Decision Making with the Aid of a Subjective Prior Distribution.
UNIVERSITY OF SOUTHERN CALIFORNIA LOS ANGELES DEPT OF QUANTITATIVE BUSINESS ANALYSIS
Pagination or Media Count:
This paper discusses a methodology to establish a subjective prior distribution via an interactive computer program on a CRT. The program asks specific carefully worded questions. The answers to these questions are checked for logical consistency and are translated into fractiles of a subjective probability distribution. The fractiles as well as other summary measures like the mean and variance are printed out. In order to find a conjugate prior distribution for a Bayesian treatment, the mathematical form of a possible conjugate prior needs to be ascertained. The paper illustrates how the parameters of a lognormal distribution might be obtained from the fractiles of the subjectively derived prior. Author
- Operations Research