Publication Date

12-8-1972

Abstract

The problem investigated is that of estimating, in a Bayesian framework, the difference between the means of two normal distributions under the restriction that the total sample size be fixed beforehand. Two-stage designs are first discussed. In terms of the relevant prior parameters, the optimal second stage allocation is given and formulae are presented for determining the optimal first stage sample size. General sequential procedures are discussed, and a theoretical characterization of the optimal sequential procedure is given. Some numerical illustrations are also provided.

Degree Name

Mathematics

Level of Degree

Doctoral

Department Name

Mathematics & Statistics

First Committee Member (Chair)

S. Zacks

Second Committee Member

Theodore N. Guinn

Third Committee Member

Clifford Ray Qualls

Language

English

Document Type

Dissertation

Share

COinS