Population specific data improves Fordisc®’s performance in Italians

Abstract

Using discriminant functions based on morphometric data is one of the most approved methods for sex and ancestry estimation on skeletons. Nevertheless, population data from various parts of the world is needed in order to reliably classify an individual into a group. Due to population variation even sex estimation is biased when there is a lack of adequate data. Software that computes discriminant functions based on morphometric data is Fordisc®. Unfortunately, the above mentioned effects reduce its applicability in countries other than the US. For improvement of this situation data collection is currently performed extensively. The present paper shows a comparison of an Italian sample from the identified modern skeletal collection of CAL (Collezione Antropologica Labanof) [1] (Cattaneo, 2018) at the Institute of Legal Medicine Milan, Italy with a Euro-American sample from the Forensic Data Bank at the University of Tennessee Knoxville, USA.Fordisc® 3.1 was used to study population differences and sexual dimorphism. The analyzes were performed on a selection of 19 highly influential measurements that are present in most individuals of both groups respectively.Italian crania show in relation to Euro-Americans wider and lower vaults with shorter cranial bases and wider faces. The degree of sexual dimorphism is similar in both groups. Yet there is a shift in the absolute value range for males and females that biases sex estimation by almost 25% when an individual is classified on the respective other discriminant function.Our results provide explanations for Fordisc®’s unsatisfying performance on non-US individuals. At the same time they show that significant improvement is achieved by adding more population samples to its dataset.