Statalist


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

st: using xtnbreg; need for overdispersion at level 1


From   "Lucas Bremer" <bremer@bwl.uni-kiel.de>
To   <statalist@hsphsun2.harvard.edu>
Subject   st: using xtnbreg; need for overdispersion at level 1
Date   Mon, 8 Feb 2010 07:32:41 +0100

Dear all,

I?m trying to run a negative binomial panel count model with random effects
(xtnbreg). In ?Multilevel and Longitudinal Modeling Using Stata?  I read
that you can only 
use this procedure if you have heterogeneity at level 2 and overdispersion
at level 1.

Can somebody tell me what is meant by ?overdispersion at level 1?. Do I have
to compute the mean and variance of my counts of each individual over the
observation periods and compare these? 
Or compare mean and variance in the cross-section at every period?

The distribution of my dependent variable (the counts) is highly skewed if I
ignore the clusters and simply compute mean and variance.

Thanks in advance!

P.s. We had some trouble with our mail server so maybe my question exists
already. 

Lucas




*
*   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/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index