Due to the importance of eyewitness identification, a large amount of research has been conducted to examine the reliability of eyewitness memory. Typically a staged crime is viewed and then the witnesses are to identify the perpetrator in a lineup. In reality the identity of the perpetrator is not known, so the police are not certain that their suspect is actually the perpetrator. This aspect of the investigation is simulated by comparing perp-present and perp-absent lineup conditions. Responses to these conditions are of interest because the choice of foils may result in a witness incorrectly identifying an innocent suspect.

Unfortunately, experiments generally use the same foils in both the perp-present and perp-absent lineups. This is done to control for any possible affect of the foils. In other words, the lineups are identical except that one contains the actual perpetrator, while the other contains the innocent suspect. Therefore foils are chosen who look like the perpetrator rather than the suspect. This results in fewer false identifications of the innocent suspect and increased identification of the foils than there would be if the lineup had utilized match-to-suspect foils. In real investigations, foils are usually chosen to match-to-suspect rather than match-to-description (perpetrator).

The same-foils design is not necessarily a good simulation of description based foil selection either. The reason being that foils are chosen to match a picture or very accurate representation of the perpetrator, rather than based upon the description of witnesses which are often lacking in detail or inaccurate. Therefore, in order for experiments to more closely mirror real investigations, perp-present and perp-absent lineups should not include the same foils. Rather they should be equated in that the foils chosen should be as similar to the suspect (innocent or guilty) in the lineup as is possible.

Problems also arise when suspect-matched lineups are produced. Basically, the innocent suspect has the most in common with the description of the perpetrator and the foils have less in common. This increases the likelihood that the suspect will be identified as the perpetrator. This is called the backfire effect. This effect may also be produced by the fact that foils will have more in common (more traits) with the suspect than they do with each other. Subsequently the witness can identify the origin of the lineup as the suspect. Thus identifying the suspect’s central tendency status. This is the basis of the current research, which is designed to examine the backfire effect prediction and the related prediction that the effect is stronger in suspect-matched lineups than for perp-matched lineups.

Materials

Procedure

• Groups of 1 to 5 participants watched a 5-minute film of a car jacking.
• Participants wrote descriptions of the perpetrator, answered questions about he film, and gave a confidence rating for their ability to identify the perp.
• Participants completed a filler task for 30 minutes: rating 30 words on positive or negative meaning, completing personality questionnaires and the California Q-sort
• Experimenter read official admonition providing warnings for viewing lineups
• Participants view the lineup and make identification (if no pick- asked if unsure or if perpetrator not in lineup)
• Participants rate their confidence in their identification.
• Participants who did not identify someone were then asked to pick someone who most closely matched his or her memory of the perpetrator

Results

• See Table 1 on page 207
• Responses classified as: ID Suspect, ID Foil, Don’t Know, Perp Not There
• Pattern of responses different for perp-present and perp-absent lineups
• Suspect-matched lineups had higher false identification rates of suspects (25.4%) than Perp-matched lineups (4.8%)
• Perp-matched lineups had higher false identification rates of foils (50%) than Suspect-matched lineups (15.9%)
• As predicted, the conditional probability of false identification is greater than chance (.167) for perp-absent-suspect-matched lineups (.615) and less than chance for perp-absent-perp-matched lineups (.088). Deviation from chance was reliable for suspect-matched but not perp-matched lineups
• Inclusion of forced choice gave the following results: see Table 2. While moderate increases in identification rates were seen across the board, the most dramatic increases were in the identification of foils in both the perp-matched and suspect-matched conditions
• Confidence – prospective confidence was reliably higher than retrospective confidence across conditions
• Confidence and Accuracy – No reliable correlation
• Similarity Analyses – raters rated similarity of 40 pictures (including the foils) to the target picture based on a 4-point scale. The perp-present and perp-absent lineups were very well equated in foil to suspect similarity. There was a difference between perp-absent perp-matched foils and suspect-matched lineups because perp-matched foils were not matched to the suspect.
• The two innocent suspects produced somewhat different results as a function of their similarity to the perpetrator. IS2, the more similar suspect, was responsible for most of the false identifications. This fits with the notion that when they are similar false identifications will increase. In IS1 the backfire effect was not noted in the suspect-matched lineup unless those witnesses who had originally not made an identification. In this case there was an increase in the chance that the suspect would be identified. Conclude that high false identification rates are result of the "backfire effect" and not the similarity of the suspect to the perpetrator.

• Results support the "backfire effect" caused by using a suspect-matched lineup, as is typically done by the police.
• Results also support the notion that using the same foils in perp-present and perp-absent lineups is not consistent with the real world. Rather the similarity between foils and suspects should be controlled across conditions.
• Experimenters should create such lineups using different experimenters
• Similarity ratings for foils and suspects for each condition should be collected and analyzed
• Standard police procedure (suspect-matched lineups) may produce conditions under which false identifications are of increased probability because the suspect has been chosen based on similarity to the perpetrator, but the foils have been chosen based on similarity to the suspect. A match-to-description procedure would likely negate the possibility of the backfire effect.