Date:         Fri, 11 Jun 2004 16:10:47 -0700
Reply-To:     cassell.david@EPAMAIL.EPA.GOV
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         "David L. Cassell" <cassell.david@EPAMAIL.EPA.GOV>
Subject:      Re: a question of genmod
Content-type: text/plain; charset=US-ASCII

Jonas Bilenas <Jonas.Bilenas@CHASE.COM> replied:
> No stepwise or other selection options in PROC GENMOD.  At least not
in
> SAS8.

Thank goodness!  I would be perfectly happy if *NO* SAS proc had
stepwise
selection options.  Of course, I have kvetched about this issue before.
(_ad_infinitem_ :-)

>        It is strange how some regression procs have selection options
and
> other don't.

Not at all.  Straight 'regression' procs have stepwise and other
selection
options, because the rules for making the drop/keep selections are based
on
the underlying regression assumptions.  The further from the standard
'linear model with continuous variables and normal i.i.d. errors'
framework
you go, the further you get from the underpinnings that make the
selection
processes reasonable to attempt.

> I like to use genmod since it has GLM-like model statement that allows
you
> to indicate interaction terms without having to define interaction
> variables in a data set.

Exactly!  This is one of my favorite features of GENMOD as well.

David
--
David Cassell, CSC
Cassell.David@epa.gov
Senior computing specialist
mathematical statistician