Assessment of Subjective Probability Distributions

The hypothesis of this study conducted 50 years ago, is that individuals presented with the same information base cannot agree on their assessments of the subjective probability distribution for a particular variable of interest. Moreover, this is caused by an inability to translate judgment and information into a subjective probability. Individuals do show good agreement regarding the central tendency of the distribution, but poor agreement regarding the entire distribution. They are also poor processors of Bayesian information. Three groups of individuals were the subjects of this study. The first group were practicing Operations Research professionals well-versed in both classical probability theory and the development and application of subjective probabilities. The second group consisted of undergraduate seniors in an industrial engineering department also well-versed in classical probability theory but had no work experience and no exposure to subjective probability. The third group consisted of a number of market research analysts in a large chemical company who had only small exposure to both classical and subjective probabilities, but who would normally represent the type of individuals called upon to assess subjective probabilities in an industrial setting