Multilevel zero truncated poisson model?

[Isabel Lechner]

Dear Stata users,

I'm using Stata 13 and I'm wondering whether there is an option to perform a zero truncated poisson regression model including random effects.
I found options for both a zero truncated and a multilevel poisson model, but not a combination of truncation and a multilevel options.
Did I miss something or is it not possible?

Many thanks for your inputs,
Isabel

[Carlo Lazzaro]

Isabel:
a quite old thread on a very similar topic (please, see http://www.stata.com/statalist/archi.../msg01309.html) did not get any reply, seeminlgy supporting that what you're after it not easy to implement or not fesible at all.
I do not know yet whether or not a hurdle multilevel regression model is feasible (in your case and in general).

[Clyde Schechter]

There is certainly no official Stata command for this. I couldn't find a user-written command specifically for this using either Stata's -search-, nor with a Google search. So I suspect there is nothing out there specifically for this.

[Joao Santos Silva]

Dear Isabel,

There may be a good reason to explain why such command is not available: the truncation of the sample will distort the distribution of the random effects and that means that such models will in general have limited applicability. If you are interested, I have an old paper that discusses that.

All the best,

Joao