In my dissertation I try to develop a new theory of partisanship that better captures the changing character of partisan identity. My presupposition is that partisanship in the 1950s reflected a society in which partisanship had more the character of a central social identity than it does today. Partisan identity was more important because local party organizations were more prevalent and important in daily life, and people were more likely to hold especially salient religious, ethnic, class, union, or regional identities that were connected to the party system in a way that effectively mobilized partisan identity. The contemporary political system, on the other hand, lacks social identities and social institutions that can connect people to parties in an analogous way. Contemporary partisanship, instead, has the character of a loose identification with a set of issue-symbols rather than a strong connection with a social group.

        My approach is based on the idea that the ANES operationalization and theory of partisan identity is not able to adequately explain this changing character of party identification. Studies using the ANES party identification conception of partisanship have found changes in the partisan balance of the PID scale and changes in the predictive power of PID in vote and candidate choice models, but they have never looked to see whether the underlying character of partisanship has changed. That is, a person may still say that she is a "Strong Republican" but what she means when she says that may be very different from her analog of 50 years ago. First, she may say she is a strong Republican and yet not exhibit evidence that this avowal expresses a strong social identity. This is the theme of Chapter 5 where I argue that since contemporary partisans do not exhibit predicted stereotype effects their partisanship must not be a salient social identity. Second, the contents of partisanship (defined in the dissertation as party stereotypes) may have changed since the 1950s in such a way that partisanship has distinctly different dynamics. This is the theme of Chapter 6 where I argue that the changing character of the partisan information makes contemporary partisanship less stable.

        These insights are simply missed if we just rely on the ANES PID scale. Our new party system has produced more than just a new distribution of partisan attachments. Partisanship itself has changed. This does not mean that partisanship is irrelevant or even less central; instead, it is central in a new way. It entails a partisanship that has new sources of socialization, different underlying meanings for partisans, and a different effect on political behavior.

        The dissertation uses the concept of party stereotypes and self categorization theory to explore the changing nature of  partisan identity.  My theoretical contribution is a critique of  the common characterization of partisanship as a cognitive heuristic.   Self categorization theory, I argue, suggests that this understanding does not adequately capture the importance of social identity to partisanship.  Using this approach I explore the changing contents of party stereotypes, the growing independence of party stereotypes from party identification, the possibility that partisanship today acts less like a social identity than in the 1950s, and the issue of growing instability in party stereotypes.   .