Date:         Fri, 18 May 2001 06:51:31 +0100
Reply-To:     Chris Evans <chris1@PSYCTC.ORG>
Sender:       "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>
From:         Chris Evans <chris1@PSYCTC.ORG>
Organization: PSYCTC.org
Subject:      Re: S-PLUS
In-Reply-To:  <001e01c0de16$8b191ee0$a8a1f9c3@JanIvanouw>
Content-type: text/plain; charset=US-ASCII

On 16 May 2001, at 16:43, Jan Ivanouw wrote:

> I am considering bying the S-PLUS as a supplement to SPSS. It has
> a full programming language which we do not have in SPSS, and it
> seems there are some interesting functions (as bootstrap and
> jacknife) which is more complicated to implement in SPSS.
>
> Does anyone have experiences with this program? Altso about using
> both programs together. From my trial version it seems S-PLUS is
> very slow with large data file in comparison with SPSS. Any
> comments would be welcome
I have been using SPSS for nearly 20 years (argh!) and S+ for
about three.  I find having both excellent and S+ has almost released
me from using SAS and SAS/IML to do things that SPSS can't do
(or won't do easily). Above all, S+ has provided really excellent
control over graphics and what S/R mavens call "trellis" plots:
graphics with a lot of graphs or plots conditioned on values or
ranges of one or more variables.

As you say, S+ is generally slower for things built in to SPSS and
there is a fair old learning curve: there is a usable GUI built on to S+
but to use that is almost completely to miss the power of thing which
is its object oriented power and its built in awareness of matrix and
list data structures (including high dimensional matrices/arrays and
ragged, potentially highly complex, list structures).  The matrix
awareness makes SPSS matrix look very limited and not very
friendly to me and even beats SAS/IML into a cocked hat.

S+ output is generally very easily fed back in as data which is great
for simulation work and jackknifing/bootstrapping.

S+ has a number of statistical approaches including linear mixed
effects multilevel modeling built right in.  It's factor analytic and
reliability powers tend to have me using SPSS there first but you
can (and I am .. slowly) roll your own.

I'd strongly recommend the combination.  There's a very supportive
S+ Email list (one authority is pretty caustic but very knowledgeable
and others are nicer!).  If you wanted to dip a toe in the water
without spending any money, you could try out the copyleft program
R which has much of what S+ offers, however, the learning curve
there is quite a bit steeper still unless you're on a unix/linux platform
and use ESS.  I _think_ information about R is still at:
http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html

Good luck!

Chris


Chris Evans <chris@psyctc.org>
Consultant Psychiatrist in Psychotherapy,
Rampton Hospital; Associate R&D Director,
Tavistock & Portman NHS Trust;
Hon. SL Institute of Psychiatry
*** My views are my own and not representative
of those institutions ***