A long-delayed study of Dallas Police Department witness identifications, two years in the making, could begin this month.

Traditionally, a detective shows a witness a photo lineup consisting of at least six pictures at one time. The study would incorporate a method known as the "sequential blind lineup" to help reduce misidentification of suspects. In this method, photos are shown to a witness one at a time by someone who does not know which picture is the suspect.

Dallas County has had 19 DNA-based exonerations – more than any jurisdiction in the nation since 2001 when a state law began allowing post-conviction DNA testing. All but one of the DNA exonerations were based on faulty eyewitness testimony, according to an investigation by The Dallas Morning News published in October.

Misidentifications have been cited as a key factor in an estimated 75 percent of the 220 wrongful convictions exposed by DNA testing nationwide since 1989.

The Dallas Police Department, the arresting agency in 13 of the 19 Dallas County exonerations, hopes the study will help determine the best method to keep witnesses from making the wrong choices in a photo lineup.

"We hope to determine what is the best practice and implement policies accordingly," Dallas Police Assistant Chief Ron Waldrop said.

Dallas would become just the eighth police agency in Texas – including Richardson, Lewisville and Haltom City, locally – to use the sequential blind lineup, according to a survey by nonprofit reform group The Justice Project.

Richardson police began using sequential blind lineups after the city had a DNA exoneration in April. Thomas McGowan was incorrectly selected from a simultaneous lineup after a woman was sexually assaulted. He served 22 years in prison after being wrongly convicted in the 1985 burglary and rape.

The Urban Institute, a nonpartisan economic and social policy research group based in the nation's capital, is in charge of the DPD study. Terry Dunworth, who is overseeing it, said the research should begin early in 2009 and continue through most of the year.

The delays, police said, were caused by funding issues and then waiting on computer equipment. Now, police say they are trying to resolve software problems and hope to simultaneously begin the study with Washington, D.C., police. Officials are trying to work out the protocols between the two departments to avoid any additional delays.

At least 800 lineups must be a part of the study, but only detectives in assaults and robberies will use the study's methods in their cases, said Lt. David Pughes, a Dallas officer working on eyewitness identification procedures.

Pughes said detectives will load six photos into laptops. One picture will be that of the suspect; the other five will be "filler" pictures of people with similar features.

A computer program will randomly determine whether witnesses should view the lineup simultaneously or sequentially. Then, the computer will decide whether the lineup will be conducted by the computer or a detective.

Gary Wells, an Iowa State University psychology professor and expert on eyewitness identification, said he developed the sequential blind method with a student in the mid-1980s. It was created to address the problem of a witness simply picking out the photo that looked most like the perpetrator when all pictures were viewed at the same time.

Wells said that published experiments in controlled settings typically show that viewing pictures sequentially results in fewer misidentifications when the perpetrator is not in the lineup.

"Presumably, this is because the sequential procedure prevents the witness from merely comparing one lineup member to another and picking the person who looks most like the perpetrator," he said.

One drawback of the sequential method, however, is that witnesses are less likely to select the right person when the culprit is in the lineup than when viewing the photos simultaneously, Wells said.

"This pattern of results is a classic trade-off in which reducing the rate of one type of error, mistaken identifications, can increase the rate of another type of error, failure to identify the culprit," he said.

Conflicting answers exist about whether the sequential blind method is best.

A 2006 analysis in Minnesota conducted by several police agencies found the method increased accuracy.

But an Illinois study the same year concluded the sequential method caused the wrong person to be picked more often than viewing lineups simultaneously. Experts, including Wells, say the Illinois study was flawed because the lineups were not conducted uniformly or randomly. He explained that simultaneous photo lineups were always conducted by detectives and sequential lineups were conducted by an unbiased administrator."The Dallas study," Wells said, "will be an attempt to do correctly the experiment that was done incorrectly by the Chicago Police Department."