Statistical Issues in Assessing Forensic Evidence

The National Academy of Science released its report, Stengthening Forensic Science in the United States: A Path Forward (NRC 2009). An important finding was the increased need for scientific research in the evaluation of methods used in forensic science, such as bias quantification, validation, and estimates of accuracy and precision in different contexts. This article illustrates, using medical clinical trials and two fingerprint studies, the value of applying statistical methods in the design of studies that are needed to evaluate inferences from forensic evidence. Because many sources can affect both the accuracy and the consistency of decisions at each stage of the process, from specimen collection to final decision, this article discusses methods for identifying these sources, as well as the statistical principles involved in the quantification of, and uncertainty in, measured error rates. By contrasting the design of medical trials with two previous fingerprint studies and with bullet lead studies, this article emphasizes the need for reduced subjectivity, the types of measurements on physical evidence that can lead to more accurate and consistent decisions, and the importance of carefully designed studies in the evaluation of forensic evidence.