Blind Sighted

Eyewitness identification doesn't always mesh with DNA evidence, and that's leading police departments to rethink their lineup procedures.

One day in the mid-1970s, a graduate student at Ohio State University was approached in the halls of the psychology department by a defense lawyer desperate to prove that one of his clients was innocent. The lawyer believed his client had been misidentified and was searching for any scientific research on the reliability of eyewitness memory.

The student he approached, Gary Wells, knew virtually nothing about the subject--and almost no one else did, either. But the encounter led to research that, 30 years later, is influencing police lineup procedures. Based on material developed by Wells and incorporated into a 1999 U.S. Justice Department report, "Eyewitness Evidence: A Guideline for Law Enforcement," a growing number of police departments and prosecutors are changing longstanding eyewitness-identification practices.

Through the use of such techniques as sequential presentation of lineup photos and double-blind presentation of those pictures, they are improving the accuracy of eyewitness identifications. It is a problem that has become increasingly obvious ever since the mid-1990s, when DNA evidence began exonerating scores of wrongly convicted prisoners. Researchers and policy makers stepped back to analyze how so many people could have been wrongly convicted and found that in about 75 percent of such cases, the conviction hinged on eyewitness testimony. "It's the single greatest confounding factor in the convictions proven wrongful by DNA," says Steve Saloom, policy director for the Innocence Project.


It wasn't as though the findings from the research--Wells' work over the years has included elaborate experiments to test eyewitness memory--led immediately to reform. Rather, it took a dramatic case of misidentification of a rapist in a state that could control procedures at the local level to move the findings out of the research realm.

The state was New Jersey and the case, which took years to wind its way through the court system, was State v. Cromedy. The facts are these: In 1992, a Rutgers University student was raped in the kitchen of her apartment. Although she could describe her attacker to police, she was unable to identify her rapist from a lineup. Eight months after the rape, however, she saw the man she believed to be her attacker walking down the street and called police. That man, McKinley Cromedy, was arrested, convicted and sentenced to life plus 50 years in prison.

Cromedy's defense lawyer, Anderson Harkov, had intended to introduce psychological evidence at the trial demonstrating the unreliability of eyewitness testimony. Specifically, he wanted to present studies showing that the accuracy of identifications drops significantly when the witness is of a different race than the suspect, as in the Cromedy case. The judge denied his request to present the evidence, and Harkov fought his client's conviction, based on the judge's refusal, all the way to the state Supreme Court.

Finally, in April 1999, six years into Cromedy's sentence, the New Jersey Supreme Court ordered a new trial and mandated special jury instructions for cross-racial identifications. That December, Cromedy was freed after DNA tests showed that his semen didn't match the rapist's.

In light of the court instructions, the state attorney general's office started to investigate its eyewitness-identification procedures and came across the report recently released by the U.S. Justice Department. New Jersey is unique among the 50 states in that the attorney general's office has control over every police department in the state. So when New Jersey decided to reform eyewitness- identification procedures along the lines of the Justice Department report, police departments had no choice but to go along.

While the Justice Department report discussed some of Gary Wells' research, it did not go as far as to recommend two of his most prominent research findings. Independently consulting with Wells, however, New Jersey decided to adopt both of them.

The first is a sequential presentation of lineup photos, rather than the traditional method of showing witnesses all the photos at once. "If the witness looks at a traditional simultaneous lineup, what essentially happens is the person does a little comparison shopping," says Nancy Steblay, a psychology professor at Augsburg College in Minneapolis. "The pictures don't usually look exactly like the memory, so the witness looks to see which one's closest."

The second is what Wells calls a "double-blind" presentation of the lineup photos. Wells found that when police officers know who the suspect is, they often--consciously or unconsciously--reward witnesses when they pick the suspect out of the lineup. Wells recommends that the person administering the lineup should be someone without knowledge of the case or the "correct" suspect.

As part of the reforms, police officers are required to tell witnesses that the perpetrator might not be in the lineup at all. They must also ask them how confident they are when they make an identification.


As information about New Jersey's efforts trickled through law enforcement circles, other states and localities also began to consider changing their procedures. Sometimes, the changes were spurred by specific exonerations. Wisconsin, for instance, changed its eyewitness-identification procedures last year in part because of the case of Steven Avery, who was exonerated by DNA evidence in 2003 after he spent 18 years in jail for a brutal rape. (He was, however, recently accused of murdering a woman.)

