Field study of police lineups suggests courts must pay attention to initial witness confidence ratings and police departments should continue using traditional, simultaneous procedure. (Boston Bill via Flickr: young woman sat talking to the detective, describing the man who had raped her just hours before. The police presented her with a photo lineup. Five minutes of clear indecision passed before she picked out a face.

“I think this is the guy,” she said.

“You ‘think’ that’s the guy?” the detective said.

“That’s him.”

Months later, Jennifer Thompson’s answers were unequivocal on the stand. And Ronald Cotton, whose face she had picked out of the lineup, was found guilty. He served a decade in prison before the DNA evidence later proved he was not the attacker.

Thompson’s identification of Cotton, and the disastrous consequences, is often pointed to as a key example of how eyewitness identification can be unreliable. But a new study says Thompson’s initial hesitation instead shows how the degree of initial confidence – before interactions with police and prosecutors reinforce an impression either way – can help investigators determine the degree of accuracy of an identification.

“Ignoring low confidence in the beginning is a grave error. The witness is telling you that there’s a good chance they’re making a mistake,” said John Wixted, a psychologist from the University of California San Diego. “To protect the innocent, it is important to realize that an initial low-confidence ID is untrustworthy. On the other hand, when lineups are fair and administered neutrally, high confidence at the start can also be quite telling.”

Some 348 photo lineups taken during investigations by the Houston Police Department’s Robbery Division included a single criteria: gauging the confidence of the witness on a three-tier scale: high, medium, and low confidence levels, in the study in the Proceedings of the National Academy of Sciences.

The high confidence grades resulted in 97 percent accuracy estimates, based on theoretical recognition memory models. But that confidence decreased to 87 percent and 64 percent at the middle- and low estimates, according to the work.

A person who is confident immediately in their recognition is much more likely to be sure of their memory, before the “contamination” from suggestions and other factors can get in the way, according to the study.

Wixted told Forensic Magazine in an interview that the doubt of eyewitness identifications coincided with the exonerations that began appearing in the late 1980s and early 1990s due to DNA advances. Since then, eyewitness identifications have become better scrutinized as reliable evidence – but the initial confidence factor needs to be considered as a better way to point an investigation.

For instance, a low-confidence ID could still be potentially valuable to a detective, Wixted told Forensic Magazine .

“The way to treat a low-confidence ID should not be to use it as a piece of evidence to convict – but to use it to get a search warrant, to do more investigation,” Wixted said.

The Innocence Project has stated that eyewitness misidentification is the single most prevalent cause of the wrong convictions in the U.S. – and they have estimated it played a role in 70 percent or more of the 330 wrongful convictions overturned by DNA since 1989.

Another part of the study was recommending simultaneous lineups, as opposed to sequential faces presented to a witness, one at a time. The reason is a group of individuals matching the same description all at the same time compels the witness to recall as much as possible from identifying features to differentiate between the faces, Wixted said.

The study follows on the heels of a National Academy of Sciences recommendation last year which called for double-blind interview and videotaped lineups, to preserve the process of identification.

William Wells, of Sam Houston State University’s Department of Criminal Justice and Criminology, worked cooperatively with the Houston police to conduct the field study.

Wixted and collaborators published a precursor study in September in the journal American Psychologist which laid out the confidence-accuracy association. Some parts of the criminal justice system, most notably the New Jersey Supreme Court, have decided that eyewitnesses are by nature unreliable – and Wixted warns that taking such a stance would be like throwing out the baby with the bathwater.

“A blanket indictment of the reliability of eyewitness expressions of confidence is wrong,” said Wixted. “It’s a huge mistake our legal system is making. At the time that they’re first making an ID, eyewitnesses can give us reliable information about their accuracy.”