Plane answers to complex questions. The theory of linear models.

*(English)*Zbl 0645.62076
Springer Texts in Statistics. New York etc.: Springer-Verlag. XIV, 380 p.; DM 84.00 (1987).

The theory of linear models is one of the main tools in applied statistics. There exist several excellent books attacking on different levels and with different approaches the mathematical problems involved. From this point of view the author’s contribution brings nothing which is new; the level is intermediate, the approach projective in a very strict sense. So you will not find any new theorem.

As a bridge between mathematics and applied statistics it serves, however, very well. It covers most of the topics necessay to answer the - not too - complex questions arising in statistical practice and provides a wealth on motivation, examples and statistical philosophy. I have rarely found a set of exercises so well suited to the sections at wich they are aimed. I missed some help in form of references for sections which treat a topic only informally, e.g. the discussion of outliers.

As a mathematician I dislike the book (is Cauchy-Schwartz inequality only a printing error?), as a statistician I like to have it in my shelf. There seems to be no plane answer to this complex dilemma.

As a bridge between mathematics and applied statistics it serves, however, very well. It covers most of the topics necessay to answer the - not too - complex questions arising in statistical practice and provides a wealth on motivation, examples and statistical philosophy. I have rarely found a set of exercises so well suited to the sections at wich they are aimed. I missed some help in form of references for sections which treat a topic only informally, e.g. the discussion of outliers.

As a mathematician I dislike the book (is Cauchy-Schwartz inequality only a printing error?), as a statistician I like to have it in my shelf. There seems to be no plane answer to this complex dilemma.

Reviewer: O.Krafft