Optimal Bayesian Design for a Logistic Regression Model: Geometric and Algebraic Approaches

by Marilyn Agin
and
Kathryn Chaloner
Technical Report No. 622
School of Statistics
University of Minnesota
August, 1997


Abstract

A simple logistic regression model with a known slope parameter provides a simple model for understanding optimal Bayesian designs. For this model the geometric approach of Haines (1995) and the algebraic approach of Chaloner (1993) for finding optimal designs for a prior distribution with just two support points are reviewed, compared and discussed. A result is given, using the algebraic approach, for a three point prior distribution and difficulties of the geometric approach are illustrated.


Click here to download the complete PostScript technical report.