Optimizing lineup construction: the impact of filler similarity, distinctive facial features, and individual differences in facial recognition on lineup performance

Roughton, Georgia L. (2023). Optimizing lineup construction: the impact of filler similarity, distinctive facial features, and individual differences in facial recognition on lineup performance. University of Birmingham. Foren.Psy.D.

Preview

Roughton2023ForenPsyD_Redacted.pdf
Text - Redacted Version
Available under License All rights reserved.
Download (4MB) | Preview

Abstract

The present thesis sought to investigate optimal lineup construction methods that enhance ability to discriminate between innocent and guilty suspects. Firstly, I highlight the need for lineup construction methods that enhance eyewitness discriminability. Next, in chapter 2, I conducted a systematic literature review of suspect-filler similarity and found that there were no standardised procedures for constructing lineups in experiments. I also highlight the impact of methodological factors on discriminability. Using the feature matching model and diagnostic feature detection theory (Colloff et al., 2021; Wixted & Mickes, 2014), I argue that low similarity lineups allow the witness to focus on the perpetrators’ unique features that are diagnostic of guilt, to make an accurate identification decision. However, I note that the low similarity lineup advantage holds only when lineup construction methods are fair (i.e., result in two memory strength distributions in witness memory: one for the perpetrator and one for the fillers and innocent suspect). In chapter 3, I conducted an experiment investigating lineup construction methods for distinctive suspects (e.g., with a facial tattoo). I compared a high similarity replication lineup, in which the distinctive suspects’ facial tattoo is exactly replicated across lineup members; a low similarity replication lineup, in which the lineup members have a similar but non-identical distinctive facial tattoo; and a do nothing lineup in which only the suspect has a distinctive feature. As predicted by the feature matching model, I found that low similarity replication lineups yield higher discriminability compared to high similarity replication and do nothing lineups. In chapter 4, I critically evaluate the Benton Facial Recognition Test and advise on the use of psychometric tools to allow for further exploration of lineup construction methods that enhance discriminability when individual differences are also considered. Finally in chapter 5, I disentangle mixed findings in the literature to date and recommend that future research thoroughly reports lineup construction methods.

Type of Work:

Thesis (Doctorates > Foren.Psy.D.)

Award Type:

Doctorates > Foren.Psy.D.

Supervisor(s):