[R-sig-ME] Logistic regression with random effect and AIC output?

Douglas Bates bates at stat.wisc.edu
Wed May 21 16:24:12 CEST 2008

On Tue, May 20, 2008 at 5:50 AM, Federico Calboli
<f.calboli at imperial.ac.uk> wrote:
> R.S. Cotter wrote:
>> Dear all,

>> I will run a logistic regression model, with sevral explanatory variables.
>> But I don't find a script that both include a random effect and also has a
>> AIC output. Have tried different search options.

> Have a look at glmmPQL in library MASS. it might be what you're looking for.

I don't think you will get an AIC value from glmmPQL.  The penalized
quasi-likelihood method doesn't provide the likelihood for the model
(because it minimizes another quantity called the quasi-likelihood).

Is there a reason not to use lmer with the optional argument family = binomial?

More information about the R-sig-mixed-models mailing list