Facial width-to height ratio (fWHR), presumed to be shaped by testosterone during puberty, has been linked with aggressive, dominant, and power-seeking behavioral traits in adult males, although the causal mediation is still being disputed. To investigate the role of mere observer attribution bias in the association, we instructed participants to draw, feature-assemble, or photo-edit faces of fictitious males with aggressive-dominant character (compared with peaceloving-submissive), or powerful social status (compared with powerless). Across three studies involving 1,100 modeled faces in total, we observed little evidence for attribution bias with regards to facial width. Only in the photo-edited faces did character condition seem to affect fWHR; this difference, however, relied on displayed state emotions, not on static facial features. Anger, in particular, was expressed by lowered or V-shaped eyebrows, whereby facial height was reduced so that fWHR increased, relative to the comparison condition where the opposite happened. Using Bayesian analyses and equivalence testing, we confirmed that, in the absence of state emotionality, there was no effect of character condition on facial width. Our results add to a number of recent studies stressing the role of emotion overgeneralization in the association of fWHR with personality traits, an attributional bias that may give rise to a self-fulfilling prophecy. Methodologically, we infer that static images may be of limited use for investigations of fWHR because they cannot sufficiently differentiate between transient muscular activation and identity-related bone structures. (PsycInfo Database Record (c) 2023 APA, all rights reserved).