Revisiting the Political Theory of Party Identification

Abstract

Recently, Lewis-Beck et al. (The American Voter Revisited, 2008b) re-created The American Voter using contemporary data. Although these scholars ultimately conclude that voters today behave in ways that are consistent with the account of voting behavior presented in The American Voter, their work nonetheless highlights the importance and value of re-examining past ideas. Given that Lewis-Beck et al. have re-tested the findings of The American Voter, it is both timely and worthwhile to re-examine Fiorina’s (Retrospective voting in American national elections, 1981) political theory of party identification, which is often seen as a critique of the theory of party identification presented in The American Voter, using newly available panel data. In this paper, I re-examine Fiorina’s (Retrospective voting in American national elections, 1981) political theory of party identification using data from the 2000–2002–2004 NES panel study. In addition to applying Fiorina’s approach to party identification to new data, as a more robust test of Fiorina’s theory, I develop a model of party identification where changes in party identification are modeled as a function of the actual changes in retrospective political evaluations. Overall, my findings are broadly consistent with the findings from Fiorina’s original model of party identification; however, my analysis suggests that the distribution of opinions in the electorate and elite signals may be important to changes in party identification.