Experimental designs that result in orthogonal data structures allow us to get the most out of both our analysis and our research budget. As a result, designed experiments have been used for everything from basic research to process optimization. These multiple roles make it crucial to understand the implicit and explicit assumptions behind the use of an experimental approach.

Consider the problem of process optimization. It seems quite simple to carry out a sequence of experiments in such a way that we can move our process toward a set of operating conditions that will optimize the critical outcomes. George Box and Norman Draper came up with the idea of evolutionary operation. Stan Deming taught classes on the use of simplex algorithms for optimization. Over the years many others have followed the same path with various experimental approaches for process optimization, and today you can find software, books, and classes on this topic.

 …