Re: st: RE: Adding "offset" to ZINB and ZIP models

["Mary E. Mackesy-Amiti" <mmamiti@uic.edu>]

An offset is generally used to control for the amount of exposure when 
the units of analysis have varying levels of time observed, that is to 
change a count into a rate.  I have never actually seen this done, but I 
guess including the total number of employees as an exposure/offset 
would change the count into a proportion.  In Stata, the -, exposure - 
option does the work of transforming the offset variable into 
ln(offset).  Alternatively, you can just include the total number of 
employees as a predictor in your model.


On 10/21/2010 8:21 AM, Cobb, Adam wrote:
> Hi all,
>
> I have a general question regarding zero-inflated negative binomial (ZINB) and poisson regressions (ZIP).  Note:  I use Stata 11 on a Windows machine.
>
> I am attempting to model the number of employees that participate in a health care plan for a thousand or so companies over several years.  I was exploring ZINB models because several firms do not have such a plan (or abandon it during the period of study) so there are a lot of 0 (but not missing) observations.  In talking with some people, a few indicated I should use the "offset" to offset the number of employees in the firm.  The rationale being that the firm can't have more participants in the plan than they have employees.  However, I didn't think that is what the offset was for and a a few other people I spoke with didn't believe that is what it was for either.  The stata documentation states that "offset" constrains the offset variable to have a coefficient of 1, but I'm not 100% sure how this would affect a model.   Basically I'm just looking for information regarding adding an "offset" variable to a model.  If anyone has any insight they can provide on using an "o!
ff!
>   set" in a ZIP or ZINB model, an example of when an "offset" is necessary, etc ... I'd be most thankful.
>
> Sincerely,
>
> Adam
> *
> *   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/

--
Mary Ellen Mackesy-Amiti, Ph.D.
Research Assistant Professor
Community Outreach Intervention Projects (COIP)
School of Public Health m/c 923
Division of Epidemiology and Biostatistics
University of Illinois at Chicago
ph. 312-355-4892
fax: 312-996-1450



*
*   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/