Gülay Özcan and Magnus Rödin
Using a unique laboratory experiment where subjects are asked to guess the test performance of candidates presented by facial portraits and voice messages, this paper explores the following questions: Are beliefs about performance affected by if a candidate is perceived to have looks that are non-stereotypical for the dominant population and do these beliefs change if the candidate has native-like versus accented speech? The experiment is conducted in Sweden and the results show that candidates not perceived as stereotypically Swedish are considered to be worse performers. These beliefs are found in within-gender but not in cross-gender evaluations and are not eliminated when additional performance-related information about the candidates is provided. When candidates are presented by both looks and speech, differential evaluations based on looks disappear. Instead, we find strong negative beliefs about performance for candidates that speak Swedish with a foreign accent implying that ethnic stereotypes associated with speech override stereotypes associated with appearance. The negative beliefs associated with foreign-accented speech are not supported by corresponding mean differences in the candidates’ actual test performance.