Uncertainty and Heterogeneity in the Issue-Basis of Party Identification.

One of the oldest and most enduring debates in American Politics involves the endogeneity of party identification to the issue preferences of American citizens. More recent interest in the nature of party identification has shifted the terms of the debate from whether one view or the other dominates individual decision-making to a consideration of factors that condition the direction of causality (Carsey and Layman 2006, p. 645). Carsey and Layman's research points to the need for more theoretical consideration of which factors condition the relationship between policy preferences and partisanship. In this paper, I seek to examine the heterogeneity in the issue-basis of party identification by drawing on another classic literature concerning the uncertainty that citizens have about politics. The empirical analysis will employ an estimation technique used by Alvarez and Brehm (1995, 1997) in which heteroskedastic regression is used to directly model the error variance of the respondent's own issue placement, which is used as a measure of individual uncertainty. After controlling for an extensive list of demographic characteristics, I argue that the error variance encompasses not only lack of knowledge about the respondent's issue position, but also the use of cues and shortcuts from the social environment. Uncertainty in this broader sense of within group variation in issue placement is expected to condition the issue-basis of party identification. The estimated error variances from the first stage regression are then interacted with the respondent's issue placement in a second-stage ordered logit regression of party identification. ..PAT.-Unpublished Manuscript [ABSTRACT FROM AUTHOR]