More than two decades ago, a sexual-assault victim from Sandy Springs, Ga., pointed to a picture of her attacker in a photo lineup.
"From zero to 100 percent, how sure are you?" a detective asked.
"I'm 120 percent sure," the woman said, as stated in court testimony.
It now appears that she was 100 percent wrong, according to the Fulton County district attorney. The result of DNA testing led to the release of Willie "Pete" Williams on Jan. 23 after nearly 22 years in a south Georgia prison for a crime he did not commit. Her mistake and his exoneration have revived a debate about the accuracy of eyewitness identifications – and their central role in persuading juries to convict.
But even as a handful of police departments from Boston to Minneapolis have voluntarily changed lineup procedures to help reduce such tragic errors, resistance to top-down reform from the majority of American police chiefs and prosecutors is pushing the debate into the legislative chamber. Last week, Georgia introduced a bill that would scrap the side-by-side police lineup, and would instead require police departments to present suspects one by one – either in person or by photo. Georgia joins Texas, New Mexico, West Virginia, and Vermont, which have similar bills pending.
Some experts argue that the new procedure limits inaccurate "relative judgments" that victims can make during a lineup. But police counter that they do not want to be forced to use a specific procedure that they say would undermine proven techniques detectives use.
"You can start to see a move now in policymaking and the decisions that prosecutors are making about how they elicit evidence," says Professor Christian Meissner, director of the Investigative Interviewing Research Laboratory at the University of Texas at El Paso. "The involvement of legislatures is an attempt to get prosecutors to realize that these are real issues that need to be addressed."
Since 1992, 194 people have been exonerated on the basis of DNA evidence, and some 75 percent of the convictions involved at least one faulty eyewitness identification, according to the Innocence Project, which works to clear those who are falsely imprisoned. A separate study published by the University of Michigan in 2004 found that 90 percent of mistakes that led to false convictions in rape cases were caused by eyewitness errors.
"The way the [procedure] is now, it's easy to pick somebody out that's not involved," says Michael Moore, a private investigator in Cairo, Ga.
To minimize errors, the new bills in the five states propose that police must employ a "blind" lineup administrator, who in a small town could be a retired police officer or even a barber. This person would not know anything about the case, and therefore would be unable to influence the witness, for example, to "take another look at No. 5." Also the measures would replace the lineup in which people or photos are shown side by side with a procedure where suspects are shown one at a time. This "blind sequential lineup" strategy reduces "relative discrimination" in which witnesses identify somebody who looks most like the perpetrator out of a group, researchers say.
"Let's learn the lessons not only about the fallibility of eyewitness identification, but also of the ... research that shows how to eliminate the inadvertent suggestiveness of traditional police lineup procedures," says Stephen Saloom, policy director at the Innocence Project in New York.
In recent years, research has showed that blind sequential photo lineups produce half as many mistakes as the traditional "parade" of suspects, but they also result in 3 percent fewer total identifications.
But the blind sequential photo lineup resulted in more false identifications when it was employed in Illinois in the squad rooms of Chicago, Joliet, and Evanston in 2005. Using this method, witnesses made positive identifications 62 percent of the time; the sequential method yielded identifications 53 percent of the time, according to a study by the Illinois attorney general.
With this study in mind, some criticize states' efforts to mandate a specific procedure that would replace the experience detectives have from "thousands of years of court with 25 years of studies on human memory," says Rick Malone, director of the Prosecuting Attorneys' Council of Georgia. "A clinical study in a research laboratory versus someone walking out of a liquor store who shoots the fellow across the street – those aren't the same conditions."
The change could also trigger a flood of appeals from convicts who say they were mistakenly identified, says Terence Campbell, a forensic psychologist in Sterling Heights, Mich., offering a reason prosecutors are against the legislation.
During his trial, Mr. Williams said he was playing cards with friends at the time of the rape, but his testimony could not overcome the jury's perception of the victim's "120 percent" certain identification. Williams wrote a letter from prison to the Georgia Innocence Project saying he was "a poor black man with a criminal record" who was convicted of a crime he did not commit. The group then used DNA technology to test the evidence again. "We are all human," Williams said after his release, as he forgave the witness for the mistake.
The story of Williams, an aspiring gospel singer who broke into a bar of "Amazing Grace" after his release, is likely to influence whether Georgia lawmakers vote to pass the bill, experts say.
"When you have someone just released from prison who was unjustly put in there, [the success of a reform bill] comes down to how much sympathy defense attorneys can garner in [the house judiciary] committee, and they could garner a lot," says Frank Rotondo, of the Georgia Association of Chiefs of Police.