[R-sig-ME] profile, confint and p-values in lme4

[Douglas Bates]

Profiling and the associated plots and confidence intervals are not in
the currently released version of the lme4 package.  They currently
only available in the development version that is called lme4a and
must be obtained from R-forge.  At some point this version will be
released as lme4 but I can't say exactly when.  Both Martin and I have
heavy teaching loads at present and are unable to make much progress
on the package.

On Wed, Nov 10, 2010 at 5:07 PM, Marie-Hélène Hachey
<marie_helene48 at hotmail.com> wrote:
>
> Hi,
>
>
>
> I've been trying to make a profile zeta plot and check the confidence interval of
> my data in lme4 but I got these messages:
>
>
>
>> profile(model)
>
> Erreur dans UseMethod("profile")
> :   pas de méthode pour 'profile' applicable pour un objet de
> classe "mer" (No method for "profile" applicable for an
> object of class "mer")
>
>
>
>> confint(model)
>
> Erreur dans object$coefficients : $ operator not
> defined for this S4 class  (Error in objects$coefficients)
>
>
>
> I found the "profile" and "confint" functions in Doug Bates e-book. I
> know he's been using lme4a for that book, so I downloaded it and installed it
> last week but then the function lmer stopped working and I uninstalled it.
>
>
>
> Any advice?
>
>
>
> Next question: I've read about p-values for lmer,
> and I understand that because of the problems with the p-values (conservative, value of p twice of what it should be),
> lme4 didn't give them before in its output. At the present time, lme4 gives the p-values. Is the problem
> solved?
>
>
>
> In other words: can I rely on those p-value or
> should I use a Monte Carlo simulation like D. Bates suggested in 2006 (in article "lmer,
> p-values and all that")?
>
>
>
> Thank you
>
>
>
> Marie-Helene Hachey
>
>
> _______________________________________________
> R-sig-mixed-models at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models
>