[R] GLM model simplification

[Martin Hoyle]

Hello R Users,

I have the GLMs:

model1 <- glm (response ~ explanatory * block, poisson)
model2 <- glm (response ~ explanatory + block, poisson)
anova(model1, model2, test="Chi")

and I find that the interaction is significant.
I am interested in the main effect of "explanatory". However, I am advised that model simplification should stop when a higher order term is found to be significant (here, the interaction).
I imagine that I should do;

model3 <- glm (response ~ block, poisson)
anova(model2, model3, test="Chi")

to test for the significance of the main effect of "explanatory". Am I allowed to do this, even though the interaction is significant?

Thanks for your attention,
Martin.


Martin Hoyle,
School of Life and Environmental Sciences,
University of Nottingham,
University Park,
Nottingham,
NG7 2RD,
UK
Webpage: http://myprofile.cos.com/martinhoyle