Re: st: Question about correlated variables in zinb regression

[Maarten buis <maartenbuis@yahoo.co.uk>]

--- On Fri, 11/3/11, rachel grant wrote:
> I am modelling counts of amphibians arriving at a breeding
> site with zinb regression. I am using two variables to
> predict zeros. These two are highly correlated with each
> other (0.8 correlation coefficient) as one is derived 
> partially from the other. If I include them both in the
> model one coefficent is negative and one becomes positive!
> it is not really possible that there can be a positive
> correlation between this variable and my zero counts, it
> makes no sense. 

The key is that by adding both to your model you are looking
at the effect of one while keeping the other constant. Think
of your effects as comparing two observations that are equal
with respect to your first variable but differ with respect
to your second variable: what difference in zero count would
you expect in that case.

Hope this helps,
Maarten

--------------------------
Maarten L. Buis
Institut fuer Soziologie
Universitaet Tuebingen
Wilhelmstrasse 36
72074 Tuebingen
Germany

http://www.maartenbuis.nl
--------------------------


      

*
*   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/