In North Carolina, Ronald Cotton was exonerated in 1995 after spending more than 10 years in jail for the rape of Jennifer Thompson. Since Cotton's exoneration, he and Thompson--who misidentified him numerous times--have improbably become friends. Thompson has spoken to law enforcement groups all over the country advocating eyewitness identification reform, including North Carolina's Actual Innocence Commission, which has recommended that the state change its procedures.

In Hennepin County, Minnesota, county attorney Amy Klobuchar heard about eyewitness-identification reform and decided to implement a pilot program. Klobuchar, who is now running for the U.S. Senate, implemented the program in four police departments ranging from Minneapolis (population 380,000) to New Hope (population 21,000).

At times, it was a tough sell. "Change is never easy in government," Klobuchar says, diplomatically. "I know that our chiefs had to do a lot of encouraging with their troops, the front-line officers."

She notes that a presentation with Wells and Thompson in front of 500 officers was remarkably effective in convincing police officers. After Thompson told her story, Klobuchar says, many of the grizzled cops had tears in their eyes.

To Joy Rikala, police chief in Minnetonka, which participated in the Hennepin pilot program, many of the reforms instinctively made sense. In her 30 years in law enforcement, "I did see many flaws in witnesses who felt like they were trying to be people pleasers, felt that they had to select someone," she says. "Now people are actually comparing the one photo in front of them to what's in their mind, not going through process of elimination."

Steblay, the professor at Augsburg College, recently conducted a study of the implementation of eyewitness reform in Hennepin County. She found that the number of mistaken identifications of "fillers," or lineup participants who were not suspects in the crime, was lower than those in experimental results with simultaneous identifications.

A study released in late March from an Illinois pilot program showed different results, however. That study found filler identifications higher in double-blind, sequential identifications than in non-double- blind, simultaneous identifications. Wells questions the methodology of that study, saying that because the study switched two variables at once, it is impossible to draw conclusions on the effect of either variable.

In jurisdictions nationwide where the new procedures have been implemented, police departments have struggled at times with the double-blind component. Sometimes in small police departments, no additional officer is available to go through a lineup with a witness. Other times, a case will be big enough that every officer is aware of the suspects.

Law enforcement officials in Wisconsin came up with a relatively simple solution to the problem: When no "blind" officer is available, they put the pictures inside folders so the officer can't see which picture is which. Other jurisdictions, such as Hennepin County and Virginia, are developing computerized identifications.

Despite the spurt of activity on eyewitness-identification reforms in the last few years, Wells estimates that only about 10 percent of the U.S. population now lives in places that practice sequential and double-blind identifications.

Wells notes that there are numerous obstacles to the adoption of new eyewitness-identification reforms, from police and prosecutors set in their ways, to the sheer number of police departments nationwide. Pennsylvania alone has about 14,000 independent law enforcement agencies.

But as county prosecutors and police chiefs hear results from other jurisdictions and listen to researchers and advocates such as Wells and Thompson, things are slowly changing. "All we have really is the persuasion process, the sharing of the science, to convince them it's the right thing to do," Wells says. "When you really have a chance to lay all this out, jurisdictions get on board. I'm optimistic that the pace is going to pick up now that there's enough success out there that they will sort of snowball. It's snowballing now, but we have to push the snowball up the hill."


Since the 1970s, Gary Wells, who is currently a professor at Iowa State, has performed elaborate experiments to test eyewitness memory. In a room full of waiting subjects, he has had "perpetrators" snatch purses and flee the scene--and then tested to see how well his subjects could identify the suspect. He has also done similar experiments using videotape, showing subjects a man planting a bomb on a roof, for example, and then checking their memory.

Overall, the results haven't been pretty. "Mistaken identifications happen rather easily, much more commonly than people think, and it's pretty easy to demonstrate," Wells says. "People pick the wrong person, and they often do so with a lot of confidence."

After his experiments were published in top scientific journals and referenced in textbooks, Wells was interested in moving beyond the academic world to put his ideas to use in law enforcement. But practical applications proved nearly impossible. "Frankly," he says, "it was hard for me to get into a local police station."

That all changed when DNA started exonerating people convicted of vicious crimes who had been identified by their victims. At the highest levels of government, officials started looking for ways to make eyewitness testimony more reliable. All of a sudden, Wells could gain entry to places a lot bigger than his neighborhood police department. "Once the DNA exonerations started coming in, suddenly I'm called by the attorney general of the U.S., Janet Reno, and brought to Washington," he says.

Wells' trip to Washington helped the Justice Department write an official document examining eyewitness testimony. But the 1999 Justice Department report, "Eyewitness Evidence: A Guideline for Law Enforcement," initially got the same reception as Wells' previous research: Not a single police department made substantive policy changes--until a dramatic misidentification case in New Jersey.