Adaptive statistical methods are statistical methods that are altered on the basis of an initial examination of the data that is being studied. They can be used to develop tests of significance, confidence intervals, and estimates that are adaptive in nature.

This book, which is part of a series on statistics and applied probability, is intended for those familiar with the fundamentals of multiple regression analysis and matrix algebra. It could be used as a supplementary book for courses on regression analysis. The objective of the book is to demonstrate the power of adaptive testing methods, when compared with traditional methods, despite the fact that they may not be suitable in some cases. As many of the adaptive methods have been widely distributed in the literature, the author has conducted a survey and assessment of techniques spanning a period of nearly three decades, and included the most beneficial techniques of the last ten years in this book. The utility of the methods described in this book is augmented by the provision of computer programs, so that they may be applied for real-world problems.

Adaptive methods offer an alternative to traditional methods, and help make confidence intervals shorter. They also help make tests of significance more powerful. The author has included examples from the pharmaceutical industry, agriculture, environmental and health sciences, and biostatistics.

The book has eight chapters. Chapter 1 introduces adaptive methods, including those based on ranks. An adaptive rank-based test known as HFR is discussed, and its performance is examined. Chapter 2 includes an adaptive weighted least squares test. Understanding this test is vital, since it embodies the fundamental concepts pertaining to modern adaptive tests. Chapter 3 discusses a general approach to adaptive testing. Testing a subset of coefficients in a linear model is the basis of several of the procedures discussed in the book. Chapter 4 describes the software required to conduct the adaptive weighted least squares test. This software is also included, as an appendix to the book. The proper way of using the software is demonstrated using examples.

Chapter 5 describes adaptive tests for paired data, and includes recommendations for carrying out tests. Paired data occur in many practical problems, hence adaptive tests for such data are helpful. Chapter 6 addresses adaptive confidence intervals. The kinship between tests and confidence intervals is described, along with numerous examples of adaptive confidence intervals. Chapter 7 introduces adaptive estimates, and includes examples illustrating these estimators. Chapter 8 is the concluding chapter. Topics included are rank-based adaptive tests, techniques for handling unequal variances, and adaptive methods for discretized data.

The book has a handy index, and an up-to-date bibliography. The references are adequate, and of interest to the reader. Some of them draw the reader to the author’s own work in the discipline. Perhaps the author could have discussed the strengths and shortcomings of adaptive methods in more detail. The book will be useful as a primer and reference book on applied adaptive statistical methods.