Benefits and Limitations of the Statistical Methodology in Determining Evidentiary Value of Latent Print Identification: A Prospective Evaluation

The process of latent print examination has been challenged in court based on the fact that the series of decisions made during the latent print examination process typically rely on observational data and on the education and experience of each practitioner. Recent research has shown that these decisions are not based on a definable sufficiency threshold, nor are they correlated to examiner’s education or experience. In addition, these research projects have highlighted the large variability in the decision-making process among multiple examiners considering the same latent print.
Of particular relevance to this research projects are the challenges related to the lack of any statistical elements to support the decision-making process within ACE-V, and more specifically:

1. Whether decisions made during the examination process are scientifically supported (i.e. based on appropriate assessment of features observed between latent and known prints);
2. Whether the weight of latent print evidence is being properly reflected when examiners report identifications in courts; and
3. Whether some friction ridge evidence is currently missed and not reported at all.

This project will rely on a large dataset of current casework latent prints examined by the DPDCL in order to study the use of a statistical model. This dataset contains unexamined latent prints collected from crime scenes and evidence in Denver. Available analytical information associated with these latent prints will be documented for this study. This will produce a research dataset of latent prints as observed in actual casework and expands on research previously conducted in controlled conditions using the model developed by Dr. Neumann.

We aim to study the possibility of using a fingerprint statistical model to:
1. Examine the potential improvement in resource management and work product resulting from the use of fingerprint statistical models in forensic casework.
2. Obtain statistical data to compare the level of agreement between the statistical model and the examiner in determining sufficiency during the Analysis stage.
3. Propose metrics for assessing the validity of statistical models by comparing likelihood ratios of both identifications and close non-matches as a result of AFIS submissions.
4. Propose guidelines for the operational deployment of fingerprint statistical models in forensic laboratories.

Overall, our research will evaluate the benefits and limitations of utilizing a fingerprint statistical model as a practitioner tool for latent print examiners.

ca/ncf