Imagine the following scenario. A thief throws a brick through a shop window, grabs the merchandise on display, and runs off before the shopkeeper sees anything. The only two witnesses to the crime were standing in the street at the time, and they come forward to help the police. One witness – let's call him Jake – says that he is very confident that the thief was dark-haired, whilst another witness – Sam – is fairly confident that the thief had blonde hair. Clearly there is a discrepancy in the descriptions, and in the confidence that the witnesses espouse in their descriptions. Should the police (and later the courts) have greater faith in the witness with the most confidence? Whilst the majority of members of the public might say “yes,” the psychological literature does not warrant such a clear-cut and positive response.

Expert psychologists were asked a generalized version of this question, over ten years ago, in a survey of expert opinion by Kassin, Ellsworth, and Smith (1989). Kassin et al. asked sixty-three experts in the area whether they felt that various forensic psychology findings were reliable enough for psychologists to present in courtroom testimony. One such issue was the weak relationship between confidence and accuracy. An overwhelming majority of experts in the area (87 percent) agreed that the effect (i.e., the lack of relationship between confidence and accuracy) was reliable enough to present as testimony, and a similar figure (83 percent) reported that they would be prepared to testify themselves (37 percent already having done so).