Re: [S] Comments on Concerns over New Versions of S-PLUS for Windows

John Maindonald (
Tue, 26 May 1998 09:46:56 +1000 (EST)

Andrew Kuhn wrote
> Some users, such as myself, may see the concentration on the
> development of the gui instead of new statistical features as the start
> of the same problem. The recent inquires on the list from Mathsoft
> for suggestions of what statistical methods should be next to add
> is encouraging.
> I think that Dr. Millard's post is dead on. However, it still makes me
> think about all the methods I would like to see S-plus add and what
> Mathsoft's attitude is about, what I consider to be, it's core users.

There are several areas where the interests of novice and more
experienced users strongly coincide. Boths classes of users
(1) must have a reliable engine underneath;
(2) must be able to trust the documentation.

I have been bitten often enough that I approach almost anything I do
with a fair amount of trepidation. But consider the novice user who
gets values from predict() that are onviously nonsense, or cannot work
out why, when missing values are present, table() has spirited away a
substantial number of observations. How does he/she respond to
statements in the documentation of factors that are wrong &/or
misleading, & which StatSci seems not disposed to correct? While I
understand that moves are afoot to do something about the factor mess,
I am concerned that statements which apply to present functions,
functions which even in deprecated are (like category) likely to stay
lurking in the code for a long time, will be left there incorrect and

John Maindonald email :
Statistical Consulting Unit, phone : (6249)3998
c/o CMA, SMS, fax : (6249)5549
Australian National University
Canberra ACT 0200
This message was distributed by To unsubscribe
send e-mail to with the BODY of the
message: unsubscribe s-news