Read more: “Judge Mental: Saving justice from the unreliable mind“
In the summer of 1981, a woman was abducted from a car park by a man with a gun, who repeatedly beat and raped her before leaving her bound and naked. A few days later, she thought she had found her attacker when she happened to see her stolen car at another apartment complex. Although the driver of the car, Robert Clark, did not completely match the victim’s initial description of her attacker, she subsequently identified him from a set of images and a live line-up. He was eventually convicted of rape, kidnapping and armed robbery.
Clark always maintained his innocence, claiming that he had acquired the car from another man a few days before. He was proved right in 2005, when the police finally investigated vaginal evidence taken from the woman at the time of the crime, and he was exonerated after 23 years in jail.
Advertisement
Such cases highlight vulnerabilities in police line-ups. In fact, according to US pressure group the Innocence Project, eyewitness misidentification is responsible for around later overturned by DNA evidence. So where does it all go wrong?
On paper, the logic seems sound enough. Line-ups present a suspect to a witness along with around eight similar-looking individuals – either in the flesh, or through photos and videos. Unless the witness has a clear memory of the suspect, they are likely to choose one of the foils, in which case the police can dismiss their judgement. Otherwise, a positive identification adds to the evidence against the culprit.
Yet there is plenty of scope for error. When members of the line-up are presented simultaneously, for instance, a witness will start to compare one with another, using relative judgements to choose the “best match” rather than relying on definite objective characteristics that match their memories. “People tend to feel under a lot of pressure to pick someone,” says , a psychologist at Royal Holloway, University of London. If that person happens to be the suspect, the decision can weigh heavily in court – even if they are not, in fact, guilty.
To get around this, some psychologists suggest that line-ups should present individuals one at a time, which should discourage relative judgement as the witness compares each person to their memory of the event. There is strong evidence that this works. Last year, at Augsburg College in Minneapolis, Minnesota, analysed the results of 72 separate studies comparing the two approaches to conclude that sequential line-ups result in around 20 per cent fewer errors ().
Simultaneous line-ups are still common in much of the US, where witnesses will often compare photos of many different people to try to pinpoint the criminal. In contrast, police in the UK increasingly opt for video line-ups, which show the candidates in sequence. Yet there may be room for improvement here, too. UK law states that a video line-up must show all of the faces twice before the witness makes a decision. While this may offer more time to decide, the practice also allows relative judgement to creep into their deliberations. “Sometimes you start showing the images and the witness says straight away: ‘no, wait, it’s that one’, but the law requires the officer to show the whole set of images twice before the witness is asked to give their decision,” says Memon, who is working with police to assess and improve the way line-ups are conducted.
Further scope for change lies in tweaking the appearance of the foils. Say a witness tells the police they saw a man with a mark on his cheek flee the scene of a crime. If the police subsequently arrest a man with a similar marking, he is likely to be the one picked out from a line-up of men without the blemish, whether or not he is the man seen at the crime scene. To try to prevent this happening, the police often try to conceal the marking in the line-up image. They would achieve far better results if they used the alternative, but currently less popular, strategy of replicating the marking on each of the decoys, according to research by Theodora Zarkadi, now at the University of Bedfordshire, and ‘s team at the University of Warwick in the UK ().