Racially Polarized Voting

Abstract

Whether voting is racially polarized has for the last generation been the linchpin question in vote dilution cases under the core, nationally applicable provision of the Voting Rights Act. The polarization test is supposed to be clear-cut (“manageable”), diagnostic of liability, and free of strong racial assumptions. Using evidence from a random sample of vote dilution cases, we argue that these objectives have not been realized in practice, and, further, that they cannot be realized under current conditions. The roots of the problem are twofold: (1) the widely shared belief that polarization determinations should be grounded on votes cast in actual elections, and (2) normative disagreement, often covert, about the meaning of racial vote dilution. We argue that the principal normative theories of vote dilution have conflicting implications for the racial polarization test. We also show that votes are only contingently related to the political preferences that the polarization inquiry is supposed to reveal, and, further, that the estimation of candidates’ vote shares by racial group from ballots cast in actual elections depends on racial homogeneity assumptions similar to those the Supreme Court has disavowed. Our analysis casts serious doubt on the notion — promoted in dicta by the Supreme Court and supported by prominent commentators — that courts should establish bright-line, vote-share cutoffs for “legally significant” racial polarization. The courts would do better to screen vote dilution claims using evidence of preference polarization derived from surveys, or non-preference evidence of minority political incorporation.

Keywords: vote dilution, racial polarization, racially polarized voting, race, voting rights, vote, voting, Section 2, Voting Rights Act, ecological inference, empirical