[R-sig-ME] glmmPQL simplification

[David Duffy]

On Sat, 22 Jan 2011, Iker Vaquero Alba wrote:

> I am trying to fit a mixed effects model with the glmmPQL function, as I 
> get an error message trying to fit the same model with lmer.
>
> When I try to simplify the model, removing some of the interactions, I get this message:
> 'anova' is not available for PQL fits"
>
> Number of Observations: 48
> Number of Groups:
>  site pair %in% site
>  10   25

I would think that a dataset with 48-50 observations is not really enough 
to support a model with 15 terms in the fixed effects part.  If this is 
the same dataset as before, I seem to recall that none of the main effects 
looked particularly impressive.  Maybe building upwards?

Just 2c, David Duffy.