Hello all,
I using lmer to develop a mixed effects model. I start with an overly
parameterized model (as suggested in Zuur et al. Mixed Effects Models and
Extension in Ecology with R) that looks something like this:
m1 <- lmer( Y ~ aS + bS + c + d + e + (c|SpeciesId) + (d|SpeciesId) +
(e|SpeciesId))
aS and bS are species level predictors an so do not vary within a SpeciesId.
However, c, d, and e are population level predictors and can all potentially
vary significantly within species.
I am trying to arrive at the best model describing Y.
I have been beginning with selection on the random effects, subtracting one
term and performing a likelihood ratio test, then another term and another LRT,
etc. I then delete the term with the lowest non-significant test statistic and
do the whole procedure again on the reduced model.
After getting the "optimal" random effects structure, I move on and perform a
similar procedure on the fixed effects.
First of all, does this sound at all sensible?
Secondly, when I do this with my data the resulting "optimal" or best fitting
model looks like this:
mFinal <- lmer( Y ~ aS + bS + c + (d|SpeciesId) + (e|SpeciesId))
d and e show up as random slopes but not in the fixed effects.Is this ok, and
if so, I'm not sure what the interpretation is...
Thanks so much,
HamishUCSD ebehamis...@hotmail.com
_________________________________________________________________
The New Busy is not the old busy. Search, chat and e-mail from your inbox.
N:WL:en-US:WM_HMP:032010_3
[[alternative HTML version deleted]]
______________________________________________
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
