Presenting Forensic Identification Findings: The Current Situation

Abstract

Forensic science practitioners asked to identify fibers, fingerprints, handwriting, blood, semen, or other trace evidence report their findings in a variety of ways. The basic findings consist of observations or measurements of features — the bifurcations in a fingerprint, the striations on a bullet, the slant of letters in a signature, the concentrations of certain elements in a fiber, the number of short tandem repeats in certain DNA sequences, and so on. Because the measurement process for these tasks varies in the degree of subjective judgment required, human interpretation often is present in producing data on the set of features that the analyst uses in comparative examinations.

Even when the measurements are incontrovertible, however, a second level of interpretation is required to resolve questions of identity. What do the measurements imply? Typically, at least two items must be compared. One sample may come from a crime-scene or a victim, and another from a known source (such as a suspect). What does the degree of similarity reveal about the origins of the two samples? Do they originate from a common source or from two different sources? The factfinder must not only consider whether to accept the data as accurate, but also must assess the extent to which those data support (or undermine) the hypothesis of a common source.

Current practices for conveying the significance of the data vary both among examiners and across evidence types. In many fields, analysts present the evidentiary implications of their observations as either definitive conclusions or as totally inconclusive. In fields such as document examination, on the other hand, qualitative expressions of subjective confidence may be used to grade the strength of the analyst's belief in the source attribution. In still other fields, analysts supply factfinders with numerical probabilities or likelihoods. This review surveys most of the modes of expression now in use and comments on their underlying logic and limitations. It was prepared for an Expert Working Group on Reporting and Presenting Qualitative and Quantitative Evaluations of Forensic Science Evidence established in 2012.

Keywords: forensic science, scientific evidence, interpretation, identification, probative value, likelihood ratio, Bayes factor, p-value