[R-sig-ME] false convergence of lmer model

[Highland Statistics Ltd]

On 24/07/2013 10:06, Tara Woud wrote:
> I'm sorry, I forgot to mention that I already centered the covariates
> (that's where the "c_" in the parameter names stands for.
>


Try standardizing.

Also..things like max temperature and min temperature are probably 
highly collinear. Simplify the model and see what happens...and then 
build up the complexity of the model up to the moment numerical problems 
appear.

Alain