Using Synthetic Tool Marks in a Likelihood Ratio Test for Forensic Comparisons

• Activity Number: 670
• Type: Topic Contributed
• Date/Time: Thursday, August 8, 2013 : 10:30 AM to 12:20 PM
• Sponsor: Section on Physical and Engineering Sciences
• Author(s): Amy Hoeksema*+
• Keywords: Profilometry ; Striae ; Correlated data ; Outlier test ; Synthetic data

Abstract

Over the last few years, several numeric methods have been proposed for comparing a field tool mark found at a crime scene to ones made in the laboratory using a suspect tool to identify a match. For comparisons resulting in a single numerical index, Hoeksema and Morris (2013) proposed the use of a likelihood ratio test to analyze the difference between a sample of comparisons of lab tool marks to a field tool mark, against a sample of comparisons of two lab tool marks. In that analysis, a standard two-sided hypothesis test was used for which the null hypothesis states that the means of the two samples are the same, and the alternative hypothesis states that they are different. The weakness of this approach is that the hypotheses are reversed from the desired analysis; we must assume that the null hypothesis is true until we can prove otherwise, which equates to assuming the tool marks were made by the same tool (i.e. the evidence supports the suspects guilt) until we can prove otherwise. Using synthetic tool marks statistically modeled from the lab tool marks, we propose a method for comparing marks that reverses the hypotheses to achieve the desired test.