What is the influence of candidate gender and ethnicity on vote choice? In recent years there has been boom in candidate choice experiments addressing this question. We take stock of the field in this meta-analysis of 37 studies in which candidate gender and ethnicity are randomized. We find that overall effect sizes for both gender and ethnicity are small to non-existent. However, almost all studies use the general population as a sampling frame, and do not assess whether candidates appeal differently to voters that share their gender and ethnicity to those that do not. Correspondingly, the majority of the publications are theoretically underpinned by research on stereotyping and discrimination by majorities. We argue that candidate choice experiments can and should be approached the perspective of minority representation. We present a research agenda in which this idea is elaborated on.
