There was a scary show on the TeeVee the other night called Frontline - The Real CSI which took on crime scene investigators, and in particular, fingerprint analysis. It was scary in the sense that people make mistakes, and that includes people who purport to be forensic experts.
Fingerprints, we have been told, are an infallible means of identifying someone. No two people have the same fingerprints, at least it was so believed until the Brandon Mayfield case. Mayfield was the Oregon lawyer who was arrested in 2004 because one of his fingerprints was found on a bag of detonators in Madrid, Spain, after a terrorist bomb blew up a commuter train killing many people. After FBI fingerprint experts declared his prints a match, and after an expert for the defense agreed with the FBI, Mayfield seemed destined to the death penalty. However, just prior to a hearing at which he was supposed to tell the judge whether he intended to testify at trial, he was informed that the real owner of the prints was another man found in Algeria.
The FBI said it wouldn't happen again. However, it was a sobering reminder that the "science" of fingerprint analysis isn't that scientific. It's subject to human error, and especially, contextual bias.
Contextual bias might occur when the fingerprint analyst is told that the prints he/she is examining belong to someone who confessed to a crime. Change the context and the analyst may come to a different conclusion.
But mistakes are made even without the context. Dr. Itiel Dror conducted a fascinating experiment. See Why Experts Make Errors. Dr. Dror's experiment was conducted as follows:
Participants were approached by the director or head of the laboratory or bureau and were asked to provide opinions on a variety of latent prints and their comparisons to tenprint exemplars. They were told that the conclusions they reached after the examination would be used for an assessment project. They were further told that the project was intended to look at problematic prints and assessments.
The prints used in the experiment were obtained from archives and had been evaluated some years before by the very same experts. But the experts didn't know that. From the article:
Overall, from 48 experimental trials, the fingerprint experts changed their past decisions on six pairs of fingerprints.
Is a 12.5% error rate good enough for government work? Maybe, but it doesn't inspire a lot of confidence in fingerprint analysts. And it certainly isn't proof beyond a reasonable doubt.
Frontline - The Real CSI covered other areas of forensics, and one interesting segment touched on the Casey Anthony case. You'll remember the outrage expressed by so many pundits over the not guilty verdict. The show noted all the junk science that was put on by the prosecution. One such witness was a "smell" expert who presumed to be able to identify "the smell of death." It seems possible that the jury found the defendant not guilty because the prosecution threw up so much unseemly evidence that the jury simply dismissed all of it.
Comments