LAUSR

Abstract Commonly used measures of authoritarian predispositions have received mixed support as a predictor of political preferences in American elections. Using new survey data (N = 1,444), we demonstrated how imprecise conceptualization and measurement of authoritarianism can obscure its relationship to candidate preferences. First, authoritarians have largely sorted into the Republican Party and self-identified as conservative, thereby attenuating the predictive power of authoritarianism when such features are used as controls or selection criteria. Second, the authoritarianism measure typically used in election studies covers a limited range of the construct, specifically focusing on the facet of authoritarianism we observed to be least associated with support for Republicans candidates in the 2016 American electoral context. We find predictive gains both from more comprehensive measurement of authoritarianism and from analyzing facet-level authoritarianism.