--We are interested in how people
are getting along these days. Would you say that you (and
your family living here) are better or worse off financially
than you were a year ago.
Is that much better/worse off or somewhat better/worse off.
l)Much better 5)Much worse.
V373 R's perception of National economyHow about the
economy in the country as a whole? Would you say that over
the past year the nation's economy has gotten better, stayed
about the same, or gotten worse?
Would you say much better/worse or somewhat better/worse?
l)Much better 5)Much worse.
V385 R's self placement on seven point ideological scale.
1) Extremely liberal 7)Extremely conservative.
V595 R's Age Actual age.
17) seventeen years 99) 99 years or older
V602 Summary of R's education.
1) 8 grades or less 2) 9-11 grades completed
3) H.S. Diploma/GED
4) More than 12 years 5) AA degree 6) BA level
7) Advanced degrees
V733 R's family income for 1985.
1) less than $2,999 22) $75,000 and over
South. l)Alabama, Arkansas, Florida, Georgia, Louisiana,
Mississippi, North Carolina, South Carolina, Tennessee,
Texas, Virginia.
0) all others
Leadership Evaluation constructed with Feeling Thermometers
(FT)


12
obstacles to participation, including poll taxes and literacy
tests.
Poor whites were stymied by the one party system in two
ways. First, the barriers designed to exclude blacks from
participation if evenly applied would have also excluded many
whites. Key dismisses the effects of legal barriers to white
participation. This view is upheld by Harold Stanley's
analysis of participation patterns since the revocation of
those barriers (Stanley 1987; for contrary views see Kousser
1974; and Rusk and Stucker 1978). More important, according
to Key, were other factors: low education, a lock by elites
on real power, the punishment of dissenters, the obfuscation
of factions caused by the one party system and race baiting.
Indeed,
[t]he presence of the black provides a ready instrument
for the destruction of tendencies toward class division
among the whites. The specter of Reconstruction, of
Negro government can still be used to quell incipient
rebellion by discontented whites. (Key 1949, p. 655)
Key paints a picture of the South in which Democratic
single-party rule operated to the benefit of upper class
whites, planters and industrialists. Cheap black labor was
exploited and blacks had little opportunity to participate in
the political arena. Poor whites (rednecks, peckerwoods and
peckerheads) enjoyed a similarly degraded position. They
farmed inferior lands and had little hope for meaningful
participation. Poor whites who did join the fray met with
stiff retribution or faced near impossible odds in the


I certify that I have read this study and that in my
opinion it conforms to acceptable standards of scholarly
presentation and is fully adequate, in scope and quality, as
a dissertation for the degree of Doctor of Philosophy.
{S- U fcA (cr-rvli H
M.Margaret ''Conway, Chairing
Professor of Political Science
I certify that I have read this study and that in my
opinion it conforms to acceptable standards of scholarly
presentation and is fully adequate, in scope and quality, as
a dissertation for the degree of Doctor of Philosophy.
Leonard Beegh
Professor of
I certify that I have read this study and that in my
opinion it conforms to acceptable standards of scholarly
presentation and is fully adequate, in scope and quality, as
a dissertation for the degree of Doctor o ^Philosophy.
James Button
Professor of Political Science
I certify that I have read this study and that in my
opinion it conforms to acceptable standards of scholarly
presentation and is fully adequate, in scope and quality, as
a dissertation for the degree of Doctor of. Philosophy.
A V CL r
V'^--\V-V AW/t vt\
WayneY.. Francis
Professor of Political Science
I certify that I have read this study and that in my
opinion it conforms to acceptable standards of scholarly
presentation and is fully adequate, in scope and quality,
a dissertation for the degree of Doctor of Philosophy.
Michael D. Martin
Associate Professor of
Political Science
as


145
racial threat. Neither of the individual terms demonstrates
significance when placed in similar models without the
partner. The variable was designed to measure the level of
plantation-style agriculture in the South and was expected to
be positively correlated with level of prejudicial attitude
primarily in the South. A tentative explanation might
include the notion that the big agricultural states at the
turn of the century are not the same states which attracted
black migrants in the middle of the century. The competition
for factory and other jobs in the urban and industrial areas
produced a predisposition to perceptions of threat. The
questions raised by this finding cannot be answered in the
present study, but are worthy of further research.
A further adjustment to the model is presented in
equation IV on Table 7-4. In this equation, two racial
composition variables from 1940 which were excluded from the
previous model are included, the state percentage black and
the quadratic term for the county level percentage black.
The addition of these statistically insignificant variables
has little effect on the other independent variables in the
model, save one. The standardized regression coefficient for
the percentage black 1940 enjoys a substantial increase in
its explanatory power. The findings reported in this and the
previous paragraph point to the importance of historic and
contextual factors in understanding contemporary racial
attitudes.


165
composition on racial scales tend to be curvilinear. The use
of quadratic terms indicates that the local racial
composition has its greatest impact on the level of symbolic
racism at lower levels (of percentage black); as the
percentage black increases, its marginal effect on the level
of symbolic racism (or slope) decreases (Hsyn03)* Levels of
symbolic prejudice increase in response to the local racial
composition to a point. This implies that the tendency to
harbor symbolic prejudice is related to exposure to blacks or
information about blacks within the community. This exposure
may increase the opportunities for the attitudes to be used
and, albeit erroneously, reinforced. But this cycle of use
and reinforcement only affects prejudicial attitudes to a
point, after which saturation sets in, preventing escalation
of the prejudicial attitude.
Similarly, the current levels of perceived racial threat
vary with the archaic racial composition (Hsyn04)* The
contemporary racial context also makes an independent
contribution to the models explaining perceived racial
threat. This relationship is also curvilinear, with
decreases in the marginal effects of the racial context
occurring as the percentage black increased beyond a certain
point. But this point is generally outside the realistic
range; the effect of racial composition of the perception of
racial threat begins to decline at about 120% black in most
models. Though quadratic estimation improves fit, the actual
effect of racial composition for most models is close to


3
1) as a class or as individuals may detract from whites'
economic well being by providing competition for
jobs that would otherwise go to whites;
2) present challenges to physical well being (i.e. that
criminal activity engaged in by blacks will endanger
their persons or property); or that
3) present a serious challenge to whites' political
standing (e.g. that black officials or officials
elected by blacks will wrest control of the ship of
state from whites and use the state apparatus to the
disadvantage of whites).
Perhaps low levels of perceived racial threat allowed some
Virginia whites to vote for Wilder. Similarly, high levels
of perceived racial threat fueled support for Duke in
Louisiana. Studies by Hertzog (1991) and Giles and Buckner
(1993) support this notion by finding a relationship between
local racial composition and support for candidates at the
district level. These studies echo the findings of earlier
scholars (Key 1949, Blalock 1957, Allport 1954, Pettigrew
1959, and Giles 1977).
While both Hertzog (1991) and Giles and Buckner (1993)
claim the relationship between racial composition and support
for these candidates is a function of racial threat, they are
unable to substantiate this because they lack individual-
level data; they do not demonstrate that whites actually feel
threatened. While from a behavioral standpoint it is only
necessary to demonstrate that individuals perceive a threat
(or feel threatened), the position adopted by Giles and
Buckner and others posits that the material interests of
whites actually are threatened; the necessity of this
contention is assessed in the next chapter. Using


78
Figure 4-2. Quadratic function with increasing slope


150
the average income for the races is to parity, the greater
the diversity or range of level of perceived racial threat.
While income equality may not tell much about the absolute
levels of threat, it may reveal something about areas in
which a wider range of attitudes exists. The implication is
that where income equality is high and hence competition more
stiff two countervailing conditions may co-exist. Some
whites may develop closer associations with those blacks with
whom they share common interests and means. Others may
experience heightened sensitivity due to competition for
scarce resources, like jobs or housing.
The effects of the independent variables other than the
mean level of perceived threat on the model for Northern
respondents is muted. Equation II indicates that the
percentage black from 1940 has a significant negative effect
on the standard deviation of perceived racial threat. Those
areas with larger black population in the middle of the
century have greater unity of opinion currently on this
racial scale. The relationship appears to be fleeting,
however. When other variables are added to the model, the
effect becomes insignificant.
The findings presented in this chapter have sought to
shed light on regional differences in the racial attitudes.
The results indicate that the measures that are efficacious
in demonstrating the differences between regions are not as
useful in demonstrating the differences between individuals
or counties within the regions. Since the non-South is more


99
War II represented the first act of the nation's civil rights
drama. After that point, the values of individuals were more
subject to influence from either side in the ensuing
struggle. Data from 1940 should provide a snapshot of the
immediate post cotton era but precede the dawning of the
national civil rights movement.
Key (1949) implies that the states have unique political
attributes. In this way the state represents more than an
(often) artificial line around a large territory. Rather,
states develop unique cultures of their own. This has much
to do with the hand states are dealt in terms of natural
resources and the like. But the states' culture also follows
from the importance of states as political units. In this
sense, state politics influence not only the political
culture in the state but even culture in general. Surely the
converse is true as well, that each state's politics are
influenced by the exigencies of the (political) culture.
Still, Key argues many common traits in the South influence
the region's and the states' political development; it is in
this sense that Key spoke of universal effects like the
'friends and neighbors vote' and regional effects like the
cotton economy within the context of state politics.
So, dirt farmers from North Alabama and North
Mississippi have a lot in common, but the fact that they hail
from different states is important. Similarly, dirt farmers
from those two Deep South states (i.e. states entrenched in
the cotton economy) have less in common with dirt farmers


137
percentage black and the level of symbolic racism is positive
until the local maximum is reached at 72%, where the slope
reverses. In other words, the marginal effect of increases
in the percentage black on the level of symbolic racism
decreases as that percentage increases. This reduction in
the marginal effect should not distract the reader from the
essential finding of a positive relationship between the
contemporary racial composition and the level of symbolic
racism.
Equations II and III in Table 7-1 include additional
variables. The relationships discussed above demonstrate
little change with the addition of these variables, none of
which have significant independent effects. The effect of
the contemporary percentage black and its quadratic term in
equation III are just shy of significance at the 0.05 level.
The removal (not shown) of the percentage black 1940 for the
state or county push those variables back into the
significant range. The marginal effects of the archaic
racial composition and the other hybridized variables are
indistinguishable from zero in this reduced sample. Change
from the national models (presented in Table 6-1) may result
from the strain on the models from the reduced number of
cases. Alternatively, the significance of those variables in
the national models may reflected the dramatic cultural
differences between the regions. Once the ultimate regional
control, the division of the sample, is exercised the effects
become markedly less dramatic. Further, the correlation


