In 2017, an appeals court in North Carolina did something largely noticed but remarkable: the court found that it was an error to admit fingerprint evidence in a criminal case.  This summer, I submitted an amicus brief to the North Carolina Supreme Court that argues the appeals court had it right.  Twenty-six leading forensic analysts, statisticians, and researchers also signed the brief, calling for stronger examination of the reliable application of fingerprint and other expert evidence.  The goal was to emphasize to the court that only expert work that is reliably applied to the facts should be admitted at trial.  The brief describes how wrongful convictions can and have resulted when forensic methods are poorly applied in criminal cases.  I  also wrote a piece in UCLA Law Review Discourse about the McPhaul case earlier this year.

North Carolina rules of evidence require that forensic evidence presented in court be “the product of reliable principles and methods” and that an expert testifying about it “has applied the principles and methods reliably to the facts of the case.” However, the fingerprint examiner who testified in the case, North Carolina v. McPhaul, made “unequivocal statements” that the defendant was the source of prints found on certain pieces of evidence. Despite repeated questioning by the defense attorney, the prosecutor, and also the trial judge, the fingerprint examiner could not explain what was done during the comparison process, for how long that process took place, or how this conclusion was reached.

The reliability of a method like fingerprinting depends on the skill, experience, and the work done by the examiner. The Presidential Council of Advisors on Science and Technology (PCAST) 2016 report concluded that while “foundationally valid,” fingerprint analysis should never be presented in court without evidence of its error rates and of the proficiency or reliability of not just the method, but the particular examiner using the method.  Fingerprint analysis depends on the experience and skill of the person doing the analysis.  For that reason, the factfinder must know how reliable the person’s work is.  

Moreover, we know that errors can happen.  The PCAST report noted that error rate studies had now been conducted.  Two methodologically sound black box studies (referring to studies that independently test experts for errors using realistic materials) were conducted.  They found nontrivial error rates.  The false-positive error rate (incorrectly matching a print to a suspect) “could be as high as 1 error in 306 cases,” based on an FBI study; or a rate of “1 error in 18 cases,” based on a study by the Miami-Dade police laboratory.

North Carolina, like many other states in the U.S., has adopted Rule 702 in its rules of evidence, following the federal courts’ approach. This ‘reliability rule’ asks the judge to ensure that the application of methods to the facts is itself reliable.  Even if the method itself is accepted, the person’s work in a given case must be reliable.  However, many state courts have not rigorously applied their reliability rules.  In an article I wrote earlier this year with Chris Fabricant, of the Innocence Project, we surveyed all state opinions applying the Rule 702 reliability rule and found that very few actually discussed the concept of reliability and hardly any excluded forensic evidence offered by the prosecution (although they sometimes did for defense expert evidence).

The North Carolina Supreme Court held oral arguments in the case and last Friday decided to let the appellate ruling stand.  That decision sends an important message that the reliability of an expert’s work matters. The amicus brief was filed as part of a series of research and education projects on forensic science at Duke University School of Law, as part of the Center for Statistics and Applications in Forensic Evidence (CSAFE).