st: RE: RE: Nested zip

["Jacobs, David" <jacobs.184@sociology.osu.edu>]

Check the econometric program called Limdep for a pooled time-series version of zip models.  I think (?) such models are implemented in that package.  And at least some multilevel models can be estimated using random-effects pooled time-series models.

Dave Jacobs

-----Original Message-----
From: owner-statalist@hsphsun2.harvard.edu [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of Nick Cox
Sent: Tuesday, December 06, 2011 1:33 PM
To: 'statalist@hsphsun2.harvard.edu'
Subject: st: RE: Nested zip

The answer depends on the question. 

I guess that you already know the answer that no -xtzip- or -xtmezip- command exists, and that you are seeking for an explanation of that fact in turn, and not merely that StataCorp haven't written it. 

This is for StataCorp for answer, except that questions of this form don't usually get answered by them (not least because they often morph into "Do have plans to work on it then?"). 

In very general terms, a command is likely to get written by StataCorp and issued with official releases depending on such factors as 

* whether someone at StataCorp has developed interest and expertise in a field (StataCorp can be as capricious as any bunch of academics in choosing to work on something or ignore it, regardless of popularity) 

* whether there is a literature on a technique

* whether it seems well used in statistical science, or at least deserving of being so

* whether an algorithm can be thought up that works well in practice (this is the one that users often under-estimate; StataCorp can put a vast amount of effort in setting up simulations and various other kinds of assessment; sometimes a technique is emphatically _not_ implemented officially because StataCorp have found all too much evidence that it often produces lousy results). 

* all sorts of other things (a massive noise term). 

In the case of -xtzip- is there even a literature? Would you expect a parameterisation in which the fraction of zeros is allowed to vary by panels (insisting on the same fraction sounds unlikely to appeal)? How would that work numerically? 

Why haven't users tried to write it either? 

Nick 
n.j.cox@durham.ac.uk 

Kris Anderson

I am trying to model a count variable (lifetime use of drugs) nested in 
schools. Given it's distribution, a zip model would be ideal (lots of 
zeros). However, it seems Stata will run an xtmepoisson but not zip. 
Does anyone know why?


*
*   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/

*
*   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/