Assessment of Bias in Police Lineups

Materials from five extant field studies were analyzed to determine the level of structural bias in police lineups. Depending on the jurisdiction. between 33% and 68% of lineups sampled from 1,548 real police lineups scored as suspect-biased using mock-witness proportion score. The suspect did not draw a fair portion of mock-witness picks in 20% of field lineups (reverse-biased lineups). Lineup fairness measures revealed that a point estimate (mean) for a set of lineups can mask significant problems in lineup construction and that any single lineup should not be assumed to be fair based on an aggregate score. A sample of 190 lineups from the field data of Wells, Steblay, and Dysart (2015) was used to conduct four new studies that examined the relationship between lineup structure and real eyewitness decisions. Our primary hypothesis-that real eyewitness decisions could be predicted by lineup bias measures-was partially supported. Suspect identifications from simultaneous (but not sequential) lineups were disproportionately linked to suspect-biased lineups. Suspect identifications from suspect-biased lineups were almost twice as frequent for simultaneous than for sequential lineups. Additional experimental tests of these field lineups using mock-witness measures demonstrated that detailed descriptions produced higher lineup bias scores than did brief descriptions, indicating that brief descriptions can hide substantial lineup bias. Mock-witnesses were able to find the suspect in some lineups via cues not readily apparent in the description alone. The data support the idea of using a framework of descriptors to improve the fairness of lineups. Study outcomes demonstrate the realities and nuances of field lineup structure that necessitate updated consideration of the use of mock-witness measures and a blended approach to lineup construction.