In this study precinct data for county, state, and federal elections between 1980 and 1996 are matched with data on the racial composition of precincts in order to analyze racial polarization and realignment in the south. Racial polarization is defined as the difference in support for the Democrats among blacks and whites (or Latinos where appropriate). Using an ecological inference program (EI) this study estimates, with confidence intervals, support for Democrats and Republicans among blacks, whites, and Latinos in every county in the south for virtually all county, state, and federal elections. Litigants and judges in cases challenging or supporting the creation of majority-minority districts are likely to rely heavily on these estimates. Unlike past studies which have focused primarily on whether polarization prevents the election of minority officials, this investigation gauges the size of the racial gap in support for the Democrats. Contrary to the conventional wisdom often expressed by the courts and the media, the principal investigator hypothesizes that racial polarization has increased as blacks continue to vote nearly unanimously for Democrats and greater numbers of whites cast ballots for Republicans. The estimates of racial polarization make it possible to test competing theories for why racial polarization varies across location, time, and office. The study compares the strength of the white backlash theory with other nonracial theories. The white backlash theory states that whites who live in areas with a high percentage of blacks usually vote more conservatively than whites in areas with few blacks, but some speculate that the economic gap between blacks and whites explains racial polarization. The investigation tests several theories about the nexus between racial redistricting and realignment. While several studies examine the partisan impact of racial redistricting at the federal level, this study is the first to explore whether racial redistricting resulted in Republican state legislative or county council gains. The data set produced by this investigation will be of immense value to other scholars interested in exploring this important topic.
