The non-orthogonal design of experiments. Discussions. Reply

The computer revolution has had surprisingly little impact on the philosophy, theory, teaching and practical use of experimental design. This absence of effect can be seen most clearly in the adherence to a small class of orthogonal structures as the principal content of our teaching and as the recipe book used by experimenters. A more opportunist philosophy of design should be taught; this can display the statistical principles of the subject better than the recipe approach. Many exciting research areas, needing computing power, are available, some of which are discussed briefly