After nearly 38 years, on Jan. 30, Malcolm Alexander walked away from a place he never should have been to begin with: the Louisiana State Penitentiary.
Earlier that day, a Jefferson Parish judge had tossed out Mr. Alexander’s conviction, partly in response to new DNA evidence proving that Alexander could not have been the armed man who in 1979 raped a woman in her antique shop bathroom.
Twenty-one when he was sentenced to life without parole, Alexander had been the victim of what the Innocence Project described as “a deeply flawed, unreliable identification procedure.”
The Innocence Project, whose advocacy led to Alexander’s release at age 58, said the flaws included interactions with police officers that strengthened the victim’s initial “tentative” identification of Alexander in a police lineup into one of certainty by the time she testified at the trial nearly a year later.
Alexander’s wrongful conviction helped convince Louisiana lawmakers to join a growing list of states in taking steps to prevent eyewitnesses from being swayed by police or other influences. Among the changes Louisiana enacted earlier this year is requiring police to use “double-blind” lineups, in which the officer conducting the lineup doesn’t know who the suspect is. Other changes are intended to reduce the chance that witnesses will make identifications through guesswork or to please police investigators.
Although research has long demonstrated the unreliability of eyewitness identifications, police agencies largely resisted making changes until DNA exonerations over the past three decades presented incontrovertible proof that the wrong person had been convicted, as was the case with Alexander.
The new policies, which have been adopted by police in half the states, largely grew out of research showing that traditional police practices and cues often induced witnesses to identify the wrong suspect.
While some police organizations resisted the new policies, calling them impractical and possibly costly, many police departments, including those in Minneapolis and San Diego, adopted new eyewitness procedures without waiting for legislation.
“There’s no question, we just didn’t know any better back then,” said William Brooks, chief of the Norwood Police Department in Massachusetts. Mr Brooks is a member of the board of the International Association of Chiefs of Police and has traveled extensively around the nation to speak in favor of the changes.
The new policies do not address a long-noted weakness in eyewitness identification: Eyewitnesses are more likely to misidentify someone of a race other than their own. Though judges often instruct juries about the problem with cross-racial eyewitness identifications, no one has figured out how to solve the problem.
“The goal,” said Rebecca Brown, policy director at the Innocence Project, “isn’t to pick someone, but to pick the right person.”
Researchers have been chronicling the inaccuracy of eyewitness identification at least as far back as the 1930s. One theme in the scholarship is that memory is not fixed. In the words of Brian Cutler, a prominent researcher into eyewitness identification at the University of Ontario Institute of Technology, memory is “not like a tape recording.”
Witnesses's memories can be influenced by later stimuli, such as police body language during the lineup.
University of Alberta psychologist Gary Wells published a seminal paper in 1978, showing the ways in which witness identification can be influenced by such factors as the way a police lineup is conducted, the instructions given to a witness, and the behavior of officers in front of a witness.
At the time, Dr. Wells and fellow researchers in the field tried to get the attention of law enforcement. It didn’t go well.
“We were largely ignored in the legal system,” said Wells, who is now at Iowa State University. Then, Wells said, it was the rare police department that even had formal procedures for conducting eyewitness identifications. Lineups were conducted differently from police department to police department, and even from investigator to investigator within the same department.
Even as scholarship continued to hammer away at the fallacy of memory in identification, police procedures remained largely the same over the next decade. The first US exoneration by DNA evidence came in 1989, and the Innocence Project was founded in 1992. It has been especially prominent in using DNA to exonerate hundreds of people serving lengthy prison sentences.
The Innocence Project found that eyewitness misidentifications were a factor in 71 percent of the more than 350 wrongful convictions overturned by post-conviction DNA.
Another organization that tracks wrongful convictions, the National Registry of Exonerations, which is run jointly by the University of California at Irvine and law schools at the University of Michigan and Michigan State University, found that mistaken identifications were a factor in 29 percent of 2,245 exonerations since 1989.
The registry includes cases that do not involve DNA evidence and surveys a broader category of crimes than the Innocence Project, whose list of exonerations is heavily weighted toward homicides and sex crimes.
