Forensic DNA analysis typically proceeds by first determining whether alleles (one of two or more alternative forms of a gene) found in DNA apparently left by the perpetrator of a crime at a crime scene (the "evidence sample") match alleles extracted from a sample of the suspected criminal's blood (the "suspect sample"). If alleles drawn from the two sources match, the next step is to provide information about the probative value of the match by estimating the probability that alleles extracted from the blood of some random individual would have matched the alleles in the evidence sample. Thinking in terms of suspect populations reveals that often there is little reason to be concerned that a defendant's ethnic group is poorly represented in a laboratory population database; however, when a defendant's close relatives are members of the suspect population, statistics based on applying the product rule to allele frequencies taken from a laboratory's population database will be misleadingly low. Until DNA databases come on line, a substantial amount of non-DNA evidence ordinarily will link to the crime a defendant whose DNA is tested. This evidence, which is unlikely to similarly implicate other members of the suspect population, is often essential for concluding that a defendant with DNA matching the evidence DNA is uniquely linked to the crime. When the non-DNA evidence is strong, neither population substructures nor the presence of relatives is likely to create a large risk of injustice. Where the non-DNA evidence is weak or where that evidence would as strongly implicate related others, as an identification based on appearance and accent might, the cautions mentioned in this article are essential if the DNA evidence is to be given its proper weight and justice to be done.
Lempert, Richard O. "The Suspect Population and DNA Identification." Jurimetrics Journal 34, no. 1 (1993): 1-7.