|
Abstract:
|
Statistical thinking and practice can make a substantial contribution to the manner in which forensic science is handled in the laboratory and the courtroom. There have been many forensic science experiments and analyses of data that have been carried out without any recognition that statistical expertise should be a necessary part of the process. We present a discussion of the current state of statistical methods for comparing forensic measurements obtained from DNA profiling from different sources. Comparisons usually involve evidence left at a crime scene and whatever information can be retrieved from a database or from test measurements. In this paper, we introduce two statistical issues that are important when computing probabilities of a match in the comparison of DNA forensic measurements. First, we correct a common misconception among forensic analysts that the so-called ``product rule'' corresponds to an assumption of independence of alleles on a given locus of DNA. Second, we show that database searches should involve a recognition of the multiple testing problem when calculating match probabilities.
|
