Date: Tue, 19 Oct 2004 15:52:59 -0400
Reply-To: nomi <email@example.com>
Sender: "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From: nomi <sajeelm@GMAIL.COM>
Subject: Best model selection criterion or test in Logistic regression
Content-Type: text/plain; charset=US-ASCII
I'm building a logistic model and I'm at the point where I have to
choose a model out of three possible models. All the three models have
the same variables the only differences between them is the number of
regressors/variables. Now, one way of comparing these models is to
look at the K.S values, but the K.S value is not that different to
give a clear cut decision.
Would any of you know of a test with wich these models could be
compared? or any statistic other than K.S that could be reffered to?
(somtheing maybe similar to the F-Drop test in OLS resgression)
Any info will be appreciated.