Using Image Manipulation to Construct Fair Lineups: The Case of the Buddy Holly Glasses

Article excerpt

ABSTRACT

Constructing and administering lineups is a necessary and important part of police practice. Guidelines for outlining proper procedures have been put forth by the U. S. Department of Justice, and many police departments are making efforts to comply. However, practical difficulties can arise in choosing appropriate photographs for use as lineup fillers. In the current case, a lineup was presented to us for consultation and evaluation. The lineup showed six White males with glasses; five of the males had wire-rimmed glasses and one male (the suspect) had thick, dark-rimmed glasses. The witness's description indicated that the perpetrator had "Buddy Holly" glasses. We evaluated the lineup and found it to be biased towards choosing the suspect. Information from law enforcement indicated that they felt they were doing the best they could do by including fillers who at least were wearing some sort of glasses. The evaluation presented here explains a method of image manipulation that could have been used to make the lineup a better test of memory. Data are provided that show that the resulting lineup reduced the bias against the suspect. We also provide some general "how-to" information on image manipulation for the construction of fair lineups and rationale for doing so.

Police lineups are a test of a witness's memory and are strong evidence for or against a suspect's innocence. As such, constructing and administering lineups is a necessary and important part of police practice and should be done with the utmost care and concern for accuracy. A review of recent false conviction cases points to faulty eyewitness identification as a major case factor. In response, the U. S. government has set forth guidelines for proper procedures of lineup construction and administration, and many police departments are making efforts to comply.

Guidelines for a photo lineup procedure are: (1) include only one suspect; (2) select fillers (the non-suspect members of the lineup) who generally fit the witness description (or in the absence of a description, should generally resemble the suspect in "significant features"); (3) when multiple photos of the suspect are available, use the photo that resembles the suspect description at the time of the incident; (4) include a minimum of five fillers members; (5) avoid using fillers who too closely resemble the suspect; and (6) create a consistent appearance between the suspect and fillers (e.g., scars, tattoos) (The Technical Working Group, 1999). The overall emphasis of the recommendations is to avoid having the suspect unduly stand out. If the suspect does stand out, a witness with a poor or no memory of the perpetrator would be able to "identify" the suspect if he or she was guessing and inclined to make a choice.

Lineups can be scientifically evaluated by subjecting them to a mock witness test to determine if they are fair (Doob & Kirshenbaum, 1973; Malpass & Lindsay, 1999; McQuiston & Malpass, 2002). A fair lineup means that each lineup member has an equal likelihood of being chosen (a chance probability) if a witness has poor memory or is simply guessing. There are two aspects of fairness to consider: lineup size and lineup bias (Malpass, 1981; Malpass & Devine, 1983). Lineup size refers to the issue of having enough reasonable choices in the lineup to adequately test memory and reduce the likelihood of chance identification. In addition, the choices should be "reasonably plausible alternatives" (Brigham, Meissner, & Wasserman, 1999, p. S74); if some members are not reasonable alternatives to the suspect, then the size of the lineup is minimized, and the chance for incorrect identification increases. Lineup bias on the other hand refers to aspects of the individuals, photo or lineup that makes one member distinctive from the rest (also called positive bias). Negative bias occurs when the suspect and the other lineup members are too similar to each other such that even a person with a good memory could not choose the suspect. …