Fig. 1.

Quantifying and explaining individual variation in face evaluation. Variance in key facial impressions explained by Face, Observer, and the unique combination of Face by Observer that reflects individual differences in facial evaluation. Variance is computed through intraclass correlation coefficients (ICCs) in a random-intercept multilevel model. Data from twin sample. N observations = 379,200 (1,264 perceivers by three traits by 50 faces by two trials). Error bars are 95% confidence intervals, fit using bootMer from lme4 in R (2,000 bootstrap samples). Individual difference components are substantial and significant.