Document Type


Publication Date



A recent report of the President’s Council of Advisors on Science and Technology questioned the validity of several types of criminalistics identification evidence and recommended “a best practices manual and an Advisory Committee note, providing guidance to Federal judges concerning the admissibility under Rule 702 of expert testimony based on forensic feature-comparison methods.” This article supplies information on why and how judicial bodies concerned with possible rules changes—and courts applying the current rules—can improve their regulation of criminalistics identification evidence. First, it describes how courts have failed to faithfully apply Daubert v. Merrell Dow Pharmaceutical’s criteria for scientific validity to this type of evidence. It shows how ambiguities and flaws in the terminology adopted in Daubert have been exploited to shield some test methods from critical judicial analysis. Second, it notes how part of the Supreme Court’s opinion in Kumho Tire Co. v. Carmichael has enabled courts to lower the bar for what is presented as scientific evidence by maintaining that there is no difference between that evidence and other expert testimony (that need not be scientifically validated). It suggests that if the theory of admissibility is that the evidence is nonscientific expert knowledge, then only a “de-scientized” version of evidence should be admitted. Third, it sketches various meanings of the terms “reliability” and “validity” in science and statistics on the one hand, and in the rules and opinions on the admissibility of expert evidence, on the other. Finally, it articulates two distinct approaches to informing judges or jurors of the import of similarities in features—the traditional one in which examiners opine on the truth and falsity of source hypotheses—and a more finely grained one in which criminalists report only on the strength of the evidence. It contends that courts should encourage the latter, likelihood based testimony when it has a satisfactory, empirically established basis.