Posted Jun 11, 2007 11:07 pm CDT
For years, social scientists have been telling anyone who would listen how to improve that age-old staple of police work, the lineup.
Showing a witness an array of possible suspects one at a time rather than all at once will reduce the chances that an innocent person will be falsely accused of a crime, researchers say. They say the sequential lineup helps ensure that the witness doesn’t by comparison just pick the person who most closely resembles the suspect.
And having the lineup administered by an official who doesn’t know who the real suspect is—called a double-blind test—will eliminate the possibility that the administrator will inadvertently steer the witness in the preferred direction.
Now social scientists are fighting a new battle, after the release of a study that suggests the old-fashioned method is superior. They are pointing to flaws in the research and suggesting the issue is far from settled.
The traditional lineup could certainly use some improvement. Mistaken identification is the leading cause of wrongful convictions, accounting for more innocent people being sent to prison than all other causes combined, according to the Innocence Project, the New York City-based legal clinic that works to free the innocent with DNA evidence.
But law enforcement officials had been slow to act on the experts’ advice. In 2001, New Jersey, with some fanfare, adopted new guidelines incorporating many of the recommended changes for the conduct of police lineups statewide.
New Jersey is still doing its lineups that way, but it has not collected any data or conducted any analysis of the results. A handful of local jurisdictions have since followed suit.
But the majority of the nation’s nearly 15,000 law enforcement agencies still do lineups the way they always have: The witness views an array of six or so possible suspects, either in person or through photos, simultaneously. And the process is almost always administered by an official—usually the lead detective in the case—who knows who the real suspect is.
Experts say there are many reasons their recommended changes haven’t been more widely implemented. One is a general lack of knowledge in the law enforcement community about what research had previously shown: Double-blind sequential lineups are more reliable than non-double-blind simultaneous ones.
Another is that nobody likes change. A third is the apparently widespread belief in law enforcement circles that acknowledging the need for change could undermine confidence in the convictions they’ve already obtained under existing procedures.
“If prosecutors everywhere suddenly started saying, ‘OK, we’ll go ahead and implement all of these proposed reforms,’ it would be sort of a tacit admission that there’s something wrong with the way they’ve been doing it all of these years,” says Iowa State University psychology professor Gary Wells, a pioneer in the field of eyewitness identification and a leading proponent of the proposed changes.
Paul Logli, Winnebago County, Ill., state’s attorney and outgoing president of the National District Attorneys Association, says he doesn’t know of any prosecutors who are opposed to changes that have been shown to work. But, he says, many prosecutors, including him, aren’t convinced that the changes being advocated by Wells and others represent an improvement over the way lineups have traditionally been done.
“The academics are the ones who are pushing these changes,” he says. “They’re not being pushed by police or prosecutors or anybody in the field.”
Doing lineups sequentially, Logli adds, seems contrary to human nature. “It’s just my opinion,” he says, “but I believe a witness can make a better identification looking at people simultaneously rather than in sequence.”
The results of the new study, purportedly the first to compare the two lineup procedures side by side in the real world, probably won’t change any minds.
The study, conducted by the Illinois state police at the behest of the state legislature, found that the sequential double-blind procedure recommended by the experts produced nearly 15 percent fewer correct suspect identifications than the simultaneous method did.
And when the witness to a sequential lineup did identify a suspect, the study showed, he or she was more than three times as likely to pick the wrong person as a witness to a simultaneous lineup. Witnesses also declined to identify a suspect in 47 percent of the sequential lineups, compared with 38 percent of the time in the simultaneous ones.
Sheri Mecklenburg, general counsel to the superintendent of the Chicago Police Department and director of the study, says she was as surprised by the results as anyone.
“To be honest, I thought the data would show what the experts said it would,” she says. “Instead, we found just the opposite.” Mecklenburg also says the findings call into question the validity of the classroom experiments on which much of the research in the field is based.
“It tells me we shouldn’t be shaping policy based on academic data that may not reflect what is actually happening in the real world,” she says.
However, critics contend the study, though well-intentioned, is fatally flawed. They say the design doesn’t permit them to draw conclusions because the sequential lineups were all conducted double-blind while the simultaneous lineups were not.
“One could argue that sequential lineups have not been proven to be superior to simultaneous lineups,” says Wells, who has published a critique of the Illinois study on his Web site. “But you’d never know whether that’s true from this study.”
Experts are also troubled by the unusually low rate of false identifications the study reported for simultaneous lineups, which they note is only a fraction of the rates found in other field studies using real witnesses. That could mean that the administrators, who knew who the real suspect was, were inadvertently influencing the witnesses’ decisions.
It could also mean that the fillers, or foils, looked nothing like the real suspect. Or it could just mean that the administrator didn’t record the fact that the witness picked the wrong suspect.
In the absence of any analysis of the data for the presence of such problems in the Illinois study, Wells says, none of those possibilities can be ruled out—either alone or in combination.
Mecklenburg, who appeared stung by the amount of criticism the study has received from Wells and others, has published an addendum to her report on the study, both of which are posted on the Chicago Police Department’s Web site.
In the addendum, she points out that the Illinois study was designed to compare the sequential double-blind procedure with current practice, not with some hypothetical, simultaneous double-blind procedure that is not being widely used.
She also takes issue with the critics’ suggestion that administrators who know the real suspect might inadvertently be influencing the witnesses’ identifications.
“There is no empirical evidence anywhere to show that real witnesses with real motivations in real cases would either notice or be willing to follow inadvertent police cues during a lineup,” she says.
Mecklenburg also says lineup administrators in Illinois are required by law to instruct witnesses that the real suspect may not be in the lineup, that they need not make any identification, and that they should not assume the administrator knows who the real suspect is—all of which should reduce the chances that the witness would make a false identification.
Logli notes that he’s no scientist but says the fact that the Illinois State Police performed the study lends credibility to it. And, he says, the results seem to suggest that simultaneous lineups are more effective than sequential ones. “The bottom line is the sequential lineups produced more false identifications and a reduction in the number of correct identifications than the simultaneous ones,” he says.
Wells of Iowa State University readily admits that sequential lineups are controversial. And he says he’s all in favor of new field studies—some of which are now in the works—that will provide a proper comparison of the two procedures.
But Wells says it would be a real shame if the controversy over sequential lineups prevented law enforcement agencies from implementing other, less controversial changes that have been shown to improve the process, such as double-blind testing; ensuring that the fillers closely resemble the suspect; and recording a witness’s level of certainty at the time he or she makes an identification.
“One of the most unfortunate things about this study is that it gives some jurisdictions an excuse not to do anything until the debate over sequential lineups is resolved,” he says. “It sort of casts a shadow over the whole idea that any reforms are necessary.”