Date: Wed, 5 May 2010 10:12:04 -0400
Reply-To: Robert Feyerharm <robertf@HEALTH.OK.GOV>
Sender: "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From: Robert Feyerharm <robertf@HEALTH.OK.GOV>
Subject: Re: interaction & main effect question
Thank you for the suggestions.
I agree that statistical significance shouldn't be the only factor in
including a main effect or interaction term. The Akaike Information
Criterion is one alternative to p-values for selecting main effects &
interactions in models.
Perhaps plotting graphs would also be helpful to identify strong