Statistical Methods for Quality Improvement, Third Edition

CHAPTER 13

Design of Experiments

In the early 1980s, when industrial personnel began to apply statistical quality control methods extensively in industry in the United States, the emphasis was on control charts. This prompted a number of prominent statisticians to recommend the use of “conversational” or active statistical tools to supplement the use of “listening” or passive statistical tools. Specifically, a control chart is a listening tool in that data are obtained from a process but no attempt is made to see what happens when the process is changed.

Assume that our objective is to control the diameter of a machined part within fixed limits. Our chances of successfully doing so will be increased if we can identify the factors that affect the diameter (temperature, humidity, pressure, machine setting, etc.), and the extent to which the diameter is dependent on each factor.

In this chapter we present experimental design procedures that can be used to identify the factors that affect the quality of products and services. The intent is to present a capsule account of the statistical principles of experimental design. We emphasize the planning of experiments, and the fact that statistical experiments should generally be viewed as being part of a sequence of such experiments, as opposed to a “one-shot-only” approach. Many books have been written on experimental design in which the emphasis is on the analysis of standard designs. Such analysis is deemphasized in this chapter, as are ...