Modern Applied Biostatistical Methods Using S-PLUS
S. Selvin, Oxford, 1998

Here are all the datasets and R code used by S. Selvin in his textbook on
biostatstical analysis with S-Plus. Code examples have been tested with R
version 2.6.2 (2008-02-08), running on Mac OS X.

I've tried to store all the datasets in separate machine readable files such
that data might be analyzed with other statistical software, if
needed. However, the open-source R package allows some flexibility that is
generally lacking in most of the software I ever tried. In particular, one is
able to do both classical statistical analysis and data manipulation
(e.g. apply a specific function on 3 dimensional dataset conditional on the
values taken by another fourth variable, simulating many statistics using
a large dataset (even if it is time and ressource consuming) is still possible
within R while one often needs to resort to other software to carry out such
analysis).

About the add-on
----------------
Some of the S functions used throughout the textbook are not available within
R. I thus adapt some pieces of code to run the examples. Such add-on appears
in the R files within dedicated quotes looking like `### -- R -`. I've tried
to keep the R code as close to the S syntax as possible, though sometimes it
may have been rewritten in a more efficient manner.

Figures have been generated using a small Bash script, also included in this
archive. Basically, it only processes the R file searching for some plot
commands to be issued in a postscript device (those marked `Fig. x.x`). Some
figures are missing with respect to the original ones (e.g. Fig. 2.22,
p. 102), due to missing data.

Overview
--------
Chapter 2: Descriptive Techniques.
The aims of this chapter is to provide the user useful tools to summarize a
dataset icnluding quantitative variable. Both numerical (dentral tendency,
dispersion, range, etc.) and graphical (histgram, barplot, box-and-whiskers
plot, Q-Q plot, etc.)  summaries are covered throughout the chapter. This also
includes more "modern" techniques like kernel density estimation.

Christophe Lalanne
www.aliquote.org
March 2008