Purpose: A popular cognitive explanation called "the law of small numbers" suffers from a trouble on its prediction power so that it is not clear to predict one' s judgement about whether the runs of event last or break. In this experiment, we examine the validity of two complementary explanations, on one of which people's making prediction depends on their belief about the process produced runs, on the other of which it depends on success and failure of their prediction of the outcome on last trial. Method: On within-subject ABAB design, 40 subjects made prediction on total 864 trials of binary outcomes, consisting of four phase of the series produced by either Bernoulli or sine with random noise process. Results : Subjects demonstrated their capacity of discrimination by two processes. They judged with negative correlation as runs extended under Bernoulli phase, where as with positive correlation under sine with noise phase The analysis of success and failure didn't show the tendency that previous study suggested. Discussion: We discuss the results can interpret from the study about the bias of subjective randomness, where as the law of small numbers showed other trouble to explain two opposing prediction.

論文