PROC COUNTREG, WARNING: The Hessian matrix is singular.

[Nayika]

Hi everyone,

I am wondering if anyone has any insight on my problem. I am running multiple models with the same variables (full sample, male sample, female sample) and on SOME of my models I am getting zeros and the warning: "The Hessian matrix is singular." Now, it is the same data, same variables, but some of the models produce results and some do not. I am not sure whether my machine has gone rogue or if I am violating some assumption...

I am running ZINB models. Anyone have any experience with this?

Thanks so much,

Nayika

[PGStats]

It might be that some of your categories (CLASSES) are empty in some data subsets, leaving no effect to estimate (i.e. your model has a isSmoker effect but all females are non-smokers).

PG