48
the attitude have its roots? Like other predisposition in
the symbolic politics paradigm, symbolic racism has its roots
in pre-adult and adult socialization. Let us leave aside for
the moment issues arising from the effects of material
conditions on socialization.
All the studies of symbolic racism draw the distinction
between the abstract, or value driven, nature of symbolic
politics and the material context in which the individual
finds him- or herself4. This abstractness principle is
generally presented in juxtaposition to the notion of threats
to personal lives (Kinder and Sears 1981, p. 416, Sears and
Citrin 1982) or the effects of personal experiences
(McConahay and Hough 1976, p. 37; McConahay 1982, p. 705n;
Sears, Lau, Tyler and Allen 1980, p. 671; Sears and Lau 1983,
p. 224) The articulation found in Sears, Hensler and Speer
(1979) is particularly behavioral in its emphasis on the
stimulus-response mechanism:
Whether or not the issue has some tangible consequence
for the adult voter's personal life is irrelevant.
One's relevant personal 'stake' in the issue is an
emotional, symbolic one; it triggers long-held, habitual
responses. (1979, p. 371)
4 The position articulated in a more general study of
symbolic attitudes is inconsistent with the statement. In
Kinder and Rhodebeck's analysis of the 1972-76 ANES panel
study (1982) they state, "These stability coefficients . .
imply both that racial beliefs are comparably durable . .
and that they are far from fixed. Racial socialization does
not end with adolescence." (p. 205)


129
Table 6-4. Logistic Regression of Decreased Federal Spending
on Racial Attitudes for White Respondents Only.
Variable
Racial Attitude
Threat
Food Stamps
-0.04 013
Aid
Unemployed
-0.05b
Aid Blacks
0.03
Space
Research
0.01
Symbolic racism
0.25***
0.18***
0.22***
0.08*
R's Characteristics
Age
0.0002
-0.02*
-0.0009
-0.003
Education
0.04
-0.002
-0.01
-0.25**
Family Income
0.08***
0.02
-0.004
0.01
National Economic
-0.10
-0.32*
-0.03
-0.02
evaluation
Personal economic
-0.06
-0.07)
-0.05
0.18b
evaluation
Conservatism
0.26
0.30*
0.36
-0.25*
Context
Percentage Black
0.30
1.38
2.67b
-0.97
Percentage employed
10.90
17.24b
9.91
12.28
in Agriculture
Income equality
0.37
0.36
-0.54
0.59
Nativity
-0.74
-1.16
-0.13
1.18
South
0.18
-0.38
0.23
-0.45
Constant
-0.67
-0.14
-0.24
0.08
N
497
519
502
518
df=13
model Chi-square/p
88.17/0.0001
42.08/0.0001
90.55/0.0001
40.27/0.0001
a parameter estimates from logistic regression indicating
support for decreased federal spending
* p<0.05 ** p<0.01 *** pcO.OOl
b p<0.05 in a one-tailed test
Sources 1986 ANES, 1980 Census.


108
y=a+piXl . +P4X4+P6Z1 . +piOZ5+rxiZi . +rxjZj+e
where:
y = either the racial threat or symbolic racism scale
Xl=Age
X2=Education
X3=Income
X4=evaluation of the national economic performance
X5=evaluation of personal economic performance
X6=Ideology
Zl=South
Z2ij=Percent black 1980,1940,1900
Z3=Percent employed in agriculture
Z4=Income equality quotient
Z5=Nativity
XiZi to XjZj = all possible combinations of Xs and Zs.
Normally one would check all the two-way interactions as
a matter of course. When interaction terms are significant,
special estimation procedures are required. Typically one
would re-estimate the model for different levels of one of
the constituent variables. Most often, one finds minor
differences in slope on the other variables and only
occasionally opposite signs. Here, one has other reasons for
the inclusion of interaction terms. For instance,
interaction terms can be an effective way of modeling
contextual variables. Boyd and Iverson (1979) provide an


11
context. This second stream will be taken up in Chapter
Three.
In this chapter, two examples of the first school will
be examined in some detail. The first is a classic with a
regional focus, and the second an important contemporary
effort which explores the significance of race for electoral
and party politics nationally. This will be followed by a
close examination of the current work in this school which
uses aggregate and individual level data. While there are
some notable debates among the proponents of this general
area of scholarship, the similarities deserve more emphasis
than the points of disagreement.
Foundations
V. 0. Key's (1949) analysis in Southern Politics is
typical of the first and older line of argument. Key argued
that understanding the role of race was essential to
understanding Southern politics (p. 5). The means through
which blacks were socially and politically subjugated
thwarted the political articulation of class differences
among whites. In particular, the one-party system common
throughout the region prevented the empowerment of both
blacks and poor whites through the political arena. Blacks
were excluded from the system altogether. Extra-legal
sanctions against those who sought change were severe.
Blacks who braved those sanctions faced a maze of legal


68
Table 4-1 (continued)
Derived Symbolic Racism Hypotheses (continued)
Hsr06: Support for safety net policies will vary with
symbolic racism.
hSR07: Support for policies which benefits African Americans
in particular will vary with levels of symbolic racism.
Synthetic Hypotheses
HsYN01: Perceptions of racial threat and symbolic racism
represent different dimensions of racial attitudes.
HSYN02! The level of perceived racial threat will vary with
the level of income equality between the races.
HSYN03: Symbolic racism will covary with the percentage
black at low levels, after which the marginal effect of the
racial composition of symbolic racism will diminish.
HSYN04: Current level of perceived racial threat and
symbolic racism will covary with archaic levels of the
percentage black.
hSYN05: Levels of perceived threat will vary with personal
contacts with out group members, as in the work place.
economy (or that one is losing relative to others) then the
sense of threat from an external source becomes more salient
than it would be under other conditions.
The assumption of abstraction on the part of the
proponents of the symbolic racism hypothesis implies
independence between symbolic racism and contextual factors
like the local racial composition (Hsroi) or the region of
residence or upbringing (Hsr02)* Likewise, symbolic racism
should not be influenced by the individual's economic outlook
because, according to its proponents, symbolic racism is


57
effects to be the same as when the criterion variable was
controlled statistically in an earlier equation. If there
were differences in the absolute levels of the dependent
variable, in this case the evaluation of a mayoral
candidates, how would one know? The marginal effects of the
independent variables should not be very different from the
equations where the criterion was included as a control, but
one would find differences in the constant of the equation.
Technically, the constant is the value of the dependent
variable when all the independent variables are zero. In a
sense, the constant gives a measure of the dependent variable
before the contributions of the independent variables are
included. In this light, the differences between the
subsamples could reside in the constant. For example, those
living in the area where busing is occurring may be so much
more prejudicial than those in other areas that the marginal
effects of the other independent variables is negligible; in
such a situation, comparing constants would expose or
disconfirm this possibility. Kinder and Sears (1981) do not
report the constant, so the comparison across sub-samples is
incomplete.7 Of course, the constants may not have been
7 In general, the omission of the constant is not an
egregious transgression. Generally, when one or more of the
independent variables lacks a meaningful zero point (i.e.
zero falls outside the accepted range for the independent
variable or the models in general) the constant may be
omitted. For example, if year is a predictor of automobile
cost, the value of the constant (the cost of an automobile in
the year zero) is of little interest. The trend in many
journals is to report the constant even when it is
meaningless, for no apparent reason. Here Kinder and Sears


51
Table 3.1 Sears's two dimensions of symbolic racism measures
Antagonism Toward Blacks' Demands
Blacks are getting too demanding in their push for equal
rights. (Agree)b,c
Blacks shouldn't push themselves where they are not
wanted. (Agree)a
Some say that the civil rights people have been trying
to push too fast. Other feel they haven't pushed fast
enough. (Trying to push to fast)d
It is easy to understand the anger of black people in
America. (Disagree)b,c
Resentment About Special Favors for Blacks
Over the past few years, the government and the news
media have shown more respect to blacks than they deserve.
(Agree)b
Over the past few years, blacks have gotten more
economically than they deserve. (Agree)b
The government should not make any special effort to
help blacks and other racial minorities because they should
help themselves. (Agree)e
Do you think blacks who receive money from welfare
programs could get along without it if they tried, or do you
think they really need the help? (Could get along)a
Do you think the Los Angeles city officials pay more,
less, or the same attention to a request or complaint from a
black person as from a white person? (More)a
a Kinder and Sears (1981).
b McConahay (1982)
c McConahay and Hough (1976)
d Sears and Allen (1984)
e Sears and Citrin (1982)
Source: Sears (1988, p. 57)


151
diverse than the South, it is reasonable to expect the more
diversity from the models of the non-South. In this sense,
the variable which measured the level of 'Southerness' in the
last chapter retain some explanatory power for states which
are like the South. The relatively more homogenous nature of
the Southern region makes it stand out in rather stark relief
from the rest of the country, but at the same time makes it
more difficult to identify the causes and explanations of
Southern peculiarities within an exclusively Southern sample.
For example, the effects of leadership in Southern
communities on the attitudes and behavior of residents is
evinced by the uneven pattern of desegregation in the South
during the 1960s. Still, localized effects like this, while
important, go unmeasured in this research. The measured
markers of the South are demonstrative in the national model,
but lackluster once the sample is broken up by region.
One thing is relatively apparent. Understanding the
roots of both racial scales requires an appreciation for both
the present and the historical context. This seems true even
for understanding symbolic racism which showed some
sensitivity to contextual effects in both the South and the
North, despite the ardent assertions by its proponents that
the racial orientation is context-free. Not only do the
individual level regression models imply such a connection
but the analysis of contextual effects on the standard
deviations of the racial scales implies the same. The racial
context at the county level seemed to influence the


59
or a decreased level of symbolic racism where redneck racism
prevails because people there would not need to veil their
attitudes. This would be especially useful if tests were
provided comparing the two forms of racism.
Of course, given the framework of symbolic politics,
there is no expected relationship between context and
symbolic predispositions. In this sense, the theoretical
problems caused by the loci of the various studies only apply
to the alternative hypotheses. Still, the use of national
level data or comparisons of absolute level of symbolic
racism across regions might provide a useful test of the
degree to which the new concept is actually symbolic. If it
were truly abstract then there should be no differences
across regions. Previous studies have not tested this. A
fuller elaboration of this hypotheses is outlined in the next
chapter.
The framework of symbolic racism rests on three pillars:
distinctiveness from known forms of prejudice; the
combination of antiblack affect and traditional values; and
the divorce of symbolic attitudes from context and self
interest. The concept has been tested against a narrowly
constructed self-interest alternative hypothesis. The data
used for those tests have been drawn from municipal and
state-wide samples, primarily collected outside the South.
In this chapter and the one before it, discussion has
focused on two major theoretical traditions in the study of
white racial attitudes in political science. In the next


65
multiple dimensions
in racial attitudes but
fail to explore
the significance of
multi-dimensionality.2
Attitude A on
horizontal axis
Attitude B
low
high
high
I
II
low
III
IV
Figure 4-1. matrix of racial attitudes
There is little reason to expect people's prejudicial
attitudes to follow a zero sum pattern, where one new
attitude replaces the other. Rather, the attitudes may
represent distinct and independent dimensions. A graphical
representation of a two dimensional model is presented in
figure 4-1. This two by two matrix is not meant to pigeon
hole the orientations into four categories. Instead, the two
continua fall on the axes. So an individual could fall
anywhere in the range of axis A, without restricting their
location on axis B.
The literature discussed in the previous two chapters
leads one to some propositions concerning the nature of
multi-dimensional racial attitudes. This discussion focuses
on the theoretical expectations in two dimensions for
clarity's sake but the arguments could be extended to more
2 For example, Sears, Hensler and Speer (1979, p. 382n) use
factor analysis to identify three racial dimensions: 1) old
fashioned racism; 2) opposition to government action; and 3)
opposition to racial protest. Still, they combine all the
items into a single racial attitude scale.


