Figure 1.
Example of female and male stimuli used in Studies 1, 2 and 3.
The faces differ in width-to-height ratio (face ratio). The lines drawn on the faces were not shown to observers and are included here to illustrate the landmarks used to measure the face ratio.
Table 1.
Descriptive statistics for ratings of female (n = 31) and male (n = 24) faces in Study 1.
Figure 2.
Bar graphs and scatterplots showing the relationship between the face ratio judgements of aggression.
A The mean Fisher's Z correlations between observers' judgements of aggression and the face ratio in Study 1, for male (n = 24) and female faces (n = 31), and, in Study 2 (shaded area), for female faces (n = 31). Error bars represent the standard error. * male faces > female faces, p<0.05. B The percent of observers whose judgements of aggression were correlated significantly with the face ratios of male (n = 24) and female faces (n = 31) in Study 1, and female faces (n = 31) in Study 2 (shaded area). C Scatterplot of the face ratio and judgements of aggression in male (n = 24) and female faces (n = 31) in Study 1. D Scatterplot of the face ratio and judgements of aggression in female faces (n = 31) in Study 2.
Table 2.
Pearson product moment correlations between the face ratio and face ratings in Study 1.
Figure 3.
Interaction plots between sex of face and masculinity, and sex of face and attractiveness in the linear regression predicting aggression ratings.
A The interaction between the sex of the face and ratings of masculinity (in males) or femininity (in females) predicting judgements of aggression when controlling for ratings of attractiveness and the face ratio. B The interaction between sex of the face and ratings of attractiveness when controlling for the face ratio and ratings of masculinity. βs are the standardized regression coefficients, representing the unique influence of each predictor when controlling for the other variables that were entered into the model. High and low scores for the plots were calculated using scores 1 SD above and 1 SD below the means.
Figure 4.
Face ratio accounted for unique variability in judgements of aggression over and above other predictors.
Mediation models were used to determine if the face ratio remained a significant predictor of judgements of aggression in male (n = 24) and female faces (n = 31) when controlling for ratings of masculinity and attractiveness. The numbers shown are standardized regression coefficients, (β weights). In the mediation model used for male faces, face ratio was entered on the first step and ratings of masculinity and attractiveness were entered on the second step. The first standardized regression coefficient between face ratio and judgements of aggression is that when the face ratio alone is used as a predictor of judgements of aggression. The second standardized regression coefficient is that when face ratio and ratings of masculinity and attractiveness are entered on the same step as predictors. For female faces, because of the high redundancy between ratings of femininity and of attractiveness, two mediation models were used to examine the unique effect of the face ratio in predicting judgements of aggression first, over and above ratings of femininity and, second, over an above ratings of attractiveness. The first standardized regression coefficient between face ratio and judgements of aggression is that when the face ratio alone is used as a predictor of judgements of aggression. The second standardized regression coefficient is when the face ratio and ratings of femininity, or of attractiveness, are entered on the same step as predictors.
Figure 5.
Mediation model and interaction plot with ratings of masculinity predicting judgements of aggression.
A A mediation model was used to determine if the face ratio remained a significant predictor of judgements of aggression in female faces (n = 31) in Study 2 when controlling for ratings of masculinity. The numbers shown are standardized regression coefficients (β weights). The first standardized regression coefficient between face ratio and judgements of aggression is that when the face ratio alone is used as a predictor of judgements of aggression. The second standardized regression coefficient is that when face ratio and ratings of masculinity are entered on the same step as predictors. B Plot of the interaction between ratings of masculinity by sex of the face in predicting judgements of aggression. Low and high values represent scores 1 SD below and 1 SD above the mean.
Table 3.
Pearson product moment correlations between the face ratio and ratings of female faces (n = 31) in Study 2.
Table 4.
Descriptive statistics for ratings of female (n = 31) and male (n = 24) faces in Study 3.
Table 5.
Pearson product moment correlations between ratings of female faces (n = 31) and male faces (n = 24) in Study 3.
Table 6.
Pearson product moment correlations between ratings of aggression, masculinity, nurturing, and femininity and with the face ratio and actual aggression across Study 1 (S1), 2 (S2), and 3 (S3).