Familiarity doesn't breed contempt: The political geography of racial polarization

Presents an intensive exploration of American racial politics, particularly the fatalistic view that polarization is greatest when two ethnic groups appear in close proximity (the “white backlash” or “group threat” hypothesis). Combining survey analysis with an ecological study of recent Southern voting returns (using Gary King's solution to the ecological inference problem), I show that the density of African-American population in a community does not influence racial views or political coalitions in a consistent manner. Rather, heterogeneity operates differently depending upon social and historical context. In particular, the traditional proximity pattern is reversed in urban areas. Predominantly white suburbs and small towns host a “white middle class” subculture generally at odds with minorities, especially when they reside close enough together to threaten each other but not close enough for mutual assimilation. Whites who reside in a more diverse urban landscape, by contrast, generally are closer to the attitudes, interests, and behavior of other races or ethnic groups. The clear but complex geographical pattern calls into question purely psychological or “symbolic” approaches to racial conflict. Racial attitudes may build on ignorance and stereotype, yet the arrangement of social forces is politically and geographically meaningful, and therefore appears tied to rational group conflict.