NEWARK, N.J. - One of the country's leading experts on police lineups says New Jersey's guidelines on eyewitness identifications of suspects are "pretty good."

But authorities can do better, Iowa State psychology professor Gary Wells said.

Wells addressed a panel yesterday conducing a New Jersey Supreme Court-mandated review into law enforcement agencies' eyewitness-identification practices.

He told the group how the state's guidelines don't account for all situations and research that shows how easily innocent people can be picked out of a lineup.

Wells said some studies show that more than 20 percent of eyewitnesses choose an innocent person in a lineup.

Many factors can influence eyewitness memory and it can be hard for investigators to tell whether a witness is lying, unaware, or mistaken - a serious problem for investigators.

Curbing mistakes

Police "find the mistaken eyewitness to be as credible as the one who was accurate," he said.

According to Wells, such things as whether a weapon was present during a crime or whether a suspect is biracial are factors known to make identifications less reliable.

He said police can help prevent wrongful convictions by giving specific instructions before a lineup, recording witness responses, and avoiding suggestive procedures, such as making sure case investigators don't conduct the lineup.

In 2001, New Jersey became one of the first states to establish lineup protocol guidelines designed to prevent mistaken suspect identifications.

They include recommendations that police use only sequential blind lineups - where mug shots are shown one at a time and the detectives displaying the photos don't know who the suspect is, which means they can't intentionally or accidentally tip off witnesses.

However, the guidelines don't address such things as "show-up" identifications, where police show up at a scene, eventually spot someone matching the suspect's description, and physically ask a witness to say on the spot whether that person is the suspect.

Experiments

"Once the witness has view of a photo or lineup, that later description of the perpetrator may be reflections of what they picked up from the photos or live lineups," rather than what they remember, Well said.

He said lineups should be seen as psychology experiments and should follow specific rules to exclude undue influence on the human memory.

That includes composite sketches, which he said are problematic because they don't often closely resemble suspects and because witnesses are more likely to render a false identification after having helped police create a composite sketch.

It will be up to retired state Appellate Division Judge Geoffrey Gaulkin to draft any recommendations for potential change in suspect-identification procedures. There is no deadline for the report.

The court will consider the nonbinding recommendation that could eventually serve other states as a legal blueprint for revamping identification protocols.

Defense lawyers argue that those guidelines now in place should become standard practice. They also argue the state should have the burden of showing that a witness identification has been made without suggestion.

Prosecutors say that while the guidelines should be adhered to, a deviation from them should not automatically mean that a witness identification can't be used.

Nationally, more than two-thirds of the 242 DNA exonerates released since 1989 were sent to prison based on witness misidentification. It's the most common element in a wrongful conviction, the center said.

Three of the five exoneration cases in New Jersey involved misidentification, according to the group.