Jim Josefson, "An Exploration of the Stability of Partisan Stereotypes in the United States," Party Politics, 6 (July 2000), 285-304.

First Paragraph:
The stability of partisanship in the USA has been a central concern of political behavior research since the development of party identification measures. The issue has aroused much-spirited debate but, despite continued controversy, the degree and sources of partisanship instability seem very far from settled due to a complicated tangle of thorny methodological issues (Erickson et al., 1998; Green et al., 1998). Given this confusion, the approach to partisanship followed by Trilling (1976) and Wattenberg (1994), looking instead at 'party images' or what I will call partisan stereotypes for evidence of partisanship stability, might be profitable. The American National Election Study (ANES) party identification measure of partisanship has always been defined in terms of long-term affective identification with a party, a facet of partisanship that should naturally be so resistant to change that it will be difficult to separate out significant variance from measurement error. The advantage of looking at partisan stereotypes, or the meanings people attach to the parties, is that the categories people use in exercising partisan reasoning should be naturally more variable, and therefore a good place to look for partisanship instability. For instance, a partisan might reliably say over time when asked that she is a strong Republican, but her notion of what it means to be a Republican and the categories that she ascribes to Republicans and herself as a Republican might easily be more variable. In 1992, for example, she might conceive of Republicans more in terms of their defense of 'family values', while in 1996 she might categorize Republicans more in terms of their policies about taxing and spending.

Figures and Tables:
Table 1: Stability of net party likes/dislikes responses in ANES panel data: correlation coefficients between waves
Table 2: Continuity in Democratic party likes/dislikes in ANES panel data
Table 3: Continuity in identical responses, democratic party likes/dislikes in ANES panel data
Table 4: Top five responses in Democratic party likes/dislikes
Table 5: Summary of cluster contents, 1952-92, by era

Last Paragraph:
The understanding of party stereotypes presented here is different from the usual notion of stereotypes as either a structure of attitudes or as short cuts for political cognition, because it allows us to conceptualize stereotypes not as the intentions behind behaviors but rather as the structure within which intentional behaviors are formed. This structure has two main characteristics, which I have used as analytical tools in this paper: normativity, which determines which understandings are most available or powerful in shaping cognition, and heterogeneity, which describes the degree of variety in the understandings available for cognition, their depth, complexity and meaning. These characteristics allow us to see that not only does it matter just what meanings citizens attach to the parties but also the character of those meanings matter as well. The character of contemporary partisan categories presents a double-edged sword for the future of partisanship stability. While the 1990s have produced new strongly normative understandings of the parties which encourage partisanship stability, the concurrent rise of negative categories and social issues as well as the continuing presence of old New Deal stereotypes have increased stereotype heterogeneity and the potential for partisanship instability. Nevertheless, the rise of new normative categories suggests that American politics has finally seen a subtle political realignment, not in the sense of a new distribution of partisan loyalties, but in the sense of a new definition of political conflict (Schattschneider, 1960; Sundquist, 1983).