Statistical Methods for Assessing Strength of Cartridge Case Evidence
Date
2024-12
Authors
Osei, Fredrick
Major Professor
Ommen, Danica
Niemi, Jarad
Advisor
Committee Member
Chu, Lynna
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
The quantification of evidence in forensics is crucial for investigating crime and its consequent verdict by a jury. Two cartridge cases can be classified as a match (fired from the same firearm) or non-match (not fired from the same firearm) using the number of similar regions obtained when they are compared. The Congruent Matching Cells (CMC) algorithm generates a score representing the number of valid correlation regions when comparing cartridge cases from different crime scenes.
The CMC score only enables the classification of whether the compared cartridge cases match or do not match using certain proposed thresholds.
That leads to a need for more necessary details as to the extent of match or non-match when comparing the defense and prosecution propositions at trial by a jury. We illustrate the application of statistical methods such as a two stage approach, score-based likelihood ratio (SLR), and Bayesian approach (posterior model probability and Bayes factor) to assess the strength of evidence that cartridge cases from different crime scenes were fired from the same firearm. These statistical methods are an improvement in using only the CMC score to classify two cartridge cases as a match or a non-match. The performance of the statistical methods revealed at least a 98\% average value of sensitivity, specificity, and accuracy based on out-of-sample cross-validation estimates using a cartridge cases scan database.
Series Number
Journal Issue
Is Version Of
Versions
Series
Academic or Administrative Unit
Type
creative component
Comments
Rights Statement
CC0 1.0 Universal
Copyright
2024