Federal Court Gets ID Science Terribly Wrong, Denies Expert Testimony

June 26, 2007

In one of the worst eyewitness ID decisions I’ve read, the US District Court for the Eastern District of Virginia denied expert testimony in US v. White earlier this month, and in the process revealed a substantially flawed comprehension of eyewitness research. US v. White, Slip Copy, 2007 WL 1768718 (E.D.Va. June 15, 2007).

The case involves a police officer as a witness, and despite a wealth of research demonstrating that cops are no better at remembering faces than anyone else — even when specifically “trained” to do so — the court relied substantially on that fact in distinguishing this case from other cases in which expert testimony on eyewitness issues had been admitted previously:

First, unlike Lester and other cases admitting expert testimony on eyewitness identification, the eyewitness in the instant case is a trained police officer. Officer Catterton served as a military police officer for the United States Army for six years. He has been trained to be constantly observant of his surroundings, especially the people he encounters on the job. He has been specifically trained to observe and remember the faces he comes across so that they may be later identified. This is in sharp contrast to the participants in Dr. Cutler’s cited studies, the majority of whom were college students.

In fact, the specific question of cops vs. college students as eyewitnesses has been studies, and all the evidence shows that trained police are no better at remembering perpetrator identities than random college students. See, for example, a 1998 study by Christianson & Karlsson: Police personnel as witnesses to a violent crime, 3 L. & Crim. Psychol. 59 (1998) (cops are no better at identifying perpetrators than civilian college students). And yet judges continue to rely on this false notion in rendering critical decisions, and in turn juries end up relying on the same false notion in their deliberations, because actual experts are prohibited from testifying to the actual scientific findings that clearly clash with the widely held common sense view.

The police-are-better-witnesses misconception was only the first of the court’s errors in rendering this decision.

Confidence and Accuracy

The court also misinterpreted the research on the relationship between the confidence of an eyewitness, and that witness’s accuracy. The court characterized the research as showing that “an extremely confident eyewitness is more likely to be accurate than a less confident eyewitness, but only by a modest amount.” In fact, the research shows that there is a modest correlation between confidence and accuracy only under very narrow circumstances, more resembling a vacuum than the real world, which rarely occur in actual practice — namely, if the confidence of a witness is recorded immediately following the identification procedure, and only when no other factors are present to corrupt the witness’s perception, such as routine suggestive police practices. In most real-world contexts, there is no correlation between confidence and accuracy, and in some cases even a negative correlation has been observed.

This case was no vacuum. By the time the officer made the identification in this case, he had already been exposed to the strong suggestion of Mr. White’s guilt — he was told that Mr. White had been arrested previously on a similar charge, was suspected of the current charge, and on those grounds the officer requested that his arrest photo be sent over. By the time the officer-witness encountered Mr. White in person, there was little doubt that he would make an identification, regardless of whether or not Mr. White was the actual culprit. There is little question that expert testimony on the lack of correlation between confidence and accuracy would have helped dispel a common misconception in the minds of jurors, and thus enhancing the accuracy of their deliberations.

Cross-Racial IDs, Mugshot Exposure, and Show-Ups

The court went on to deny expert testimony on the cross-race impairment, namely that members of one race are more likely to mis-identify members of another race, despite the fact that the phenomenon is well-documented in the scientific literature and has been shown to be misunderstood by juries.

The court made similar errors with respect to the mugshot commitment effect, and the suggestivity of show-up procedures.

As the exonerations continue to show, juries are going to keep getting it wrong until judges allow science into the courtroom to undo the widely held errors of common sense on the reliability of eyewitness evidence. In the meantime, at least there’s Tennessee.


Recent (disturbing) cites to Manson v. Brathwaite

February 12, 2007

Manson v. Brathwaite is the seminal Supreme Court eyewitness ID case, which set out the constitutional standard governing the admissibility of eyewitness evidence in criminal courts. A couple of recent cites to Manson caught my attention, insofar as they highlight a couple of the myriad problems with the status quo re: admissibility standards and judicial reasoning on the subject of eyewitness ID reliability. Cites and quick summaries below.

In the first, a federal judge inexplicably relies on the eye contact skills native to the work of an insurance agent to find heightened reliability of an ID, with no empirical basis to speak of:

U.S. v. Welch, Slip Copy, 2007 WL 119954 (E.D.Pa., Jan 09, 2007)

Photo show-up in which federal agents arrived at witness’s workplace and showed single photo (driver’s license) to witness, and asked if witness knew the person. Witness said yes, that’s the guy who opened the fraudulent account you happen to be investigating. Court found that anticipated in-court ID was sufficiently reliable and not the result of a tainted process, citing the “benign motive” of the agents in conducting the photo show-up (they only haphazardly discovered that the office where the witness worked was across the street from the house they were searching, so decided to “stop on by” given that they were “unexpectedly” there during business hours). Court also relied on opportunity to view (5-10 minutes in well-lit room), and the “skill and appreciation for the importance to make eye contact” in the work of an insurance agent.

In the second, a state appellate court relies on the myth that law enforcement officers are better at identifying/recalling culprits than other people:

State v. Biggs, — S.W.3d —-, 2006 WL 4007009 (Tenn.Crim.App., Jul 21, 2006)

Photo show-up two months after incident, in which witness was undercover narcotics agent. Fourteen months after incident, witness was unable to ID defendant in court. Despite needless suggestivity, court found ID reliable, citing three opportunities to view (longest being two minutes), and also relied on the erroneous observation that cops are better at IDing than lay people.

How do we know this judge’s assumption is wrong? Because decades of research tell us so. Here are a few quick examples:

Christianson & Karlsson, Police personnel as witnesses to a violent crime, 3 L. & Crim. Psychol. 59 (1998) (cops are no better at identifying perpetrators than civilian college students).

Stanny, Claudia & Johnson, Thomas, Effects of stress induced by a simulated shooting on recall by police and citizen witnesses, 113 Am. J. of Psychol. 359 (2000) (no difference between cops and civilians on identification accuracy).

Woodhead, Baddeley, & Simmonds, On training people to recognize faces, 22 Ergonomics 333 (1979) (three-day training course on face recognition had no effect on identification accuracy).