##
**Regression. New views on an old statistical method.
2\(^e\) éd. rév.
(La régression. Nouveaux regards sur une ancienne méthode statistique.)**
*(French)*
Zbl 0788.62058

Actualités Scientifiques et Agronomiques de L’INRA. 13. Paris: Masson. 188 p. (1992).

There are only few subjects in statistics that have been so intensively described and studied as regression analysis. And still, this new French book on the subject will be welcommed by many users, who are generally no great experts in the mathematical foundations of this technique but like to understand what they are doing.

Indeed, rather than giving many mathematical developments to justify the estimation of the different regression coefficients and other parameters, the authors have wittingly chosen to put the emphasis on the significance and interpretation of the formulae. And in doing this, they are going very progressively at work, starting with the simple linear regression (chapters 1 and 2) before introducing the multiple linear regression (chapter 3) and qualitative rather than quantitative variables (chapter 4); this progression peaks in the presentation of some relevant facts and methods concerning nonlinear regression (chapter 5). A final chapter is devoted to some specific problems of the multiple regression analysis such as orthogonal regression and the best choice of the (number of) variables. Great efforts are developed by the authors to explain all the results and to illustrate them with appropriate and real-life examples.

The total approach of this work makes it a very valuable book for non- specialists, more interested in understanding the reasons for all the different aspects of regression, rather than their mathematical developments. It will be most helpful to all users of regression programs, keen on understanding what they are doing rather than blindly applying some formulae. As a pedagogic tool it is certainly a very valuable book.

Indeed, rather than giving many mathematical developments to justify the estimation of the different regression coefficients and other parameters, the authors have wittingly chosen to put the emphasis on the significance and interpretation of the formulae. And in doing this, they are going very progressively at work, starting with the simple linear regression (chapters 1 and 2) before introducing the multiple linear regression (chapter 3) and qualitative rather than quantitative variables (chapter 4); this progression peaks in the presentation of some relevant facts and methods concerning nonlinear regression (chapter 5). A final chapter is devoted to some specific problems of the multiple regression analysis such as orthogonal regression and the best choice of the (number of) variables. Great efforts are developed by the authors to explain all the results and to illustrate them with appropriate and real-life examples.

The total approach of this work makes it a very valuable book for non- specialists, more interested in understanding the reasons for all the different aspects of regression, rather than their mathematical developments. It will be most helpful to all users of regression programs, keen on understanding what they are doing rather than blindly applying some formulae. As a pedagogic tool it is certainly a very valuable book.

Reviewer: E.Trauwaert (Mol)

### MSC:

62J02 | General nonlinear regression |

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

62J05 | Linear regression; mixed models |