Study Finds Interactive Lineup Improves Accuracy of Eyewitness Identification, But Does It Also Increase Likelihood of False Identifications?
A team of researchers at the University of Birmingham’s School of Psychology in the U.K. developed interactive lineup software that allows witnesses to view lineup photographs from different angles. They tested the software to see if it improved eyewitness identification and concluded that it may improve identification accuracy. They also discovered that witnesses viewing the lineup would automatically adjust the angle of viewing to comport with the angle in which they viewed the suspect. However, the data also indicate that this may increase the likelihood of false identifications.
Police photo-lineups generally consist of an array of photographs containing a photograph of the suspect and other photographs of people who are supposed to look similar to the suspect called fillers. Traditionally, the photographs in the lineups are taken from the front. However, the witness may not have viewed the suspect head-on. To test whether viewing a suspect in a photo-lineup from the same angle as the suspect was viewed during the crime and encoded into the witnesses’ memories can increase the accuracy of eyewitness identifications, the researchers designed two experiments.
Both experiments used thousands of people recruited online using Amazon Mechanical Turk (also known as MTurk). In the first experiment, participants viewed a 14 second video of a crime with the suspect head on or in right profile and perpetrator’s face in view for nine seconds. The participants were then shown a brief cartoon and asked to solve an anagram as a distraction. Then they were allowed to sequentially view a nine-person lineup of head-on or right profile photographs either in the same pose as the suspect was in the crime, in a different pose, or in both poses. Some lineups did not have the suspect’s photograph at all. People in the lineup were viewed sequentially for an unlimited time, but once the viewing of a specific person was complete, the participant could not back up to that person again.
Participants were asked to identify the suspect or reject the lineup if the suspect was not present. For participants viewing different poses, the suspect was correctly identified 49% of the time. For the same pose, it was 66%, and for both poses, it was 65%. A filler was incorrectly identified 18% of the time for different poses, 16% for the same pose, and 15% for both poses.
For different poses, 33% of the lineups containing the suspect were rejected as not having the suspect in them. For same poses, it was 18%, and for both poses, it was 20%.
In different pose lineups without the suspect 58% were correctly rejected and 42% of the participants falsely identified a filler. For same poses, it was 62% correctly rejected and 38% having false identifications of fillers. For both poses, it was 65% correctly rejected and 35% false identifications.
The overall correct identification rates were 49% for different pose, 66% for same pose, and 65% for both poses. The false identification rates for any particular filler were 5% for different poses and 4% for the others. However, the lineups contained either eight or nine fillers so the probability of some false identification in any particular lineup was much higher.
An analysis of the data from the experiment that considered the degree of certainty of the identification concluded that same-pose lineups were significantly more accurate than different-pose lineups.
In the second experiment, participants were shown the same crime video as the first experiment but only showing the suspect in left-profile or right-profile. After their distraction task, the participants used an interactive program to view a lineup of six persons. The program allowed the participants to change between left-profile, right-profile, and head-on photographs. Participants could view as long as they wished but could not back up to previously viewed persons in the lineup.
When using the program, participants who had seen a left profile in the video correctly identified the suspect 67% of the time, had a false identification 12% of the time, and incorrectly rejected the lineup 22% of the time. Surprisingly, the numbers for those who had seen a right profile were 74%, 16%, and 10%, respectively. The reason for the difference is unknown.
The left-profile encoding participants rejected a lineup that did not show the suspect 59% of the time, falsely identifying a filler the remaining 41% of the time. For right-profile participants, it was 57% and 43%, respectively. These figures are close to those of the first experiment participants who viewed different poses and worse than those who view the same pose or both poses. This raises the question of whether giving witnesses agency over the viewing of lineup profiles also pressures them to identify someone, even when the suspect is not present.
Further analysis showed the correct identification rates were 67% and 74% for the left-profile and right-profile encoding conditions, respectively. The false identification rates for any specific individual filler for both groups were 7%, considerably higher than the three groups in the first experiment.
Again, since there were five or six fillers in each lineup, the probability of a false identification in any given lineup is much higher.
The study concluded that, as expected, matching the pose seen in a photo-lineup with the pose a criminal suspect is seen in during the crime may increase the accuracy of eyewitness identification. Unmentioned by the report’s authors, but evident from the data, is that it may also increase the likelihood of false identification. How can both be true at the same time?
The data indicate that both correct identifications and false identifications are being increased at the expense of admitting an inability to identify anyone in the lineup as a suspect. Thus, this technique of allowing witnesses agency to manipulate lineup poses bears further investigation before being adopted by law enforcement.
Sources: nature.com/scientificreports, forensicmag.com
As a digital subscriber to Criminal Legal News, you can access full text and downloads for this and other premium content.
Already a subscriber? Login