Olivola and Todorov (Elected in 100 milliseconds: appearance-based trait inferences and voting. J Nonverbal Behav, 2010) provide a convincing demonstration that competence ratings based on 1-second exposures to paired photos of US congressional candidates predict election outcomes at better than chance levels. However, they do not account for variation in competence judgments. In their analysis, Olivola and Todorov show that attractiveness, familiarity, babyfacedness and age are proximal predictors of vote choice, but find that after controlling for competence these factors no longer reliably influence the margin of electoral victory. Drawing on well-documented halo effects of attractiveness on character-based inferences and the extensive literature on mere exposure effects, we re-organize Olivola and Todorov's analysis into a simple path model to explore the causal ordering of these factors. We find that spontaneous assessments of attractiveness and familiarity occur prior to attributions of competence, and thus exert a downstream effect on judgments of competence.