If we are talking about empirical models, then yes, we should (usually)
include x1 if x1^2 is significant. If we have mechanistic knowledge of the
model, then we do not have to do so. For example, if we are measuring the
weight of spheres as a function of radius, we would write:
Weight = constant*radius^3
and including lower order terms would be silly.
-------------------------------------------
Emil M Friedman, PhD
2304 Richmond Road
Beachwood, OH 44122
[log in to unmask]
216-591-1750 (voice)
775-249-6744 (fax)
----- Original Message -----
From: "Kim Pearce" <[log in to unmask]>
To: <[log in to unmask]>
Sent: Tuesday, June 03, 2003 5:34 AM
Subject: Interactions and main effects
> Dear all,
>
> Am I correct in thinking that when we form a model containing
interactions,
> the main effects making up these interactions should be included in the
> model regardless of their significance? i.e. if the term x1^2 significant
> and in the model then x1 should also be present even if x1 is non
> significant and if x1*x2 is significant and in the model, then x1 and x2
> should both be present even if one or both of them is non significant.
>
> Many thanks for your help on this,
> All the Best,
> Kim.
>
|