Skip to content

Statistical Thinking for Forensic Practitioners Short Course – Session 3

Session 3 of this three-part CSAFE short course was held on April 23, 2021.

Presenter:

Hal Stern
Provost and Executive Vice Chancellor, Chancellor’s Professor, University of California, Irvine
CSAFE Co-Director

Presentation Description:

The National Academies report in 2009 and the PCAST report in 2016 encouraged consideration of quantitative approaches to assessing forensic evidence. In this session, we review two of the most often-suggested quantitative approaches, the two-stage approach and the likelihood ratio (Bayes factor) approach. Specific topics include:

  • The two-stage approach to assessing forensic evidence
    • The role of statistical tests in assessing the similarity of two samples
    • Approaches for assessing the relevance of observed similarities
  • Introduction to the likelihood ratio approach
    • Definition and interpretation of the likelihood ratio/Bayes factor
    • Possible applications to different types of evidence (DNA, trace, pattern)
    • Score-based likelihood ratios
    • Sensitivity of the likelihood ratio to modeling choices
  • Results of studies comparing different ways of expressing source conclusions

Related Resources

Towards a likelihood ratio approach for bloodstain pattern analysis

Towards a likelihood ratio approach for bloodstain pattern analysis

In this work, we explore the application of likelihood ratio as a forensic evidence assessment tool to evaluate the causal mechanism of a bloodstain pattern. It is assumed that there…
Does image editing improve the quality of latent prints? An analysis of image-editing techniques in one crime laboratory

Does image editing improve the quality of latent prints? An analysis of image-editing techniques in one crime laboratory

Field research within latent print comparison has remained sparse in the context of an otherwise growing body of literature examining the discipline. Studies examining how ACE-V procedures are implemented within…
An Open-Source Implementation of the CMPS Algorithm for Assessing Similarity of Bullets

An Open-Source Implementation of the CMPS Algorithm for Assessing Similarity of Bullets

In this paper, we introduce the R package cmpsR, an open-source implementation of the Congruent Matching Profile Segments (CMPS) method developed at the National Institute of Standards and Technology (NIST)…
A Data-Driven Approach to Model the Generation Process of Bloodstain Patterns

A Data-Driven Approach to Model the Generation Process of Bloodstain Patterns

This poster was presented at the 106th International Association for Identification