[R-sig-ME] difference in lmer results with REML = FALSE and refitML

[Ben Bolker]

 I've logged this at https://github.com/lme4/lme4/issues/202 .

Henrik Singmann <henrik.singmann at ...> writes:

> 
> Dear list,
> 
> based on a question on the sister list for linguistics
> (http://thread.gmane.org/gmane.comp.lang.r.linguistics/844/focus=846) 
> I came across some weird
> inconsistencies when comparing a lmer fit with REML = FALSE 
> and the same model first fitted with REML and
> then refitted with ML. The strange behavior is that both ML
>  fits do not agree. This seems to be somewhat unexpected.
> 
> See case below.
> 
> Cheers,
> Henrik

   I do at least get convergence warnings (with my current
development version of lme4) for the second, suboptimal fit.
I should look at slices to see what the surface looks like
(i.e. where the suboptimal stopping point is), and how this
behaves with other optimizers.

  thanks
    Ben Bolker