It is one of the paradoxes of criminal justice: Eyewitness testimony isn't really the gold standard of proof that years of courtroom dramas have taught TV viewers to expect.
Faulty identification, for instance, played a role in the convictions of more than half of the death-row inmates exonerated since the death penalty was reintroduced, according to a 2001 study by the Center on Wrongful Convictions at Northwestern University in Chicago.
In 1999, the US Justice Department issued a guide calling for major changes in the way police ask eyewitnesses to identify criminal suspects. The message: the traditional lineup could be improved.
With more than 19,000 autonomous law-enforcement agencies spread throughout the country, however, introducing any wholesale procedural changes is usually difficult. But in New Jersey, the attorney general can unilaterally order police departments to change the way they operate.
So the Garden State, which introduced new procedures last October, is emerging as a proving ground for the new approach. The initial response from prosecutors is positive.
Under the old system, a row of suspects would stare blankly ahead as a witness tried to pick out the perpetrator. Or a crime victim would sift through a string of mug shots arrayed on a table.
Now, witnesses look at only a single suspect or mug shot at a time. And, where possible, the lineup is conducted by an officer not involved in the investigation; this helps ensure that police don't pressure a witness, subtly or otherwise, into fingering their own prime suspect.
In Dover Township, where a police force of 144 officers patrols a suburban community, Chief Michael Mastranarde says he didn't think his department's lineup procedures needed any fixing.
But a presentation last year for New Jersey police chiefs by Gary Wells, a professor of psychology at Iowa State University, helped convince Chief Mastranarde that the changes made sense. Wells's research found that witnesses confronted with a lineup where the actual culprit isn't present often pick the person who looks most like that culprit. When witnesses look at only one picture or witness at a time, however, "they have to dig deeper and make an absolute decision," Wells says.
Research also shows that the way a detective behaves during a lineup often determines how sure witnesses are about their selections.
Dover Township detective David McCallum says he was always careful to avoid giving any verbal cues or body language that might have led witnesses to an answer. But Wells says even the best detective may unwittingly send cues.
Most lineups are conducted using mug shots rather than live people. In fact, Lt. Steven Henry, who commands the department's detective bureau, says he hasn't participated in a single live lineup during his entire career as a detective.
Most identifications here occur in a windowless interrogation room. An officer not involved in the investigation who may be just a patrol officer sticks strictly to a script.
The officer tells the witness, "You should not conclude that the person who committed the crime is in the group merely because a group of photographs is being shown to you." Then six photos are shown, one at a time, as the officer looks on silently. Witnesses can change their selection after looking at all six pictures, but the officer is required to write down all responses and note anything the witness changes his or her mind about. Officers shuffle the pictures if a witness wants another look. "For [the witness], it's better," says Detective McCallum. "It either is or isn't the suspect."
So far, detectives say they don't notice a difference in the number of identifications under the new system. But prosecutors say the new system certainly will produce identifications that will be subject to fewer challenges by defense attorneys during trials.
Technology is also helping improve lineups. The Dover Township police department has purchased software that picks photos out of the database with similar characteristics. Detectives will no longer have to go through piles of pictures and codes to select photos to match the actual suspect.
Eventually, technology could remove officers from the lineup process altogether, says Ronald Fisher, a psychology professor at Florida International University who has studied lineups. Fisher says outside influences could be eliminated if a witness viewed pictures on a computer that could record witness behavior, including time taken to pick out a suspect.