Shows like CSI have taught us that a lot can be gleaned from clues left at a crime scene, and painstakingly pieced together by investigators. What they don’t often show is how fallible evidence can actually be if it’s not properly obtained. More and more, prosecutors and police are coming to question the ways in which they’ve typically done business.
Say a murder happens. Police arrive at the scene. But the shooters are long gone. How they figure out who done it often comes down to one thing: a witness. It’s intuitive. If you want to know what happened, who committed a crime, ask someone who witnessed it.
“Well a lot of people think that eyewitness identification is the most accurate,” says Jess Adachi, San Francisco’s Public Defender. “The reality is very different.”
Adachi has handled thousands of criminal cases. “And I have to say,” he says. “Probably one of the most common sources of error that I've seen is misidentification.”
Research backs that up. A 2004 analysis by San Francisco Magazine looked at 200 wrongfully convicted people. Of those 200, more than half, 60 percent, were the victims of misidentification.
“Why is that?” Adachi says. “Because we're human beings, and human beings and the human eye is fallible.”
What science is learning about memory helps explain why the eye can lie. The old way to think about memory treated it like a picture stored in your brain. We now know that isn't true. Memories are actively created. Remembering an event is more like painting a picture than looking at a photograph.
“That's what makes it scary,” Adachi says. “Because, often a witness will be 100 percent sure. It's just that their perception is incorrect.”
A crime scene could be dark, a witness’s adrenaline high. And a witness may only get a fleeting glimpse of the suspect. All these factors – experts call them estimator variables – will make it harder to recall what the perpetrator looked like. But perception is only half the story.
“When you talk about how a person is identified, often there are suggestive procedures that are institutionalized,” Adachi says.
Suggestive procedures can be subtle and they can occur when police or prosecutors interview a witness. They can also happen during a police lineup.
“I've seen cases where the officer who was administering the line-up has a bias and that comes off clearly,” Adachi says. “We've seen TV shows where the officer coughs or does something to indicate who the suspect is, and yes that does happen some time in real life.”
Obviously, you don't want police officers telling witnesses who to pick. But the much deeper problem is when officers unconsciously indicate who the witness should choose. This can happen in all sorts of ways, from the way an officer asks a question, to the officer's body language. These behaviors are the system variables. And they are the target of a nationwide effort to reform police lineups.
“There are a small number of errors that happen in criminal cases,” says District Attorney Jeff Rosen of Santa Clara County. “The criminal justice system is a human system.”
The question for Rosen is how to think about those errors. “In the criminal justice system, and when i say errors, I mean someone being convicted who didn’t commit the crime,” he says. “Is it like being struck by lightning, or is it like an airplane crash?”
Lightning, Rosen says, is just an accident. A plane crash, however, is probably preventable. Likewise, if that conviction is the result of system variables in the lineup, then police can control for errors. In 1999 the Department of Justice released recommendations for police line-ups. There were two main suggestions. The first was that the officer conducting the line-up not know who the suspect is. This is called double blind administration and it's how much scientific research is done. What it does is prevent the officer from subtly guiding the witness to choose the person the officer knows is the suspect. The second recommendation is sequential line-ups. This means that the witness is shown potential suspects one at a time, instead of all at once. This forces the witness to be sure of his or her choose – and not just choose the closest match.
“This was not that difficult to get our police chiefs in the county to agree to,” Rosen says.
Santa Clara County adopted the new procedures in 2002. Sequential line-ups and double blind administration became standard throughout the county. Recently, the county has also added video recording of witnesses, to improve reliability.
“So we have slightly fewer positive identifications, but they’re much stronger,” Rosen says.
Despite this success, California cities have been slow to adopt the new protocol. San Francisco did so last year. Across the Bay, Alameda County still hasn't done so. A recent study found only 8 percent of police departments in the state are in compliance with Department of Justice recommendations.
“There's always some resistance to change, particularly when something has been done for a long period of time,” Rosen says. “And within the DA's office, there were certain prosecutors, ‘Do we really need to make this change? What does it matter?’"
In Sacramento, California legislators have tried several times to pass bills mandating improved line-up protocol. Each times the bills have failed. Police departments say the procedure is more expensive. Reformers disagree. The real issue, they say, is inertia. The way things have been, is the way they tend to stay. Those pushing for reforms say that change will come when police departments and district attorneys get on board. Though the process is slow, signs are that may be happening. Until then, we might do well to be a bit more skeptical of what we think we've seen.
Audio available after 5pm PST on February 13, 2012.