[R-sig-ME] random effects specification

[Ken Beath]

On 01/05/2008, at 11:28 PM, Sebastian P. Luque wrote:

>
> which I presume is due to the fact that the data are unreplicated,
> i.e. individuals get each treatment only once.

No. Due to not having a random component for id in the simulated data.

Try

rid <- rep(rnorm(20),each=3)
dta$n <- dta$n+rid

Then refit the model.

>
> Are there any gotchas in
> the interpretation of the results after this warning?
>
>

Probably not, but with real data I would remove the random effect from  
the model as it is 0.

Ken