Test Approaches to Population Affinity/Sex Classification Using Fordisc 3.1 Considerations for Best Practice Recommendations for U.S. Forensic Casework

Abstract

Fordisc 3.1 is the primary quantitative method of estimating population affinity for the biological profile in forensic anthropology. The creators, Drs. Jantz and Ousley, put forth program use guidelines with two different approaches to population affinity estimation: simultaneous estimation with sex using all 13 available reference samples (Approach 1) or using reference samples from only one sex based on a priori sex estimation (Approach 2). Ousley and Jantz (2012) further recommend reducing the reference samples iteratively down to four (four-way) to two groups (two-way) for the final classification. Using cranial data of known demographic individuals (n = 133), this research tested both reference sample selection and reference group reduction approaches, with and without stepwise selection, to set forth recommendations for U.S. forensic casework. For Approach 2, sex was estimated qualitatively using MorphoPASSE and ordinally scored cranial features. Overall accuracy was higher in Approach 1 compared to Approach 2 by 1.5% to 4.5%, depending on group reduction method and use of stepwise selection. This suggests that Fordisc 3.1 should be utilized for simultaneous population affinity/sex estimation, rather than relying solely on qualitative methods for a priori sex estimation. While accuracy rates were similar between the four-way and two-way analyses, the correct population affinity/sex group was more likely to be retained in the final group comparison for misclassified individuals in the four-way analyses. Finally, while stepwise selection did modestly improve accuracy, larger samples are needed to more critically evaluate the impact on individual population affinity/sex subsamples.