Date: Wed, 17 Dec 2008 11:27:55 -0500
Reply-To: Kevin Viel <citam.sasl@GMAIL.COM>
Sender: "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From: Kevin Viel <citam.sasl@GMAIL.COM>
Subject: Re: Regression: do you always need main effects with interactions?
On Wed, 17 Dec 2008 10:33:14 -0500, Peter Flom
>I don't think it's a strict dichotomy. I don't think any data analysis is
>purely exploratory, and very few are purely confirmatory. We almost
>always have *some* idea what we are expecting, and we almost never have a
>very exact idea.
A worthy and interesting discussion, but we have drifted from the original
The few analysis that are confirmatory, or intended to test a specific a
priori hypothesis, may be dominated by experiments. Anyone who has
conducted an experiment knows that error and other forms of variation
occur, thus introducing the potential for additional factors :)
Either way, I think the term parsimonious needs to mean that lower order
terms and base factors of interactions are included in the model.
PS FWIW, I have argued for the past three years that the talent in science
is acquiring the data with which to test (or to explore <vbseg>)
hypotheses. Most students can adequately analyze a research question given
data. I think this is true even for ongoing studies; obtaining permission
and funding to collect the samples and analyze them is an art...