Scientific Heat about Cold Hits
Keith Devlin · 2007
This paper presents a study, from a mathematician’s perspective, of the current ongoing debate as to how to calculate the significance of a DNA profile match in a “Cold Hit” case, where the match is the result of a search through a DNA database, and what statistical information about the database identification may be presented in court as evidence. At present, such evidence may be (though often is not) excluded from court proceedings in jurisdictions that adhere to the 1923 Frye ruling that only scientific evidence may be admitted that uses methods on which the scientific community has reaches a consensus. Much of the legal debate has centered around the government’s insistence that it present the RMP (random match probability) in cold hit cases, a position that we argue strongly against. We use a particular current case as an illustrative example, but the focus of the paper is the general mathematical issues involved in such cases. The paper is in part expository, written both to make the mathematical community more aware of this important legal issue, and to provide lawyers and others in the legal profession with the necessary background to understand the main statistical issues involved. We do however present a proposed resolution to the key disagreement between the two main protagonists in the statistical debate — the adherents of the procedures recommended in the National Research Council’s 1992 and 1996 reports on DNA profiling evidence, and those who favor the Bayesian approach advocated by Balding and Donnelly.