ABSTRACT
Response surface designs are normally used at the last stage of experimentation-if they are required at all. The important factors have already been determined by screening experiments, experience, or sound theory. The goal at this stage of experimentation is to describe in detail the relationship between the factors and a response. It is now known (or strongly anticipated) that a simple linear model – even with interactions – is not good enough to adequately represent that relationship. If we look at Figure 9.1, we see that we want to move far to the right on our Knowledge axis.
