Objective

Our objective was to examine how Amendola and Wixted (A&W, 2014) arrived at their conclusion that eyewitness identifications of suspects from simultaneous lineups were supported better by corroborating evidence than were identifications from sequential lineups. Their cases came from a randomized field experiment by Wells et al. (2014).

Methods

We gathered information from the A&W article, examined an earlier, more complete report by Amendola et al. (2013), and then confirmed our numbers with Amendola.

Results

We discovered that the small subsample (n = 52) on which A&W’s entire conclusion was based was unrepresentative of the larger set of cases (N = 236) in a way that was heavily biased in favor of the simultaneous lineup. Specifically, although the larger data set showed that simultaneous and sequential lineups produced the same rate of adjudicated guilt, their small subsample of 52 cases was highly imbalanced: Among the 30 sequential cases selected, 16 were drawn from the adjudicated-guilty set and 14 were from the not-prosecuted set; among the 22 simultaneous cases selected, 17 were drawn from the adjudicated-guilty set and a mere five were from the not-prosecuted set. This problem could not be known from the article itself.

Conclusions

Because adjudicated guilty cases had more corroborating evidence than not-prosecuted cases and because simultaneous and sequential lineups produced equivalent rates of adjudicated guilty outcomes, the small sub-sample of 52 should have reflected this same equivalence. Instead, the sub-sample was stacked against the sequential and in favor of the simultaneous and A&W’s conclusion is not warranted.