On the evening of June 30, 1985, Virdeen Willis Jr., an off-duty official at a state prison, was drinking with two women in a bar on Chicago's South Side. As he and his companions left, someone approached and shot him fatally in the neck. Four days later, police arrested Steven Smith, a convicted killer who had served time in the facility where Willis worked. Smith denied any involvement, the police couldn't produce the murder weapon, and no physical evidence tied him to the crime. About all the prosecution had to offer was a witness, Debrah Caraway, who said she saw Smith shoot Willis.
The prosecution's case was far from ideal, and Caraway was one of its big liabilities. The day of the murder, she had been smoking crack. She was across the street when Willis was shot. Her boyfriend had been considered as a suspect before Smith was arrested. The evidence suggesting Smith wasn't the culprit was also strong. The two women who were standing next to Willis when he was shot couldn't identify Smith as the gunman. Several witnesses said Smith had left the bar with two friends before Willis walked out—not alone and afterward, as Caraway claimed. But its one witness apparently was all the state needed. Two different juries found Smith guilty, and both times he was sentenced to death. Today, however, Smith is not only alive but free: In 1999, the Illinois Supreme Court vacated his conviction for lack of evidence, barred a retrial, and ordered him released.
Since restoring capital punishment in 1977, Illinois has executed 12 inmates. During that period, 13 other death-row inmates have been exonerated. This amazing record of fallibility was what prompted Republican Gov. George Ryan to impose a moratorium on executions two years ago. He also appointed a commission to examine the Illinois system of capital punishment and to offer proposals for its reform. Among the recommendations made by the commission in a report released last month were a handful aimed at a problem that has contributed to several innocent men being sentenced to death: false identifications by real or alleged eyewitnesses. But as the commission took pains to note, this is a problem that plagues the entire criminal-justice system, not just death penalty cases.
False eyewitness identifications are certainly most conspicuous in erroneous capital convictions. The Center on Wrongful Convictions at the Northwestern University School of Law looked at 86 death-row inmates who were cleared and found that faulty eyewitness identifications played a role in more than half of them—making them "by far the most ubiquitous factor." More troubling, in 38 percent of these cases, eyewitness testimony was the sole evidence against the accused.
But this type of evidence is commonplace in all sorts of cases. The spread of DNA analysis has exonerated many convicted rapists who had been wrongly identified by victims. Of 108 post-conviction exonerations by DNA identified by the Innocence Project at the Benjamin N. Cardozo School of Law, some 97 involved sexual-assault and rape charges, and most of these convictions hinged on faulty eyewitness testimony. Iowa State University psychology professor Gary Wells, who has written extensively on the subject, says unfounded eyewitness identifications are the greatest single cause of wrongful convictions.
What makes eyewitness identifications especially mischievous is that they are not only often inaccurate, but usually they can be quite convincing to juries, regardless of their veracity. One of the most powerful forms of evidence a prosecutor can deploy is a person who was at the scene of the crime, who will point to the defendant in court and say, "That's the man who did it." Juries often believe such witnesses even when loud alarm bells are warning them away. After a woman was raped in her New Jersey apartment in 1992, the victim couldn't find her assailant in police photos. But eight months later, she saw McKinley Cromedy on the street and implicated him—even though she had passed him over in the original photo lineup. Fingerprints from her apartment didn't match his; neither did hairs or blood samples recovered by police. He was convicted anyway and sentenced to 60 years in prison, five of which he served before a DNA analysis cleared him.
Wrongful death sentences have come to light mainly because capital cases attract special scrutiny, and false rape convictions have emerged only because DNA analysis can irrefutably clear a defendant. But it's anyone's guess how many inmates are behind bars on robbery or assault charges because the victim or an eyewitness identified someone who wasn't actually there. Wells says that every year 77,000 people become criminal defendants after being identified by a crime victim in a police lineup.
Why are eyewitnesses so prone to errors? They may not get a very good look at the criminal, or they may have been focused on the gun or knife he was carrying rather than on his face. Cross-racial identifications are particularly unreliable because people are known to be less adept at distinguishing among members of other racial groups. Human memory is a malleable commodity, capable of being unconsciously reshaped to fit the needs of the moment. And crime witnesses are sometimes so eager to see the culprit punished that they may pin the tail on any available donkey.
In speaking on the subject, Gary Wells sometimes shows a videotape of a man on a roof, his face visible, apparently putting something in an air shaft. Wells informs his listeners that the man planted a bomb, shows them mug shots of six different men, then asks how many people in the audience think each of them was the guilty party. "When I'm done going through the list, virtually everyone in the room has raised their hand," he says. "Then I say: 'You're all wrong. It was none of them.' " Witnesses, according to Wells, are usually accurate in picking out the perpetrator if he is present in the lineup. "But if he's not there, they tend to pick somebody anyway," he says. In that situation, they'll tend to pick whomever looks most like the culprit.
Other errors are not entirely the identifiers' fault. One man who had seen a Chicago murder picked James Newsome out of a lineup only after two police detectives asked the witness to consider suspect No. 3. (Newsome was convicted and then, after he had served 15 years of a life sentence, cleared.) Even if police are not trying to rig the results, they may inadvertently give signals that nudge the witness toward the person they've arrested. And if the witness picks the "right" person, a cop or prosecutor may buttress his confidence by praising the choice. A witness who is tentative at the lineup stage may be absolutely certain by the time she takes the stand.
It would be too much to ask courts to do without relying at all on a type of evidence that can be very useful. Fortunately, the criminal justice system doesn't have to choose between abandoning eyewitness testimony and falsely convicting lots of innocent people. The Illinois death-penalty commission made several recommendations that would reduce the risk of mistakes. Police lineups, it said, should be administered by someone who doesn't know which member of the lineup is the "true" suspect, to avoid tip-offs. Witnesses should be shown each member one at a time, rather than all together, which studies indicate lowers the rate of false identifications without reducing the number of correct identifications. And before viewing the lineup, the witness should be informed that the actual criminal may not be in it.
These are not new ideas. The National Institute of Justice, an agency of the Justice Department, recommended changes along these same lines in a law-enforcement manual it published in 1999. The reforms are simple, requiring little in the way of money or trouble. But few police departments have adopted them even though mistaken identifications only ultimately help guilty people go free. New Jersey mandated these sorts of changes last year, but thus far it's the only state to have done so. Another antidote is admitting trial testimony from experts who can explain to juries why eyewitnesses can't always be believed. The federal courts generally allow the defense to call such experts, but few state courts have followed suit. In most courts, whether or not to allow such testimony is strictly at the whim of the judge.
The case against traditional eyewitness identification practices may sound like it was concocted by Groucho Marx, who once said: "Who are you going to believe? Me or your own eyes?" But it's time the criminal justice system recognized that, in some instances, the very last thing you can believe is what you saw.