Regression analysis: Theory, methods, and applications.

*(English)*Zbl 0714.62057
Springer Texts in Statistics. New York etc.: Springer-Verlag (ISBN 0-387-97211-0/hbk; 978-1-4612-8789-6/pbk). ix, 347 p. (1990).

This is a book on linear least squares regression. In the opinion of the authors their treatment of the theory is the most complete of any book at this level. This statement is correct, if one admits sufficiently many levels. In my list of positive and negative features just the treatment of the theory belongs to the latter.

It is a book highly to be recommended to practitioners with some basic knowledge of mathematics and statistics. It has numerous, very instructive examples, it is clearly written and organized and probably the first text on regression with sufficiently many hints to what can be done by existing computer packages. A floppy disc helps to work out exercises with large data sets. (Unfortunately, with only one exception (p. 274) there are no hints to sources of errors when using such packages without intelligent care.) The material comprises almost all relevant methodology, the list of references is up to date. Some parts (e.g. outliers, multicollinearity) touch however only the problems and refer to more specialized literature. Nothing is said e.g. on how to design regression experiments.

Citing the authors again - and here I would agree - when the reader has mastered the contents he will have gained the experience necessary to competently practice this valuable craft, i.e. linear regression.

It is a book highly to be recommended to practitioners with some basic knowledge of mathematics and statistics. It has numerous, very instructive examples, it is clearly written and organized and probably the first text on regression with sufficiently many hints to what can be done by existing computer packages. A floppy disc helps to work out exercises with large data sets. (Unfortunately, with only one exception (p. 274) there are no hints to sources of errors when using such packages without intelligent care.) The material comprises almost all relevant methodology, the list of references is up to date. Some parts (e.g. outliers, multicollinearity) touch however only the problems and refer to more specialized literature. Nothing is said e.g. on how to design regression experiments.

Citing the authors again - and here I would agree - when the reader has mastered the contents he will have gained the experience necessary to competently practice this valuable craft, i.e. linear regression.

Reviewer: O.Krafft

##### MSC:

62J05 | Linear regression; mixed models |

62-01 | Introductory exposition (textbooks, tutorial papers, etc.) pertaining to statistics |

62-04 | Software, source code, etc. for problems pertaining to statistics |