Re: st: random-slope model xtnbreg or other in stata

[Scott Baldwin <baldwinlist@gmail.com>]

I know that Mplus can fit a multilevel negative binomial model (I
think you need Mplus 5 for this). PROC GLIMMIX in SAS (included in
9.2) also can fit it. In Stata, I've followed the recommendation
provided on Statalist to add another level to the model consisting of
the observations themselves (see:
http://www.stata.com/statalist/archive/2009-07/msg00142.html). This
can be a bit slow.

Best,
Scott



On Sat, Apr 24, 2010 at 10:09 AM, Stas Kolenikov <skolenik@gmail.com> wrote:
> With Mplus unfortunately there is no telling which interaction of
> options and features will work, and which won't. So I wouldn't put
> very hopes on it.
>
> In fact, the negative binomial regression can be derived as Poisson
> regression with gamma random errors. So in a sense the negative
> binomial model is a particular case of an overdispersed Poisson model.
> You can try to overdisperse it even further with additional random
> effects... but since the sum of two independent gammas is a gamma,
> you'll still be a step away from Poisson. So I would just fit -gllamm-
> with Poisson and won't bother.
>
> Of course that's my personal opinion, and I am a strongly opinionated
> guy :)). You'd want to look elsewhere for advice on this, such as the
> multilevel list at jiscmail.ac.uk (make sure that your question is
> answered by Cam McIntosh :)).
>
> On Fri, Apr 23, 2010 at 8:10 PM, Michael Norman Mitchell
> <Michael.Norman.Mitchell@gmail.com> wrote:
>> Dear Kevin
>>
>>  Based on the link http://www.unc.edu/~skolenik/stata/gllamm-demo.html it
>> seems that gllamm (within Stata) will not fit such a model.
>>
>>  HLM will fit random slope poisson models, but I do not see anything about
>> it fitting random slope negative binomial models.
>>
>>  I am pretty sure Mplus will fit random slope negative binomial models,
>> although I have never done so. The link
>> http://www.statmodel.com/verhistory.shtml indicates Mplus does fit negative
>> binomial models, and for Mplus it is common that pretty much any parameter
>> can be treated as a random slope.
>>
>>  You can download a demo version of Mplus at
>> http://www.statmodel.com/demo.shtml and you can see example Mplus programs
>> at http://www.ats.ucla.edu/stat/Mplus/dae/ and annotated outputs at
>> http://www.ats.ucla.edu/stat/mplus/output/default.htm .
>>
>>  Best luck,
>>
>> Michael N. Mitchell
>> See the Stata tidbit of the week at...
>> http://www.MichaelNormanMitchell.com
>>
>> On 2010-04-23 5.38 PM, Kevin Wolff wrote:
>>>
>>> I am currently trying to run multilevel count models using the xtnbreg
>>> command and cannot seem to find any resources that use this
>>> command to estimate a random-slope model. Is there a package in stata
>>> which is capable of running a negative binomial regression and
>>> estimating a random-slope or random-coefficient model? If this is not
>>> possible in stata,
>>> are there any other statistical packages that anyone is aware of that
>>> would
>>> be capable of estimating this model?
>>>
>>> Thank you very much for your time.
>>>
>>> Kevin Wolff
>>>
>>> *
>>> *   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/
>>
>
>
>
> --
> Stas Kolenikov, also found at http://stas.kolenikov.name
> Small print: I use this email account for mailing lists only.
>
> *
> *   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/