Sequential lineup presentation promotes less-biased criterion setting but does not improve discriminability.

Abstract

When compared with simultaneous lineup presentation, sequential presentation has been shown to reduce false identifications to a greater extent than it reduces correct identifications. However, there has been much debate about whether this difference in identification performance represents improved discriminability or more conservative responding. In this research, data from 22 experiments that compared sequential and simultaneous lineups were analyzed using a compound signal-detection model, which is specifically designed to describe decision-making performance on tasks such as eyewitness identification tests. Sequential (cf. simultaneous) presentation did not influence discriminability, but produced a conservative shift in response bias that resulted in less-biased choosing for sequential than simultaneous lineups. These results inform understanding of the effects of lineup presentation mode on eyewitness identification decisions.

Receiver operating characteristic analysis of eyewitness memory: Comparing the diagnostic accuracy of simultaneous versus sequential lineups.

“…The analysis reported by Palmer and Brewer (2012) represents an important step forward in that it used signal-detection concepts to clearly separate accuracy-that is, how far an ROC falls above the diagonal line of chance performance-from response biasthat is, where a point falls on an ROC (Ebbesen & Flowe, 2002;Meissner et al, 2005). However, actually performing ROC analysis would be a far better way to investigate this issue because it is not dependent on detailed theoretical assumptions, whereas the analysis reported by Palmer and Brewer is dependent on numerous minimally tested assumptions that eyewitness memory researchers are free to dispute (e.g., the target and lure distributions are assumed to be Gaussian in form and to have the same variance, an integration decision rule is assumed instead of an independent observations decision rule, the decision rule is assumed to be the same for both lineup procedures, etc.).…”

“…In other words, theoretical assumptions are used to estimate from a single HR-FAR pair what the rest of the ROC would look like had the ROC data actually been collected. Using this approach, Palmer and Brewer (2012) recently fit a theoretical signal-detection model to single pairs of hit and false alarm rates produced by simultaneous and sequential lineups in previously published studies. They found that both procedures yield approximately the same d=, from which one might infer that the two procedures would yield approximately equivalent ROCs (contrary to the idea that there is a sequential superiority effect) and that they differ only in that the sequential procedure yields more conservative responding.…”

Why eyewitnesses fail

“…Three decades of such studies have quantified the effects of many of these variables on identification accuracy (6)(7)(8)(9)(10)(11)(12)(13)(14)(15)(16)(17)(18)(19)(20). Simply put, data on estimator variables tell us something about the probative value of eyewitness testimony-and thus the degree to which law enforcement and the courts should be circumspect-given the conditions under which a crime was witnessed.…”

A signal-detection analysis of eyewitness identification across the adult lifespan.

“…The model accounts for all identification decisions: suspect identifications, foil identifications and lineup rejections in both target-present and target-absent lineups. Therefore, the model fitting helps us to understand witnesses' decision-making processes and illustrates how willingness to make identifications (i.e., placement of the decision criterion) changes with differences in discriminability (Palmer & Brewer, 2012 µ innocent distributions and reflect poorer discriminability (see Figure 4).…”