Police in the United States might be doing something right, at least compared to their peers in United Kingdom.
By Nathan Collins
(Photo: Joe Gratz/Flickr)
At its worst, eyewitness testimony is a deeply flawed form of evidence. Consider the case of Ronald Cotton, who was convicted of rape after police subtly (perhaps unconsciously) nudged a witness toward a false identification. But even when police take steps to avoid such outcomes, as some did following Cotton’s exoneration, our memory for faces is not always that great. Now, researchers have taken another step toward improving eyewitness IDs—and, it turns out, the police in the United States may have been doing one thing right all along.
Eyewitness misidentification is one the main reasons innocent people end up going to jail for crimes they didn’t commit, as misidentification contributes to around 70 percent of wrongful convictions. Fortunately, there are some ways to improve the process—for example, double-blind identification, where the detectives administering the procedure don’t know who the suspect is and therefore can’t influence witnesses.
There are some aspects of the procedures that receive less attention: the number of people that witnesses see in a line-up, whether witnesses see photographs or videos, and whether those images are presented all at once or sequentially. Such choices might seem innocuous, but given the inherent limitations on our visual memories, all could have meaningful effects on who a witness identifies as the perpetrator, researchers Travis Seale-Carlisle and Laura Mickes write in Royal Society Open Science.
At its worst, eyewitness testimony is a deeply flawed form of evidence.
To see exactly how much influence such procedures had, Seale-Carlisle and Mickes first had 2,205 undergraduate students at the University of California–San Diego watch 20 seconds of staged surveillance video showing a young white male stealing things from an office. Later, students were shown a U.S.-style line-up of six photos all at once or a United Kingdom-style line-up of nine videos presented one after the other. Finally, each student either chose a perpetrator or rejected the line-up, meaning they didn’t feel the criminal was in the company presented to them. Finally, the participants stated how confident they were in their choices.
Seale-Carlisle and Mickes found that the U.S. system was better at producing correct identifications than the U.K. method. Specifically, the researchers broke up the data by students’ confidence in their judgments, then plotted the rate of correct versus false identifications, in which students identified someone other than the person in the surveillance video. For a given rate of false IDs, the researchers found, the U.S. system had a higher correct-ID rate—though it’s worth noting that even the correct-ID rate never exceeded 40 percent.
Because the team compared the two approaches head to head, they can’t say exactly why the U.S. system outperformed the U.K. version. “Whatever the reason for the difference, these findings underscore the importance of directly comparing line-up procedures in terms of their ability to discriminate innocent from guilty suspects,” the researchers write.