In the process of preparing for a course I am teaching next month
on the negative binomial model at both the FDA and in Boston (ASA
sponsored Learnstat program), I have created a number of hurdle
models that I have sent to Kit for inclusion on the SSC site. Moreover,
among other programs, I also programmed a zero-truncated geometric
regression (which is intrinsic to geometric hurdle models) and have updated
the censored Poisson.
All models are written in verison 9.1 code, allowing the full range
of maximize and survey options. See help maximize and help svy to
get a feel of what is available.
Hurdle models include:
========================
poisson-logit
poisson-cloglog
geometric-logit
geometric-cloglog
negative binomial-logit
negative binomial-cloglog
Additional models
========================
zero-truncated geometric
censored Poisson
I have also prepared simulated data sets with which to test the hurdle
models. One for each model, they were included with my submission to Kit,
as well as help files. See Allen McDowell's Stata Journal article on hurdle
models [SJ (2003) Vol 3, nu. 2, pp-178-184] for background.
Each hurdle model also comes with an AIC GOF test statistic. The
zero-truncated
geometric has associated AIC, BIC, and deviance (and Chi2-based p-value)
statistics.
These are helpful when assessing the comparative worth of a model.
Kit should have these models posted to the SSC site within the next couple
of
weeks. If someone just can't wait for a particular model, let me know and
I'll send it to you.
Joe Hilbe
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
