Racial context of voting

How does the racial composition of a voter’s neighborhood change his or her voting behavior? In a time of swiftly changing racial context, this question looms as large as ever. Two major theories of racial context in voting have been dividing the field for decades: Contact theory states that actual experiences with people of other races voters to ignore racial prejudice and to judge candidates on their merits. Group threat theory, on the other hand, expects voters to regard a significant number of people of different race as a threat and to vote for candidates of their own race as a way to counter that threat. Since Carsey (1995) the clash of the two theories has been largely led by advances in aggregate-level data and methodology. The existing studies are however invariably either done on a high level of aggregation, where the assumption of direct contact is at best shaky or, their methods do not enable us to distinguish between the effects of racial context and racial composition. We advance this line of research by using newly available data from the US congressional elections in a geographically weighted model. We are able to link Census and election data at a low level of aggregation, the voting tabulation district: This allows us to assume contact between a voter and a significant proportion of the population within his or her unit, as well as the population of neighboring units. At the same time, we use both local and geographically weighted variables, which makes it possible to distinguish between compositional and contextual effects. By studying the effects of non-local variables, we are then able to provide evidence for either of the two theories and their limitations.