Abstract
Sex estimation based on cranial non-metric traits remains one of the most widely used methods in forensic and anthropological practice. Logistic regression equations based on five non-metric traits, first proposed by P. Walker (2008) and subsequently modified by many researchers for the use for populations of different origins, have gained considerable popularity. In the present study, we tested the effectiveness of several formalized methods based on the analysis of five traits: glabella (supra-orbital ridge), mastoid process, external occipital protuberance, superior orbital margin, and mental eminence. The research was carried out employing cranial collections from the Kirov Military Medical Academy and the Peter the Great Museum of Anthropology and Ethnography, Russian Academy of Sciences (242 skulls). Record data on the biological sex were available for all specimens. Each of the authors has evaluated the degree of trait development independently using the five-grade scale published by A. Klales and S. Cole (2018). The influence of interobserver variability in trait scoring on the classification accuracy was analyzed. The results show that the actual performance of most published equations is lower than originally reported and is comparable to, or even lower than, the accuracy of visual sex estimation based on overall researcher’s impression. Interobserver error in scoring two of the five traits (mastoid process and mental eminence) significantly reduced reproducibility. The magnitude of this effect is comparable to the level of interpopulation differences in sexual dimorphism among different European populations. Our findings highlight the necessity of performing/employing blind tests and developing population-specific models based on stricter approaches to the determination of biological sex using cranial morphoscopic traits.