Is facial width-to-height ratio reliably associated with social inferences?

Theoretical considerations and early empirical findings suggested facial width-to-height ratio (fWHR) may be relevant to person perception because it is associated with behavioral dispositions. More recent evidence failing to find fWHR-behavior links suggests that mismatch or byproduct hypotheses may be necessary to explain fWHR-based trait inferences; however, these explanations may not be needed because it is not clear that fWHR is reliably associated with trait inferences. To investigate the robustness of fWHR-inference links, we conducted secondary analyses of a cross-national dataset consisting of ratings by 11,481 participants across 11 world regions who judged 60 male and 60 female faces on one of 13 social traits (ns per trait range from 760 to 975). In preregistered analyses—and exploratory analyses of a subset of traits in the larger sample of 597 faces from which the 120 faces were drawn—we found mixed evidence for fWHR-based social judgments. In multilevel models, fWHR was not reliably linked to raters' judgments of male faces for any of the 13 trait-inferences but was negatively associated with ratings of female faces' dominance, trustworthiness, sociability, emotional stability, responsibility, confidence, attractiveness, and intelligence. In exploratory analyses of a subset of traits using the larger sample of faces, fWHR was associated positively with perceptions of meanness and aggressiveness in male but not female faces, negatively with attractiveness and dominance in female but not male faces, and negatively with trustworthiness in male but not female faces. We interpret these mixed findings to suggest that (1) fWHR-inference links are likely to be smaller and less reliable than expected from prior research; (2) fWHR may play a larger role in perceptions of female faces than would be predicted from the theory underpinning fWHR hypotheses; and (3) future research should more closely examine the extent to which robust fWHR-inferences reflect mismatch in the reliability of fWHR-behavior links between ancestral and modern environments versus byproducts of other person perception mechanisms.