Designing experiments for binary data using search algorithms

In some experiments in the physical and biological sciences, a binary response is of primary interest and is often described by a generalized linear model. Examples include experiments in food technology and studies in chemistry where the outcome is whether or not a salt is formed in a chemical reaction. For such experiments, designs that are efficient under the assumption of a linear model may be inadequate for the description and prediction of the response. The generation of designs using a search algorithm is addressed for completely randomized designs when a generalized linear model describes the response. A method of assessing the designs is discussed and illustrated by examples.