[R] lmer in package of lme4

Ben Bolker bolker at ufl.edu
Fri Feb 15 16:32:44 CET 2008


WANG <ping.wang <at> newcastle.ac.uk> writes:

> I use lmer to extract model in your package of lme4. It seems works
> well. But the problem is when I use anova/summary the extracted model,
> no p-value is shown at all. In previous version(nlme), I mainly use
> p-value to judge which term is significant or not, and then make a
> decision to keep this term or not. Does it means that sth wrong with my
> installation of package/R? or you use other value to judge the
> significance of terms?

   As it turns out, assessing the significance of fixed-effects terms
in a mixed model is potentially a huge problem.  Essentially, Doug
Bates no longer feels that the simple denominator-degrees-of-freedom 
approach that was used in nlme is reliable.

http://wiki.r-project.org/rwiki/doku.php?id=guides:lmer-tests&s=bates%20mixed

   After reading all of this (and checking out mcmcsamp in the lme4 package),
you may need to turn to r-sig-mixed at r-project.org for further assistance ...

  good luck,
    Ben Bolker



More information about the R-help mailing list