Copy the page URI to the clipboard
Clark, Clifford
(2010).
DOI: https://doi.org/10.21954/ou.ro.0000f244
Abstract
Witnesses are routinely interviewed as part of police investigations in order to obtain a description of the offender. This description can take the form of a verbal, written statement or a visual statement that can consist of a facial composite. This thesis investigated the construction of facial composites and explored a variety of techniques aimed at improving the accuracy of the likeness produced.
A survey of E-FIT operators in the UK was used to identify a variety of techniques which had the potential to affect the accuracy of the composite produced. The three most promising techniques were the use of an initial interview prior to composite construction, working through a list of facial descriptors with the witness prior to construction and instructing the witness to image the face of the offender during construction. The utility of these techniques was tested in two experiments conducted using trained police personnel. The results of these experiments showed that neither prior interviewing or use of facial descriptors appeared to affect the accuracy of the composites produced but that composites produced by witnesses instructed to image were identified less often than when no such instruction was employed.
The negative effect of imaging during composite construction was explored in two further experiments by separating the instruction to image from seeing a facial composite, which revealed that imaging a face negatively affected recognition performance but that seeing a composite of a face had no effect on recognition. The results are considered in light of previous psychological research and theory and their potential impact on police procedures.