[R-sig-ME] residuals in lme4

[Douglas Bates]

On 8/26/07, lunina at access.unizh.ch <lunina at access.unizh.ch> wrote:

> Hello

> I want to get the residuals of a heteroscedastic linear mixed-effects-model.
> How can I fit such one ond how is it possible to extract the residuals?

> I saw that the the usual function "resid(.)" works for lmer, but not for
> lmer2. But no one of these two shows any change in the model values if I
> define some "weights". Is "weights" not yet implementet? And is there
> another possibility to calculate a weighted linear mixed-effects-model?

The weights argument to lmer() is for fixed weights, as in the lm()
function, not for variable weights, as in lme().  So the answer is
that if you want weights that, for example, depend on the fitted
values, then that is not implemented in lmer().

> I already tried it with "lme(.)" of the package "nlme", but the models with
> "lme" fit much worse than those with "lmer", also if I compare the
> unweighted ones...
>
> Thanks a lot
>
> Monika
>
> _______________________________________________
> R-sig-mixed-models at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models
>