Abstract
Sixty adults from the city of Linz took part in this experiment. They read a storyβThe War of the Ghostsβin pairs, and were required to reproduce it either jointly, in dyads, or singly, and either immediately following or one week later. In addition they were asked a number of specific questions about the story, and gave various ratings of confidence and completeness. As predicted, social (dyadic) performance increased confidence, completeness and the incidence of implicational errors (errors which do not contradict the original). Delay significantly diminished confidence. Analyses of confidence for right and wrong answers indicated that subjective testimonial validity was substantial in all conditions. Analyses of objective testimonial validity showed that dyads are more trustworthy than individuals when they are correct, but are less trustworthy when they happen to be wrong. They overall conjidence I accuracy correlation across conditions of +0.6 masks the fact that the greatest obstacle to valid testimony is inappropriate confidence in wrong answers, especially in dyads, and especially immediately after the event. The incidence of implicational errors was highly related positively to measures of confidence and accuracy, whereas confusional errors were independent. Implicational errors are regarded as an especially important element in schematic recall, and are an important aspect of the superiority of social performance.