Search

This paper deals with the frequently encountered problem of estimating the probabilities of future events when there are no appropriate historical data with which to compute the probabilities. If adequate data were available, the estimation of probability distributions would be obtained from histograms or relative frequency counts. But in many instances the necessary historical data are not available, and the needed information must be obtained from estimates provided by knowledgeable experts. We include in this paper an approach for obtaining these subjective estimates. The strategy is based on the results of the research work of a number of investigators who have studied this problem, and it differs from the strategy that one might be tempted to use on intuitive grounds. In dealing with this subject, quite unfamiliar and therefore suspect to most readers, we have referenced some of the relevant psychological research. In particular we have drawn heavily upon the reviews by Peterson and Beach (1967). Heinrich (1971), and Slovic and Lichtenstein (1971). Two questions arise: Are subjective probabilities accurate? How can subjective probabilities be obtained? In this paper we examined some practical suggestions for obtaining subjective probabilities. We feel confident, if we may use such a casual phrase in the context of the present paper, that subjective probabilities will be much more frequently used by consultants in the near future than they are now. Our reasons are the following: (1) The days when people forced the problem to fit the technique are gone, and since many problems are probabilistic and do not fit the classical OR techniques. we will see much greater use of techniques that require probability distributions. (2) The rate of social and technological change is accelerating, and thus historical data bases appropriate for estimating probabilities of future events will be less frequently encountered. (3) More and more professionals are accepting t... [ABSTRACT FROM AUTHOR]