Lineup Fairness: Optimizing Filler Selection Strategies

Eyewitness misidentifications are a primary source of erroneous convictions and a primary safeguard against such errors is the use of fair lineups. Department of Justice guidelines tell officers to construct lineups with fillers similar enough that the suspect won’t stand out--but not so similar as to reduce positive eyewitness identifications. At present police “eyeball” arrays to gauge whether they are fair enough. In appellate court cases judges also “eyeball” lineups and consider two “fairness” comparisons: They compare members to witness descriptions to assure that everyone is consistent with the description and compare members of arrays to assure they are generally similar in appearance. Some eyewitness researchers argue that assuring all lineup members are consistent with witness descriptions is sufficient to assure fairness. They argue that other variations in lineup members’ appearance will aid witnesses in distinguishing the guilty from the innocent—whereas increasing the similarity of suspects and innocent “fillers” will undermine witness performance. Among the questions addressed in the proposed research are does matching to description alone produce optimal witness performance or does additional suspect-filler similarity further enhance performance? The project will also test whether recently-developed face-similarity software enables us to quickly and reliably construct fair arrays which produce high levels of performance and does it improve on “eyeball” judgments? The primary objective of the research is to improve police lineup selection procedures and judicial evaluations of lineup fairness. Although there are indications from the prior research that it is possible to construct lineups that are unfair, fair, and too fair, we do not have a very clear picture of the levels of suspect-filler similarity that produce those outcomes. At present there are no methodologically sound studies with reliable similarity measures that directly compare strategies that at least some theorists and researchers believe will produce optimal levels of witness performance (using measures such as d’). To bring order to these findings we will experimentally pit theoretically- and empirically-driven selection strategies against one another under controlled conditions--using match-to-description vs match-to-suspect strategies employing varying degrees of suspect-filler similarity. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.