Over two studies, participants (total N=642) rated photos of Black, White, and East Asian males' faces on traits related to approachability (i.e., friendliness, trustworthiness, and threat). The purpose was to see if the commonly-used within-subjects research design, in which each participant evaluates targets of different groups (e.g., ethnicities), may lead to socially desirable responding to avoid appearing biased. We predicted that participants in this design would normalize their approachability judgments across their ratings of ethnic groups in comparison to participants who rated photos of only one ethnic group, which was the between-subjects design used in the second study. Results showed that for all three trait ratings, there was a significant participant x target ethnicity interaction for the between-subjects study suggesting an in-group bias, but for the within-subjects study, only threat ratings produced a significant participant x target ethnicity interaction. These results suggest the importance of incorporating a diversity of methods. Additional results and conclusions are discussed.