R Documentation(s)

Kjetil Kjernsmo <kjetil.kjernsmo at astro.uio.no> writes:

On Tue, 2 May 2000, Emmanuel Paradis wrote:

To me, the greatest issue is the audience targeted by a documentation. Many
specialists are potentially interested by R, from the top-statisticians to
the users who hardly knows about statistics but need to run some tests (at
least to publish their results). It seems to me that some queries on the
list have more to do with "using statistics" than with "using R". As long
as non-statisticians use R (like me), this seems inevitable.

I just occured to me that something that would probably boost the number
of users, is a compendium on "How to do the things one usually does in a
beginners statistics course for non-statisticians". Many are introduced to
statistical software in such a course, and I can imagine many will never
use any other software than they were introduced to in the course. 
Such a compendium might catch new users and their instructors at the very
start. 

Here at my university, the standard package has been MINITAB. I have had
one single session of MINITAB myself, and it appeared to me as an endless
quest for the right dropdown menu point or dialog-box point. I switched to
S-plus, and to R after learning about it the week after. The reason I
mention this is the seemingly irrational urge many have to have drop-down
menus, even though the path to what you need is considerably longer than
just typing. If the compendium is able to show that the things you need
are not at all far away, and much more logical than drop-down menus, I
think many will be drawn towards R.

Perhaps it is not exactly a "compendium" in the sense you describe but
there is an R package called Devore5 containing all the data sets
for a specific introductory statistics text, Jay Devore's "Probability
and Statistics for Engineering and the Sciences (5th ed)".  Because
there are facilities for documenting data sets in R and because you
can use
 example(topic)
to run the example section from a manual page as a script in R, I can
document all the examples in the text and show the R code that
corresponds to the example.

This approach is tied to the specific text but, for someone who has
access to that text, it provides a ready answer to questions like "How
do I do a two-sample paired t-test in R?".  They go to the appropriate
section of the text, see that Example 9.8 does such a test, and then
run
 library(Devore5)
 example(xmp09.08)
to see the commands and resulting output.

It is interesting that it is extremely difficult to package these
kinds of demonstrations for students when using a GUI-based system
like Minitab.  I had previously used Minitab in the course where I use
this text and I found that I either explained the command-line
interface or I did an in-class demonstration with the GUI or I was
spending my time describing things like "you click on the Statistics
menu and that gives you an option of Analysis of Variance (or maybe it
is called ANOVA, I can't remember) and that brings up a panel where
...".  Those kinds of explanations are very hard to follow.  

With the command line interface in R I can put up a transparency and
show the class "type this and you get that".

Douglas Bates                            bates at stat.wisc.edu
Statistics Department                    608/262-2598
University of Wisconsin - Madison        http://www.stat.wisc.edu/~bates/
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._