ABSTRACT
It is usual to attach a numerical weight to a match between DNA obtained from a crime sample and DNA taken from a sample given by a suspect. Adding the crime stain profile and the suspect’s profile to the sample before constructing allele sample frequencies neither accounts for the sampling variation induced by finite samples nor corrects for the presence of population substructure. Part, and maybe the smaller part, of the uncertainty in the estimate is often referred to as sampling error. The word error does not refer to an analytical error but rather the variation that would occur if a different sample of individuals were taken to create the population database. An estimate of sampling error was calculated for the match probabilities estimated from each database. The concept of a minimum allele probability replaces zero or very small allele probabilities derived by counting from some database with some minimum probability.
