Reconsidering the Measure of Partisan Ambivalence

Abstract

Ambivalence is an important concept in the study of American public opinion and voting behavior, particularly given an increasingly polarized political class. However, some studies of ambivalence are based on a problematic measure. Because many survey respondents do not answer the questions used to measure partisan ambivalence, the resulting measure does a poor job of distinguishing between ambivalence and indifference (or apathy) toward the political parties and their presidential candidates. The challenge is determining who closely related the two concepts are. We re-examine a common measure of political ambivalence based on data from the American National Election Studies (ANES). We model the survey response as a two-step process in which the respondent first decides whether to share an opinion and then reveals the direction and magnitude of that opinion. We find strong evidence of selection bias among non-respondents to the ANES items commonly used to measure ambivalence. More importantly, our analyses suggest that some ambivalent citizens may be hiding their evaluations of the parties and their presidential candidates.