Janet Reno, the US attorney general under former President Bill Clinton, summoned Wells to Washington in the late 1990s to discuss his work. She appointed a task force that recommended a series of changes to police identification procedures. Wells traveled the country to police forces trying to sell the reforms. “It had almost no effect on anyone,” he said.
The problem, Wells said, is that most US police agencies are local – independent of state and federal agencies. Few are interested in being told how to run their criminal investigations.
New Jersey is different. Unlike other states, it has a unified, integrated law enforcement system. The state attorney general is also the chief law enforcement officer. Two decades ago, then-Attorney General John Farmer was upset by the revelation, based on DNA evidence, that a New Jersey man was wrongfully convicted of a 1992 rape after the victim misidentified him as her assailant.
In 2001, Mr. Farmer ordered police departments in the state to adopt new procedures for conducting lineups. Those requirements mirrored recommendations from the Reno commission, the National Academy of Sciences, and the International Association of Chiefs of Police.
After New Jersey, progress by the states in adopting the reforms was fitful but steady. The pace has picked up considerably since 2014, when the National Academy of Sciences released a report supporting the new policies, and the Innocence Project intensified its advocacy of the changes. A dozen states have adopted the new policies in the last three years. Police in half the states now have the new policies in place.
The California House sent a bill to the Senate earlier this year. Iowa and New Mexico are expected to take up legislation next year.
“The reality of witness identification is that it is one of the least-reliable pieces of evidence, and yet we put great weight on it,” said state Sen. Scott Wiener, a Democrat and sponsor of the California bill. “We’ve designed a system to encourage people to make inaccurate identifications, and we should have a system that does everything in its power to get accuracy.”
Louisiana state Sen. Wesley Bishop, a Democrat and a criminal lawyer, said he agreed to sponsor the bill in his state after he was approached by the Innocence Project in New Orleans before the 2018 legislative session. He said he was also prompted by the exoneration of Alexander and others who were wrongly convicted in part because of misidentifications.
Senator Bishop recalled in an interview that his allies at the Innocence Project told him they weren’t optimistic about passage, anticipating opposition from law enforcement. But Bishop had worked closely with police and prosecutors in the past, he said, and in the end, the Louisiana Sheriffs’ Association and the Louisiana District Attorneys Association supported the bill without changes.
The key change is adopting double-blind lineups, whether live or by photos. If the witness understands the officer doesn’t know who the suspect is, Wells said, the witness won’t try to read cues from the officer’s behavior, demeanor, or words.
The witness should be instructed that the perpetrator might not be in the lineup. Ontario’s Dr. Cutler said that instruction is intended to relieve the witness of feeling pressured to make an identification.
The lineup should include no more than one suspect and at least five others. The so-called “fillers” should bear some resemblance to the perpetrator as described by the witness.
The officers presenting the lineup should record the level of confidence with which the witness makes an identification.
Police departments in Santa Clara County, Calif., adopted the changes 15 years ago. They even adopted another of Wells’s recommendations: They record video of the identification from start to finish so that jurors can view it later.
The head of the Conviction Integrity Unit for Santa Clara, Assistant District Attorney David Angel, said that police were skeptical about the new procedures but now embrace them. He said he’s unaware of any false identifications under the new policies.
“The nature of science is to change as new evidence comes in,” Mr. Angel said. “We want to make sure we change with it.”
Not all police resistance has vanished. The California Police Chiefs Association, among other California law enforcement organizations, opposes Senator Wiener’s bill.
Its position is that in gang-related crimes, officers develop trust with potential witnesses who may object to being passed on to unfamiliar officers for lineups. According to the association, the bill as written would make it impractical for lineups that must be conducted in the field, when police are pursuing a criminal who presents an imminent danger to the public.
It would also require recording video of the identifications and storing those video recordings, which according to the association could be too costly for some small police departments.
Nevertheless, Jonathan Feldman, head of legislative affairs for the association, said he expects the organization will be able to reach a compromise with Wiener. “We just haven’t gotten there yet.”
This story was reported by Stateline, an initiative of the Pew Charitable Trusts.