Closed-circuit television: How effective an identification aid?

Two experiments simulated identification of suspects from CCTV recordings: Expt 1 simulated identification from whole body shots while Expt 2 showed close-up pictures of targets' faces. The results are consistent with those of Bruce rt ai: (1999) and Kemp, Towell, and Pike (1997) in showing surprisingly high rates of error in person identification from CCTV material. In Expt 1, participants (N = 80) searched for a suspicious incident in a 25-minute surveillance film which was shown either in colour or monochrome. They then attempted to describe and identify the persons involved. Accuracy of identification was just 15 % from memory with a 60 % false alarm rate. When judgments were made with access to a still-frame image of the target, accuracy rose to 30%, with a 65 % false alarm rate. Colour did not improve identification, but did prompt more description of the targets' clothing. In Expt 2, participants (N = 80) observed video-clips which included close-up pictures, of four targets before identifying each in target present or absent arrays. Identification errors were lower on this occasion, averaging 21 % when judgements were made from memory and 13 % with continuous reference to the recording. The theoretical and practical implications of these findings are discussed.