Statistical Methods for Rates and Proportions

It is always nice to see a book’s second edition (2E) that highlights in its Preface the improvements that were indicated as deŽ ciencies in the Technometrics review of its Ž rst edition (1E). Patch (1991) noted that “conŽ dence intervals and graphical methods do not receive enough attention in the book” (p. 376). The authors state (Preface, p. xix) that “the most extensive revision is the greatly increased attention given to measures of effects.” They also claim “an increased use of graphics.” The primary attributes listed by Patch (1991) for the book were “thorough, well-written, and uses a somewhat nonstandard approach” (p. 345). The latter comment relates to the book’s subtitle, “A Model-Comparison Approach.” In addition to the usual introductory chapters, the chapters of the 1E were divided into two sections. The Ž rst of these sections, “Model Comparison for Between-Subject Designs,” encompassed the usual topics that one encounters in the design of experiments. The second, “Model Comparison for Design Involving Within-Subjects Factors,” had four chapters that focused more on the designs applicable in the behavioral, animal, and health sciences. The primary addition in the 2E is a new third section of chapters that extends the material in the second section of the 1E. This new section in the 2E, “Alternative Analysis Strategies,” comprises two chapters. The Ž rst, “An Introduction to Multilevel Models for Within-Subjects Designs,” deals with the models for longitudinal data. The focus is primarily on the models that can be estimated using PROC MIXED in SAS. Although the chapter is more about models than designs, there is also considerable discussion on the set up of good designs. The second new chapter, “An Introduction to Multilevel Hierarchical Mixed Models: Nested Designs,” also focuses on the capabilities of PROC MIXED for analyzing the nested designs. All of this chapter is about modeling. Patch (1991) summarized all 14 chapters of the 1E. In the Preface to this 2E, the authors note the speciŽ c improvements they made to each of these original 14 chapters. Another excellent feature is the CD-ROM that accompanies the book, which includes a basic statistics tutorial, a regression analysis tutorial, and data Ž les in SAS, SPSS, and ASCII formats. The programming codes for SAS and SPSS, as well as the project descriptions for all of the datasets, can be found at the book’s website.