A simulation for exploring the effects of the "trait list" method's subjectivity on consistency and accuracy of ancestry estimations

The nonmetric "trait list" methodology is widely used for estimating ancestry of skeletal remains. However, the effects of the method's embedded subjectivity on subsequent accuracy and consistency are largely unknown. We develop a mathematical simulation to test whether variation in the application of the "trait list" method alters the ancestry estimation for a given case. Our simulation explores how variations in (i) trait selection, (ii) number of traits employed, and (iii) ancestry choice thresholds affect the ancestry estimation of an unidentified skeleton. Using two temporally and geographically diverse samples, the simulation demonstrates that trait selection, trait quantity, threshold choices, and the exclusion of high-frequency traits had minimal effect on estimation of general ancestry. For all data sets and Runs, Accuracy(AS) was maintained above 90%. The authors close with a discussion on the logistical issues present when choosing traits, and how to avoid ancestry bias.