September 25, 2006
The Weekly Detail, a newletter for the latent fingerprint community, reported a talk by Steve Scarborough in today's edition. The talk, the first of two parts, was given to a large group of latent examiners. You can subscribe to The Weekly Detail here. Since it appears to be taken from an oral presentation, the talk rambles somewhat and is hard to appreciate from a distance. Nonetheless, it is instructive, since it represents some of the thinking (or lack thereof) that is going on in this community. His main argument appears to be in response to those who doubt the "uniqueness" of fingerprints. Along the way, he seeks to defend claims of "infallibility" in the process of fingerprint identification, though he argues that such claims do not mean that examiners don't make mistakes. The two issues of uniquenenss and error rates must be kept separate, though Scarborough does not always accomplish this well.
First of all, on the question of uniqueness. Most critics do not challenge the uniqueness of fingerprints. It is a red-herring; it is both scientifically uninteresting and legally irrelevant. The uniqueness of fingerprints says absolutely nothing, NOTHING, about the ability of examiners to reliably and validly make fingerprint identifications between partial latent fingerprints and the known prints of a suspect or defendant. In theory, every person's face is unique, but this fact, if it is so, does not tell us whether reliable and valid identifications can be made when comparing the nose and ear of an unknown person to the full face of a known person. Empirically, the hypothesis that fingerprints are unique is separate and largely unrelated to the hypothesis that fingerprint identifications can be made from partial latent prints.
Please, can we all just stop talking about whether fingerprints are unique. Nothing follows from the fact, if it is so, that they are unique. I, for one, willingly (nay, enthusiastically), concede for the sake of all further argument that fingerprints are unique. Now, let's move on.
The second issue presented, and the one of great legal significance, is the error rate of latent print examination. Claims of "infallibility" pertain to whether fingerprint examiners are 100% accurate. To his credit, Scarborough admits that fingerprint examiners make mistakes, despite apparent comments to the contrary:
In all the training classes and presentations and testimony, the FBI has never once said that there are no mistakes made by fingerprint experts. In fact the FBI, in warnings about effective verification, mentions mistakes that they are run across in submitted cases from local agencies. The FBI has always promoted verification, consultation and double checks to assure that no mistakes in fingerprint identifications are made.
The fact that mistakes are possible, then, requires some level of quality control, which might or might not be effective. It should also result in attempts to measure the rate of those errors. On the general issue, he states as follows:
The FBI instructors stress verification and other quality control measures. They promote and teach verification to prevent mistakes. If people didn’t make mistakes with regard to Fingerprints ... then we wouldn’t need verification. But we all know that human beings make mistakes, and it goes without saying that humans are not infallible. The assumption that when the FBI fingerprint expert says that they are 100% certain about the ID, that they are implying that they don’t make mistakes, is a grand leap of logic.
At the end of the above quote, the waters get muddied. Scarborough is saying that a claim of 100% certainty by a latent print examiner is not a claim that they are 100% accurate. Fair enough. But what, then, is the claim of 100% confidence based upon, if not some idea that the error rate associated with the technology he or she used is "vanishingly small." Indeed, I don't know of any bona fide scientist who would claim 100% confidence in a technique/process/machine that itself did not have 100% accuracy. But, in any case, how do examiners know that their error rates are so low that they can have 100% confidence in their conclusions? It cannot be made on the basis of the uniqueness-of-fingerprints hypothesis, since that hypothesis has nothing to do with the hypothesis of latent examiner validity (see above). It is not based on published research, since precious little exists. It cannot be based on experience -- other than casual anecdote -- since no systematic attempt has been made to catalogue errors.
So, I am very heartened to see that a prominent latent examiner has admitted the existence of measurable error rates associated with latent fingerprint procedures. It's about time that researchers began to actually measure those error rates.
September 25, 2006 | Permalink
TrackBack URL for this entry:
Listed below are links to weblogs that reference Clueless Fingerprints:
The one study on latent print ID error rates that I found was published in the Journal of Forensic Science in 1996. Here are some excerpts:
"Of the 156 respondents, only 68, or 44% had correctly identified the five latent impressions as well as correctly noted the two eliminations. The remaining 88 participants provided responses which differed from the correct answer. What is revealed within the 56% who provided divergent answers is the distinct outline of incredulous truth...
Six respondents failed to identify any of the latent prints that the manufacturer and reviewers believed should be identified. It is presued, therefore, that crime scene prints of similar quality submitted to these six individuals or agencies would escape proper detection. Valuable evidence would be negated. Justice simply would not be served.
More alarming, however, was the fact that within the 156 answer sheets, a total of 48 erroneous identifications were recorded. Misidentifications occurred on all seven latent prints, including 13 made on the five latents which could be correctly identified to the supplied suspects.
If presented in actual casework, one in five participants would have provided damning evidence against the wrong person. PAC comments fightfully noted this as a serious concern requiring immediate action.
Reaction to the results of the CTS 1995 Latent Print Proficiency Test within the forensic science community has ranged from shock to disbelief. Errors of this magnitude within a discipline singularly admired and respected for its touted absolute certainty as an identification process have produced chilling and mind-numbing realities. Thirty-four participants, an incredible 22% of those involved, substituted presumed but false certainty for truth...
If one in five latent print examiners truly possesses a knowledge, skill, or ability level below an acceptable and understood base line, then the entire profession is in jeopardy.
Once begun, the assumption of absolute certainty as the only possible conclusion has been maintained by a system of societal indoctrination, not reason, and has achieved such a ritualistic sanctity that even mild suggestions that its premise should be re-examined are instantly regarded as acts of blasphemy. Whatever this may be, it is not science."
Posted by: michael | Sep 28, 2006 11:36:03 PM