ABSTRACT
The current paper examines the accuracy of existing binary logistic regression equations for sex prediction based on pelvic and cranial traits in a modern Greek assemblage and proposes new equations with the aim of improving correct classification rates for Balkan material. Our results suggest that existing equations based on pelvic traits perform very well on the Greek material, which can be attributed to the fact that sexual dimorphism in the pelvis results from common evolutionary forces across populations. In contrast, equations based on cranial traits are highly dependent upon the populations based on which they were developed, stressing the need to produce more population-specific functions. Our proposed equations achieve correct sex classification in 92.59% of the females and 95.79% of the males for pelvic traits, while these percentages rise to 97.53% for females and 98.95% for males when the vertical femoral head diameter is included in the models. Our functions based on cranial traits produced correct classifications in up to 92.59% of females and 88.42% of males, and when the cranial scores where combined with the vertical femoral head diameter, the correct classification rates increased to 93.83% for females and 94.73% for males. Prior to the generalization of the use of these functions, further research examining their accuracy in other groups is required, but our results appear promising.
