Mistaken for a criminal: Eyewitness identifications are often wrong

A man walks into a crowded restaurant and snatches a handbag off the bar. A dozen patrons witness the crime, but can they recognize him just minutes later?

With guidance from experts in eyewitness identification, the TODAY show set up such a scenario at a suburban New York bar and grill with our hidden cameras rolling. The answer surprised us: When it comes to faces, our memories can often be quite flawed.

“A lot of people believe that memory might work like a video recorder. They can close their eyes and kind of replay things, but that's really not how it works,” said Jennifer Dysart, a professor at John Jay College for Criminal Justice in New York who observed our experiment. “A lot of things actually influence our memory. Many outside influences can come in and really create a problem."

When it comes to identifying criminal suspects, many of us conjure up an image of the police lineup of old: a group of men standing side-by-side behind a one-way mirror as victims or witnesses examine them. But most police departments now ask observers to pick perpetrators using photos.

There are two types of photo lineups: the “simultaneous,” where witnesses view arrays of six photos at once, and “sequential,” where they look at the faces one at a time. We used the “simultaneous” method, which is the most commonly used method, for our experiment.

It turns out, though, that identifications made that way are more likely to be erroneous.

In a recent field study , researchers analyzed the results of nearly 500 photo lineups at four local police departments around the country, and found that when detectives showed eyewitnesses a "simultaneous" array, 41 percent of those who made an identification picked an innocent person. If they were shown the photos sequentially, or one at a time, the number of mistaken identifications were reduced by 11 percent, meaning that for every 100 line-ups, 11 fewer innocent people were picked out as suspects.

"The consequences are severe. Because, once the eyewitness makes a mistake, it's very difficult to stop the process," said Dysart. "Law enforcement can use the eyewitness identification for arrest. Prosecutors use it to charge individuals. And jurors can use it to convict."

Hundreds of people have been thrown in prison in the United States for crimes they didn't commit partly based on faulty witness identifications.

In fact, of the 314 DNA exonerations since 1989, misidentification played a role in 72 percent of the wrongful convictions, according to the Innocence Project, which works to exonerate wrongly convicted people through DNA testing.

In Texas, David Wiggins was convicted of a sexually assaulting a 14-year-old girl, a crime he did not commit. While prosecutors were unable produce any physical evidence connecting Wiggins to the crime, the victim picked out his image in a photo lineup and wrote "looks familiar" next to his name. Wiggins served 24 years of a life sentence before being exonerated in 2012 based on DNA evidence.

In Virginia, Thomas Haynesworth spent 27 years in prison for a series of rapes he didn't commit. Five rape victims identified him in photo arrays as their attacker. He was later cleared by DNA evidence in two of the cases and all five convictions were eventually overturned.

"Our clients really represent the tip of the iceberg," said Karen Newirth, an attorney with the Innocence Project, which provided the examples. "If you think of all the kinds of cases that would turn on an identification, like robberies or a burglary, it's unlikely there would be any DNA left behind, those people really don't have much recourse."

For a number of years, reform advocates have been pushing police departments to use the "sequential" procedure rather than the “simultaneous” arrays.

And some states including New Jersey, Connecticut, Texas, Florida, and North Carolina have adopted or recommended procedural reforms aimed at reducing mistaken identifications, including the use of sequential lineups administered by detectives who are not aware of who the actual suspect is, to reduce the likelihood that they could inadvertently influence a witness.

Experts say such changes can have huge impact.

Dysart, who co-authored the field study, attributes the difference between the two procedures to the tendency of witnesses to treat the simultaneous photo arrays like a multiple choice test. In fact, we saw that play out in our hidden camera experiment when we decide to take our perp out of the array entirely.

"He's too dark, he's too dark. Something about him ... He's too light. So yeah, he was the one," said one woman pointing to various photos in an array that didn't even include our thief.

Another witness who got the same array said he was 100 percent confident that the "innocent" man in our lineup was our perp. The witness is a retired police detective.

"If you listen to the way that they describe the process, those witnesses essentially narrowed down their choice and just picked the best one," commented Dysart.

So how can you be a better witness if you see a crime go down? Here are some tips from the experts:

•            Put safety first.

•            After you are safe, write down everything you remember, as memories can fade quickly.

•            Focus on unchangeable and/or unique features of the perpetrator, such as a scar or skin complexion.

•            Relate the perpetrator(s) to someone you know. How are they similar? How are they different?

•            When speaking with police, avoid making inferences. If you don’t know the answer, say you don’t know.

•            Request best practices in your case. Ask for a sequential double-blind lineup with pre-lineup instructions.

•            Don’t assume the perpetrator is in the lineup that the police show you.

Have an idea for an upcoming edition of Rossen Reports? Email us.