Are defendants guilty if they were chosen in a lineup?

Abstract

Courts overbelieve witnesses who choose suspects in lineups. The extent of the problem depends on the probability of defendants who were chosen actually being guilty. According to Bayes' theorem, the probability of their guilt depends as much on the relative number of guilty who are chosen [p(C/G)] as on the number of innocent suspects [p(C/not G)]. Evidence is presented, based on both experimental data and archival reports of real eyewitness cases, that p(C/G) = 0.29 and p(C/not G) = 0.098 are conservative estimates. This leads to 0.247 being the probability of innocence if chosen (assuming no a priori presumption of guilt or innocence). The problem, then, is serious. Potential remedies are discussed.