49
Two statements by Kinder and Sears (1985) reinforce this
point.
Resistance to change in the racial status quo would
seem, there-fore, to have little to do with the direct
tangible threats blacks might pose to personal life, and
a great deal to do, instead, with prejudice and values,
(p. 1141)
We say that personal life is usually compartmentalized
away from political life; that any real or imagined
threats posed by blacks to whites' private lives seldom
spill over into antiblack political positions, (p. 1143)
Given the evidence of regional variation presented in
the previous chapter, it is difficult to accept uncritically
the assertion that racial attitudes are divorced from
context. Still, the proponents of symbolic racism (speaking
for themselves) make their position abundantly clear.
Methods and Measurement
Given its ad hoc development, symbolic racism did not
start out with a set of theoretically derived measures.
Rather the concept was deduced from measures meant to tap
general racial attitudes in the Los Angeles studies (Sears
and Kinder 1970, 1971; Kinder and Sears 1981). Since that
time, efforts have been made to develop better measures.
These efforts have met with varying degrees of success.
While critics of symbolic racism point to shortcomings in the
measures used in a number of studies employing national level
data, Kinder (1986) and Sears (1988) argue that these studies
were not designed to test symbolic racism per se. Rather,
they argue that the studies were examinations of general


4
individual-level data to construct a racial threat scale
would facilitate the study of the relationship between
perception of racial threat and political behavior. A
number of other studies support the salience of white
perceptions of racial threat and the effects of local racial
composition. (Fossett and Kiecolt 1989, Giles and Evans
1985, 1986)
The difficulty encountered by those who hold theoretical
stock in the racial threat concept is understandable.
Relatively few of the national surveys contain items that
would allow measurement of such a notion. Still, the 1986
American National Election Study (ANES) contains an extensive
battery of racially oriented questions. Many of these items
may be used to measure perceptions of racial threat. (See
the Appendix for actual text and marginals). The theoretical
ground staked out by the proponents of the racial threat
framework may be measured, surveyed or tested with individual
level survey data which is widely available.
Symbolic Racism
Despite a number of papers indicating the salience of
perceptions of racial threat and local racial composition in
influencing white attitudes and voting behavior, the lion's
share of research on white racial attitudes has situated
itself in the symbolic racism paradigm (Kinder and Rhodebeck
1982, Kinder and Sears 1981, 1985, McConahay 1982, McConahay
and Hough 1976, Sears, Hensler, and Speer 1979). These


29
unable to support Giles' (1977) Southern exceptionalism
explanation. Fossett and Kiecolt's (1989) results differ
because of a slight difference in the way they specified
their models. Using data from the 1976 and 1977 GSS, they
find that support for integration was inversely related to
the percentage black in all regions. They included a measure
of the respondent's community size, because 'urbanness'
should produce greater tolerance (Anderson 1962; Jackman
1978; Thomlinson 1969). Indeed, their data support the
ameliorating effect of 'urbanness.' Moreover, the inclusion
of urbanness in the models reveals the relationship between
racial composition and perceived threat. Presumably,
respondents from urban areas are so much more tolerant in
general than those residing in rural areas that the predicted
linear effect of racial composition is obscured unless
community size is controlled.
Fossett and Kiecolt's (1989) rebuff of Giles' (1977)
lies in their finding that the direction of the relationship
is the same for both southern and non-southern sub samples.
This they take as support for "Lieberson's (1980) hypothesis
that regional differences in white racial attitudes result in
part from regional differences in the distribution of certain
contextual variable." (Fossett and Kiecolt 1989, p.829) This
may be true in part, but their own data indicate that even
with several control variables a South dummy variable remains
statistically significant in the prejudicial direction.
Similarly, the effects of percentage black and a measure of


LIST OF TABLES
Table 5-1. T-test for race of interviewer bias on
whites 93
Table 5-2. Factor analysis for racial items among
whites with white interviewers 94
Table 5-3. Black Concentration in 1900, 1940, and 1980
and the levels of Perceived Racial Threat and Symbolic
Racism in 1986 for White Respondents Only 102
Table 6-1. Symbolic Racism Regressed on Contextual and
Individual Level Variables for White Respondents
Only 112
Table 6-2. Perceived Racial Threat Scale Regressed on
Contextual and Individual Level Variables for Whites
Only 118
Table 6-3. Regression of Evaluation of Candidate and
Group Objects on Racial Attitudes for White
Respondents Only 123
Table 6-4. Logistic Regression of Decreased Federal
Spending on Racial Attitudes for White Respondents
Only 129
Table 7-1. Symbolic Racism Regressed on Contextual and
Individual Level Variables for White Southern
Respondents Only 136
Table 7-2. Perceived Racial Threat Scale Regressed on
Contextual and Individual Level Variables for White
Southerners Only 139
Table 7-3. Symbolic Racism Regressed on Contextual and
Individual Level Variables for White Northern
Respondents Only 141
Table 7-4. Perceived Racial Threat Scale Regressed on
Contextual and Individual Level Variables for White
Northerners Only 143
Table 7-5. Standard deviation of Symbolic Racism
Regressed on Contextual Variables in Each County 147
vi


77
churches, which in turn were able to recruit and maintain
sophisticated leaders (e.g. educated ministers). The same
elements that allowed increased concentrations of blacks in
the cities to generate successful social movements may
aggravate white hostility to the extent that movement
objectives and tactics threaten the economic and political
positions of whites.^ in this light, the higher percentage
black would elevate the sense of threat among whites.
Indeed, because the processes described by McAdam are largely
cumulative one would expect a nonlinear relationship between
the percentage black and the level of threat. Graphing this
relationship would require a quadratic function with an
increasing slope (see figure 4-2).
Similarly, the relative ability of blacks to mobilize
other resources may influence the degree to which whites feel
threatened. McAdam (1982) and others in the social movements
paradigm (particularly the resource mobilization wing) pay
particular attention to the ability of nascent movements to
generate money as a measure of that movement's strength. In
this sense, one might expect the relative wealth of the black
community to correlate directly with the level of threat
perceived by whites. Blalock (1967) argued that the sense of
competition was most heightened among those closest to each
other in class position (Hrt08)* The relative or comparative
level of black income vis-a-vis white income should be an
6 For example, Bobo and Gilliam (1990) found that black
empowerment in major city governments was linked with white
frustration with those governments.


157
lies in two central areas. First, people have the ability to
use more than one form of racial attitude. Symbolic racism
and the perception of racial threat are represented in
different dimensions. In measurement models, people may
posses varying levels of each orientation. Indeed,
individuals may access more racial orientations than this
study could measure.
The second main innovation of the synthetic theoretical
positions lies in the attention paid to the effects of
cultural and historical traits on racial attitudes. People
do not live in a vacuum, so their attitudes, racial and
otherwise, are subject to influence by the local political
culture. The history of a locality influences the ways in
which current residents see themselves and the community. In
particular, past patterns of race relations or past economic
structures, such as plantation agriculture, are thought to
have particular explanatory value in addition to being
measurable.
The measurement of the historical and cultural
differences between regions relies on data from past censuses
as predictors of current racial attitudes. Data are drawn
from the 1900 and 1940 censuses to represent the effects of
past economic and demographic influences on current
attitudes. Demographic and economic patterns established in
an earlier age influence today's political culture.
Similarly, data drawn from the 1980 census provides
information about the current local context.


166
linear. This implies that whites in areas with significant
or even majority black population are more likely to resort
to or harbor perceptions of racially based threat. Racial
animosities are likely to be aggravated when jobs are scarce.
Discussion in the next section assesses the value and
substantive significance of these findings and raises issues
for future research in this area.
Conclusions
The study just completed suffers from any number of
problems afflicting empirical social science research in
general. For example, the theoretical and measurement models
alike are oversimplified and are not able to capture the full
complexities of life's rich pageant or the depths of its
evils. Similarly, even the simplified models suffer from
statistical ailments. The low number of cases strains
analysis at times. The colinearity between contextual
variables obscures relationships. Despite these limitations
and the concomitant caveats, a number of conclusions may be
reasonably drawn.
The matter of symbolic racism is addressed first.
Evidence from this study indicates that some of the claims of
symbolic racism's proponents are untenable. The concept is
measured using widely accepted indicators, yet its behavior
is remarkably inconsistent with the claims of its authors.
Most notably, little support is found for the notion that
symbolic racism, or the scale used to measure it, is actually


133
along without welfare if they tried harder, or that 'blacks
could get ahead like the Jews, Italians, or Irish' might make
similar statements about anyone on welfare or any immigrant
group. One recalls the quip about Archie Bunker, "Hes not a
bigot, he hates everybody the same."
In the next chapter, discussion focuses on regional
difference in the salience, construction and distribution of
racial attitudes.


69
independent of personal life (Hsr03)* Indeed, the proponents
of symbolic racism are clear that symbolic racism should be
completely independent of all environmental factors or
personal interests. (Kinder and Sears 1981, p. 416; Kinder
and Sears 1985, p. 1143; McConahay and Hough 1976, p. 37;
McConahay 1982, p. 705n; Sears, Hensler and Speer 1979, p.
371)
The two dimensions should differ in their traits.
According to the findings discussed in Chapter Two,
perceptions of threat should be more pronounced in the South
given its tradition of sensitivity to such threats. On the
other hand, symbolic racism should not vary regionally,
according to its proponents, because of its independence from
the individuals' material context.3 According to symbolic
racism theory, one would not expect variation among any of
the regions including the most likely suspect, the South. So
aggregate levels of perceived racial threat should be highest
in the South (Hrt04)^ while aggregate levels of symbolic
racism will not vary by region (Hsr04)*
In terms of the effects of the two racial dimensions,
one might expect a different picture. The perception of
3 Historical conditions in the South and previous research
lead one to expect living in the South to have some effect
(input) into the make-up of an individual's racial attitudes;
this notion is supported by the threat literature, although
some scholars see Southern exceptionalism or the uniqueness
of Southern culture and politics as fast fading (e.g.
Stanley). As a trait, however, geographic distribution must
be considered independently form arguments based on the
impact of southernness on the racial attitude. This
difference is especially salient given the abstractness
assumption of symbolic racism.


