TABLE OF CONTENTS Several techniques have been developed, based on the general design principles presented in the first 15 chapters of this book and on other well-known statistical methods. These approaches are not discussed in detail because they are presented very well in the references. It is hoped that a background in experimental design as given in the first 15 chapters will make it possible for the experimenter to read and understand the references.
The methods to be discussed are covariance analysis, response surface experimentation, evolutionary operation, analysis of attribute data, incomplete block design, and Youden squares.
Occasionally, when a study is made of the effect of one or more factors on some response variable, say Y, there is another variable, or variables, that vary along with Y. It is often not possible to control this other variable (or variables) at a constant level throughout the experiment, but the variable can be measured along with the response variable. This variable is referred to as a concomitant variable X because it "runs along with" the response variable Y. Thus, to assess the effect of the treatments on Y, one should first attempt to remove the effects of this concomitant variable X. This technique of removing the effect of X (or several X's) on Y and then analyzing the residuals for the effect of the treatments on Y is called covariance analysis.
Several examples of the use of...
