Date: Wed, 2 Jul 2008 13:15:15 -0400
Reply-To: Peter Flom <email@example.com>
Sender: "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From: Peter Flom <peterflomconsulting@MINDSPRING.COM>
Subject: Re: Making you think (was Re: A powerpoint slide show comparing R
with SAS and SPSS_
Content-Type: text/plain; charset=UTF-8
Nat Wooding <Nathaniel.Wooding@DOM.COM> wtote
>I don't recall anyone commenting on the issue that it appears that a lot of
>functionality can be added to R by other users but you would appear to be
>at the mercy of the testing abilities of the author and anyone else who has
>used the code. On the other hand, with SAS, you know that there is rigerous
>testing of the procs and functions and yes, there are bugs but you can
>learn of these from the usage notes. As I recall the story, SAS once found
>a bug in one of the Intel chips when they were testing a stat proc and
>found that there was a decimal error in the results for something like the
>hundreth variable in the set that was being analyzed.
This is definitely an issue. I've mentioned occasionally, but not on this thread. You are at the mercy of the programmer.
OTOH, this is one reason that new statistical methods come out first in R.
>And, with SAS, you get really good technical support. Yes, sometimes the
>answer isn't the best but as far as I'm concerned, it sure beats what you
>get from the other packages unless you are willing to pay for on-damand
>support that many offer.
R has no tech support.
I had S-Plus for a while, and it has good tech support.
The R-help list can be helpful, but it is not as friendly as SAS-Help, certainly for newbies. OTOH, R-help has more statistical expertise than SAS help does. There are a few people on SAS-L who *really* know statistics at a deep level. On R-help, there are many. (This is different from being able to be helpful..... I would *not* count myself as someone who *really* knows statistics, but I do try to be helpful).
>Also, isn't IML supposed to give the statistician the ability to do the new
>gee-whiz analysis straight from the pages of the latest stat journal?
Yes, but 1) I see very few people *using* IML to do anything. and 2) When they do, there is no mechanism for sharing.
Peter L. Flom, PhD
www DOT peterflom DOT com