"intmethod(ghermite)" option. Are coefficients reliable?

[Gianfranco Di Gennaro]

Dear all,
I'm running a multilevel model that doesn't converge in any way except with the "intmethod(ghermite)" option, nonadaptive Gauss-Hermite quadrature.

I have tested the fit of the model and it looks good.

I wonder if I can trust the coefficient estimates.
Frankly, they seem a bit exaggerated to me.

Thanks so much, as always.

Gianfranco