Stas,
thanks a lot for your comments.
2009/4/3 Stas Kolenikov <[email protected]>:
> If you have figured out -gllamm-, then you might be able to use it to
> set up a mixture/zero-inflation model with two-point distribution of
> the latent variable, using -ip(f) nip(2)- options for the relevant
> part of the model. I would probably be more convinced if you had full
> panels that consist of zeroes, and other panels that have a mixture of
> 0s and non-zeroes, rather than each panel having 5 zeroes and one or
> two non-zeroes, since zero-inflation models are essentially stating
> that an individual is either in "don't-do-it" class with zero outcome,
> or "do-it-sometimes" class with zero outcomes coming in a random way
> along with non-zeroes. See -zip- for a canned routine doing this in
> official Stata.
I was very tempted this approach, and I had already in the past
written something along these lines but only as a pooled estimator.
The difficulty is that individual heterogeneity in my application is
not just "don't do it" vs "do it sometimes", but rather "don't do it",
"do it increasing with x1, but less so towards the last time period",
and "do it increasing with x1, no matter what time period". There is
also evidence of some "do it always" people, but those are very few,
and far between, and I am happy to ingnore them for the time being.
> And btw it might be worth looking at -xt[me]poisson- if your data are
> integers. See if interpreting your dependent variable as a count is at
> least an approximately reasonable interpretation in your application.
Being able to use -xtmepoisson- would be lovely, as it's a ready-made
multilevel mixed-effects regression and therefore exactly what I need.
However, analysing my variable as a count is probably strechting it a
bit.
> As far as I know (and you should not over-rely on this :)), it is the
> tobit model that is usually behaving in a weird way, as it is quite
> fragile to the violations of normality assumptions.
And I had just gotten my hopes up... I will certainly check out the
GLM approach again, with the references provided to me. However it's
important that I can estimate the variances and covariances of the
random effects (and not simply APEs), as this is very much what I am
interested in.
Thanks again,
Eva
*
* For searches and help try:
* http://www.stata.com/help.cgi?search
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
