Race/Ethnicity in Candidate Experiments: a Meta-Analysis and the Case for Shared Identification

Abstract

Does race/ethnicity affect how voters assess political candidates? To address this question, we pooled data from 43 published candidate experiments from the last 10 years with a combined N of 305,632. We distinguish three different schools of thought that authors apply: unjust stereotypes, useful stereotypes and shared identification. Voters use “unjust stereotypes” and discriminate against candidates of color or use “useful stereotypes” that inform them of the policy positions they expect candidates to defend. Scholars increasingly apply a “shared identification” perspective and study the effect of congruence between voter and candidate characteristics on assessments. The results show that voters do not assess racial/ethnic minority candidates differently than their majority (white) counterparts. This does not hold for Asian candidates in the US: voters assess them slightly more positively than majority candidates, although this effect is small (0.76 percentage points). Shared identification matters enormously: when voters share the same race/ethnicity as a candidate they assess them 7.9 percentage points higher than that they assess majority candidates. This effect is substantively meaningful and significant for all most researched (US-based) races/ethnicities. This indicates that the underrepresentation of racial/ethnic minority citizens cannot be explained by voting behavior, but possibly by supply side effects.