The Use of Correlated Binomial Distribution in Estimating Error Rates for Firearm Evidence Identification

Area of Science:

• Forensic Science
• Ballistics
• Statistical Modeling

Background:

• Estimating error rates in firearm evidence identification is a significant challenge.
• The congruent matching cells (CMC) method offers a quantitative approach to improve ballistic identification accuracy.
• Existing probability models for CMC often assume independence between cell pair comparisons, which may be invalid.

Purpose of the Study:

• To propose novel statistical models for firearm evidence identification based on dependent Bernoulli trials.
• To develop a methodology for parameter estimation within these new models.
• To evaluate the effectiveness of the dependent Bernoulli trial model using real-world firearm data.

Main Methods:

• Development of statistical models utilizing dependent Bernoulli trials.
• Application of parameter estimation techniques to the proposed models.
• Analysis of an actual dataset of fired cartridge cases to demonstrate model performance.

Main Results:

• The study proposes statistical models that account for dependence in cell pair comparisons within the CMC method.
• Methodology for parameter estimation is provided for the new models.
• Application to fired cartridge case data demonstrates the potential improvements offered by the dependent Bernoulli trial model.

Conclusions:

• The assumption of independence in CMC cell pair comparisons may not hold, necessitating alternative models.
• Dependent Bernoulli trial models offer a more statistically sound approach to estimating firearm evidence error rates.
• The proposed methodology shows promise for enhancing the accuracy and reliability of ballistic identifications.