An Evidence-Based Imperative to Videorecord Eyewitness Lineups

In a large-scale, high-impact experiment (N = 1,496), we videorecorded witness-participants as they completed lineup procedures in the presence of a condition-blind research assistant. A second team of condition-blind research assistants then reviewed these videos and scored witnesses on a large number of behaviors that reflect recognition strength and fluency. We then subjected the resulting data to machine learning analyses to determine whether witness behaviors could discriminate accurate from inaccurate witnesses. We found that a strong and fluent recognition experience implied accuracy and a weak and disfluent recognition experience implied inaccuracy. Furthermore, the behaviors that witness’ displayed while completing lineups discriminated between accurate and inaccurate decisions over and above confidence and decision speed. These same patterns held true both for positive identifications and for lineup rejections. Finally, these data also suggest that the behavior of a witness at the time of a lineup can be used to diagnose high-confidence mistaken identification errors. When law enforcement personnel adhere to science-based best-practice recommendations, high-confidence mistaken identifications are rare. But they do happen, and they have catastrophic consequences to the innocent persons that are forced to shoulder the burden of these identification errors, including the potential for wrongful conviction. That the behavior of a witness at the time of a lineup can be used to diagnose these errors is a groundbreaking discovery that creates a moral imperative to mandate the videorecording of eyewitness identification procedures.

Link: https://www.researchgate.net/publication/394079255_An_Evidence-Based_Imperative_to_Videorecord_Eyewitness_Lineups