The Problem of a Hammer: Eyewitness Identification Research Relies on the Wrong Comparisons

• Published in: Journal of applied research in memory and cognition 2021-09, Vol.10 (3), p.351-355
• Main Author: Hyman, Ira E.
• Format: Article
• Language: English
• Subjects: Contextual Associations; Crowdsourcing; Experimental Subjects; Eyewitness identification; False identification; Memory; Methodology; Sample Size; Signal Detection (Perception); Signal detection theory; Social Influences; Social Interaction; Source monitoring; Statistical Reliability; Unconscious transference; Witnesses
• ISSN: 2211-3681; 2211-369X
• DOI: 10.1016/j.jarmac.2021.06.004

Comments on an article by M. B. Kovera and A. J. Evelo (see record 2021-79725-001). Kovera and Evelo have identified important concerns regarding recent developments in the study of eyewitness identification. They are concerned that the growing reliance on signal detection theory, and particularly Receiver Operating Characteristic (ROC) curves, is leading eyewitness identification research in a dangerous direction. As they noted, the power needed for ROC studies of eyewitness identification requires incredibly large numbers of participants. Signal detection theory and ROC curves always need a lot of data. In many domains, this is not a problem. In particular, Kovera and Evelo expressed the concern that online studies remove the social interaction that is central to the identification process. When people make identifications, they are engaged in a series of social interactions. Witnesses have observed an event that involved people, perhaps several people beyond the culprit and the victim. They most likely have discussed their memory with others and encountered additional information. The identification process is also a social interaction involving the people administering the test and potentially other witnesses. Unfortunately, these social interactions are challenging to recreate in online contexts. (PsycInfo Database Record (c) 2022 APA, all rights reserved)