124
appendix for details). The measure of candidate evaluation
takes the difference between the only nationally prominent
black politician mentioned in the survey and the average of
two other politicians, one Democratic and one Republican.
Similarly, the measure for group evaluation takes the
difference in feeling thermometer rating between blacks and
the poor. The independent variables for both models are the
same. They include the two racial scales, a group of
individual level variables, and measures of the current
context. The results of the models are found in Table 6-3.
Note that the parameter estimates presented are
unstandardized, since the dependent variables in both models
have the same metric.
Table 6-3 presents models of the respondents'
evaluations of a prominent black leader, namely Jesse
Jackson, and of blacks as a group. The variables are coded
such that the dependent variable represents a measure of the
respondents disdain for the attitude object. Accordingly,
Equation 1 is labeled Jackson Antipathy; higher levels of the
dependent variable indicate that respondents rate Jackson
lower than other national political figures. Similarly,
Equation 2 is labeled Group Antipathy: higher levels of this
variable indicate that blacks as a group were rated lower
than the poor.
Equation 1 presents the results of the model predicting
Jackson Antipathy. While neither of the racial scales is
significant at the (two-tailed) 0.05 level, the racial threat


72
alternative hypothesis, namely narrow self interest defined
in largely objective terms. Yet recent scholarship argues
for a broader conception of interest and identity. For
example, a burgeoning social movements literature sees the
development of group consciousness (identity with a group
beyond the self) as an essential element in explaining
movement genesis (McAdam 1982; Zald and McCarthy 1977).
Veritable cottage industries have sprung up to explain
different kinds of consciousness from gender (Rhodebeck 1992)
to race (Jackson 1981; Shingles 1981; Miller, Gurin, Gurin
and Malanchuk 1981), even to identification with businessmen
(Miller, Gurin, Gurin and Malanchuk 1981); these studies
found significant links between group identity and political
behavior and various forms of political participation.
Similarly, much of the extensive group conflict literature
(while maintaining that objective material interests between
groups are at odds) argues variously that group membership or
identity are essential elements of that conflict (e.g. Baker
1975; Blauner 1972; Isaacs 1975; Schemerhorn 1956; Van den
Berghe 1981; Wilson 1973). Giles and Evans (1985) found that
levels of racial hostility were positively related to levels
of white and Anglo-Saxon identity.
Given the weight of theoretical and empirical evidence
documenting the importance of group identification and
consciousness, one cannot conclude that self-interest
represents the sole conduit for the threat mechanism.
Indeed, one expects that individuals would be sensitive not


130
The second equation modeled in this table has as its
dependent variable support for cutting funding for aid to the
unemployed. Like the previous equation, the racial scales
split on this issue. Those with higher levels of threat are
more likely to resist spending cuts for unemployment
benefits, while those with higher ratings on the symbolic
racism scale are more likely to support cutting funds for aid
to the unemployed. (Discussion of this phenomenon follows
below.) The control variables hold few surprises.
Conservatives support cutting spending. Those with
pessimistic outlooks on the national economy and the older
Americans resist spending cuts for the unemployed. Older
Americans might remember hard times, while the most
pessimistic might be more likely to envision the usefulness
of such programs than those with more optimistic outlooks.
Equation three has the respondents' attitudes on funding
for programs that aid blacks as the dependent variable. In
this equation the split between the two racial attitudes
evaporates. As before, those with high symbolic racism
scores are more likely to desire funding cuts in this area
than those with lower scores on that scale. But in this
model the effect of perceived racial threat is reversed. In
the previous two equations respondents with higher racial
threat scores were more resistant to cutting funding for
programs that assisted the unemployed or provided food stamps
for the poor. In this equation, those with higher levels of
perceived threat are more like to support cuts in programs


21
threatened and could more easily cast their votes for the
candidate they preferred for some reason other than race.
Those who felt threatened had more difficulty in casting a
vote for Wilder, other things being equal.
Of course, Hertzog (1991) is not able to substantiate
the intervening variable, or mechanism, of threat without
individual level data. It is entirely possible that the
relationship he observes, albeit a strong one, is spurious.
It is possible, but unlikely, that any other unmeasured
demographic or social characteristic would correlate well
enough with racial composition to produce such a pronounced
effect. Indeed, Wilder's extremely narrow victory, despite
wider margins in opinion polls, implies the importance of
latent prejudices of which individuals were loath to speak.
Other scholars bring a similar sort of analysis to bear
on a most different case. Giles and Buckner (1993) examined
the pattern of support for David Duke in the 1990 Louisiana
race for a seat in the U.S. Senate. Duke, widely perceived a
racist, ran as a Republican and found himself in a run-off
with Bennett Johnston, a Democrat. Duke, former leader of
the Ku Klux Klan, has been characterized by some as a
purveyor of thinly veiled racist jingoism. Less benevolent
observers call him an overt racist. Duke tends to deny that
he is a racist, saying he has changed since his days with the
Klan. Still he does consider himself an advocate for white
people's rights; he even founded an organization called the
National Association for the Advancement of White People


Context and the Changing Meaning of Racial
Composition 75
The Archaic Significance of Racial Composition 83
CHAPTER 5 89
DATA 89
Data 89
Dependent Variables 91
Contextual Variables 95
1980 Census 96
1900 and 1940 Censuses 98
Endogenous variables 104
CHAPTER 6 107
NATIONAL MODELS OF PREJUDICE AND ITS EFFECTS 107
Procedures 107
Models Explaining the Origins of Racial Attitudes Ill
Racial Attitudes and Other Political Orientations 121
CHAPTER 7 134
REGIONAL ANALYSIS 134
Southern Models of Racial Attitudes 135
Northern Models of Racial Attitudes 140
Regional Differences in Attitudinal Unity 146
CHAPTER 8 152
CONCLUSIONS: FARTHER ALONG 152
Racial Threat 152
Symbolic Racism 154
The Synthetic Theoretical Position 155
Assessing the Derived Racial Threat Hypotheses 157
Derived Symbolic Racism Hypotheses: 160
Synthetic Hypotheses 162
Conclusions 165
APPENDIX 172
Racial Threat scale items: 172
Symbolic Racism scale items: 173
Coding of other items: 174
Data Derived from the Census 175
BIOGRAPHICAL SKETCH 184
v


176
Mario Cuomo=CFT, Robert Dole=DFT, Jesse Jackson=JFT
((CFT+DFT)/2)-JFT=Leadership evaluation
Group evaluation constructed with FTs
Poor People FT-Blacks FT=Group score
Federal spending questions-
If you had a say in making up the federal budget this year,
for which of the following programs would you like to see
spending increased and for which would you like to see
spending decreased. Should federal spending on be
increased, decreased or kept about the same?
1) Decreased 0) all others
Data Derived from the Census
Percentage black the black population divided by the total
population for each SMSA (where available) or county.
Percentage employed in agriculture the number employed in
agriculture, forestry and fishing divided by the total
population for each SMSA (where available) or county.
Nativity the number of residents born in the state they now
reside in divided by the total population for each SMSA
(where available) or county.
Income quotient the income earned by blacks divided by the
income earned by whites for each SMSA (where available) or
county.


84
As McAdam (1982) points out, the economic system in the
South underwent a series of revolutions in the early part of
this century. New machinery reduced the need for cheap black
labor. New pesticides had a similar effect. Most
importantly the collapse of the international cotton markets
made the system of cotton dominated monoculture less (or un)
profitable by the beginning of World War II. Even as Key
(1949) wrote, the basis of his argument was changing (or had
changed). Despite the revolution in the Southern economy,
changes in the political system and the culture lagged behind
by forty years or more. What accounts for this lag? As
Tullos (1989) and Scott (1985) document in widely divergent
cases, changes in economic systems (in particular
agricultural revolutions) may change the class system but not
necessarily the relative positions of actors (or families)
within the new system. Powerful families may translate or
transform their advantages (education, capital and other
assets, prestige, and political power) from the previous
system into advantages in the new system. A revolution in
the economy does not produce an immediate social revolution.
This conservation of advantage forestalls the changes one
might expect after such a radical transformation as the
Southern economy experienced.
The structure of the political system and the entrenched
positions of its inhabitants would have delayed immediate
changes in the political system. The lack of a grass roots
organization in the Republican party made (make) challenges


80
Figure 4-3 Parabolic quadratic function


42
Finally, like other symbolic predispositions in the
symbolic politics paradigm, symbolic racial attitudes were
thought to be completely abstract. While older forms of
racial attitudes had been thought to be related to the
context within which the individual found him or herself,
this new set of attitudes was thought to have no relationship
to the individual's context or material interests.
In sum, the essence of symbolic racism, as argued by its
proponents, is characterized by three points. First, it is a
new racial attitude, distinct from previously examined
prejudicial attitudes. Second, it is comprised of two
central elements, antiblack affect and traditional values.
Third, it is wholly abstract, completely unrelated to an
individual's context or material interests. These three
characteristics seem to represent the best and most widely
accepted 'definition' of the concept. The proponents of
symbolic racism have stood by these elements in the face of
some of their more trenchant critics (see the Sniderman and
Tetlock (1986a) exchange with Kinder 1986, or the chapters by
Sears and Bobo in Katz and Taylor, 1988). These exchanges
provide a more detailed history of the concepts development
and weaknesses than space permits here. A more detailed
discussion of the three central elements follows.
Newness or Distinctiveness from Old Fashioned Racism
Early articulations of symbolic racism maintained that
the new combination of anti-black affect and traditional


22
(NAAWP). There is little doubt about the racial content of
Duke's appeal.
Giles and Buckner's (1993) findings are consistent with
those of Hertzog. They find a strong relationship between
local racial composition and the level of white support for
Duke2. Hertzog's estimates of white support for Wilder were
complicated by that candidate's significant black support.
Giles and Buckner assume that Duke garnered a negligible
number of black votes. This reasonable assumption allows
them to more fully examine the central relationship and to
introduce important controls in a multivariate analysis.
Despite the fact that they have only aggregate level
data, Giles and Buckner (1993) include a number of controls
by including census data at the SMSA and parish level. These
controls provide a convenient opportunity to assess the
current impact of theoretically important variables from
earlier work. Included as controls are two measures of
social status, median white income and the percentage whites
with a high school education. Additionally, the percent of
whites unemployed, the percent living in urban areas, the
percent of white inmigration, and the percentage of the
population coming of age after 1961 are also included.
Even with all these controls, their measure of racial
composition proves efficacious in predicting both the
2 Operationally, Giles and Buckner use the percentage of all
registered voters who are listed as black. They argue that
this better represents the threat to white political hegemony
than a population based figure. Of course, the two figures
are so highly correlated that it makes no difference in the
analysis.


