Chapter 1 - Introduction
V.O. Key (1984) claimed, in his classic work on Southern Politics in State and Nation, that the glue binding the Solid South was "the Negro." The South's political institutions enforced a rigid social code, one that promoted the political, cultural and economic oppression of a cheap labor force. The champions of segregation were not yeomen farmers scraping alone at the pine hills of northern Alabama. They did not make a living digging coal from Tennessee mines. They were not Cajun fishermen poling floating shacks down the bayous of South Louisiana. That is, they were not ignorant of the African American. Rather, Southern segregationists lived among the very people the system was intended to oppress, knew them intimately, and often carried out the subjugation with their bare hands. The South's distinctive political culture grew out of the localized effort to maintain white supremacy.
One implication of Key's thinking was that regional pathologies should appear most strongly in states and counties with the greatest concentrations of African Americans. Racial conservatism, in particular, should shift with a community's black density. Observing his own time period, this is exactly what Key found (1984, chap. 15). "The hard core of the political South-and the backbone of southern political unity-is made up of those counties and sections of the southern states in which Negroes constitute a substantial proportion of the population," he concluded (1984, 5). Conversely, "departures from the supposed uniformity of southern politics occur most notably in those states with fewest Negroes and in those sections that are predominantly white" (1984, 668).
Key's book offered readers a sophisticated exploration of the mid-century South's political institutions. He couched his generalization about race relations firmly within the context of regional history, and his attempts to test the hypothesis constituted but one sliver of a multifaceted research design. Nevertheless, the simple idea that racial diversity might unify members of the majority, creating a hostile "white backlash," was destined to capture the fancy of a multitude of researchers who would follow. They attributed to Key a fatalistic theory of intergroup conflict, one that neglected the spirit but retained the letter of his concise, exceptionally memorable, summary.
A Fatalistic Theory of Intergroup Relations
Cut off from the historical setting that spawned it, Key's description has taken on its own life as a general approach to race relations. Drawing in large part on the observations in Key's book, as well as on related explorations of other Southern data, social scientists have derived what is sometimes called the "group threat" hypothesis (e.g., Giles and Evans 1986): Other things being equal, racism or prejudice or discrimination against an out-group increases as that group becomes a larger (and presumably more threatening) proportion of the community. Gordon Allport, in perhaps the most influential work on intergroup hostility, lists this generalization as one of the "sociocultural laws of prejudice" (Allport 1979, 221).(1)
The white backlash hypothesis is not limited to studies of intergroup conflict, or even to academic research (although when found in popular writings it usually does not possess such a formal name). People frequently, almost reflexively, attribute the presence of conflict to a community's heterogeneity. Mixing peoples is a recipe for trouble, by this logic. The view is not limited to racist or ethnocentric whites. Among some policy analysts and political activists-especially civil rights leaders and others on the left-the goal of integration has faded precisely because of the view that it engenders too much debilitative hostility. The paltry benefits of proximity simply do not match the nasty backlash that it apparently produces.
Why is this facile theory of intergroup relations so durable and so widespread? It possesses a base plausibility that is impossible to deny. Racism is not possible without at least two culturally defined "races." Prejudice requires someone to prejudge. The majority cannot discriminate without a minority to suffer discrimination. By definition, that is, we will never observe prejudice within a community that is entirely homogeneous. And until the minority group reaches a certain density, bigotry is unlikely to become a serious social problem because opportunities for expression are limited. At its extreme, the logic must be correct.
Beyond White Backlash
Nevertheless, the goal of this research is to challenge the white backlash concept's prominent place in social science, at least as commonly represented. Beyond the face validity that it possesses, the theory rarely draws empirical support. I test the observable implications of the hypothesis in both aggregate-level voting returns and individual-level survey responses-in both cases, the political expression of racial antipathies does not respond much to physical proximity. White backlash fails to characterize patterns of racial polarization even within Southern states, from which the idea first evolved.
More generally, the expectation that heterogeneity will breed group hatreds simply ignores the bulk of anecdotal evidence. History is replete with instances of diverse groups living together in peace,(2) as well as of persecution leveled at extremely small minorities. The social and historical context mediates intergroup relations to such an extent that we cannot, and should not, draw broad lessons from a single slice of time.
In addition to its empirical weaknesses, the white backlash hypothesis also flounders as theory. Not only does it beg the question of what makes one race or ethnicity an out-group when others are not, it fails to settle any issue that would matter to a practical policy maker. What causes racial conflict? What preconditions mitigate it? Under what conditions does it go away altogether? What fuels Mister Backlash, and where should we look for him? All we know is that blacks and whites will fall to fighting when brought into proximity, and the conflict will become increasingly salient as local diversity increases-a terribly fatalistic view, absent qualification.
The Contributions of This Project
Social science should not cast off Key's central insight that particular forms of intergroup relations may produce unique geographic patterns of political behavior. The problem is only that, until now, the theory was so simple that it always pointed researchers toward a single expectation: with diversity comes trouble. It always presumed the same sort of (ill-defined) mutual threat. In this manuscript, I attempt to refine what has been an hypothesis grounded in the Solid South into a more complex, but still useful, framework within which to understand intergroup relations. I set out several theoretically distinct forms of intergroup conflict that might characterize black-white relations in the United States (and especially the South), and show how they would produce empirically distinct patterns of political behavior.
