45 1 Introduction In increasingly diverse democratic societies political candidates of color struggle with competing evaluations by voters. On the one hand they face racism, on the other hand voters increasingly value diversity (Dancygier et al., 2021; Dancygier, 2017). Does race/ethnicity affect how voters assess political candidates? A meta-analysis on gender demonstrates that voters assess women candidates more positively than men candidates (Schwarz and Coppock, 2022). We conduct a similar meta-analysis on race/ethnicity. Drawing on all articles published in political science journals that conduct candidate experiments we re-analyze 43 studies published between 2012 and 2022. Researchers present respondents with profiles of fictional political candidates and randomize race/ethnicity. They ask respondents about their vote choice and/or to evaluate the candidate(s). We retrieved the original datasets of 81 percent of the studies and recoded the results in a consistent manner, of the remaining 19 percent we read the results from the publications. To understand the theoretical underpinnings of the publications in our selection, we first analyze the theoretical frameworks. We identify three dominant schools of thought: unjust stereotyping, useful stereotyping and shared identification. Some studies stress the unjust nature of stereotyping and link it to bias, discrimination and prejudice. Others emphasize the utility of stereotypes in voting, and mention cues, heuristics, low-information elections and shortcuts. Shared identification refers to the dynamics of in-group voting behavior, such as (unconscious) in-group favoritism (Hogg et al., 2012; Tajfel and Turner, 1979) or the expectation that descriptive representatives better represent group interests (Cutler, 2002). The selection includes over 305,632 observations. The analyses demonstrate an overall effect of racial/ethnic minority candidates of 0.235 percentage points, this effect was not statistically significant. Meta-analyses of Black and Latinx candidates reveal similar results while the meta-analysis of Asian candidates reveals voters prefer them over white candidates (0.76 percentage points, statistically significant). Shared identification has a strong impact on voters. When voter and candidate share the same race/ethnicity, they evaluate that candidate a significant 7.9 percentage points higher. As we do not find any evidence for voter discrimination against racial/ethnic minority candidates, we explain political underrepresentation by pointing towards supply-side effects (Norris and Lovenduski, 1993). Key references and dominant theoretical schools What are the theoretical underpinnings of the experimental research on the effect of candidate race/ethnicity on voters? We reveal three main schools of thought in the selection: 1) ‘unjust stereotypes’, 2) ‘useful stereotypes’ and 3) ‘shared identification’ (see Table 1) . First, the ‘unjust stereotypes’ school underlines the negative consequences
RkJQdWJsaXNoZXIy MjY0ODMw