Bayesian Experimental Design: A Review

Kathryn Chaloner and Isabella Verdinelli

Abstract:

Non Bayesian experimental design for linear models has been reviewed by Steinberg and Hunter (1984) and in the recent book by Pukelsheim (1993); Ford, Kitsos and Titterington (1989) reviewed non Bayesian design for nonlinear models. Bayesian design for both linear and nonlinear models is reviewed here. We argue that the design problem is best considered as a decision problem and that it is best solved by maximizing the expected utility of the experiment. This paper considers only in a marginal way, when appropriate, the theory of non Bayesian design.

Here is the full postscript text for this technical report. It is 428 kbytes.