Garrett Helps Overhaul Virginia's Model Policy for Police Line-ups, Eyewitness Identification
Marvin Anderson, who served 15 years for rape before he was exonerated by DNA testing in 2001, was convicted in Hanover County, Va., in 1982 on the basis of the victim's eyewitness testimony.
The victim had picked Anderson out from a photo array, in which his was the only color photo, as well as the only photo from an employer I.D. He was also the only person from the photo array who was then included in a lineup.
The investigators never instructed the victim that her attacker might not be present in the array or line-up. As it turned out, a photo of the real perpetrator was included in the array but was not identified by the victim.
Including Anderson's case, 11 of the 14 post-conviction DNA exonerations that have occurred to date in Virginia involved eyewitness misidentification.
Now, to minimize the chances of future wrongful convictions, Virginia is rolling out an overhauled model policy for police lineups and eyewitness identification that was crafted with the help of University of Virginia School of Law Professor Brandon Garrett.
"We've known for some time that eyewitness misidentifications are an important cause of wrongful convictions," said Garrett, a leading expert on wrongful convictions. "Virginia DNA exonerations, like the DNA exonerations around the country, typically involve rape cases, in which the victim misidentified the perpetrator. Decades of research show how unsound lineup procedures can cause such tragic errors."
Garrett's 2011 book, "Convicting the Innocent: Where Criminal Prosecutions Go Wrong," examined the cases of the first 250 people exonerated across the country by post-conviction DNA testing. Of the 250 cases, 190 — or 76 percent — involved eyewitness evidence, he found.
The Virginia State Crime Commission began efforts to improve local police lineup procedures in 2005 after having reviewed a growing body of social science research, listening to experts and holding hearings on best practices.
The commission asked the Department of Criminal Justice Services to draft model lineup procedures but did not recommend legislation that would require police departments to follow them. Instead, the commission recommended that the General Assembly enact a law requiring police departments to have written procedures, without mandating what those procedures should be.
"As these exonerations started happening in Virginia, and also around the country, there was a growing awareness that we need to make sure that lineups are being done right," Garrett said. "There was no real guidance from Virginia on how to do that, and there was some indications that many police departments in Virginia didn't have any written guidelines on how to do lineups. They were just doing line-ups based on whatever process some senior detective had passed on informally."
Five years later, the commission conducted a follow-up survey to see what policies the police departments had established. The results, Garrett said, were "shocking."
"Although some big departments had good lineup procedures — like in Roanoke and Charlottesville — many others didn't," he said. "They found that about a third of departments still had no written policy at all."
Even more troubling, Garrett said, only a tiny number of police departments had procedures requiring blind line-ups, in which the investigator does not know the suspect's identity, thereby eliminating the chance of verbal or non-verbal cues to the eyewitness.
In many cases, he added, the departments' new written procedures may have been more error-prone than what they had before, calling for sequential line-ups that were not double-blind, meaning that the investigator knows the suspect's identity as he shows the photos one at a time to the eyewitness.
"When you show pictures one at a time, that prolongs the encounter and there's more of an opportunity for unintended suggestion to affect the identification," he said. "Because you're laying down one photo at a time, and in effect asking a witness, 'Is this him? Is this him?' There's more of an opportunity for the detective to give even inadvertent cues."
Following its survey, the crime commission decided to recommend that DCJS revise its model policy for police lineups and eyewitness identification to clarify the best practices. Garrett served as a member of the commission's Law Enforcement Work Group that examined the issue.
The new policy, which was published in November, is a "detailed, practical, useful document" for police departments, Garrett said.
It explains why it is important for police to conduct double-blind eyewitness identifications. It also addresses a top concern of small police departments worried that they lack the manpower to spare an investigator for a double-blind identification.
"The solution is the folder system," Garrett said. "You just put the pictures, one in each folder, and shuffle the folders with a couple extra folders that are blank. And the witness can open the folders and look at the pictures without the detective being able to see what they're looking at. Unless you're looking over their shoulder, the detective can't see which photo the witness is looking at. It's a really simple way to make a line-up both sequential, and double-blind — two of the main reforms."
It is important for double-blind lineups to also be sequential, Garrett said, rather than in a traditional "six-pack" photo array.
"The problem with the six pack is that social scientists have found a comparison shopping effect," he said. "Even if you tell them the suspect may or may not be present, like you're supposed to, [the eyewitness] may still assume the suspect is there and pick the one who looks the most like the person who did it."
Garrett's expertise was key as the agency crafted the new model policy, said Gary Dillon, manager of the Department of Criminal Justice Services accreditation center. Dillon's role is to maintain the model policies and bring Virginia's recommended lineup procedures in line with contemporary practices.
"Professor Garrett was extremely helpful in putting this policy together," Dillon said, calling him "instrumental in reviewing the drafts all the way through final approval at DCJS."
Virginia's police accreditation standards will also require departments to have a written policy for lineups and eyewitness identifications, and it will reference the state's new model policy, Garrett said. DCJS is developing a detailed curriculum for training to explain best practices and the new model policy to law enforcement statewide.
Garrett gave a training talk on eyewitness procedures and the state's new model policy to the Charlottesville Police Department in January.
"We're fortunate to have Professor Garrett as a resource to our police department and broader criminal justice community," Charlottesville Police Chief Timothy Longo said. "His willingness to share his research in the area of best practices related to eyewitness identification has influenced change in police policy, practice and training in this critical area."
It is in the interest of law enforcement to implement best practices for lineups and eyewitness identifications, Garrett said. Doing so minimizes the chance that an eyewitness will choose a "filler," or someone in a lineup who is not the actual suspect. Picking too many fillers can hurt a witness' credibility at trial, and damage the case.
"I'm not sure to what degree a lot of departments realize how often poor lineup procedures harm them," Garrett said. "Even if it's not one of these rare wrongful convictions that comes to light years later, departments all know that a lot of fillers get picked out by eyewitnesses. Some studies suggest it happens in up to a third of lineups.
"Whenever a filler gets picked up, that harms the investigation," he continued. "That makes some departments reluctant to even use eyewitnesses because if they pick a filler out, they don't seem very credible. Using these improved techniques can really reduce filler identification rates because they cut down on guessing. Police don't want witnesses to just be guessing."