Methods & Meta-science

Comparing three analyses for face preferences

What difference does it make to analyse data on face perception by observer, by item, or using mixed models? I will present a dataset of 1429 observers rating 103 faces for attractiveness. I will present by-observer, by-item, and mixed model analyses for two simple questions. First, do men and women rate male and female faces differently? I will test the hypothesis that female faces will be rated as more attractive than male faces, but that this effect will be larger for male observers. Second, how does face age predict attractiveness? I will test the hypothesis that age will be negatively correlated with attractiveness, but this correlation will be more negative for female faces rated by male observers. All data and analysis files are available at