Making sense of sequential lineups: An experimental and theoretical analysis of position effects

Abstract

Abstract As part of a criminal investigation, the police often administer a recognition memory task known as a photo lineup. A typical 6-person photo lineup consists of one suspect (who may or may not be guilty) and five physically similar foils (all known to be innocent). The photos can be shown simultaneously (i.e., all at once) or sequentially (i.e., one at a time). Approximately 30% of U.S. police departments have moved to using the sequential lineup procedure over the last 30\u202fyears, yet its theoretical underpinnings remain poorly understood. A simple signal detection model makes several unexpected predictions about how the sequential lineup procedure should affect the ability of eyewitnesses to discriminate innocent from guilty suspects. For example, empirical discriminability (area under the receiver operating characteristic) should decrease as the position of the suspect in the lineup increases. In addition, under some conditions, a fair sequential lineup should not yield higher discriminability than a single-person (non-lineup) recognition test known as a showup. The results of two experiments reported here confirmed these predictions. Counterintuitively, even though empirical discriminability decreased as the suspect’s sequential position increased, a signal detection model fit to the data indicated that theoretical discriminability exhibited a small effect in the opposite direction (increasing with the sequential position of the suspect). The latter result is consistent with diagnostic feature-detection theory of eyewitness identification.