##
**An introduction to multivariate statistical analysis.
3rd ed.**
*(English)*
Zbl 1039.62044

Wiley Series in Probability and Statistics. Hoboken, NJ: Wiley (ISBN 0-471-36091-0/hbk). xx, 721 p. (2003).

This is the third extended edition of the very famous book on multivariate statistical analysis; its first edition appeared in 1958, see the review Zbl 0083.14601. It reacts on recent developments in the area. The book traditionally concentrates on methodology. The third edtion covers newly or more extensively the following items:

1. Patterns of dependence and graphical models (Chapter 15); 2. Measures of correlation and tests of independence; 3. Reduced rank regression, including the limit-information maximum-liklihood estimator of an equation in a simultaneous equations model (Chapters 12 and 13); 4. Elliptically contoured distributions. The elliptically contoured are introduced as a generalization of normal distributions and many results for normal distributions are extended to this type of distribution – most of the chapters contain one section with such an extension.

As the author admits, due to space limitations this edition does not reflect fast developments in some areas, as e.g., resampling methods and nonparametric techniques in multivariate analysis.

The book belongs among classical famous texbooks and in its present form takes into account developments during the last 20 years. The book traditionally concentrates on methodology. It is assumed that the reader is familiar with basic statistical concepts. The book is really well written. This edition will be certainly welcomed not only by professional statisticians and students of statistics but also users of statistical methods.

1. Patterns of dependence and graphical models (Chapter 15); 2. Measures of correlation and tests of independence; 3. Reduced rank regression, including the limit-information maximum-liklihood estimator of an equation in a simultaneous equations model (Chapters 12 and 13); 4. Elliptically contoured distributions. The elliptically contoured are introduced as a generalization of normal distributions and many results for normal distributions are extended to this type of distribution – most of the chapters contain one section with such an extension.

As the author admits, due to space limitations this edition does not reflect fast developments in some areas, as e.g., resampling methods and nonparametric techniques in multivariate analysis.

The book belongs among classical famous texbooks and in its present form takes into account developments during the last 20 years. The book traditionally concentrates on methodology. It is assumed that the reader is familiar with basic statistical concepts. The book is really well written. This edition will be certainly welcomed not only by professional statisticians and students of statistics but also users of statistical methods.

Reviewer: Marie Huškova (Praha)