SAMPLE SIZE AND THE REVISION OF SUBJECTIVE PROBABILITIES.