Accuracy and reliability in sex determination from skulls: a comparison of Fordisc® 3.0 and the discriminant function analysis

Abstract

Identification in forensic anthropology and the definition of a biological profile in bioarchaeology are essential to each of those fields and use the same methodologies. Sex, age, stature and ancestry can be conclusive or dispensable, depending on the field. The Fordisc(®) 3.0 computer program was developed to aid in the identification of the sex, stature and ancestry of skeletal remains by exploiting the Forensic Data Bank (FDB) and computing discriminant function analyses (DFAs). Although widely used, this tool has been recently criticised, principally when used to determine ancestry. Two sub-samples of individuals of known sex were drawn from French (n=50) and Thai (n=91) osteological collections and used to assess the reliability of sex determination using Fordisc(®) 3.0 with 12 cranial measurements. Comparisons were made using the whole FDB as well as using select groups, taking into account the posterior and typicality probabilities. The results of Fordisc(®) 3.0 vary between 52.2% and 77.8% depending on the options and groups selected. Tests of published discriminant functions and the computation of specific DFA were performed in order to discuss the applicability of this software and, overall, to question the pertinence of the use of DFA and linear distances in sex determination, in light of the huge cranial morphological variability.