Answers are grounded in this paper and cite their sources.
Suppose that we have a sample of observations on a continuous random variable with a distribution depending on one unknown parameter θ. We consider the problem of making inferences about θ when there is no prior information about the value of the parameter. Some examples are discussed for which it is not possible to find Bayesian methods possessing the properties required by frequentist statisticians. The magnitude of the difference between comparable Bayesian and frequentist probability statements is calculated. It is found that although the difference is not usually large, even in small samples, it may be serious in extreme cases. We discuss the reason for the discrepancy between the theories and suggest that it is a consequence of restricting attention to fixed sample size experiments. The argument is supported by a detailed discussion of an experimental situation common in industrial life testing.
No takes yet. Share an insight, caveat, or question.
D. J. Bartholomew (1965) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: