Skip to content

GLMM Question

1 message · Ben Bolker

#
These are all good questions.  I'm taking the liberty of forwarding
it to r-sig-mixed-models, as that's a public venue where multiple
people can ask questions, answers are publicly viewable, and answers
are archived.

  Short answers: drop1() (my preference) or the anova method from the
'car' package are good for testing whole factors (you may also want to
see the ?pvalues help page in recent versions of the lme4 package).

  For translation to probabilities, you may want
predict(...,type="response").  You may also find the effects package
useful.

 http://glmm.wikidot.com/faq can also be helpful; if you have
particular suggestions for material to go there, you can add it
yourself or suggest it on the r-sig-mixed-models list and someone
(maybe me) might add it ...

  good luck
    Ben Bolker
On 14-03-14 11:18 AM, Brad White wrote: