Beyond Anova, Basics of Applied Statistics by Rupert G. Miller (Book Review)

When an advisory editor of a Wiley series publishes a book, it is natural to set a high evaluation standard. I am pleased to say that this book more than meets such a strict criterion. It is innovative in its method of presentation, scholarly in its references and pragmatic in its suggested analysis. The general approach, for a particular problem, is to present the analysis based on normal theory and then consider the effect of nonnormality, unequal variances and dependence. Many references (approximately 335) are used to document the analyses and the properties of various procedures The author carefully compares the positive and negative aspects of the alternatives. Then he presents a recommendation which is usually based on theoretical properties and reasonableness in many actual situations. For example, his choice for one bioassay problem (p.190) "gives a good approximation,...,always well-defined, more intuitive and easier to explain".