Skip to content
Prev 12640 / 20628 Next

Error message

I would try it using adaptive Gauss-Hermite, by setting nAgQ=3 or more and
seeing how that works. It really should be your first option when fitting a
GLMM, and something that should be checked anyway. In your case with binary
data and approx 2 per group the Laplace approximation is almost certainly
poor.

On 5 November 2014 22:55, Luciano La Sala <lucianolasala at yahoo.com.ar>
wrote: