Extracting the truth from conflicting eyewitness reports: a formal modeling approach

Eyewitnesses often report details of the witnessed crime incorrectly. However, there is usually more than 1 eyewitness observing a crime scene. If this is the case, one approach to reconstruct the details of a crime more accurately is aggregating across individual reports. Although aggregation likely improves accuracy, the degree of improvement largely depends on the method of aggregation. The most straightforward method is the majority rule. This method ignores individual differences between eyewitnesses and selects the answer shared by most eyewitnesses as being correct. We employ an alternative method based on cultural consensus theory (CCT) that accounts for differences in the eyewitnesses' knowledge. To test the validity of this approach, we showed 30 students 1 of 2 versions of a video depicting a heated quarrel between 2 people. The videos differed in the amount of information pertaining to the critical event. Participants then answered questions about the critical event. Analyses based on CCT rendered highly accurate eyewitness competence estimates that mirrored the amount of information available in the video. Moreover, CCT estimates resulted in a more precise reconstruction of the video content than the majority rule did. This was true for group sizes ranging from 4 to 15 eyewitnesses, with the difference being more pronounced for larger groups. Thus, through simultaneous consideration of multiple witness statements, CCT provides a new approach to the assessment of eyewitness accuracy that outperforms standard methods of information aggregation.