A SIMPLE BAYESIAN MODIFICATION OF D-OPTIMAL DESIGNS TO REDUCE DEPENDENCE ON AN ASSUMED MODEL

D-optimal and other computer-generated experimental designs have been criticized for being too dependent on an assumed statistical model. To address this criticism, we introduce the notion of empirical models t hat have both primary and potential terms. Combining this idea with th e Bayesian paradigm, this article proposes a modification of the D-opt imal approach that preserves the flexibility and ease of use of algori thmic designs while being more resistant to the biases caused by an in correct model. These designs provide a Bayesian justification for reso lution IV designs. Several theoretical examples and a practical exampl e from the literature demonstrate the advantages of the proposed metho d.