179
Hertzog, Mark W. 1991 White flight in the voting booth:
The racial composition of localities and partisan voting
in Virginia in the 1980s. Paper presented to the
Southern Political Science Association November 7-9,
Tampa, Florida.
Huckfeldt, Robert and Carol Weitzel Kohfeld 1989 Race and
the decline of class in American politics Chicago:
University of Chicago Press.
Isaacs, H. 1975 Idols of the tribe. New York: Harper &
Row.
Iverson, Gudmund R. 1991 Contextual analysis. Newbury
Park, California: Sage.
Jackman, Mary R. 1978 General and applied tolerance: does
education Increase Commitment to racial Integration
American Journal of Political Science 22:302-324.
Jackson, Byran 0. 1981 The effects of racial group
consciousness on political mobilization in American
cities. Western Political Quarterly 34:631-646.
Jacobs, Davids 1982 Competition, scale and political
explanations for inequality: an integrated study of
sectoral explanations at the aggregate level. American
Sociological review. 47:600-14.
Joint Center for Political Studies 1988 Black elected
officials: a national roster Washington D.C.: Joint
Center for Political Studies Press.
Jones, Charles E. and Michael L. Clemons 1993 A model of
crossover voting: An assessment of the Wilder victory in
Georgia Persons (ed.) Dilemmas of black politics:
issues of leadership and strategy. New York: Harper
Collins.
Katz, Phyllis A. and Dalmas A. Taylor (eds.) 1988
Eliminating racism: profiles in controversy New York:
Plenum Press.
Key, V.O. 1949 Southern politics in state and nation. New
York: Vintage.
Kinder, Donald R. 1986 The continuing American dilemma:
white resistance to racial change 40 years after Myrdal.
Journal of Social Issues. 41:151-171.
Kinder, Donald R. and D. Roderick Kiewiet 1979 Economic
discontent and political behavior: the role of personal
grievances and collective economic judgments in


171
policy implications. The first and most obvious lies in the
consistent effect of education on the level of prejudicial
attitudes in all models. Policy makers wishing to blunt the
force of prejudice should turn their attention to the
schools. This study show a clear relationship between the
length of education and the amelioration of prejudicial
attitudes. Finding ways to keep youth of America in school
would seem to reduce the likelihood of their adopting
extremist positions. The content of education probably has
some effect on the propensity of individuals to adopt less
prejudicial attitudes, but the findings of this study do not
speak directly to that point.
Calling for educational improvements represents a
particularly un-controversial policy recommendation for any
study read largely by educators. The second central
recommendation flowing from this research may be more
controversial. The willingness of 'threatened whites' to
support programs which aid the disadvantaged may point the
way to the formation of political coalitions which may elect
officials able to influence the factors causing prejudice.
In other words, poor whites and poor blacks face many of the
same challenges. Black and white candidates may enjoy
success in some areas by appealing to these common challenges
and building multi-racial electoral and governing coalitions.
The study presents a number of questions for future
research. As noted above, unraveling the relationship
between income parity and prejudicial attitudes presents a


Table 7-6. Standard Deviation of Perceived Racial
Threat Regressed on Contextual Variables in Each
County
149
Vll


17
management dichotomy no longer represents the sole axis of
ideological placement. New left issues such as environmental
concerns and social change issues (e.g. feminism, gay rights,
etc.) don't fit on the old continuum and represent a
different dimension.1 So, the Democratic coalition of the
New Deal era is complicated by the addition of a new
ideological dimension, and old-style labor liberals find they
have less and less in common with new-left-lifestyle
liberals.
The influx of black voters into the Democratic party
after the passage of the Voting Rights Act of 1965 has
further complicated the task of coalition building in the
party system. In particular, working class whites have
become less supportive of the party. When one confronts this
issue in terms of electoral coalitions, it becomes clear that
the inclusion of blacks displaces working class whites
disproportionately. In other words, the larger the
proportion blacks voters comprise of the Democratic vote, the
smaller the proportion of whites in that same Democratic
electorate. This would seems merely tautological but for the
fact that the reduction in the white contribution (in terms
of Democratic votes) comes disproportionately from the
1 It should be noted that while Huckfeldt and Kohfeld
identify the salience of the changing landscape of American
ideological orientations, they distance themselves from those
who would couch those changes in terms of the development of
"post-material values." Indeed, they demonstrate many of the
shortcoming of post-materialist theory.


33
The discussion of the role of race in politics by Key
(1949) and Huckfeldt and Kohfeld (1989) implies that race is
not primordial, that it does not represent an unresolvable
cleavage. For Key (1949) the fundamental struggle of
Southern politics was class based. Race was used, in part,
as an instrument to thwart insurgent actions by the have-
nots. Key (1949) characterizes the net effect of racial
politics as anti-democratic because the system prevented the
recognition of objective interests and action based on those
interests. In this sense race is an unnatural basis for the
formation of group cleavages. Rather, tenant farmers and
yeoman farmers had common interests regardless of race. Race
baiting prevented the recognition of these common interests
and other forms of repression discouraged participation not
only of blacks, but of poor whites as well. Huckfeldt and
Kohfeld's analysis is similar. Race represents an unnatural
cleavage that prevents the formation of class based political
parties.
While Giles and Evans (1985,1986) do not test the
validity of their assumption that whites actually are
threatened by local blacks or cite empirical studies which
support it, there may be reason to think that there is
something to the notion. For example, the social change
movements of the 1960s and 1970s created an environment in
which racial and gender discrimination was made more
difficult. As a result, women and blacks entered the job
market and began taking some of the jobs that would have


121
black in 1940 indicates not that the level of perceived
racial threat in the south is lower than in the North;
analysis presented in the next chapter reveals that precisely
the opposite is true. Rather, its marginal effect within the
South is zero, because the interaction term cancels out the
effect percentage black in 1940, controlling for all the
other variables in the model. The local racial composition
in 1940 explains some of the variation in contemporary racial
prejudice nationally, but is not able to explain marginal
variations within the South alone. Re-estimation of these
models by region in the next chapter permits a more accurate
interpretation.
After controlling for the South, the marginal effects of
variables already in the model change. The effect of the
racial composition 1940 increases dramatically as does the
impact of that same variable at the state level. This seems
a potent endorsement of Wright's (1977) finding that archaic
racial composition influences political attitudes and
behavior today. Additionally, the impact of the contemporary
racial context comes into play in the predicted direction.
The introduction of more accurate control mechanisms for
'region' allow the revelation of the expected theoretical
relationships.
Racial Attitudes and Other Political Orientations
Having demonstrated differences in the origins of the
two racial attitudes, discussion now turns to evaluation of


97
value of the dollar. The lower the quotient the greater the
income disparity in favor of whites.5
Modernization is expected to have an impact or racial
attitudes. One expects to find a variations in context among
different types of communities. Button (1989) has
demonstrated that the pattern of race relations in southern
communities often varies along a New South/Old South
continuum, with New South communities reflecting more
progressive patterns. It is not possible to replicate
Button's qualitative approach when analyzing national
surveys, but it is possible to construct proxy variables
which quantitatively tap the national differences
corresponding to the New South/Old South continuum, like
urbanity.
But measuring urbanity presents thorny issues. For
example, the suburbanization of America may render measures
based on population density moot. People living in sprawling
and elite suburban communities may have little in common with
farmers or small town folk in areas with identical population
density or belt codes. To address problems like this, two
variables which measure theoretically distinct elements of
modernity and urbanization have been designed: the percentage
of the labor force employed in agriculture and nativity.
5 To expand the example, if the mean incomes in Columbus
were $11,000 and $11,500 the difference is $500 and the
quotient is .956. If the mean incomes in Los Angeles were
$33,000 and $34,000, the difference would be $1,000 and the
quotient .971.


76
system with a particular set of class relationships.5 Are
these interpretations reasonable in the contemporary context?
Is the racial composition a threat in and of itself or is it
a measure of something else? Operationally it provides a
shorthand measure to describe the relative number of blacks
and whites living in a given geographic area. From this, one
can not necessarily conclude that composition in and of
itself represents some threat. But the local racial
composition may be correlated with other salient but
unmeasured details of the locality.
The contemporary racial composition is correlated with
(see later chapters) racial attitudes in the expected ways
but not necessarily for the same reasons that group conflict
proponents argued. The first element of the contemporary
significance of the local racial compositions may found in
its representation as a resource for the mobilization of
black interest groups (McAdam 1982, Blalock 1965). McAdam,
in particular, details the effects of the urbanization of
black America as a resource for the mobilization of interest
groups and social movements. In a nutshell, the
concentration of blacks in the urban setting allowed for the
recognition of common interests, higher incomes than those
available on the farm, and the formation of civic groups and
5 Key (1949) wrote before the great migration was complete;
racial composition was correlated with the ruralness of
Southern counties. Giles' (Giles 1977; Giles and Evans 1985,
1986, Giles and Buckner 1993) work addresses a post-migration
America where the percentage black may be more correlated
with level of urbanity.


172
scholarly puzzle yet to be solved. At this writing the
ultimate fate of minority access districts is unknown. There
is merit in understanding the effects of contextual factors
on the electoral coalitions established in those and other
districts with significant minority populations. Similarly,
the approach taken in this study may be useful in exploring
other areas of mass political behavior. For example, one can
easily see how an area's economic history and context may
influence a variety of political orientations: ideological
dispositions, economic policy preferences, or receptivity to
taxation to name a few. Finally, the findings of this study
imply that future research should continue to find innovative
ways to asses the effects of historical and cultural factors,
whether that research take quantitative form or not.


109
extensive treatment of this technique (see also Iverson 1991,
Goertz 1992).
In the research reported here, most of the two-way
interaction terms are theoretically and statistically
insignificant. One interaction term has particular
theoretical significance and is included in the models
estimated and presented in this chapter, the product of a
South dummy variable and the percentage black 1940. As noted
above the appropriate treatment for statistically significant
interaction terms involves re-estimation of the model at
fixed levels of one of the constituent variables. This
treatment is carried out and discussed in Chapter Seven.
In addition to the tests for interaction effects three
other variables are included in the analysis of the origins
of the racial orientations. The first is a measure of the
degree to which the state economies were dominated by
plantation style agriculture in 1900. This variable is
constructed by taking the product (multiplying) the average
farm size at the state level in 1900 by the percentage black
at the state level for the same year. The second variable,
as discussed in Chapter Four, is the square of the percentage
black in 1980.
The third additional variable is the amount of change in
the percentage black at the county level over 40 year periods
between 1900 and 1940 and 1940 and 1980. For example, this
is computed by subtracting the percentage black in 1900 from
the percentage black in 1940. Both of the possible


82
Figure 4-4. Quadratic function with gradual decay


86
areas with traditions of heightened racial tension,
especially in the South. Of course, the contemporary and the
archaic racial compositions are not uncorrelated. In this
sense, the relationship between the contemporary local racial
composition and racial hostilities does not merely reflect
the current material context, but also reflects cultural
factors which may be unique to the region, state or
community.
The patterns of racial antagonism Key (1949) described
(using racial composition) continue to affect the current
state of race relations in the South and elsewhere. Of
course demographic changes and migration and the
transformation of the Southern economy have the effect of
altering the roots of antagonism in the South. The
information revolution and increased access to higher
education have muted the transmission of prejudicial values
form generation to generation. Still, one expects some
degree of continuity within states, counties and local
communities. Those raised in communities with traditions of
prejudice (or a prejudicial political culture) could be
expected to have more prejudicial attitudes than those raised
in communities without such traditions because those
traditions influence their socialization even if the
transmission of values is attenuated because of the factors
noted above.
Similarly, the local or state political culture
represents a crucial contextual factor for those who may


118
Table 6-2. Perceived Racial Threat Scale Regressed on
Contextual and Individual Level Variables for Whites Only,
entries are unstandardized coefficients from OLS regression
Variables
I
II
hi
IV
%black 1940
0.04
0.04
0.04
0.18***
State %black 1940
0.0000003
0.0000005
0.0000005
0.000003
%black 1980
0.04
0.04
0.04
0.13b
%employed in agriculture
-0.10
-0.22
-0.22
-0.09
nativity
0.02b
0.004
0.004
-0.02
income equality
-2.79*
-3.81**
-3.80**
-3.64*
v595 age

