Despite the centrality of party identification to our understanding of political behavior, there remains remarkable disagreement regarding its nature and measurement. Most scholars agree that party identities are quite stable relative to attitudes. But do partisans defend their identities, or does this stability result from Bayesian learning? I hypothesize that partisans defend their identities by generating “lesser of two evils” justifications. In other words, partisan identity justification occurs in multidimensional attitude space. This also helps to explain the weak relationship between attitudes toward the two parties observed by proponents of multidimensional partisanship. I test this hypothesis in an experiment designed to evoke inconsistency between one’s party identity and political attitudes. To establish generalizability, I then replicate these results through aggregate level analysis of data from the ANES.
展开▼
