Tuesday, October 02, 2007

Possible methodology flaw in Dallas eyewitness ID study

Last week I praised the Dallas PD for participating in a study of police lineup procedures; while I'm still glad they're doing so, the Eyewitness ID Blog warns that the study described by the Dallas News sounds like it may not be a sound test:
Unfortunately, the synopsis provided by the Dallas Morning News isn't encouraging on this front:
The method that leads to the fewest identifications of people who are not the suspected guilty party is expected to be the preferred method for conducting lineups.
In other words, if the procedure where the cop is pointing at the suspect and salivating like Pavlov's dog more often results in that person being picked -- with no knowledge of whether or not that person is actually the perpetrator -- then that procedure will be deemed superior. Hopefully this apparent fundamental flaw is in the conception of this pilot project is either a reporting mistake, or if an actual reflection of the intended assessment metric, will be cured before the first the first tax dollar is wasted.

Another component being tested is the sequential presentation of the photos, which social scientists recommend, but caution should only be done if the lineup is also conducted blindly. Sequential presentation of lineup photos by an administrator who is aware of the identity of the suspect is more dangerous than even the traditional "six-pack" photo array, because the inadvertent influence of the administrator takes on heightened power when the suspect's photo is displayed by itself. Nonetheless, the Dallas project apparently intends to spend tax money testing this obviously flawed format as well.

The rational approach would be to treat blind not as a variable, but as a baseline. Blind procedures are scientific; non-blind are not. One protects against bias; one does not. It should be that simple, yet we see another study design that refuses to acknowledge this fundamental problem that continues to infect the reform dialogue.
I couldn't agree more with Ben Hiltzheimer that "no one disputes that the most basic reform measure, namely blind procedures, is unequivocally less likely to result in wrongful convictions. Or at least no one does so with a straight face." I join Ben in hoping the study won't result in another "squandered opportunity" to test improved police lineup procedures.


Anonymous said...

Grits, you crack me up!! Not to change the subject but two weeks ago on an episode of COPS, I witness two police officers salivating like Pavlov's dog as you put it. I turned the channel and said those police officers are way too happy about such a minor violation. It was shocking that they allowed that on TV. Thanks for the humor.

Gritsforbreakfast said...

That was actually Ben's turn of phrase, wish I could take credit for it, but glad to provide the laugh. :)