New policy urged for police lineups

Techniques address false confidence

FAYETTEVILLE -- It's not just memories that can be false. There's also false confidence, said James Lampinen, a distinguished professor of psychological science at the University of Arkansas at Fayetteville.

He began his career examining how people can be overly confident in faulty memories: "How is it that they can be convincing to people, even though they're false?" Lampinen said.

He soon began applying his research to eyewitness identification within the criminal justice system. Mistakes from eyewitnesses have contributed to more than seven out of 10 convictions overturned because of DNA evidence, according to the Innocence Project, a national legal and public policy organization that fights wrongful convictions.

This year Lampinen began helping with a model policy from the Arkansas Association of Chiefs of Police. The policy outlines the best ways for police to conduct lineups for eyewitnesses. Lampinen, a certified law enforcement trainer, read about the effort after the association's president, J.R. Wilson, wrote an opinion piece explaining the importance of eyewitness procedures.

Now, Lampinen -- who co-wrote a book in 2012 on eyewitness identification -- has partnered with the group and begun attending law enforcement gatherings to present information about the model policy.

Wilson called Lampinen invaluable to the group's effort. The group said it's unknown how many departments have a policy for lineups, but Wilson spoke about the need for wider adoption.

"I am confident we will be successful in getting law enforcement agencies to voluntary adopt this critical policy," Wilson, Hope's chief of police, wrote in an email. "However, if we are not, [the police chiefs association] will support legislative efforts to accomplish this critical goal."

Out of the millions of cases investigated by police every year, only in a small percentage will police utilize these lineups, according to a 2014 report from the National Academy of Sciences.

For those cases, police most commonly use a photo array, according to the academy report. A number of photographs, typically six to nine, can be shown one at a time to a witness. Or, a matrix can be created -- often six holes cut in a folder, with the photos then taped to the back. This allows the photographs to be displayed together.

But aside from the defining characteristics of a photo lineup, questions remain. Should the officer with the most knowledge about the case conduct the lineup? Besides the suspect, how are the other photos chosen? What instructions should be given to a witnesses before they begin looking at the photos?

Confidence turns out to be a key consideration, Lampinen said.

A witness may first pick someone hesitantly. But if a detective "then says, 'Great, that's our guy,' suddenly the witness's confidence in their decision is inflated," Lampinen said.

Recent guidelines developed by various groups, including the National Academy of Sciences and the Arkansas police chiefs association, call for "double-blind" procedures. A photo array that includes the suspect should be presented by a police officer unfamiliar with the actual suspect's identity, guidelines state.

Otherwise, "the interviewer who knows who the suspect is may inadvertently cue the witness as to who they think the suspect is, and of course you want the witness to be relying on their own memory," Lampinen said.

When it comes to choosing "filler" photographs, Lampinen said the best practice is to choose photos of people who closely match the description given by a witness. He said it's also important to not have a suspect "stand out as being different" in the photo presentation, such as with a different background.

Lampinen described how various guidelines have tackled the idea of what is best to say to a witness.

"There's a couple of decades of research on how you instruct witnesses and what impact that has on their accuracy," Lampinen said.

Studies show that simply asking someone to pick out "who robbed you," for example, often leads to mistaken identifications, Lampinen said.

"The average, kind of across studies where that kind of instruction is given, is 65 percent of the time the witness will pick somebody" when it's known the suspect is not included in the array, Lampinen said. The studies vary in how they're done by researchers, but some involve elaborate simulations of crimes, Lampinen said.

By simply telling a witness to keep in mind that the perpetrator may or may not be present, that error rate drops to 40 percent in a photo array with no suspect present, Lampinen said. The model Arkansas policy includes this instruction.

If the rate seems high, Lampinen said it's important to consider that the instruction given to witnesses is "part of a package."

In "pristine" conditions with all best practices followed, the rate of witnesses picking a suspect out of a lineup of only innocents becomes much lower -- about 5 to 8 percent, Lampinen said.

Even in the best of conditions, when a lineup does contain the actual suspect, controlled studies have shown witnesses able to pick out the correct person "around 50 to 55 percent of the time," Lampinen said.

Confidence again comes into play when assessing the reliability of eyewitnesses' identifications, Lampinen said. Looking only at "those witnesses who are really confident," in the best conditions, about 35 correct suspect identifications will take place for every identification of a clearly innocent person, Lampinen said.

"That's still not perfect, but it's pretty good, given how fallible memory is," Lampinen said.

He praised the model Arkansas policy for stating that police should also record when a witness doesn't identify the suspect in a lineup.

"The Arkansas model policy does a really nice job of saying, listen, you, as an investigator, have a responsibility to record what the witness said, whether the witness picked the suspect, whether the witness rejected the lineup, whether the witness picked a 'filler,'" Lampinen said.

The Arkansas policy originated from work done around 2010 by the Indiana-based Legal and Liability Risk Management Institute, Wilson said.

Lampinen said he's submitted a few ideas for revisions. In promoting adoption of the policy, he said he's gotten positive feedback so far in talks to law enforcement leaders.

"The chiefs of police that have been in these audiences and come up and talk to me later, they talk about the fact that they recognize that witnesses aren't always accurate, and that it's really helpful to have an idea of what the scientific principles are to help make eyewitness identification more accurate," Lampinen said.

Metro on 09/05/2016

Upcoming Events