The Effects of Lighting and Distance on the Accuracy of Eyewitness Identification

Abstract

This Master’s thesis examines the reliability of eyewitness identification, a common method used in criminal investigations. An identification event was simulated with a live experimental paradigm where all participants (n=180) viewed eight unfamiliar human targets. Each human target was viewed at a unique distance: 6, 8, 10, 12, 14, 16, 18, and 20 meters. One-third of the participants (n=65) viewed the targets in 0.7 lux (“starlight”) lighting, one-third (n=57) in 10 lux (“twilight”) lighting, and one-third (n=58) in 300 lux (“daylight”) lighting. Each target was presented for approximately 20 seconds, and each presentation was followed by an immediate identification via a tablet with no time limit. The identification was taken from an 8-person photo line-up where the image of the viewed target was displayed simultaneously with the images of seven other persons. The identification outcome was binary: Correct/incorrect. The data were analyzed using a logistic multi-level model (GLMM). The results indicated that the maximum viewing distance where identification still will have probative value (i.e., better than a guess) is reached at less than 20 meters when the lighting condition corresponds to starlight (0.7 lux). Additionally, the results indicated that when the lighting condition corresponds to daylight (300 lux), the accuracy of eyewitness identification stays relatively intact until 16 meters and reduces from that point forward. Therefore, the relationship between distance and identification accuracy may not be fully identical for different lighting circumstances. In light of these results, when lighting conditions are below 1 lux and distance is 20 meters or more, eyewitness identifications should be treated much more cautiously in criminal investigations.