0.001
0.001
0.00004
v602 education

-0.39***
-0.39***
-0.38***
v733 family income

CM
O
O
1
-0.02
-0.02
v356 personal finance

0.29b
0.29b
0.36*
v385 conservatism

0.49***
0.49***
0.45**
south


0.04
2.44b
%black 1900* average farm
size 1900



-0.0006*
%black 19802



-0.005b
south*%black40



-0.19***
%black 1940-%black 1900



0.07
constant
7.46***
7.87***
7.86***
8.20***
adj R2
0.05
0.11
0.11
0.14
N
682
517
517
502
Source: 1986 ANES; 1900,1940 and
1980 Censuses
b p-value < 0.10
* p-value < 0.05
** p-value < 0.01
*** p-value < 0.001


123
Table 6-3. Regression of Evaluation of Candidate and Group
Objects on Racial Attitudes for White Respondents Only.
Entries are unstandardized coefficients from OLS regression.
Jackson
Group
Antipathy
Antipathy
Variable
Racial Attitude
Equation 1
Equation 2
Perceived threat
0.57a b
0.64**
Symbolic racism
0.39
0.04
Rs Characteristics
Age
0.40***
0.22**
Education
-1.83
-0.79
Family Income
0.42
-0.29
National Economic
evaluation
-3.70*
1.12
Personal economic
evaluation
-1.92
0.62
Conservatism
2.56*
-0.54
Context
Percentage Black
-29.42
7.61
Percentage employed
in Agriculture
-294.94b
97.73
Income equality
-12.68
11.90
Nativity
-17.21
9.93
South
3.12
6.06*
Constant
43.79*
-4.01
N
254
493
Adjusted
.15
.10
F/p
4.39/.0001
5.00./.0001
a OLS parameter estimates
b pco.10 p<0.05 ** p<0.01
Source: 1986 ANES, 1980 Census.
pcO.OOl


162
demonstrates no significant explanatory power in predicting
the respondents' evaluation of either a national political
figure (Jesse Jackson) or blacks as a group (Hsr05),
controlling for other variables. The scale does a better
job, however, in predicting the respondents' federal spending
priorities. The level of symbolic racism is positively
related to the likelihood respondents would prefer spending
cuts in all issue areas presented in Table 6-4. This finding
is consistent with the positions laid out in Chapter Four.
Because of adherence to traditional values, symbolic racism
was expected to demonstrate an inverse relationship with
support for safety net policies (Hsr06) A similar inverse
relationship was expected between symbolic racism and
programs which benefit blacks in particular (Hsr07)*
However, the consistency of the effects for all forms of
federal spending raises questions about the usefulness of the
concept. Why, for example, would the racial attitude predict
the respondents' desire to cut funding for the space program?
Similarly, why, given the racial content of the concept, does
the symbolic racism scale not have more explanatory power in
predicting attitudes concerning programs that help blacks
specifically than it does the less racially specific program
of food stamps?
The pattern implied in Tables 6-3 and 6-4 indicates that
the measures, if not the concept, of symbolic racism are
flawed. The symbolic racism scale does not have more
explanatory power when the objects of evaluations or policies


101
categories were provided for people of racially mixed
ancestry each of which had a technical definition (e.g.
mulatto, griffe, quadroon, octoroon) based on the
distribution of the mixture. By today's (culturally
dependent) standards and those used universally in the
research discussed in the previous chapters, all those
categories are subgroups of the general category black.
The 1900 census has a category called 'Colored' which
refers to people of color from places other than Sub-Saharan
Africa. These people might be thought of as Asian, Semitic,
Mestizos, Latin Americans, Filipinos or any of a number of
(typically) nationality-based titles not valued at the time.
The same category name in the 1940 census refers to people
with any (visible) African ancestors and as such is
consistent with contemporary standards of racial distinction.
The percentage black is calculated at both the county and
state level.
An additional variable is drawn from the 1900 census to
gauge in an additional way the effect of the cotton economy
on racial attitudes in the South in particular. The average
farm size at the county and state level should give
information about the prevalence of large farms in those
units. When combined (as an interaction term) with the local
racial composition the result should provide a measure of
plantation-economy-ness or latifundization. This will prove
particularly important for the discussion of the Southern
region in Chapter Seven.


CHAPTER 3
SYMBOLIC RACISM: THEORY AND MEASUREMENT
The role of race in politics has stimulated much
controversy in political science. In the last chapter,
discussion focused on approaches which point to the
perception of threat as the operative mechanism of white
racial antagonism. While scholars disagree as to the verity
of the threat, many see its perception as instrumental in
explaining, in part, larger political phenomena. This school
of thought finds its recent roots in the literature of
Southern politics and related changes to the party system
since the civil rights era. Its methods are largely
aggregative, although a handful of individual level studies
exist. Some of these were discussed.
In this chapter, discussion centers on different
explanations of the role of race in politics. Symbolic
racism1 represents perhaps the dominant paradigm in the
political science literature on white racial attitudes. This
1 "Symbolic racism" refers to a school of thought in the
literature on American racial prejudice. Other names for
this school have been bowed, new racism and modern racism.
These alternatives have all the flaws and none of the
strengths of the original term. Symbolic racism represents
the most universally recognized moniker for this particular
theoretical tradition. Many have attacked the expression on
a number of grounds. The use of the term in this study is
not meant as an endorsement of its content or its
appropriateness as a label for that content. Rather, its use
is adopted, for the purpose of clarity only, to refer to this
school of thought.
36


9
in America; likely as not, they do. It is instructive to
recall R. K. Merton's conception of the relationship between
prejudice and discrimination (1957). In a nutshell, Merton
pointed out that the two concepts represented different
dimensions; imagine a contingency table with prejudice on one
axis and discrimination on the other. It is possible, Merton
hypothesized, that one can be highly prejudiced and not
discriminate, never acting on that prejudice. Similarly,
nonprejudiced individuals may engage in discriminatory
activities.
At times provocative expressions are used. For example,
the expression (not originated by this author) symbolic
racism carries some unnecessary shock value. The expression
is really a misnomer, for the concept refers to a particular
theory of prejudice and need not carry the pejorative
implications of racism. Branding people as racists may have
few positive results, like kicking the dog. Nonetheless,
this term is used in this research because it is used by
other scholars in the field.
This study provides neither apology for nor an
admonition of prejudice per se. Rather, the author hopes to
provide insights into the causes and effects of racial
prejudice. It is hoped that the development of a more
sophisticated understanding of prejudice may contribute to
the amelioration of social conditions which have ill effects
for whites and blacks alike; if we are to defeat the forces
of ignorance and prejudice, we must first understand them.


30
the respondent's Southern origins are significantly different
in models estimated for Southern and non-Southern subsamples.
So, in addition to "differences in the regional distribution
of certain contextual variables," Southernness make a
significant and unique contribution. The regional difference
is one of degree not of kind, but it remains a significant
difference.
As might be expected, the various measures of
Southernness have their effect in the more prejudicial
direction. This points in the direction of another
unrealized finding from the data presented by Fossett and
Kiecolt (1989). The effect of Southern origins is
insignificant in the non-southern sample. This
insignificance may be the result of a paucity of cases.
Alternatively, the insignificance of this variable may be
like the dog that did not bark of Sherlock Holmes fame. The
fact that Southern transplants in the North are not
significantly more prejudicial that their neighbors may point
to the importance of context in structuring political
attitudes. This interpretation is supported by [or
supports] contextual arguments made by Wright (1977) and
others, namely that the most pronounced and important
contextual effects are mediated through one's primary group
processes. In other words, "contextual characteristics
operate by influencing the types of people with whom one is
likely to interact." (Wright 1977, p.505) The behavior of


70
threat should not have a uniform effect on attitudes, issue
positions or the evaluation of candidates across the
political spectrum. Rather, it would affect only on those
issues and evaluations which were related to one of its two
fundamental components: race and economic vulnerability. For
example, the perception of threat would not be efficacious in
predicting the relative evaluations of say, two white
candidates in a contest in which the race card was not played
by one. It may figure more prominently in situations in
which one of the candidates race baits his or her opponent by
saying the opponent will give white jobs to blacks. Even
then, the effectiveness of the tactic depends on the
particular skills of the candidates and other circumstances
surrounding that election. One would not expect a measure of
threat to predict evaluations without that disposition being
made salient. Similarly, the perception of threat might come
into play in the evaluation of a black candidate (touching on
the other fundamental element) even if economic threats were
not particularly salient(Hrt05) Symbolic racism would
covary with support for candidates who appeal to traditional
values (Hsr05)
Additionally, attitudes related to threat should guide
issue orientation when salient to the fundamental elements.
So, if one feels economically threatened one might be more
supportive of programs to help the unemployed, in general,
because this safety net would benefit the threatened
individual or his or her reference group (Hri>06)* On the


2
Court decisions have made it easier to discriminate and have
whittled away at affirmative action programs. Extremist
groups, such as those led by Tom Metzger and manned by skin
heads, seem on the rise. Indeed, the number of hate groups
nation-wide increased 27% between 1990 and 1991. (Southern
Poverty Law Center 1992) The qualified electoral success of
David Duke in Louisiana seems to scare everyone, except of
course his supporters.
While these events point to an increasingly engaged
white backlash, other events seem to offer contradictory
signals. Douglas Wilder was elected Governor of Virginiaa
feat which required substantial white support. Throughout
America the number of African-American elected officials has
been increasing steadily since 1970, largely through the
efforts of black candidates and voters. (Joint Center for
Political Studies 1988)
Perceptions of Threat
What explains these seemingly contradictory trends? One
starting point is the white mind. Both Wilder and Duke have
been supported by white voters. This support may spring from
the same pattern of racial orientation. The notion of racial
threat is exemplified by whites who feel that their jobs,
admission to schools, or lifestyles are in jeopardy because
of African-American advances. More generally, one may define
the perception of racial threat as the belief on the part of
whites that blacks:


46
the Protestant ethic (p. 41; also McConahay 1982), and has as
it "most important psychological factor . political and
economic conservatism . ." (p. 37)
A similar approach is taken by Sears, Hensler and
Speer's (1979) articulation of symbolic politics' "anti
busing sentiment as primarily derived from long-standing
political predispositions, specifically racial prejudice and
political conservatism, which in turn can be traced to
political socialization in preadult life." (p. 372). Kinder
and Sears (1981) point to symbolic racism's "blend of
antiblack affect and the kind of traditional American moral
values embodied in the Protestant ethic." (p. 416; also
endorsed in Kinder and Sears 1985; Kinder 1986; Sears 1988)
The formulation by Sears and Citrin is similar, "the
conjunction of two separate underlying dimensions: a
specifically anti-black attitude, and conservative value
priorities, which stress individual effort and responsibility
rather than government activism." (1982, p. 168)
Nowhere in the literature on this topic is there a clear
articulation of how the elements fit together, why they
should not be treated as distinct dependent or independent
variables or how they are different from garden-variety
prejudice and conservatism. One gathers that the
relationship is mutually reinforcing. A dedication to the
values embodied in the work ethic reduces sympathy with the
unemployed, underemployed, or otherwise less successful folk
one might hear aboutperhaps poor blacks on the television


