The number of fillers may not matter as long as they all match the description: The effect of simultaneous lineup size on eyewitness identification

Abstract

According to the Diagnostic Feature-Detection (DFD) hypothesis, the presence of fillers that match the eyewitness's description of the perpetrator will boost discriminability beyond a showup, and very few fillers may suffice to produce the advantage.We tested this hypothesis by comparing showups with simultaneous lineups of size 3, 6, 9, and 12. Participants (N = 10,433) were randomly assigned to one of these conditions, as well as target-present (TP) versus target-absent (TA) lineup. As predicted by the DFD hypothesis, lineups were superior to showups, and there was no advantage with increased lineup size beyond a 3-member lineup. The confidenceaccuracy (CA) relationship held a similar pattern. The only effect of increased lineup size was a lower likelihood of choosing a suspect (guilty or innocent). We conclude that police should focus more on the quality rather than quantity of fillers. K E Y W O R D S diagnostic feature-detection hypothesis, eyewitness identification, lineup size, showups, simultaneous lineup

Do sequential lineups impair underlying discriminability?

“…This suggests that the difference in discriminability between sequential and simultaneous presentation should be greatest at sequential position 1 and should reduce over the course of the lineup. Because position 1 in a sequential lineup is equivalent to a single-item show up, this result is also consistent with the robust finding that the simultaneous lineup outperforms the single-suspect show up (Gronlund et al, 2012 ; Neuschatz et al, 2016 ; Wooten et al, 2020 ).…”

Testing encoding specificity and the diagnostic feature-detection theory of eyewitness identification, with implications for showups, lineups, and partially disguised perpetrators

“…The lack of response bias consideration confounds the data and can result in misleading conclusions regarding the ID performance of the eyewitnesses. Because of this issue, among others, ROC is the preferred method for analyzing ID performance in the eyewitness domain (National Research Council 2014 ), and its use has quickly accelerated over the last decade (e.g., Carlson and Carlson 2014 ; Carlson et al 2019 ; Colloff and Wixted 2019 ; Gronlund et al 2014 ; Jones et al 2020 ; Meisters et al 2018 ; Mickes et al 2017 ; Wetmore et al 2015b ; Wooten et al 2020 ). 7 ROC analysis has been used to demonstrate several DFT predictions, such as the superiority of fair simultaneous lineups compared to showups (e.g., Wooten et al 2020 ), sequential lineups (e.g., Carlson and Carlson 2014 ), and biased lineups (e.g., Wetmore et al 2015b ).…”

“…Diagnostic information is that which differs between the guilty and the innocent (e.g., the perpetrator had blue eyes, but the innocent suspect and fillers have brown eyes); non-diagnostic information is shared by the innocent and the guilty (e.g., the perpetrator had a beard, and so too does the innocent suspect and all fillers). Researchers primarily have tested DFT by discounting non-diagnostic facial information, with four different approaches, comparing: (a) showups with fair simultaneous lineups (e.g., Wooten et al 2020 ), (b) fair simultaneous lineups with fair sequential lineups (e.g., Carlson and Carlson 2014 ), (c) fair simultaneous lineups with biased simultaneous lineups (e.g., Colloff et al 2016 ), and (d) showups with showups with non-diagnostic information removed or discounted (e.g., Colloff et al 2018 ). These four methods discount non-diagnostic information in different ways, such as adding good fillers or explicitly covering a non-diagnostic feature.…”

“…Based on recent eyewitness ID studies applying ROC analysis to lineup data (Colloff et al 2016 ; Wooten et al 2020 ), we sought at least 1000 participants per cell. This might sound like an unusually large number of participants, but each was tested only once, in keeping with the majority of the eyewitness ID literature.…”

“…Previous researchers (e.g., Carlson et al 2019 ; Colloff and Wixted 2019 ; Wooten et al 2020 ) have tested DFT using empirical discriminability, finding better performance for conditions in which witnesses are more readily able to notice diagnostic information and discount non-diagnostic information (e.g., a fair simultaneous lineup), compared to conditions where this process is more difficult (e.g., a showup, a sequential lineup). For example, Wooten et al ( 2020 ) found that surrounding a suspect with as few as two fillers resulted in a higher pAUC than a showup. Following several of these studies, our goal is also to test DFT predictions concerning theoretical discriminability with ROC analysis, which reveals differences in empirical discriminability.…”