Facial Asymmetry and Averageness Measurements: An Exploratory Methodological Comparison

Investigations into judgments of facial shape and features frequently include links between facial metrics obtained from images and subjective assessments of these faces regarding various traits (such as attractiveness, dominance, or health) and the somatic states of the portrayed individuals. Studies interpreting facial traits as a signal of underlying biological quality are numerous. This study aimed to establish the levels of correlation between asymmetry and averageness scores obtained from different measurement techniques. We implemented four commonly used techniques, including manual and automated delineation, and calculated facial asymmetry and distinctiveness (distance from the sample average). Overall, for each standardized photograph ( N = 279 Polish women; Mdn age = 60 years, IQR/2 = 8 years), five symmetry and three distinctiveness scores were computed. Spearman correlations between automated and manual delineation within the same measurement method were (very) strong for both facial traits (rs = 0.78–0.92, all ps < 0.001). Importantly, correlations between three methods differing in landmark number, placement, and analysis were between weak and moderate for asymmetry (rs = 0.38–0.56) and strong for distinctiveness (rs = 0.64–0.70, ps < 0.001). Correlations for asymmetry scores were lower than for distinctiveness. Our data present further evidence for the suitability of automated landmark placement for the derivation of face-shape measures but also suggest that differences in the specific landmarks used and/or the number of landmarks employed have potentially important effects on facial metrics.