Bayesian Inferences on Functions of the Parameters of Univariate Distributions using Estimator Distribution.

摘要

In the Bayesian approach to determining inferencing information the likelinood function is often used as the conditional distribution of sample outcome given the population parameters. For the more difficult problems involving quantities which are functions of more than one population parameter,use of the likelihood function can lead to very tedious computations. Computational efficiency can be improved in many instances if the distribution of estimators is used rather than the likelihood function. The use of the estimator distribution in determining Bayesian intervals is discussed with application being made to the Weibull mean. Exactness from a classical frequency viewpoint of the Bayesian intervals assuming uniform priors was also studied. (Author)