NCJ Number
225667
Journal
Forensic Science International: Genetics Volume: 3 Issue: 1 Dated: December 2008 Pages: 50-53
Date Published
December 2008
Length
4 pages
Annotation
This article reports on the results of empirical testing of forensic DNA match probabilities for Croatia.
Abstract
The study found no observed partial matches beyond 7/8 and accordingly the high end of the X-axis of the graph consisted of observed counts of zero and low values of expected counts. This best fit occurs at 24 pairs of siblings, 1,434 pairs of cousins, and 37 parent-child pairs, with the remaining 31,916 pairs being treated as unrelated. The fit of observed and expected for the raw product rule with no accommodation for relatedness was inferior to the optimal fit with both relatedness and population substructure. The primary factor in modeling match probabilities, in this Croatian dataset at least, is to adequately model relatedness. The classes of relationship modeled were unrelated, siblings, parent/child, and cousins. For each relationship, the authors assumed that alleles might have been identical because of chance, because they were identical by descent from the modeled relationship, or because they were identical by descent from the background co-ancestry, theta. The effect of co-ancestry was modeled by using the approach of Balding and Nichols, which is the analog of the National Research Council recommendation 4.2 equations 4.10. This is the model used in much of Europe and Australasia, but not the United States. 2 tables, 2 figures, and 22 references