[R-sig-ME] lmer book draft

[Douglas Bates]

I should have warned users about that.  The mcmcsamp function has not
been written for the new formulation.  Part of it is the old problem
of how you sample from the distribution of the variance-component
parameters appropriately accounting for the possibility that variance
components often have a non-negligible probability of being zero.
Part of it is just finding the time to write the code.  For the time
being I would suggest using Jarrod Hatfield's MCMCglmm package.  As a
interim fix we may be able to collaborate on an interface between
lme4a and MCMCglmm to take a fitted lmer model and use it to define
the MCMCglmm model and plausible starting parameter values for the
chains.

On Tue, Feb 23, 2010 at 8:11 AM, Dieter Menne
<dieter.menne at menne-biomed.de> wrote:
> With lme4a, profile and friends works for me, but mcmc seems to be broken (works
> for lme4)
>
> R version 2.10.1 (2009-12-14)
> i386-pc-mingw32
> ...
> other attached packages:
> [1] lme4a_0.999375-45  Matrix_0.999375-37 lattice_0.18-3
>
>
> library(lme4a)
> sessionInfo()
> (fm1 <- lmer(Reaction ~ Days + (1|Subject) + (0+Days|Subject), sleepstudy))
> set.seed(101)
> samp0 <- mcmcsamp(fm1, n = 1000)
>
> #Error in function (classes, fdef, mtable)  :
> #  unable to find an inherited method for function "mcmcsamp", for signature
> "lmerenv"
>
> _______________________________________________
> R-sig-mixed-models at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models
>