Date:         Thu, 21 May 1998 10:51:58 -0500
Reply-To:     "William P. Eveland, Jr." <weveland@FACSTAFF.WISC.EDU>
Sender:       "SPSSX(r) Discussion" <SPSSX-L@UGA.CC.UGA.EDU>
From:         "William P. Eveland, Jr." <weveland@FACSTAFF.WISC.EDU>
Subject:      Re: Interaction
In-Reply-To:  <D79469854F42D1119A4900805F15154032A487@NYMAIL>
Content-Type: text/plain; charset="us-ascii"

Good question.  The answer is no.  The product term only tests the
interaction when the main effects (the components of the product term) have
been partialled first.  So, if you drop the main effect variable, you are
no longer testing the interaction with the product term...
 
Good luck on your research.
 
William "Chip" Eveland
 
>When the effect estimate of a variable is not significant, while its
>interaction with another var is in the regression, should the variable
>be dropped and only the interaction term be kept?
 
 
 
 
 
Look at the results of our WWW research on The Why Files web
site at: http://whyfiles.news.wisc.edu/welcome/research.html
============================================
William P. Eveland, Jr., Ph.D.
Associate Researcher and Team Co-Leader
Communicating with Mass Audiences Team
National Institute for Science Education
University of Wisconsin-Madison
5056 Vilas Communication Hall
Madison, WI 53706
(608) 263-3069
============================================