Assessment of the associations of age with the classic skeletal sex indicators

Abstract

Forensic anthropologists are often required to establish a biological profile from skeletal remains, which includes the estimation of biological sex. The Walker and Phenice traits are morphoscopic methods frequently utilised to estimate sex from unknown skeletal remains. The methods employ five traits located on the skull and three on the os coxae and are considered the classic morphological skeletal sex indicators. Previous literature has stated that the growth and deterioration related to aging that occurs throughout an individual’s life may affect the classic morphological skeletal sex indicators and therefore the estimation of sex. More specifically, females are reported to become more robust with older age, while younger males appear more gracile. This overlap in sexual dimorphism leads to greater misclassification and decreased accuracy in medicolegal casework. The aim of this study was to assess if age needs to be considered when attempting to classify sex using the classic morphological sex indicators. The sex indicators were scored on a sample of 453 skulls and 429 os coxae of modern South Africans. The scores were then compared among ten different age cohorts. Separate age-specific classification models were created to assess if age affects the positive predictive performance of the sample. Significant differences were identified for the nuchal crest between age cohorts younger and older than forty years of age with the sample pooled. None of the other cranial traits demonstrated any significant differences. Regarding the classification models, accuracies increased for the younger group when separated from the older group, but the opposite was shown to be true for the older group. The os coxae did not show significant differences between the ten age cohorts. However, when separated into three large cohorts (younger, middle-aged, older) differences in accuracy was apparent between the cohorts. The results indicate that preselection of age prior to sex estimation is not necessary in a South African population. Further research is required to identify additional explanations for misclassification with sex estimation, as age does not appear to be a major influential factor for the classic morphoscopic sex indicators.