Incorporating human judgments based on a linear combination of random variables

This paper deals with the subject of incorporating human judgments under uncertainty. If a decision maker has little knowledge with regard to the parameters of interest, he will consult a number of experts and obtain their judgments in the form of subjective probability distributions on some related parameters. It this is the case, the decision maker must combine the distributions assessed by the experts and form a single distribution. First, the authors review the two approaches to combine the individual opinions: (1) the opinion pool and (2) the Bayesian approach. Second, they propose a new approach to the problem of incorporating human judgments based on a linear combination of random variables. Finally, the authors prove that the generic model structure of the proposed new approach is exactly the same as the structure of the Bayesian model under the given assumptions. [In Japanese.]