Sequential Monte Carlo Methods in Practice

approach, but also the tests that detect higher-order moment differences. It also presents a comprehensive treatment of modeling ties. Although its subject is highly technical, the book somehow maintains a good balance between theories and applications. The excellent Appendix is self-contained and very easy to read. Due to the nature of the book, it may not be suitable as a textbook for any graduate courses. But it would be suitable as a reference book for such graduate courses as categorical data analysis, as well as nonparametric statistics at the doctorate level. Chapter 1 serves as an introductory chapter, including an interesting discussion on parametric or nonparametric tests, a worked instructor’s example, the outline and scope of the book, and applications of nonparametric methods to sensory evaluation. The instructor’s example is quite interesting, illustrating points concerning standard nonparametric tests and the detection of linear and quadratic effects. Analysis of sensory data is extremely important, particularly in consumer market research and food science. I would have liked to see an expanded discussion on this topic, covering statistical methods as well as applications. Chapter 2 focuses on ties, how they have been traditionally treated and how different models lead to different treatments. The context is mainly what is for many the Ž rst nonparametric test, the sign test. The novel idea of decomposing a test statistic into its components to give a more detailed scrutiny of the data is also introduced. Chapters 3–8 focus on tests for which the data may be given in twoway contingency tables. Standard tests presented assess mean or correlation departures from the null. Because the traditional correlation is the (1, 1)th bivariate moment, all are essentially Ž rst-order moment tests. All these tests are extended to detect higher-moment departures from the null. In these chapters, several standard nonparametric tests, including Pearson’s chi-squared test, the Kruskal–Wallis test, the generalized median test, Page’s test, Yates’s test, sign test, Gart’s test, Friedman and Cochran tests, Stuart’s test, Durbin’s test, and Spearman’s test, are overviewed and linked to the tests based on models for data presented in contingency tables. Extensions to these nonparametric tests are also provided. The idea of decomposing a test statistic into its components is further explored and illustrated using various real data examples. These chapters would be easier to read and follow if some of the mathematical symbols such as the square root had been displayed better. Chapter 9 gives an overview of oneand S-sample smooth tests of goodness of Ž t. Chapter 10 includes a discussion of recent work on partially parametric testing. This has grown out of the work on one-sample smooth tests of goodness of Ž t. The probability function of the distribution tested for is nested in a rich family of distributions. Extensions of some current two-way table work to multiway tables are also included. Overall, this book is an excellent addition to the statistical literature. It contains a large number of examples and applications. The book could be greatly enhanced by introducing a set of exercises at the end of each chapter and by providing statistical computing codes in either SAS or StatXact for the worked examples. Another area needing more coverage is the power study of the tests presented in the book.