Cover Story

In 1984, 22-year-old restaurant worker Ronald Cotton was fingered in a police lineup and convicted of raping a Burlington, N.C., woman. He was sentenced to life, plus 54 years in prison. Case closed.

But the victim had picked the wrong man.

Nevertheless, Cotton served nearly 11 years in prison before police discovered that the perpetrator's DNA didn't match Cotton and released him. Cotton's case was among the first in which DNA evidence indicated a lineup misidentification, thus overturning a conviction and implicating the real perpetrator--who was already serving prison time on a different charge.

To help police departments prevent such costly errors, several psychologists who study lineups are recommending new and revised lineup procedures--based on empirical research (see sidebar). At the forefront of these efforts is psychologist Gary L.Wells, PhD, a distinguished professor of psychology at Iowa State University who travels nationally and internationally providing talks and training to police.

Among his recommendations are that police should limit feedback to witnesses choosing suspects from lineups and avoid presenting several photos or suspects simultaneously--a practice most of the nation's police departments use. Wells has been encouraging departments to instead use a sequential technique--in which police present photos to witnesses one at a time--he developed with psychologist Roderick C. Lindsay, PhD, in 1985 and that his research indicates cuts down on witnesses mixing and matching different suspects' faces.

"What we're trying to do is take scientific methods from psychology and make the lineup more scientific," explains Wells, who argues police should have a standard protocol for conducting lineups just as psychologists have the experimental method for conducting research. "The components that make a good experiment are the components that we should apply to a lineup."

Double-blind and sequential lineups

Two key scientific components, Wells notes, are separating witnesses so they don't influence one another and not leaking officers' perpetrator hypotheses to witnesses. Police can accomplish the first, he says, through the sequential method of having witnesses decide whether a particular photograph depicts the perpetrator before seeing another. They can accomplish the second through the double-blind technique, where they don't know themselves which photographs are filler and which are of the suspects, so as to not sway witnesses.

While many departments continue to favor the unblinded simultaneous photo spreads, Wells and his colleagues are starting to get their message across. In 2001, New Jersey became the first state to adopt statewide guidelines for double-blind and sequential lineups. Wells has also been working with prosecutors and police departments in Massachusetts, Minnesota, North Carolina and Wisconsin to implement such procedures. But New Jersey is the only state in which the attorney general has jurisdiction over all the state's police departments to effect such changes.

Though encouraged that a handful of states are embracing these techniques, "with over 16,000 police departments across the country, I doubt if we've covered more than 3 percent of the population with these new procedures," Wells says. "I don't know where it has to get until it gets to a snowball effect. But we're going to keep pushing the ball."

New lineup methods

While research shows sequential and double-blind lineups reduce false identifications, they're also not fail-safe: Studies show that one in five suspect identifications from sequential lineups still may be wrong.

As a result, some psychologists are exploring alternative lineup procedures to help further reduce that number. For example, Lindsay is conducting experiments to determine whether treating physical characteristics--such as face, body and voice--as independent variables and presenting them in several different lineups may provide better eyewitness evidence for prosecutors. In what he calls "multiple-identification lineups," witnesses identify just the face of the perpetrator from photos presented sequentially. Then, witnesses are given separate sequential photo lineups of the body from the shoulders down or tape recordings of a voice one at a time.

"If a witness chooses from just one lineup, it can be weak evidence," says Lindsay, a psychology professor at Queen's University in Kingston, Ontario. "But if a witness can choose the suspect from two, it's stronger. From three, it's stronger yet....Multiple identifications leading to the same conclusion becomes more convincing."

To test this theory, Lindsay set up an experiment in which 90 participants viewed a staged crime and sought to identify the perpetrator by viewing multiple identification lineups: 70 percent of participants made correct facial identifications, 23 percent made correct body identifications and 13 percent made correct voice lineup identifications.

In a similar study, Lindsay found that most of 147 participants were able to correctly identify the perpetrator's body build or voice; less than 20 percent could only accurately identify the perpetrator's face. Lindsay's first published study on this appeared in February's Journal of Applied Psychology (Vol. 89, No.1).

The study revealed a 3 percent false-identification rate for multiple-identification lineups, whereas false-identification rates hover at about 27 percent for simultaneous lineups and at 9 percent for sequential lineups. Despite the promising results, Lindsay says it's still too early for the method to be used by police departments.

"It probably will require about five more years of research before we're ready to say for sure if police should be doing this or not," Lindsay says.

Israeli psychologist A.M. Levi, PhD, has also been testing an alternative lineup procedure--a multiple-choice or modified sequential lineup (MSL), which uses large sequential lineups, such as a 40-person lineup, and allows witnesses to pick more than one person. So far, his studies show such methods help lower the probability that a witness will choose a filler from the lineup. His MSL procedure is described in the fall 2002 issue of The International Journal of Cognitive Technology (Vol. 7, No. 2).

Influencing witnesses

No matter which lineup methods police use, psychologists warn those who administer them that what they say can also affect eyewitnesses' confidence in the suspects they select.

Researchers are studying the extent that confirming or disconfirming feedback can inflate witnesses' confidence and, as such, undermine the reliability of whom they identify. Witnesses who receive feedback about the accuracy of their choices more often become overconfident in their identifications than witnesses who receive no feedback, according to a study in April's Journal of Applied Psychology (Vol. 89, No. 2) by Wells and researchers Carolyn Semmler and Neil Brewer, PhD, both of Flinders University in Australia.

To uncover other potentially prejudicial influences, Wells is also studying how witnesses respond to composite photos--in which artists develop perpetrator sketches based on eyewitness reports. So far, he has found that witnesses who first view such composite photos are less likely to pick real perpetrators from lineups.

"When picking out noses and eyes for a composite, it seems to cause problems in memory because what you end up with never exactly looks like the person, but it replaces the witness's memory of the perpetrator," Wells says.

Drawing from such research, psychologists are working on lineup improvements and interpretation with officials in the judicial system, as well as with police officers. Even judges need to be educated on lineup procedures, Lindsay says. He warns Canadian judges he works with to beware of witnesses who look at simultaneous lineups and make comments such as, "It's between one and four."

He and other researchers hope that through such science-based education and application they can one day help fix a major law-enforcement shortfall by ensuring that the Ronald Cottons of the world aren't imprisoned when innocent.