A Bayesian approach to validating STR multiplex databases for use in forensic casework

A previous paper in this journal has described the conventional statistical analysis of three databases (Caucasian, Afro-Caribbean and Asians from the Indian subcontinent) where individuals are typed at six short tandem repeat (STR) loci. This paper presents a Bayesian analysis of the same data and the approach is centered on the concept of estimating coancestry coefficients from mixed databases. Posterior distributions for the three databases are presented and discussed and the consequences of implementing bootstrap estimation procedures an also shown.