Actual vs. perceived eyewitness accuracy and confidence and the featural justification effect.

Abstract

This article documents a contradiction between objective eyewitness accuracy and perceived eyewitness accuracy. Objectively, eyewitness identification accuracy (and the confidence-accuracy relationship) is comparably strong when a lineup identification is accompanied by a justification that refers to either an observable feature about the suspect ("I remember his eyes"), an unobservable feature ("He looks like a friend of mine") or just a statement of recognition ("I recognize him"). There is, however, a weaker relationship between confidence and accuracy and an increase in high confidence errors for identifications that are accompanied by references to familiarity than by references to any other type of justification. With respect to perceived accuracy, we document a robust cognitive bias-the featural justification effect-that causes eyewitnesses to be regarded by others as less accurate and less confident when they justify their identification by referring to an observable feature as compared to when they give any other kind of justification, except for a reference to familiarity. (PsycINFO Database Record (c) 2018 APA, all rights reserved).

Do incompatible alternatives in a testimony affect witness credibility?

“…A similar conclusion can be drawn by a different research line on the featural justification effect: participants presented with a witness identification and an estimation of confidence accompanied by a featural justification (e.g., I recognize his chin) underestimate the actual confidence intended by the witness (Dobolyi & Dodson, ; Dodson & Dobolyi, , ). This featural justification effect is explained by the fact that in a fair lineup a particular physical feature will likely not differentiate between faces and, thus, will be perceived by participants as nondiagnostic of the accuracy of the identification.…”

The language of accurate and inaccurate eyewitnesses

“…In what we detail below, we follow the footsteps of Dobbins and Kantner (2019) and use their machine-learning methodology, which was originally applied to list-learning recognition data, to translate the language of an eyewitness into a classifier. Using this classifier, we are able to capture the diagnostic value contained within eyewitness confidence statements in richer detail than the broad categorizations used in past research (e.g., Behrman & Richards, 2005;Dobolyi & Dodson, 2018;Mansour, 2020). For example, we are able to identify the individual words that are most predictive of an accurate (and inaccurate) identification.…”

“…While there is much information about the value that confidence holds when a rating scale is used, less is known about the diagnostic value that verbal confidence statements hold (Behrman & Richards, 2005;Cash & Lane, 2017;Dobolyi & Dodson, 2018;Dodson & Dobolyi, 2015;Grabman, Dobolyi, Berelovich, & Dodson, 2019;Klobuchar, Steblay, & Caliguiri, 2006;Mansour, 2020). This is why Wixted and Wells (2017) highlighted this as a priority for future research, about which they asked: Should a confidence statement be taken in the witness's own words (as in Klobuchar et al, 2006), or should confidence be recorded using an explicit 3-point rating scale (as in Wixted et al, 2016) -or should a 100-point scale be used?…”

“…CAC analysis revealed a nearly identical confidenceaccuracy relationship regardless of whether participants expressed their confidence in their own words or selected their confidence from the rating scale. Dobolyi and Dodson (2018) also investigated whether verbal confidence statements yield a strong confidence-accuracy relationship, but they binned the statements based on the features described in the statements. In Experiment 1, participants studied a list of faces and were later presented a series of 6-person, simultaneous lineups half of which contained a target from the study list.…”

“…To investigate the relationship between the type of confidence statement and the accuracy of an identification, Dobolyi and Dodson (2018) fit a generalized linear mixed effects model to the "chooser" data, which included correct identifications from target-present lineups as well as filler identifications from target-present and target-absent lineups. The model predicted a strong confidence-accuracy relationship when participants justified their identification by referring to an unobservable feature (e.g., "I am positive that is him").…”