Chapter 11 - Introduction to Design of Experiments: This chapter is largely independent of previous chapters, although some previous vocabulary is used here

Abstract

This chapter introduces designs of experiments. More information can be gained in some cases by designing experiments to use the paired t-test rather than the unpaired t-test. In many other cases, there is a similar advantage in designing experiments carefully. It may be that some factors affect the results but are of no prime interest: They are interfering factors or lurking factors. Often these interfering factors cannot be controlled at all, or perhaps they can be controlled only at considerable expense, i.e, some of the factors interact in the sense that a higher value of one factor makes the results either more or less sensitive to another factor. We have to consider these complicating factors in planning the set of experiments. The chapter compares experimentation with the use of routine operating data and covers a scale of experimentation. Experiments should be done on as small a scale as will give the desired information. It discusses one-factor-at-a-time compared to the factorial design with examples and also discusses in detail the replication, the interfering factors that will bias the results, and the fractional factorial designs.