Message-ID: <42F1CDB9.21518.DAA94B@localhost>
Date: 2005-08-04T06:11:37Z
From: Bernd Weiss
Subject: Multilevel logistic regression using lmer vs glmmPQL vs.gllamm in Stata
In-Reply-To: <0IKN007Y34QX8U@mail.fudan.edu.cn>
Am 3 Aug 2005 um 18:02 hat ronggui geschrieben:
> >On Wed, 3 Aug 2005, Bernd Weiss wrote:
> >
> >> I am trying to replicate some multilevel models with binary
> >> outcomes using R's "lmer" and "glmmPQL" and Stata's gllmm,
> >> respectively.
[...]
> the glmmPQL and lmer both use the PQL method to do it ,so can we get
> the same result by setting some options to the command?
>
Thanks to Prof. Ripley and ronggui for their answers.
To verify my findings I tried other datasets and simulated some data
and compared the results between R and Stata. Everything works fine,
no differences -- except for the xerop-dataset.
Having a closer look to the R output I found some unusual values for
AIC, BIC and deviance, see below:
AIC BIC logLik deviance
1.797693e+308 1.797693e+308 -8.988466e+307 1.797693e+308
I assume I have to change some of the lmer-parameters but have
absolutely no idea which one.
Again, I would appreciate any help.
Bernd