Little is known about the voting behavior of blacks aside from the observation that they are likely to vote in blocs for black candidates. In this study, we analyze fifty-two Atlanta area elections in which black and white candidates opposed each other. While Atlanta voters usually vote for candidates of their own race, this practice is more varied and less extensive than in the rural southern communities which others have analyzed. The extent to which voters support candidates not of their race varies greatly. Multivariate models can explain about 75 percent of the variance in black crossover voting and almost 60 percent of the variance in white crossovers. Incumbency and newspaper endorsements are especially important in attracting crossover votes. Three models, derived from differing expectations about the degree of prejudice, for estimating the share of votes cast for black candidates were tested. Ninety percent of the variance in election outcomes in Atlanta can be explained by the crossover model which includes terms for crossover voting of both races, black turnout, and black registration. Models which assume no crossovers or that only blacks crossover are less successful.
Bullock, C. S. (1984). Racial Crossover Voting and the Election of Black Officials. The Journal of Politics, 46(1), 238–251. https://doi.org/10.2307/2130442
