Predictive value, Redundancy, and Confidence

Confidence in judgements for prediction should be related to the amount of information sampled, with the stipulation that each piece of information sampled is independent of others. If, in forming judgements, the same information is encountered twice, it should not be treated as two independent items of information. Much information encountered will be redundant in the sense that part of the information is related, somewhat imperfectly, to other parts. Much of the information will also be consistent, indeed, if it is redundant it is almost bound to be consistent. However, consistency of information sources which are not independent adds little to predictive validity. Consequently, consistency of dependent sources is not a good criterion for determining confidence in judgement.

(see also: Illusion of validity, Multiple end point fuzziness or Nostramdamus effect)

» Return to belief, judgement, and clear thinking »