rT V HE DOMINANT concept in the study of voting behavior is partisanship, especially its operationalization by the Survey Research Center/Center for Political Studies as party identification. Given its strong impact on behavior, perceptions of the political world, and other political attitudes, partisanship has been the subject of intense scrutiny. It has been empirically examined and theoretically analyzed both as a dependent variable1 and as an independent variable.2 Other research has concentrated on a thorough treatment of the problems surrounding the conceptualization and measurement of party identification.3 As originally formulated by the authors of The American Voter (Campbell, Converse, Miller and Stokes, 1960), party identification is a psychological concept, serving psychological functions. These functions would include the reinforcement of vote choices, and perceptual screening of political actors and events. But one school of research has attempted to describe and analyze the political content and functions of party identification, by focusing on the policy bases of partisanship.4 This approach is useful