Dissociation between confidence in a prediction and the success of the prediction

An experimental attempt was made to dissociate a person’s confidence in his prediction from the objective success of that prediction. High school students were administered 85% partial reinforcement for specific predictions, for estimates of likelihood of occurrences, or for combinations of the two. Ss reinforced for high levels of estimate made higher estimates than Ss reinforced for low levels, though levels of particular predictions did not differ for the two groups. Reinforcement for predictions influenced S’s estimates more than reinforcement for estimates influenced prediction. Awareness was found to influence performance of compound responses under partial reinforcement.