Conditioning on the number of bands in interpreting matches of multilocus DNA profiles

Abstract

There are clear and well-recognised problems in assessing the evidence provided by a match between a multi-locus DNA probe from a scene-of-crime sample and an identical profile obtained from a search of a data bank of profiles. Rarely, if ever, will the data bank be known to contain the profile of the criminal. Unless the scene-of-crime sample is unusual, a large data bank may contain a fortuitous matching profile. Thus, careful calculations similar to those used in assessing matches between individual DNA or protein sequences and DNA or protein sequences found by searching relevant data banks are needed if the occurrence of the match is to be used as evidence of involvement in the crime (Mott, Kirkwood & Curnow, 1989, 1990). Certainly, the match may assist the police in identifying potential culprits, but any use of the evidence of the match must be carefully scrutinised.