\documentstyle[11pt]{report}
\begin{document}
\begin{center}
{\bf                       
Statistical Models in S 

                               edited by 

                 John M. Chambers and Trevor J. Hastie 

                                 1992 

                              608 pages  

                          ISBN 0-534-16765-9 

             Wadsworth and Brooks/Cole Advanced Books and Software 

}
\end{center}

Contents:
\begin{enumerate}
\item An Appetizer
    (by J.M. Chambers, T.J. Hastie)
\item Statistical Models 
    (by J.M. Chambers, T.J. Hastie)
\item Data for Models
    (by J.M. Chambers)
\item Linear Models
    (by J.M. Chambers)
\item Analysis of Variance; Designed Experiments
    (by J.M. Chambers, A.E. Freeny, R.M. Heiberger)
\item Generalized Linear Models
    (by T.J. Hastie, D. Pregibon)
\item Generalized Additive Models
    (by T.J. Hastie)
\item Local Regression Models
    (by W.S. Cleveland, E. Grosse, W.M. Shyu)
\item Tree-Based Models
    (by L.A. Clark, D. Pregibon)
\item Nonlinear Models
    (by D.M. Bates, J.M. Chambers)
\end{enumerate}
\begin{itemize}
\item
Appendix A. Classes and Methods: Object-Oriented Programming in S
    (by J.M. Chambers)
\item
Appendix B. S Functions and Classes
(Formal documentation.)
\end{itemize}

New programming functionality has been added to the {\em New S} language
since the publication of the {\em New S} manual in 1988 (The New S language:
A programming environment for data analysis and graphics, by R.A. Becker,
J.M. Chambers, and A.R. Wilks).
By using this extension to the {\em New S} language, the ten authors of this
manual are able to develop a
unified approach to the fitting and analysis
of a fairly complete collection of response models (traditional and recent).
The book represents the first major effort in this area.
I highly recommended it to anyone
interested in using New S, or
in applying more recent response models, or
in research in statistical computing.

The book is well organized and reads more like a
single treatise than a collection of papers.
The editors and the authors are to be congratulated for a remarkable job.
Each chapter is organized into four primary sections.
The first describes the statistical methodology, the second how to use
the S functions and data structures, the third how to extend or
specialize the given software, and the fourth contains more detail on the
computations.
Consequently, by reading only the first two primary sections of
each chapter one comes away well equipped to use the software in
a host of applications.
For most readers this will be enough.
As interest and circumstance demand, the remaining sections of any chapter
can be read with profit.
Although early chapters are required reading for later chapters,
chapters 7 through 10 can be read independently of one another.

The language extension has moved New S in the direction of
object-oriented programming.
For S users this is an important and exciting development and will expose
many of them to some of the ideas of object-oriented programming for the first
time.
Because of this it is a pity that the extensions of the language take New S
such a small
step towards realizing the power of object-oriented programming.

As it stands, New S is quite different from what is generally understood
to be an object-oriented programming language.
As is pointed out in Appendix A of the book, New S has much in common with
object-oriented languages {\em but differs in a number of respects
related to the nature of S} (p. 457).
Chief among these differences is that while New S purports to be a {\em
class based} object-oriented programming language, it falls short of
satisfying any definition of one.
Despite the impression given to the casual reader,
there is no such thing as a {\em class} in this New S; there is merely an
attribute called {\em class} which can appear on any S data structure.

Specifically,
New S's functional programming style has been extended so that the
user can write functions which dispatch to other functions
(called {\em methods}) depending only
on the value of the {\em class} attribute of one of its arguments.
True, this makes it possible to write functions which are {\em generic}
but in my opinion it is not quite object-oriented programming.
Instead,
the function dispatching ({\em method lookup}) in the New S model has been
confounded with the definition of a class.

By contrast, in a class-based
object-oriented programming language, classes are data structures
which can themselves be manipulated.
Minimally, they can be related one to another
through the notion of inheritance.
No such data structures exist in New S.
Because New S's class attribute only determines the method lookup to be used for
a particular instance it might be better understood by the reader had it been
named the {\em method-precedence} attribute.

In summary, the book and the attendant software are interesting, valuable,
and important.
The book should be of interest to a wide audience.
The authors are to be congratulated for an excellent well organized book that
represents the first major effort towards a unified presentation of statistical
response models
and which takes New S an important first step towards object-oriented
programming.
My only caveat to the consumer is that
as an object oriented programming system the extended New S system is
unusual and in my opinion falls short of fulfilling the
promise of object-oriented programming.
\begin{center}
R.W. Oldford\\
Department of Statistics and Actuarial Science\\
University of Waterloo\\
Waterloo, Ontario\\
N2L 3G1\\
Canada
\end{center}
\end{document}