Revise Eyewitness Procedures to Convict Guilty Only: View

(Update in first and ninth paragraph incorporating Troy Davis case.)

If there is cause to doubt the justice of the execution this week of Troy Davis by the state of Georgia, it stems in part from the fact that seven witnesses recanted after Davis’ trial. A study released Monday provides additional confirmation that rules on eyewitness evidence in criminal trials must change.

The study, co-sponsored by the Innocence Project, the American Judicature Society and the Police Foundation, follows a unanimous decision last month by the New Jersey Supreme Court, which ruled that the state’s procedures for handling eyewitness evidence must be upgraded to reflect the findings of decades of social science research.

As Chief Justice Stuart Rabner wrote, current rules on eyewitness evidence provide neither “adequate measure for reliability,” nor a sufficient deterrent to “inappropriate police conduct.”

The failings have been exposed through some 2,000 studies of eyewitness identifications over the past three decades. As investigations of human psychology, the studies are fascinating; as assessments of American justice, however, they are deeply disturbing, documenting an extraordinarily high prevalence of error in eyewitness identifications in both lab experiments and cases in the field. As a report from New Jersey’s court-appointed special master concluded, “At a minimum, almost one third of witnesses who make identifications are wrong.”

If that seems implausibly high, consider this: Three-quarters of convictions overturned on the basis of DNA evidence involved eyewitness identifications. In more than a third of those cases, multiple eyewitnesses identified the same innocent suspect. (There is no way of knowing how high the rate of eyewitness error might be in cases where DNA is not a factor, though there is no reason to think it is lower.)

Witness Stress

Some witness errors result from faulty memories that have been further clouded by the stress that often accompanies seeing a crime. Witnesses are especially prone to error when identifying a suspect of a different race. Other misidentifications are a product of everyday human frailty combined with substandard -- yet widespread -- police procedures. This is where the New Jersey court has rendered an important service, demanding that police adopt higher standards to maintain the credibility of eyewitness evidence.

For example, police lineups should be supervised by officers who are not involved in the investigation. Instructions for witnesses should be purposefully neutral, because research has shown that witnesses can be swayed by even unconscious, nonverbal cues from officers.

The study released Monday makes a strong case for organizing suspect lineups sequentially, so witnesses are shown photos one by one. Such an arrangement discourages witnesses from selecting a suspect on the basis that he or she simply looks more like the perpetrator than any of the others. (In the Davis case, police in Savannah showed Davis’s photo to some witnesses even before the lineup and re-enacted the crime in front of four witnesses.)

Witnesses should be instructed that a suspect may or may not be in a given lineup and that they shouldn’t feel compelled to make a choice. Officers should avoid giving any kind of confirmatory feedback after a selection is made, in order to avoid transforming an uncertain witness into a highly confident

Jury Instructions

Yet the changes have been hard to put in place given the decentralization of U.S. law enforcement, which includes 16,000 independent agencies.

In California, courts have used their leverage to prod some police departments to adopt better procedures. Departments that use substandard procedures risk a California trial judge informing a jury of just how unreliable eyewitness evidence obtained under sub-optimal circumstances can be. If judges go too far with such instructions, they can prejudice a jury against legitimate evidence. But the risk of poor police procedures leading to contaminated evidence is at least as great.

By producing statewide guidelines for handling eyewitness evidence, New Jersey has issued a challenge that should be answered nationwide. Police departments across the U.S. must now adopt the necessary changes. These sensible measures are not efforts to hamstring police and prosecutors; they are designed to prevent the conviction of the innocent, the effect of which is to let the truly guilty go free.

To contact the senior editor responsible for Bloomberg View’s editorials: David Shipley at