Tuesday April 25, 2006

That One?

How accurate are eyewitnesses? A senior psychology major will present his research at a national conference as the debate grows over methods to identify suspects.

Eyewitnesses are more likely to not identify anyone or choose the wrong person out of a police line-up when they're shown pictures of suspects one at a time instead of several at once, according to a new study conducted by the Chicago Police Department.

That finding doesn't entirely surprise Dario Rodriguez, a senior psychology major at the University of Dayton who, with his thesis adviser Melissa Berry Cahoon, will present a poster session on eyewitness research at the Association for Psychological Science's national convention in New York City on May 27.

"I raised some of the same questions in my honors thesis experiment. I was initially surprised to find higher identification accuracy in simultaneous conditions than sequential - though the difference wasn't statistically significant. I also found trends in non-identification between the two groups that were similar to those reported in the Chicago Police Department study - that is, I found a higher rate of non-identification in sequential arrays, but, again, not significantly higher," said the Queens, N.Y., native. "These results highlight the need to determine the precise conditions under which each type of lineup is likely to yield higher accuracy rates."

As part of his study, 138 introductory psychology students at the University of Dayton watched an eight-second videotaped mock crime of a young man vandalizing a wall. Each was then asked to identify the suspect from among seven photos presented either simultaneously or sequentially. Participants were given either confirming, disconfirming or no feedback about their decision.

Rodriguez found that more of the mock eyewitnesses accurately identified the perpetrator when shown photos all at once, rather than one at a time, which contradicts much of the existing research. Like in the Chicago study, he recognized the importance of making sure the lineup administrator was not connected to the case to avoid potentially steering the witness to the suspect the police believe is guilty. To explore a variation of this potential effect, Rodriguez manipulated the feedback given to the eyewitnesses after they identified a crime suspect in a photo lineup, as police administrators sometimes do. When he provided feedback about how many other witnesses had supposedly made the same identification, witnesses' certainty ratings about their selections changed.

"Those who received confirming feedback reported an increase in identification certainty," he said. "It's important because jurors often perceive the certainty with which eyewitnesses testify as an indicator of accuracy, which research has shown is not necessarily the case."

Rodriguez and other researchers see a need to document immediately how certain witnesses are when they first identify a suspect from a lineup. Doing so makes it possible to track whether the confidence level of a witness changes between the time of the identification and when the witness provides courtroom testimony. Some advocates, such as Barry C. Scheck, co-director of the Innocence Project, have expressed frustration that such straightforward changes haven't been implemented.

According to an April 19 story in The New York Times, DNA evidence has proved nearly 200 wrongful convictions, with three-quarters of them resulting from eyewitness misidentifications. Such statistics underscore the importance of eyewitness research studies.

"I'm happy to see that these issues are getting press attention because they are incredibly important," said Rodriguez, acknowledging that it will be challenging for psychologists to find a way to compare sequential and simultaneous lineups in the real world. Some factors -- such as memory abilities of various witnesses, whether the witness and perpetrator are of the same race and the length of exposure to the suspect during the crime -- cannot be adequately controlled.

Rodriguez also notes the importance of distinguishing between suspect-present and suspect-absent lineups. In laboratory settings, sequential lineups are shown to yield higher accuracy when the guilty person isn't there, while simultaneous lineups tend to yield more correct identifications when the guilty person is present. "This difference largely disappears when real-world conditions are approximated," Rodriguez said.

"I guess the big question is which do we value more -- accurate identification of the guilty perpetrators or accurate non-identification of innocent suspects?"

Contact Dario Rodriguez at (937) 627-6223 or (718) 662-6919.