The mugshot-exposure effect (i.e., seeing a mugshot before a lineup) increases false identifications made by child and adult eyewitnesses, but police continue to use this practice. Thus, there is a need for a method that can help police discern guilty from innocent identifications. The multiple independent lineup (MIL) technique requires eyewitnesses to identify the target on several lineups (e.g., face, body, clothing) and these additional identifications are used to reflect suspect guilt (reflector variable). The present study explored the MIL technique with 204 children (6-11-years-old) who, after witnessing a live event, were exposed to a mugshot (either the correct or ‘innocent’ target) before making a lineup decision containing that same target on the following day. Results indicate that an increasing number of target identifications across multiple lineups reflects target ‘guilt’–suggesting the MIL technique may be a mechanism to mitigate the memory harm associated with the mugshot-exposure effect.

Influence from a lineup administrator has been identified as a key factor producing eyewitness misidentifications. To prevent lineup administrators from directing eyewitnesses toward the police suspect, researchers have recommended the implementation of double-blind lineup procedures. However, double-blind administration may not protect against all forms of social influence during identification procedures. Specifically, when blind administrators are motivated to obtain an affirmative identification, they may engage in behaviors that encourage choosing generally. Biased lineup composition may exacerbate the effect of these encouraging behaviors, such that a disproportionate proportion of identifications fall on the innocent suspect. In a collaborative experiment, two research teams examined the effects of administrator motivation and suspect bias on administrators’ behaviors, eyewitnesses’ decisions, and administrators’ and eyewitnesses’ self-reports of the identification procedure. Our findings suggest that there may be value in using self-administered lineups that more effectively eliminate social influence.

This research first conducted a meta-analysis of experiments that compared fair lineups versus biased lineups. Based on the meta-analysis, we examined the expected utilities of all ROC operating points of both fair and biased lineups. Expected utility jointly considers the prior probability of guilt, likelihoods, and utilities. These parameters capture different stages of police investigation: the prior probability of guilt is affected by the amount of evidence that police collect before initiating a lineup; likelihoods are affected by factors shaping witness responses and police use of eyewitness evidence during a lineup; utilities are affected by the consequences of police actions after a lineup. The expected utility analysis reveals that the discrete operating points on a lineup ROC curve could maximize expected utility of investigator classification decisions at different ranges of the prior probability, which may raise concerns about the legitimacy of conducting lineup procedures across various circumstances.

Effective size is a measure of fairness for eyewitness lineups. For a 10-person lineup, it ranges from 1 (biased) to 10 (completely fair). We created lineups for 16 suspects and examined the accuracy of suspect identifications for lineups of varying effective sizes. Participants completed an eyewitness identification task, involving a 10-person lineup and post-identification confidence assessment. Using the distribution of lineup choices in a culprit-absent lineup condition, we found that the effective size of the lineups ranged from 5.92 to 9.49. For participants who identified the suspect in the lineup, there was a positive correlation between effective size and identification accuracy (0.30, 95% CI [-0.23, 0.69]). The increase in suspect identification accuracy for lineups with higher effective size was most pronounced for eyewitnesses who reported low confidence identifications and less pronounced for high confidence identifications. This suggests that the confidence-accuracy relationship for suspect identifications is affected by lineup fairness.