New research identifies ways to improve eyewitness identifications

For the first time, a study of real-world police lineups shows that witnesses more accurately identify perpetrators when they see photos of suspects one at a time rather than all at once. Iowa State University psychologist Gary Wells, PhD, led the study, and released preliminary findings through the American Judicature Society, which funded the research.

"This study is a notable step forward in eyewitness field experiments," says law and psychology expert Steven Penrod, PhD, a professor at City University of New York. "The procedures and the data are the strongest to date and the results confirm laboratory findings showing that sequential procedures can significantly reduce identification errors."

Years of laboratory research have supported the use of so-called sequential presentations in showing witnesses photos. But a 2006 study conducted by Chicago police, and published in Law and Human Behavior in 2008, raised doubts about using sequential presentation in the real world.

However, researchers, including Princeton's Daniel Kahneman, PhD, and Harvard's Daniel Schacter, PhD, quickly found flaws in the Chicago study design, including the fact that police officers administering the lineups were not blind to who the real suspect was. Based on those criticisms, the American Judicature Society set out to design a foolproof, real-world comparison of the two techniques, led by Wells and psychologists Nancy Steblay, PhD, of Augsburg College, and Jennifer Dysart, PhD, of the John Jay College of Criminal Justice. The team used state-of-the-art methodology agreed upon by a team of eyewitness scientists, lawyers, prosecutors and police. It was a randomized, controlled trial conducted with real witnesses and criminals within the Austin, Texas; Charlotte-Mecklenburg, N.C.; Tucson, Ariz.; and San Diego police departments. Police used laptop computers to administer photo lineups to ensure that they followed the study procedures accurately and reliably.

The sequential lineups led to fewer false identifications: 12 percent versus 18 percent for the all-at-once method.

Wells sees his study as a vindication of the sequential procedure as well as laboratory-based studies. "This was a very important study to convince some of the hold-outs who have dismissed this whole area of science because it's based on lab experiments," he says.

—B. Azar

Read the study online.