Blinding Us With Science: The Trouble with Forensic Sciences

October 7, 2014

Twenty odd years ago in Mississippi, Eddie Lee Howard, Jr. was sentenced to death for the murder and rape of an 84-year-old woman. There were no fingerprints linking Howard to the scene of the crime, nor was there any hair or semen recovered from Georgia Kemp’s body. His conviction was based mostly on the testimony of Dr. Michael West, a Mississippi dentist who identified Howard’s teeth as matching the bite marks on the victim’s body. Dr. West was a frequent witness for the prosecution in the 1980s and 1990s, testifying that like fingerprints, bite marks were unique to an individual and that flesh wounds provided sufficient samples to generate matches. But the scientific basis of this type of forensic evidence has been called into question by the National Academy of Sciences in a 2009 report. According to the report, bite marks change over time and cannot necessarily be linked to particular individuals. Moreover, like other pattern recognition evidence, such evidence is subject to bias when the analyst is asked to match the sample to an existing suspect. The Mississippi Innocence Project has filed an appeal challenging Howard’s conviction.

Howard’s case—and numerous others—reveal an ongoing crisis in the use of the forensic sciences in criminal cases. First, the so-called “pattern recognition” forensic sciences (finger printing, tire marks, hairs and other fibers, to name a few) do not meet widely recognized legal standards for scientific reliability. The National Academy of Sciences has identified several ways in which forensic sciences depart from widely accepted scientific practices. Many methods of analysis for pattern recognition evidence have not been subjected to rigorous evaluation. Forensic experts do not have the means to estimate error rates and so cannot describe systematically how confident they are of the results. The research culture in the forensic sciences is underdeveloped, in part because there is little funding to support such research. Apart from publishing information about new techniques, forensic scientists rarely submit their findings for peer review. These are just some of the practices that legal authorities usually require to call evidence “scientific.”

Moreover, pattern recognition evidence is particularly susceptible to biased interpretations. When an analyst is asked to match evidence to a particular suspect, cognitive biases make it more likely that the analyst will find a match. In an influential set of experiments, psychologist Itiel Dror drew on the famous misidentification of Brandon Mayfield’s fingerprints in an FBI investigation of terrorist attack on a Madrid train station in 2007. In the first iteration of this experiment, Dror presented five fingerprint analysts with what he said were Mayfield’s misidentified prints and asked them whether they would have made the identification. In fact, he gave them matches they had made in their own past cases. Three of the experts reversed their prior identifications, and one said it was inconclusive. Only one expert stood by the prior judgment. Dror has replicated this finding, finding these cognitive biases common among fingerprint analysts.

If forensic scientists are susceptible to suggestion, this bias can only be aggravated by their close connection to the criminal justice system. Forensic practices were developed by law enforcement for law enforcement, to assist in criminal investigations. Crime labs are often housed in police departments, staffed by police officials. These connections to police and prosecutors compromise their independence and create frequent opportunities to inject biases into their conclusions. So ironically, forensic sciences derive their legitimacy and authority not from the recognition they receive in the legal system, but from their close connection to law enforcement, which undermines the scientific basis of their findings.

Even with all of these problems, forensic scientific analysis would still be acceptable under current legal standards, given that judges are reluctant to be the final arbiter of what constitutes “science.”  The only requirement is that that the evidence be subjected to rigorous cross-examination in the courtroom. Yet we know that most evidence in criminal trials does not receive that kind of scrutiny. By most estimates, 95% of criminal cases are resolved with plea bargains and never reach trial. In such cases, police and prosecutors may have access to expert evidence, but defendants may never even see the evidence, let alone raise serious challenges to its validity.

In the few criminal cases that get to trial, prosecutors often put on experts, and we know that jurors are very deferential to these “experts.” A study in 2000 showed that most challenges to the scientific reliability of evidence occurred in civil, not criminal cases. And most challenges in criminal cases failed; the evidence wound up being presented at trial. As a result, questionable evidence—like bite marks and odor comparisons—are admitted into evidence to juries who may be unduly influenced by the appearance of apparent “scientific” expertise.

Even further, defendants often lack the resources needed to counteract the effects of such evidence. Defense lawyers, particularly for poor defendants, may be underprepared to challenge the scientific basis for the admission of such evidence. If they are prepared, they may not have the funds to hire opposing experts. And judges may admit poor quality evidence because they’ve always done it before, or because the expert has a long track record of testifying in other cases.

There are signs that the forensic science community is responding to these challenges. For example, the FBI no longer allows its experts to testify that fingerprint analysis is infallible, thus acknowledging that fingerprint identifications are subject to error. The city of Houston recently created the Houston Forensic Science Center, “a local government corporation created to provide independent forensic services to the Houston Police Department, other law enforcement organizations, and others engaged in the justice system.”  In its mission statement, the Center states that it is committed to “adhering to the highest standards of quality, objectivity, and ethics.”  Indeed, the funding for the Center comes from the City and not the Houston Police Department. But it remains to be seen how independent forensic scientists can be when their only clients come from law enforcement.


Anna-Maria Marshall is Associate Professor at the University of Illinois, Urbana-Champaign, where she also has an appointment in the College of Law. She is a member of LOTL’s Advisory Panel. 

Image: Jack Spades via photopin