Abstract
Understanding population diversity is necessary for the development of forensic anthropology methods that address population affinity. The commonly used age-at-death estimation methods for adult skeletal remains are based on American data sets, which include individuals with both European and African population affinities. Evaluating these methods on different skeletal collections worldwide is hampered by our incomplete understanding of population diversity. Hence, this study evaluated three methods of estimating age-at-death using the pubic symphyseal surface, auricular surface, and acetabulum regions on the os coxa. A total sample size of 143 os coxae from Black South African and White South African population groups was examined in KwaZulu-Natal (KZN). The accuracy (chi-squared test of independence), bias, absolute error (mean absolute error, Wilcoxon signed-rank test), interobserver, and intraobserver correlation coefficients (Cronbach's alpha test) were recorded for each method. The auricular surface method scored the highest accuracy; however, the reliability of this method is still questionable, as it also scored the highest absolute error and bias among the three methods on both population groups. Compared to the pubic symphyseal surface and auricular surface methods, the acetabulum method showed promise, as it maintained lower bias and absolute error and was consistent when re-evaluated in both populations. Laterality was insignificant for any of the three methods. Sex was insignificant for the acetabulum and auricular surface methods, but significant for the pubic symphyseal surface method in the White South African samples and insignificant in Black South African samples. These results demonstrated the need for creating population-specific standards, including precisely defined age ranges and reference samples catered to the diverse South African populations.