[R] R code for "Statistical Models in S" ?

[Charilaos Skiadas]

On Mar 1, 2007, at 4:36 PM, Robert Duval wrote:

> You might want to start looking at the FAQ's
>
> http://cran.r-project.org/faqs.html
>
> in particular
>
> http://cran.r-project.org/doc/FAQ/R-FAQ.html#R-and-S
>
Thanks I must admit that I had not looked at the FAQ's, but I have  
now and though it might become relevant a bit down the line, it  
doesn't seem to answer my questions, unless I've missed something in  
there. The only relevant bit I found was this phrase:

"Apart from lexical scoping and its implications, R follows the S  
language definition in the Blue and White Books as much as possible,  
and hence really is an “implementation” of S."

The question (one part of it at least) had to do with the data sets  
and functions used, and the fact that some of these data sets are not  
there in R. In the book they refer to a "data" package for instance,  
which seems to contain things different than R's "datasets" package.  
So the question was if the necessary data sets are available somewhere.

The second part was in particular about a call to "plot", namely "plot 
(solder.balance)", which in S according to the white book is supposed  
to produce the graph in the top of page 3, for those having the book,  
the caption of the plot being: "A plot of the mean of \textbf{skips}  
at each of the levels of the factors in the solder experiment."
I have now found out, thanks to "A handbook of statistical analyses  
using R", that the corresponding call in R would be: "plot.design 
(solder)".

I understand of course that not every difference between  
implementations in S and R should be documented, but I was hoping  
that other people who have already gone through this book would have  
documented these differences. I guess not, and I will be doing so now.

> robert

Haris Skiadas
Department of Mathematics and Computer Science
Hanover College