Stat Cafe - Dr. Alicia Carriquiry

1 minute read

Header

Statistics and the evaluation of forensic evidence

Time: Monday 2/26/2024 from 11:40 AM to 12:30 PM
Location: BLOC 448
Pizza and drinks provided
Recording

Description

When a crime is committed, CSIs often recover evidence from the scene. Evidence may include biological samples (e.g., blood), trace material (e.g., hair, clothing fibers) or what is known as patterns. Patterns include finger prints, blood spatter, striations on a bullet, a handwritten note, and other such evidence that can be represented in the form of an image. The question of interest is whether the defendant may have been the source of the evidence at the crime scene.

At present, forensic examiners carry out a visual comparison of the crime scene items and of corresponding items obtained from the defendant (for example, a print of the defendant’s fingers). If examiners find “enough” similarities, they may declare a “match”. Otherwise, the conclusion is either an exclusion of the defendant as the source, or an inconclusive finding. This subjective, data-free approach is not accurate, repeatable or reproducible. Can we do any better?

At CSAFE (www.forensicstats.org) we have spent years developing statistical and algorithmic methods to quantify the similarity between images and to interpret findings in the context of the probative value of the evidence. For illustration, we describe principled approaches for the analysis of firearm evidence and of handwriting evidence, and discuss the statistical framework for drawing inference about the source of the evidence found at the crime scene.