Statistical Evaluation of Surface Metrology Data in Firearm Identification

• Activity Number: 418
• Type: Topic Contributed
• Date/Time: Tuesday, August 11, 2015 : 2:00 PM to 3:50 PM
• Sponsor: Section on Physical and Engineering Sciences
• Author(s): Tracy Morris* and Deion Christophe
• Companies: University of Central Oklahoma and University of Central Oklahoma
• Keywords: forensic science ; cluster analysis ; logistic regression ; firearms ; tool marks ; ballistics

Abstract

The need for an objective method to improve interpretations in the forensic firearms and toolmark community has been profoundly stated in the literature. The potential impact of surface metrology instrumentation on this forensic science community is significant. One such system, EvoFinder, uses a 2D+D frame fragment technique to acquire surface data from a bullet. The EvoFinder can search up to 360º in rotation for the best match position. For each unknown/known bullet pair, the EvoFinder produces a set of measurements based on proprietary formulae. It is hypothesized that, even without knowing how the measurements are produced, they can be used to identify bullet matches. In this study, a set of bullets was inter-compared by the EvoFinder. A cluster analysis was implemented to separate the matching from the non-matching pairs. Following the cluster analysis, a list of true matches was provided. A logistic regression model was then developed using the measurements in an attempt to estimate the probability of a match, and to establish ranges for matches, inconclusives, and non-matches.