The observable implications of these competing approaches are falsifiable, and I turn to that task in the empirical portion of the research. Based upon such an exploration, I argue that the American racial context has changed over the last several decades, and with it the political geography of racial polarization. Conflict is no longer centered within geographically defined units, but reaches across them. The widest political gap is not between whites and blacks where the races mingle, as it was during the Civil Rights era, but between blacks and the whites who live apart from them, especially those who are "close but not touching." Predominantly white suburbs and small towns have become a bunker in the culture wars, at least as they are played out in the political arena, one that sets off a white subculture no longer identified by either tolerance or middle-brow moderation. So I've introduced a third option into the traditional debate over whether race still matters: yes it does, but in a qualitatively different way from that observed only a few decades before. The policy lessons of the past no longer apply, and different historical experiences have become most relevant.
Aside from the controversial question of whether and how "race matters" in American politics, this research informs two other debates in the social-science literature. The first emanates from Southern politics. Many commentators at the start of the Second Reconstruction were optimistic about where the region's politics were headed. Once the racial caste system collapsed and party competition revived, observers promised, the South would begin to exhibit a latent progressivism that could propel the nation forward. This surge in liberalism has failed to materialize, despite atrophy in the Black Belt counties that once formed Southern conservatism's bedrock. My work provides a clear answer for why these sanguine predictions did not come to pass: all-white Southern enclaves have filled the gap left by rural segregationist strongholds.
The second debate centers in the literature on political behavior, and especially the study of racial attitudes. Scholars have debated for decades whether American citizens use a rational calculus to derive their policy views-in particular whether they operate under some informal means of assessing the payoff associated with policies or candidates. On first blush, the argument I am foreshadowing here might seem to stress the irrationality of racial politics. The whites who joust with America's black voting bloc are those with the least real-life exposure to diversity; their attitudes are laced heavily with stereotype and even sheer ignorance about African Americans. Yet my findings imply just the opposite. Because racial conservatism follows predictable patterns, socially and geographically, I am forced to conclude that polarization works in service of real group interests. The stakes are often obscured to the individuals involved, but nonetheless are institutionalized in the class structure and partisan politics of the United States. Racial polarization does not stem from sheer irrationality.
An Outline
Justification for this project hinges, in part, on the widespread application of the backlash hypothesis. Chapter 2 provides a more complete discussion of its appearance in Southern politics research. Chapter 3 expands the intellectual history to include other political science, touching on the study of national voting behavior as well as the policy discrimination and elite actions that stem from how Americans vote. Chapter 4 ends the survey by concentrating on inconsistencies found across the backlash literature, and then provides an abstract framework for discussing proximity effects that would promote better accumulation of social-science knowledge. A quick treatment of Southern race relations illustrates the necessity of this less parsimonious framework, and introduces an explanation why changing conditions should have undermined the old white backlash paradigm. Chapter 5 turns to the interdisciplinary literature on racial attitudes, to show that dominant narratives in the subfield do not permit clear attempts at falsification in geographical data. The chapter ends by distilling from that work a series of "approaches" to American racial conflict, expressed within the Chapter 4 framework, that do possess observable implications.
Part II tests for these implications in aggregate data. Chapter 6 discusses the difficulties posed by using modern elections to study Southern politics. The expansion of civil rights clouded the meaning of voting returns, and popular estimation methods were not capable of clarifying racial behavior. However, as Chapter 7 makes clear, combining Gary King's solution to the ecological inference problem with precinct-level election data eliminates much of the barrier that once prevented aggregate-data studies. I illustrate by showing his method's success estimating racial voting behavior in two Louisiana data sets for which the truth is known. Chapters 8-10 apply his method to precinct-level data from Louisiana, Georgia, parts of Florida, and Kentucky. For each I chose campaigns that would bring out racial conservatism in its purest electoral form. Polarizing elections make a community's racial attitudes more salient, yet still operate within the constraints of the political system rather than within the ethereal realm of racial prejudice. Test after test indicates that familiarity does not breed contempt; no simple county-level or neighborhood-level white backlash pattern appears. The Louisiana case also allows me to test Chapter 5's more complex approaches to polarization, providing strong evidence that white enclaves buck the backlash logic.
Part III turns to survey data, focusing on a Kaiser Foundation race poll that made gauging a respondent's orientation on race-based policy issues particularly easy. By observing how respondents differ based upon the communities where they reside, I again am able to test the complicated approaches from Chapter 5, showing that proximity does not help explain racial polarization or conflict in urban areas. I conclude with a discussion of the implications of my findings.
1. Notably, Allport's examples to illustrate this "sociocultural law" stem from the Jim Crow South (Allport 1979, 227-29).
2. Usually we do not think of them as different "races" precisely because no such conflict has sprung up.