Monday, June 29, 2009

Sorry, CSI-Watchers

I just got called for jury duty (although due to a reschedule, not until April, which is far enough away to not actually exist), so I'm intrigued by court-related things at the moment.

I found this interesting article in Annual Reviews (subscription required) on Failed Forensics: How Forensic Science Lost Its Way and How It Might Yet Find It.

It talks about the "nonscience forensic sciences" that underlie a lot of the evidence presented in court cases, explaining that techniques of 'individualization' (or precisely determining that a single finger, bullet, shoe, set of teeth, etc. and no other in the world was responsible for a mark) are not actually very closely related to the testable, falsifiable theories that we think of as science.

The article more than once links forensic techniques with religious ritual, as well as with discredited sciences like phrenology. It seems that handwriting identification and other forensic techniques are based on subjective judgments and personal interpretations more than on any testable theories, and are admitted in court more because they have been admitted in the past than because they are objectively good sources of information.

It says that forensic fields often claim the possibility of providing an error rate of zero; that there is no chance that the results could be mistaken. Obviously we want to keep the chance of mistaken identification as low as possible, but I don't know, claiming infallibility seems a little bold.

The article talks about how the science of forensic identification has been shaped by the needs of the courts, as well as by the objective attempt to narrow down a pool of potential suspects. Plaintiffs need doubt, and prosecutors need certainty, and everyone is interested in convicting criminals, so there's a lot of interesting potential for pressure.

Another concern is that lawyers and judges, like many of us, are not particularly well versed in the scientific method, and lack the understanding of statistics, error rates, and other things that would allow them to distinguish science from impressive nonscience. It's therefore easy to see how things might be allowed as evidence despite not actually being well grounded in the scientific method.

I don't have any say over what's admitted as evidence, but I'll try to remember that things like fingerprint analysis, which we tend to assume are pretty soundly scientific, may not be as ironclad as we think.


Failed Forensics: How Forensic Science Lost Its Way and How It Might Yet Find It
Annual Review of Law and Social Science
Vol. 4: 149-171 (Volume publication date December 2008)
(doi:10.1146/annurev.lawsocsci.4.110707.172303)
First published online as a Review in Advance on July 8, 2008

1 comment:

erinserb said...

Aren't lie detector test results inadmissible in a court of law? Perhaps, at one time, lie detector machines electrically shocked some unassuming victims to tell the truth - :-) I know that is nonsense of course