For decades now, humans have been convicted and punished on the basis of fingerprint evidence. Society has come to accept the fact that police will take fingerprint impressions of those they arrest. Citizens largely realize the government will analyze those fingerprint impressions by comparing them to unidentified fingerprints left at crime scenes. These unidentified fingerprints, known as latent prints, have been introduced at trial for a long time and from the perspective of many jurists, latent prints constitute nearly irrefutable evidence.
In September, 2017, a group working for the American Association for the Advancement of Science (“AAAS”) performed a study which, “concluded that latent print examiners should avoid claiming that they can associate a latent print with a single source and should particularly avoid claiming or implying that they can do so infallibly, with 100% accuracy.” Specifically, the study advised against latent print examiners using terms denoting certainty, such as “match,” “identification,” and “individualization.”
The group, consisting of a forensic scientist, an academic statistician, a psychologist, and a biometric engineer, composed a report with the purpose of making prosecutors and officers admit to the shortcomings of latent print analysis. In offering their research, the group stated explicitly that their findings only added to an important 2009 National Research Council study, which found that “most forensic disciplines have not been subjected to rigorous scientific study.”
In part, the AAAS report provided a brief summarization of the past research into the reliability of latent print evidence. Overall, the studies found that false negative identifications, where an examiner incorrectly distinguishes two identical prints, are generally more common than false positive identifications, where an examiner associates two different prints. A scale of results from the study listed false positive identification error rates from 1.5% to 2.6%, while those that measured false negative identification error rates ranging from 2.2% to 8.7% percent.
The error rates found through fingerprint accuracy studies suggest the American judicial system should be very hesitant to allow the word “match” to be used in reference to fingerprint identification.
From the latent print accuracy studies, the group identified three factors which affected examiner accuracy. First, trained examiners achieved significantly better results than untrained examiners. Second, the more difficult the comparison (regarding the quality and detail of the available print), the higher the error rate. Third, examiner accuracy depended largely on the types of print features they identify as important.
Overall, the report listed a variety of reasons for its finding that latent print examination could not provide error-free results. While the group acknowledged that examining latent prints could rule out most of the population from being the source of a latent print, but determined there was not enough data on the distinctiveness of identifying characteristics to present latent print evidence as foolproof. As possible sources for error, the report also pointed to the subjective bias and cognitive bias of print examiners, varying levels of training, and deficient examining procedures.
After the group provided evidence of the potential for error in regards to latent prints being used as evidence, the group made fourteen recommendations for how the process of latent print examination could be improved. A few key suggestions were: (1) “examiners should convey the high level of scientific uncertainty [underlying] the analysis they are presenting in court and make clear the findings are subjective and not grounded in evidence”; (2) further research into whether latent print analysis is reliable enough to be admitted as evidence; (3) additional examination into how those involved in courtroom procedures evaluate and understand fingerprint evidence; (4) as a way to reduce bias, the group advised that when evaluating latent prints and their potential matches, fingerprint examiners should be secluded away from case evidence accompanying the prints; (5) the report counsels police to slip known-prints in with the unidentified prints “to test accuracy rates in real work settings”; and (6) to improve in both subjective and cognitive bias, the report recommended the improvement of automated fingerprint identification systems.
While there have been many studies over the years documenting the error rates of latent print identification, none criticized the practice as harshly as the present group. The question is whether there will be any response to the group’s findings.