115
gathers that as the levels of percentage black (1940)
increase the levels of symbolic racism decrease. Re
estimation of the total models for south and non-south in the
next chapter indicates that in general southerners have more
prejudicial attitudes than non southerners. The significance
of the negative parameter in this case lies in understanding
the marginal effect of the percentage black (1940) within the
southern states. Namely, the interaction term cancels out
the effect of the percentage black in 1940 in the model,
controlling for all the other variables in the model. In
other words, even though the South has higher levels of the
both prejudice and percentage black 1940, the marginal effect
of the local percentage black in 1940 in the South is nil.
The variable has explanatory power for the nation as a whole,
but is less useful in explaining variation within the South
alone. Other factors explain the variation in levels of
prejudicial attitudes within the South. The analysis
presented in the next chapter sheds some light on these
issues.
Taken together the block of hybrid variables should
represent a more effective control of Southernness than the
south dummy variable. After all, the essence of Key's
arguments about southern political life rests on the economic
and political factors, not on the precise latitude and
longitude on the globe. Put another way, living below the
Mason Dixon line has more to do with the (central tendencies
of the) people and the economy than with the fact that a spot


93
items. T-tests for some of these items are displayed in
Table 5-1. Note that three of the items yielded significant
differences by the interviewer groups. Accordingly,
respondents with non-white interviewers were excluded from
the analysis.
The questions used to construct these racial scales are
highly charged. One may wonder if respondents would be
comfortable revealing their positions on these touchy issues.
The frequencies for these items are reported in the Appendix.
There we see that between 12.5% and 32.8% felt comfortable
enough to choose the most extreme positions. Typically 41.4%
to 66.4% selected one of the two most extreme responses.
These figures imply that the respondents do
Table 5-1. T-test for race of interviewer bias on whites
Civil rights push
white interviewer
non-white interviewer
mean
2.60
3.06
pooled t-
value
-2.27
P
.02
separate
t-value
-2.24
P
.03
white admission
white interviewer
non-white interviewer
3.05
3.61
-2.05
.04
-2.04
.05
family admission
white interviewer
non-white
interviewer
3.32
3.83
-1.88
.06
-2.03
.05
white promotion
white interviewer
non-white interviewer
2.95
3.35
-1.58
.11
-1.43
.16
family promotion
white interviewer
non-white interviewer
3.36
3.67
-1.11
.27
-1.10
.28
Source: 1986 ANES


122
the effects of the two attitudes on other politically
relevant variables. The focus on the racial attitudes shifts
from explaining the origins of the attitudes themselves to
exploring how the racial attitudes explain and predict other
aspects of political behavior. Two types of questions are
addressed. First, how do racial attitudes influence the
evaluation of candidates and groups? Second, how do racial
attitudes influence the respondent's policy preferences, in
this case federal spending preferences? Discussion now turns
to gauging the effect of the racial attitudes on evaluations
of candidates and groups.
The design of dependent variables presents some problems
for this portion of the study. First the evaluations of
candidates is problematic. Ideally, one would use the
evaluation of a widely known and well publicized presidential
candidate. Since the data were collected during an off year
election, no evaluations of presidential candidates are
available. Other political figures with national reputations
are used, even though they were not actually campaigning at
the time. Second, the split form survey forces amendment of
standard procedures for evaluation of feeling thermometers.
These two factors work together to reduce the response rates
on the feeling thermometer evaluations. Normally, one takes
the difference between the rating of the group or person of
interest and the average of all the others. This technique
is problematic when respondents rate few candidates and
groups. Here a more conservative approach is taken (see the


24
Individual Analyses
Despite the well-developed literature linking the size
of minority populations and various measure of racial
inequality, there are only a few studies using survey data
which seek substantiate the connection between local racial
composition and the attitudes and behaviors of individuals.
Really, only a handful of recent studies have systematically
confronted the assumptions emanating from the aggregate
literature discussed above (Fossett and Kiecolt 1989; Giles
and Evans 1985, 1986; and Giles 1977). All of these studies
performed secondary analysis of nationally sampled data
collected by the two main academic research centers, the
General Social Survey (GSS) conducted by National Opinion
Research Center (NORC) and the American National Election
Studies (ANES) by the Institute for Social Research (ISR).
Despite the obvious attraction of testing hypotheses with a
nationally drawn sample, the use of such data is not without
costs. Finding valid measures for the theoretical concepts
of interest usually proves difficult and is often impossible.
The creative operationalizations used by Giles and Evans and
Fossett and Kiecolt represent serious, if not altogether
successful, attempts to overcome the limitations imposed by
the data. Though none satisfactorily address the
intervening threat mechanism, they do merit discussion. A
brief discussion of these recent studies follows. Three of


73
only to personal threats (challenges to the individual's job,
economic status, security, or political influence) but to
similar threats to their families and even their race in
general. Further, the perception of a more general threat
should predict other political attitudes and behaviors.
Measures of symbolic racism have not been tested against more
general orientations or representations of threat.
Synthesis
The empirical and theoretical controversies discussed
above impels an effort at reconciliation. For example, some
of the research discussed in the second chapter argues that
white hostility toward blacks results from real economic
threats posed by blacks. Others argue that instrumentally
marshaled racial hostility prevents a poor white-black
coalition from upsetting the status quo.
While neither of these propositions are tested in this
study, the mechanism underlying them is. The belief that
blacks pose economic, status or power threats, rather than
the essential material conditions, may lead individuals to
have hostile attitudes or to act on existing prejudicial
attitudes. Material conditions could make believers out of
many, but those conditions and consciousness of them need to
be disaggregated.
For example, imagine four individuals working for a firm
likely to change the composition of its work force in
response to a boycott by a minority group: the first worker's


32
variable, as above, is significant in the prejudicial
direction only in the Southern sample; Southerners in exile
are not less likely to support integration that their
neighbors. Further, the difference of these two variables
alone accounts for a 13.4% increase in the amount of variance
explained in the Southern sample over the Northern sample.
These difference reinforce the importance of context in
allowing both access to socialized values and in allowing the
use of those values in determining policy preferences.
Reconciliation
This discussion of the interpretation of regional
contextual variable points to another question deserving
further discussion. Giles and Evans (1985, 1986) treat
racial composition as a measure of external threat. This
interpretation is consistent with some of the sociological
literature on 'ethnic conflict' or competitive ethnicity (see
Giles and Evans 1985, 1986 for a review of this literature
and citations). In the 1985 article they make the
distinction between external threat and perceived threat.
Perceived threat was measured by asking the respondents
questions while the percent black represented the sole
measure of external threat. Giles and Evans (1985) may be
correct in making this assumption. In this way, they
distinguish themselves from Marxian and modernization
theorists who characterize race as an ascriptive identifier
and of diminishing salience in developed societies.


18
working class. Indeed, there is a strong inverse
relationship.
So, according to Huckfeldt and Kohfeld (1989), middle
class and Working class blacks make strange Democratic
bedfellows with new left upper class whites and some residue
of the working class whites who do not convert or demobilize.
The Republican coalition is similarly uncomfortable with its
combination of upper class fiscal conservatives and working
class whites. The authors also provide anecdotal data of the
splintering effects of race in particular contests. Again
the assumption concerning the "unnaturalness" of the
coalitions lies in the class differences between the various
groups.
Theoretically, working class whites and working class
blacks have compatible material interests. Similarly,
according to Huckfeldt and Kohfeld (1989), middle class
whites and blacks are presumed to have the same material
interests. Still, working class whites tend to flee the
Democratic column in inverse proportions to black
contributions to that column, even though it is in their
material interest to pursue like policies and candidates.
They flee because rightly or wrongly they feel as though
their interests are incompatible. They flee because they
feel that black politicians or politicians whose agenda gains
obvious black supports pose some threat if elected.
Despite the potent effect race can have on electoral
coalitions, the argument put forth by Huckfeldt and Kohfeld


182
Schuman, Howard and Jean Converse 1971 Effects of black and
white interviewers on black response in 1968. Public
Opinion Quarterly 35:44-68.
Schuman, Howard, Charlotte Steeh and Lawrence Bobo 1985
Racial attitudes in America: Trends and interpretions.
Cambridge, MA:Harvard University Press.
Scott, James C. 1985 Weapons of the weak: everyday forms of
peasant resistance. New Haven : Yale University Press.
Sears, David 0. 1988 Symbolic racism In P. Katz and D
Taylor (eds.) Eliminating racism, pp. 53-81. New York
Plenum Press.
Sears, David 0. and H.M. Allen, jr. 1984 The trajectory of
local desegregation controversies and whites' opposition
to busing. In N. Miller and M. Brewer (eds.) Groups in
contact: the psychology of desegregation. New York:
Academic Press.
Sears, David 0. and J. Citrin 1982 Tax revolt: Something
for nothing in California. Cambridge: Harvard
University Press.
Sears, David 0., C.P. Hensler, and L.K. Speer 1979 White's
opposition to "busing": self-interest or symbolic
politics? American Political Science Review. 73:369-
384.
Sears, David 0. and D.R. Kinder 1970 The good life, 'white
racism,' and the Los Angeles voter. Paper delivered to
the annual meeting of the Western Psychological
Association, Los Angeles.
Sears, David 0. and D.R. Kinder 1971 Racial tensions and
voting in Los Angeles. In w.z. Hirsh (ed.), Los Angeles:
Viability and Prospects for metropolitan leadership, pp.
51-89. New York Praeger.
Sears, David 0. and D.R. Kinder 1985 White's opposition to
busing: On conceptualizing and operationalizing group
conflict. Journal of Personality and Social Psychology.
48:1141-1147.
Sears, David 0., R.R. Lau 1983 Inducing apparently self-
interested political preferences American Journal of
Political Science 27:223-252.
Sears, David 0., R.R. Lau, T.R. Tyler, and H.M. Allen, jr.
1980 Self interest vs. symbolic politics in policy
attitudes and presidential voting. American Political
Science Review 74:670-684.


114
While the ability of the southern dummy variable to
capture or suppress the effects of the cultural and
historical proxy variables in the models may seem dramatic
enough, it pales in comparison to the effects on the model
when the hybridized variables are added. Of these variables
one in particular is of note in the models explaining
symbolic racism. The interaction term of the south dummy and
the percentage black 1940 not only has a significant inverse
effect but it reveals effects suppressed by the southern
dummy variable alone. Again this adaptation of the model has
little influence on the effects of the individual level
explanatory variables. Education continues to be associated
with the amelioration of the prejudicial attitude while
conservatism remains associated with its aggravation, and
both of these effects are of consistent magnitudes across all
models.
The interpretation of models with complex interaction
terms pushes scholars into hazardous territory, mainly
because it is impossible to interpret the marginal effects of
the constituent elements of the interaction terms.2 In this
case, however, the task may be less perilous because the
interaction term contains a dummy variable. The term
demonstrates the direct effect of racial composition in the
south, circa 1940, on symbolically prejudicial attitudes.
Remarkably the relationship is inverse. On its face, one
2 Re-estimation of less complex models upholds the
substantive interpretation presented here.


