The 82nd Annual Meeting of the American Association of Physical Anthropologists (2013)


The Lubischew’s test, an useful statistical tool for Forensic Anthropology

JUAN MANUEL JIMÉNEZ-ARENAS1,2, JOSÉ ANTONIO ESQUIVEL1 and ALEXIA SERRANO-RAMOS3.

1Department of Prehistory and Archaeology, University of Granada, 2Institute of Peace and Conflicts Research, University of Granada, 3Master of Archaeology, University of Granada

Friday All day, Clinch Concourse Add to calendar

The Lubischew’s test (LT) is a parametric statistical tool, based on the Student’s t-test, developed in 1961 by a Russian biologist (A.A. Lubischew) to evaluate the degree of overlap between two data sets of single variables. The coefficient of discrimination is expressed as K and is calculated as the quotient of the difference between the respective means squared (numerator) divided by the pooled variance multiplied by two (denominator). LT presents significant differences regarding the main statistical tool used for sex discrimination in Forensic Anthropology, the Discriminant Function Analysis (DFA): (1) LT emphasizes the inter-groups differences; (2) LT requires of normal homocedastic distributions; (3) LT is more size-independent than other tests and well suited for small sample sizes because in the K coefficient the number of observations (n) is removed from the numerator.

We have applied these two statistical techniques to Medieval cranial remains from a necropolis located in Southern Spain (Arenas del Rey, Granada). The sex of all individuals was reliably assigned based on the qualitative and quantitative features of the coxal bone and cranium. As an example, LT can improve the % of discrimination obtained by DFA in more than 7% (biasterionic breadth), what can represent a useful contribution of Forensic Anthropology to the criminal justice system. Although LT provides the highest discrimination, it does not always perform above the DFA on a variable-by-variable basis. Therefore, LT is a useful tool for sex discrimination by itself or combined with DFA.

Tweet
comments powered by Disqus