Facial Features Eyewitness

Extensive research has explored whether a suspect’s facial features can influence an eyewitness’ identification decision and how this may affect lineup construction. The two main issues are: how particular facial features can influence an identification decision and how the suspect’s features are distinctive compared to other fillers within the lineup. Research has found that identifying members of a different race is more difficult than those of one’s own race (Alley & Schultheis, 2001). It also suggests that bias occurs in low-similarity suspect-filler lineups (Fitzgerald, Price, Oriet, & Charman, 2013) and if the suspect has distinctive facial features (Carlson, 2011). These issues affect lineup construction and a biased lineup may result
Studies show that witnesses identify those from one’s own race better than those of a different race (Alley & Schultheis, 2001). Smith, Stinson and Prosser (2004) suggested the suspect’s race can affect clarity of memory and confidence of witness identification decision. Knuycky, Kleider and Cavrak (2014) suggested different races have stereotypical facial features, which may be perceived and remembered differently. These could lead to feelings of familiarity and misidentification (MacLin & Malpass, 2001). To overcome this cross-racial effect, Wells and Olson (2001) suggest police investigators could use same-race lineup constructors or more fillers. However, most of these studies were conducted with a facial recognition paradigm, which is not as ecologically valid as a field study (Sporer, 2001). Ecological validity is important as it demonstrates real-world applications and the study’s relevance to the
Dissimilarity of facial features can result in distinctiveness and influence identification decisions. The Technical Working Group for Eyewitness Evidence (2003) recommended that whilst the suspect should not stand out from known innocent members, these innocent members should also not be too similar to the suspect. Fitzgerald, Price, Oriet and Charman’s (2013) meta-analysis on 17 experiments found that regardless of a suspect’s guilt or innocence, suspect identifications were more likely to occur with low-similarity fillers. Concern for lineups with high similarity to target would reduce correct identifications of target was unsupported by the meta-analysis (Fitzgerald, Price, Oriet & Charman, 2013). Fitzgerald, Oriet and Price’s (2015) literature review also did not support this concern but used morphing software, so replication using fillers from police databases is needed. Moreover, there is little operationalisation of suspect-filler similarity or report of similarity ratings in the literature (Fitzgerald, Price, Oriet, & Charman, 2013). Therefore further studies should be conducted before drawing stronger