156
whites' current evaluations of blacks are based on the
symbolic predispositions galvanized in youth.
More to the point, the proponents of symbolic racism
argue that contemporary prejudicial attitudes result from a
combination of anti-black affect, a generalized distaste for
blacks, and traditional values exemplified by the Protestant
work ethic. Whites may be disposed to dislike blacks
generally, but the whites' belief that blacks fail to live up
to the standard set by the work ethic aggravates the disdain
dramatically. Further, this violation of these cherished
values provides a socially acceptable avenue for the
articulation of prejudice; whites are free to vent their
prejudicial attitudes as long as they couch their disdain in
terms of blacks' failure to 'work hard'.
A final ingredient of symbolic racism is found in the
assumption of abstraction. The proponents of the concept
argue that this form of prejudice is independent of an
individual's personal life, self interest, or economic,
political or social context. So, environmental factors found
significant in explaining prejudice in the racial threat
literature are thought not to explain symbolic racism.
The Synthetic Theoretical Position
Chapter Four presented a synthetic position which
accounts for some of the incompatibilities between the two
schools of thought discussed above and includes elements
missing from both. The novelty of the synthetic position


116
in the South lies one inch or one hundred miles below that
line.
While some of the hybrid variables are not statistically
significant in and of themselves, their inclusion in these
models provides a more sophisticated and accurate control of
the factors which determine 'Southernness'. Having taken
Southernness into account the interpretation of the first
block of contextual variables becomes more clear. For
example, once the South is fully controlled the percentage
black circa 1940 is directly and strongly related to the
level of symbolic racism. The prima face interpretation of
this rests on the notion that those areas in the north which
had larger black populations in 1940 demonized the new black
migrants.
Other important changes occur in the model with the
addition of more effective southern control variables. For
example, the effect of income equality is revealed as
significant. Where the incomes of African Americans and
whites are closer to parity, the levels of symbolic prejudice
are lower.
The interpretation of these models reveals some striking
findings. First, even though the symbolic attitudes are more
closely tied to individual level characteristics, they are
sensitive to contextual effects. This is evident despite the
repeated insistence of the theory's proponents that their
symbolic nature is completely abstract (see chapter 3 for
examples). Symbolic racism is clearly related to the


75
In this chapter, discussion has drawn attention to ways
in which political and racial attitudes may be sensitive to
and distinct from actual material conditions (as with the
worker who incorrectly perceived her job was threatened).
People's perceptions of their condition may diverge from the
actual circumstances, but they are not divorced from those
circumstances. In the above example, the climate of change
led the actually unthreatened worker to perceive that his or
her job was less secure than it was. The context, in which
co-workers with whom he or she worked were likely to lose
their jobs, heightened that worker's sensitivity to the
perception of threat. This sense of threat occurs even
though there may be no actual threat to the worker's job.
Behavior is predicated on perceptions that may be inaccurate.
And those perceptions, with their attendant 'measurement
errors,' develop within a context. In the next section,
discussion focuses on conditions which inform the context of
racial attitudes, paying particular attention to the changing
interpretations of local racial composition.
Context and the Changing Meaning of Racial Composition
Some of the studies discussed in the second chapter
point to different theoretical significances of the local
racial composition. To some (e.g. Giles and Evans 1985) the
local composition represents a contemporary and actual threat
to the lives and livelihood of whites. To others (e.g. Key
1949), racial composition represents a proxy for a labor


LIST OF FIGURES
Figure 4-1. Matrix of racial attitudes 65
Figure 4-2. Quadratic function with increasing slope 78
Figure 4-3. Parabolic quadratic function 80
Figure 4-4. Quadratic function with gradual decay 82
viii


147
Table 7-5. Standard deviation of Symbolic Racism Regressed
on Contextual Variables in Each County
entries are unstandardized coefficients from OLS regression
white Southern respondents only
Variables
I
II
Mean Symbolic Racism
0.48**
0.42*
%black 1940

-0.001b
State %black 1940

0.0004
%black 1980

0.001
%employed in
agriculture

-0.003
income equality

-0.002
constant
-0.01
-0.004
adj R2
0.30
0.18
N
27
26
white Northern respondents only
Variables
I
II
Mean Symbolic Racism
0.45***
0.44**
%black 1940

-0.0005
State %black 1940

-0.001
%black 1980

0.001*
%employed in
agriculture

0.001
income equality

0.004
constant
-0.01
-0.02
ad j R2
0.31
0.35
N
78
70
Source: 1986 ANES; 1900, 1940 and 1980 Census


LIST OF REFERENCES
Anderson, Barbara A., Brian D. Silver and Paul R. Abramson
1988a The effects of the race of the interviewer on
measures of electoral participation by blacks in the SRC
national election studies. Public Opinion Quarterly
52:53-83.
Anderson, Barbara A., Brian D. Silver and Paul R. Abramson
1988b The effects of the race of the interviewer on
race-related attitudes of black respondents in SRC/CPS
National Election Studies. Public Opinion Quarterly
52:289-324.
Anderson, Neis 1962 The urban way of life. International
Journal of Comparative Sociology. 3:175-188.
Allport, J.D. 1954 The nature of prejudice. Reading, MA:
Addison-Wesley.
Baker, Donald G. 1975 Race, power and white siege cultures.
Social Dynamics. 1(2):143-157 .
Blalock, Hubert M., jr. 1957 Percent non-white and
discrimination in the South. American Sociological
Review 21:583-5883.
Blalock, Hubert M., jr. 1965 Toward a theory of minority
group relations. New York: Wiley.
Blalock, Hubert M., jr. 1967 Causal inferences, closed
populations and measures of association. American
Political Science Review 61:130-136.
Blauner, Bob 1972 Racial oppression in America. New York
Harper and Row.
Bloom, H.S. and H.D. Price 1975 Voter response to short run
economic conditions: the asymmetric effect of prosperity
and recession. American Political Science Review
69:1240-1254.
Bobo, Larry 1983 Whites' opposition to busing: symbolic
racism or realistic group conflict? Journal of
Personality and Social Psychology 45:1196-1210.
Bobo, Larry 1988 Group conflict, prejudice, and the paradox
of contemporary racial attitudes. In Phylis A. Katz and
177


155
the stiffest competition for jobs and the benefits of modern
society. Additionally the research on the effects of racial
threat has found regional differences in the level and effect
of threats and prejudice; respondents living in the South
tend to be more prejudicial. While the intuitive reaction to
this finding lies in appeals to the influence of the South's
confederate and segregationist past, racial threat scholars
has argued that the effect is an indicator of the South's
traditionally lower levels of education or other individually
measurable variables.
In the next section a brief review of the second major
school of white prejudice, or symbolic racism, is presented.
Symbolic Racism
The notion of symbolic racism represents an extension of
the symbolic politics paradigm. The theory and measurement
of symbolic racism is discussed in detail in Chapter Three.
In general, proponents of the paradigm argue that cognitive
and affective processes drive political behavior, rather than
material conditions or an individual's political and social
environment. Affective orientations or symbolic
predispositions are set during pre-adult socialization.
These predispositions are then used as benchmark for the
evaluation of later stimuli and form the basis of political
attitudes and behaviors. The proponents of symbolic racism
argue that racial attitudes follow this same pattern. So,


74
job is threatened because of the changes and he or she knows
it; the second one's job is threatened, but he or she thinks
(wrongly) it is secure; the third one's job is secure, but he
or she thinks (wrongly) that it is threatened; the fourth
worker's job is secure and he or she knows it. Despite the
difference in the reality of threat posed to the workers, it
is evident that there is no reason to expect workers one and
three to respond differently to the perceived threat to their
jobs. Other things being equal, one would expect similar
response among those who have the same perceptions concerning
the threats (real and imagined) to their job status.
What expectations are reasonable concerning the two
workers who perceive their status as secure? Narrow self
interest predicts those workers would not harbor hostility
toward the minority group. After all, their interests remain
unaffected by the shake-up in the company (at least by the
standard of self-interest employed by the proponents of
symbolic racism); they still have their jobs. If, on the
other hand, interests are more closely related to group
identification, then despite the perceived (rightly or
wrongly) security of their positions, they may harbor hostile
attitudes. They may harbor those attitudes because the
shake-up threatens, not them personally, but people like
them, people with whom they identify. Or they may feel
threatened because the environment has simply become less
secure; they could lose their job next.


181
anti busing attitudes in Louisville. Journal of
Politics 44:692-720.
McConahay, John B. and J. C. Hough, jr. 1976 Symbolic
racism. Journal of Social Issues. 31:2345.[32(2):23
45]
Merton, R.K. 1957 Social theory and Social Structure. New
York: Free Press.
Miller, Arthur H., Patricia Gurin, Gerald Gurin and Okasana
Malanchuk 1981 Group consciousness and political
participation. American Journal of Political Science
25(August):494-511.
Omi, M. and Howard Winant 1986 Racial formation in the
United States. New York: Routledge and Kegan Paul.
Page, Benjamin 1977 Elections and social choice: the state
of the evidence. American Journal of Political Science.
21:639-68.
Page, Benjamin and Calvin Jones 1979 Reciprocal effects of
policy preferences, party loyalties and the vote.
American Political Science Review 73:1071-90.
Pettigrew, Thomas F. 1959 Regional variations in Anti Negro
prejudice. Journal of Abnormal and Social Psychology.
59:28-36.
Rhodebeck, L.A. 1992 The politicization of gender
consciousness: a pooled cross-sectional analysis. Paper
presented to the annual meeting of the American
Political Science Association, Chicago, IL.
Riker, William D. and Peter C. Ordeshook 1973 An
introduction to positive political theory. Engelwood
Cliffs, N.J.: Prentice Hall.
Rokeach, Milton 1960 The open and closed mind. New York:
Basic Books.
Rusk, Jerrold G. and John J. Stucker 1978 The effect of the
Southern system of election laws on voting
participation: A reply to V.O. Key, Jr. in Joel H.
Silbey, Allan G. Bogue and William H. Flanagan (eds.)
The history of American electoral behavior. Princeton:
Princeton University Press.
Schermerhorn, R.A. 1956 Power as a primary concept in the
study of minorities. Social Forces. 35:53-56.


37
school of thought grows out of the more general orientation
of symbolic politics. Since this theoretical tradition is
psychological, its method is exclusively individually
oriented. Although a few experimental designs have been
conducted, most work in this area has exploited survey data.
The chapter opens with a brief discussion of symbolic
politics and the role of racial attitudes in that theory of
politics. This is followed by a description of the
development of the essential elements of the theory and its
landmark studies. The chapter closes with an account of the
major critiques of the approach and its methods.
Symbolic Politics in a Nutshell
The symbolic racism orientation grows out of the
symbolic politics tradition2. A thumbnail sketch of this
orientation will provide the foundation for the later
treatment of symbolic racism. The symbolic tradition argues
that political behavior is driven by cognitive and affective
processes. Pre-adult socialization and later resocialization
creates a number of affective orientations or symbolic
predispositions which provide an enduring standard for the
evaluation of new stimuli. Among these affective
orientations are numbered racial attitudes, ideology and
2 The brand of symbolic politics advocated by Sears (see
Sears 1988 for a summary), in particular, shares some common
ground with Edelman's (1971) articulation, but features one
dramatic point of departure. While Edelman left room for the
influence of self interest, Sears and his colleagues dismiss
such influences.