Match probability calculations for multi-locus DNA profiles

The paper considers aspects of the match probability calculation for multi-locus DNA profiles and a related calculation which aims to assess the probability that a pair of profiles is concordant for the presence and absence of bands. It is suggested that levels of allelism and linkage for multi-locus profiles may be higher than reported in previous studies, and that comparison of bandsharing values between different studies is problematic. Our view is that the independence assumptions which underpin the calculations have not been established. The effect of ignoring (local) heterogeneities in band frequencies may be non-conservative. Concerns thus raised about the match probability calculation could be important in practical casework. The speculative nature of some aspects of the concordance probability calculation would seem to render it inappropriate for use in court.Editor's commentsThe author continues the investigation of multilocus probes begun introduced by Curnow in this volume. He contrasts bandsharing estimates from Jeffreyset al. (1991) and the UK Forensic Science Service. Much of the difference is attributable to the greater resolving power of Jeffreys' gels (J.S. Buckleton, personal communication). Tests for independence of multilocus probe bands, and studies of the joint possession of subsets of bands within populations, were conducted by the UK Forensic Science Service, but not published (J.S. Buckleton, personal communication). In spite of the complexities of multilocus probe profiles, it is interesting to note that, just as when several single-locus probes are used, the same profiles are not found for unrelated people in large populations.