Why the 1980s Measures of Racially Polarized Voting Are Inadequate for the 1990s

In this paper, we attempt to clarify some of the confusion that surrounds the measurement of racially polarized voting. This clarification is necessary because the determination of whether or not racially polarized voting exists is often a critical component of the evidence presented in Voting Rights Act (Section 2) litigation. We first show that the correlation coefficient should never be used to measure voting polarization by relating the statistic to the individual behavior that it is supposed to be describing. We then compare the estimates of polarized voting that are provided by other commonly used measures with individual behavior in order to show that the Voting Rights disputes of the 1990s will require different and more carefully specified measures than are currently in use.