We computed symmetry preference as the number of symmetric faces chosen converted to a percentage score (100%=all symmetric faces chosen). This was computed separately for European and Hadza faces. A repeated measures ANOVA was conducted with ‘face-race’ (European/Hadza) as a within participant factor and ‘rater-race’ (European/Hadza) and ‘rater-sex’ (male/female) as between participant factors. This revealed a significant main effect of rater-race (F1,116=4.05, p=0.046). No other effects of interactions were significant (all F1,116<0.46, p>0.498). From , it can be seen that Hadza judges had stronger preferences for symmetry than European judges. One sample t-test comparing preferences for symmetry, collapsing across both sex of judge and race of faces, against chance (50%) revealed significant preferences for symmetry in both the UK (mean=56.5, s.d.=12.8, t77=4.52, p<0.001, Cohen's D=1.03) and Hadza (mean=61.9, s.d.=15.7, t41=4.93, p<0.001, Cohen's D=1.54) samples.
Percentage of preferences for symmetry (±1 s.e. of mean) in opposite-sex faces split by European (N=78) and Hadza (N=42) judges.
To examine variability in preference, a composite symmetry preference score was calculated by taking the average score for both European and Hadza faces. Note that sample size differs in these calculations as complete data were unable to be collected for all participants, Pearson product moment correlations were performed for several traits relevant to male quality and symmetry preferences. These correlations revealed a significant positive correlation between a man's perceived hunting ability (N=17, r=0.544, p=0.024) and their preferences for symmetry in female faces and a close to significant correlation between perceived attractiveness (N=17, r=0.466, p=0.096) and preference for symmetry. Correlations did not reach significance between health (N=17, r=0.041, p=0.877), good father (N=13, r=0.290, p=0.336) and time spent with children (N=13, r=0.284, p=0.346) and preferences for symmetry. We found no significant correlations between symmetry preference and male age (N=21, r=−0.274, p=0.277).
For female Hadza participants, we examined relationships relevant to reproduction. With a composite symmetry preference score, we found no significant correlations between symmetry preference and female age (r=0.159, p=0.492). We selected Hadza women below 41 years of age to restrict the sample to reproductively active women. Small sample sizes meant we did not calculate statistics comparing married with unmarried (N=2) women. Three women reported to be pregnant and six women were nursing and so we compared the preferences of those pregnant/nursing (N=9) with those who were not either (N=9) in a repeated measures ANOVA with race of face (European/Hadza) as a within-participant variable and pregnant/nursing (yes/no) as a between-participant variable. There was a close to significant interaction between pregnant/nursing and race of faces so that those women who were pregnant/nursing preferred symmetry more than those who were not when judging Hadza faces (F1,17=4.34, p=0.053). No other effects or interactions were close to significance (F1,17<0.45, p>0.510). Independent sample t-tests revealed that preferences were close to being significantly higher for those who were pregnant/nursing than those who were not for Hadza faces (yes mean=68.9, s.d.=10.5, no mean=54.0, s.d.=21.2, t17=−1.90, p=0.074, Cohen's D=0.90) but not for European faces (yes mean=60.0.0, s.d.=17.3, no mean=68.0, s.d.=14.0, t17=1.11, p=0.281, Cohen's D=0.52).