Transformations to Normality

Abstract

Several analyses in this book have been improved by using a transformation of the response, rather than the original response itself, in the analysis of the data. For the introductory example of the wool data in Chapter 1, the normal plot of residuals in Figure 1.9 is improved by working with log y rather than y ( Figure 4.2). The transformation improves the approximate normality of the errors. The transformation also improves the homogeneity of the errors. The plot of residuals against fitted values for the original data, also given in Figure 1.9, showed the variance of the residuals increasing with fitted value. The same plot for log y, given in Figure 4.2, shows no such increase.