[R-sig-ME] is multicollinearity of fixed effects resolved by random effects
reinhold.kliegl at gmail.com
Tue May 20 09:50:21 CEST 2008
Two additions to my previous comment:
> Finally, you specify various random-effects parts and try to
> understand what they mean (see previous posts to this list), for
> (1|GENUS) + (1|SITE)
> (1|GENUS) + (1|GENUS:SITE) equivalent to (1|GENUS/SITE)
If there are no big discrepancies in goodness of fit between two
plausible random-effects specifications, go with the representation
that makes best sense theoretically to you. Usually, they do not make
a difference for the fixed-effects part anyway. People subscribing to
this list should get away from simply looking for the smallest or
largest value of something.
I forgot: Check the residual distribution of the final model.
More information about the R-sig-mixed-models