Lineup Composition Effects on Eyewitness Identification
Abstract
Whenever lineups are prepared for eyewitness identification, the investigator constructing
the lineup must decide how to choose lineup members to appear in the lineup with the
person under investigation. Accordingly, three research projects were conducted to
inform lineup construction practices. The first project is a meta-analysis of lineup
similarity effects. Results of the meta-analysis suggest that biased lineups yield the
highest rate of suspect identifications, regardless of whether that person is guilty or
innocent. When lineups with moderately similar members are compared with lineups
containing highly similar members, increasing similarity reduces innocent suspect
identifications and has only a small and nonsignificant effect on correct identifications.
Two experimental studies were then conducted using morphing software to
systematically manipulate the degree of similarity between lineup members who are
known be innocent (fillers) and the lineup member who is under investigation (the
suspect). In the first experiment, lineups with moderately high similarity fillers yielded a
higher correct identification rate than did lineups with very high similarity fillers. When
comparable procedures were used in the second experiment, fillers of low and moderately
low similarity to the culprit yielded nearly identical correct identification rates. In both
experiments, increasing suspect-filler similarity led to a decrease in innocent suspect
misidentifications. The accumulation of evidence from the three research projects
suggests lineups would be best constructed with fillers of moderate similarity to the
suspect.