In order for the observations from a forensic examination to be calibrated, they must be properly interpreted AND integrated with other forms of evidence. In this talk I discuss two experiment that demonstrate that scientists have some work to do to improve how likelihood ratios are intepreted. For fingerprint comparisons, the data suggest that the term “identification” is associated with a likelihood ratio of around 8000. This is a far cry from some of the values we observe for majority ID comparisons, which can be as low as 10. This demostrates that the field needs to start having a conversation about the use of the term Identification and when it should be used.

This talk was given in January, 2024 to the Department of Psychological and Brain Sciences at Indiana University.

The video below is a presentation that explains how to use the Ordered Probit model to compute likelihood ratios for forensic black box studies. This work was published as:

Busey, T., & Coon, M. (2023). Not all identification conclusions are equal: Quantifying the strength of fingerprint decisions. Forensic Science International, 343, 111543.

Bayesian analyses in forensics sound complicated, but I’ve explained the basic principles in the video below.

Our NIJ grant was funded!

This three-year grant is designed to help us understand whether the words we use to make conclusions in fingerprint comparisons are aligned with the strength of the evidence.