Knowledge that Transforms

The important question of whether confidence and accuracy are related in eyewitness identifications involves a number of methodological considerations peculiar to the eyewitness setting. Inconsistencies in the literature are attributed in part to these methodological factors. Distinctions between four types of identification error (i.e., false identification, identification of foil in perpetrator-present lineups, identification of foil in perpetrator-absent lineups, and false rejection) and two types of accuracy (accurate identification and accurate rejection) are critical to an applied analysis. Eyewitness confidence in foil identifications, although of potential theoretical interest, should not be included in the forensically relevant calculations of confidence-accuracy relationships. Choice and nonchoice responses of witnesses should be analyzed separately in terms of whether confidence has predictive utility. Applied generalizations should take account of routine real-world events such as attorney-witness briefings and other extra-memorial factors that can have robust effects on confidence but not accuracy. Attempts at generalization should also take account of the imperfect relationship between witnesses ' self-rated confidence and jurors ' ascriptions of confidence. The common practice of squaring correlations to assess predictive utility is criticized and an alternative statistic is proposed. Does the confidence with which an eyewit-ness proffers identification testimony bear a useful relationship to the accuracy of the identification? This nascent issue has pro-found practical significance, especially in light of recent studies indicating that eyewitness confidence is a strong determinant of the credibility that people ascribe to identification testimony (e.g., Lindsay, Wells, & Rumpel,