Fig. 4
From: ROC curve analyses of eyewitness identification decisions: An analysis of the recent debate
Bayesian measures of eyewitness identification decisions. Lower panel: Posterior probability of guilt after suspect identifications (solid curves) and filler identification or lineup rejections (dashed curves) for two levels of true discrimination and various response criteria. Upper panel: Information gain (posterior probability of guilt − base rate) for the same data. Bias is defined by the location of the decision criterion in standard deviation units (e.g., d‘ + 1 indicates a criterion located 1 standard deviation above the mean of the guilty suspect distribution mean)