Statistical Considerations for the Analysis and Interpretation of Forensic Evidence

Abstract

The potential of forensic science to assist criminal investigations has become well-known to the general public through the popular television show CSI. Forensic science does assist investigators but the scientific foundation for claims made about some kinds of forensic evidence has been questioned in a 2009 report of a National Academy of Sciences’ panel and a 2016 report by the President’s Council of Advisors on Science and Technology. In addition to the questions raised by those reports, misapplication of forensic science was identified as a contributing factor in nearly half of the more than 350 cases in which DNA helped the Innocence Project to exonerate wrongfully convicted individuals. These findings have led to recent efforts to improve the practice of forensic science. Statistics with its focus on study design, reliable measurement, uncertainty assessment, and decision-making under uncertainty has emerged as a critical part of these efforts. This chapter reviews three different ways in which forensic evidence can be analyzed and interpreted (expert opinion, two-stage approach, likelihood ratios) and for each describes the key statistical issues that need to be addressed for forensic evidence to satisfy the requirements of the federal rules of evidence.