[R-sig-ME] variance-covariance matrix of the estimations of the variance-covariance matrix of the random effects

[Douglas Bates]

Variances and covariances of estimators of variance components are
usually of little value because the distribution of the estimators are
often highly skewed.  See the notes at
http://lme4.R-forge.R-project.org/slides/2011-01-11-Madison/3ProfilingH.pdf
for illustration of such cases.  Notice Figure 4 in particular.

In theory you could evaluate the information matrix for the profiled
deviance and use the "delta method" to determine an information matrix
for the variance-covariance parameters but it would take considerable
effort and probably not produce useful information.

On Thu, Jan 13, 2011 at 3:51 PM, Resche Rigon Matthieu
<matthieu.resche-rigon at paris7.jussieu.fr> wrote:
> Dear all,
>
> I try to obtain the variance-covariance matrix of the estimations of the
> variance covariance matrix of the random effects. I know how obtain it using
> lme but I would like to obtain it with lmer(). Is it possible?
>
> Moreover do you know which formula is applied by ranef() to extract the
> conditional modes and the conditional variance-covariance
>  matrix of the random effects?
>
> Thanks in advance
>
> Matthieu
>
> _______________________________________________
> R-sig-mixed-models at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models
>