Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.

# RE: st: RE: zinb for Zero-inflated negative binomial regression

 From Caroline Wilson <[email protected]> To "[email protected]" <[email protected]> Subject RE: st: RE: zinb for Zero-inflated negative binomial regression Date Thu, 3 Oct 2013 23:50:48 +0000

```Many thanks, Dave! I will try that.

----------------------------------------
> From: [email protected]
> To: [email protected]
> Subject: st: RE: zinb for Zero-inflated negative binomial regression
> Date: Thu, 3 Oct 2013 21:56:41 +0000
>
> One possibility that's worth trying concerns your offset term. If it is something like population--which tends to be a large number in comparison to the numeric size of your other explanatory variables--you might solve your problem by dividing by a constant like 100,000 or some such to bring the numeric size of your offset variable closer to the same general size as that of your other variables.
>
> More generally, the count estimators and particularly the zero-inflated versions often have difficulty converging, but the above trick has worked for me.
>
> Hope this simple remedy helps.
>
> Dave Jacobs
>
> -----Original Message-----
> From: [email protected] [mailto:[email protected]] On Behalf Of Caroline Wilson
> Sent: Thursday, October 03, 2013 3:05 PM
> To: [email protected]
> Subject: st: zinb for Zero-inflated negative binomial regression
>
> Hi everyone,
>
> I'm new to using the zinb command in Stata (for a Zero-inflated negative binomial regression) and had a few questions about my code:
>
> This is my current code:
>
> zinb test var_m var_l var_c, inflate(var_m var_l var_c, offset(var_l)) vce(cluster group1)
>
>
> 1. I'm trying to add an offset variable to the model (var_l).  I have added it by including ", offset(var_l)" in the code. However, when I run the model including the offset terms it just runs indefinitely (while it runs fine when I don't include this). Is there something wrong with how I have specified this?
>
> 2. I'd like to look at residuals for some of the linear terms in the model. How might I do this with zinb?
>
> 3. I'd like to account for clustering of people within group1. Is adding "vce(cluster group1)" the correct way to handle this?
>
> Many thanks!
> Caroline
> *
> * For searches and help try:
> * http://www.stata.com/help.cgi?search
> * http://www.stata.com/support/faqs/resources/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/faqs/resources/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/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/
```