Statalist


[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

st: RE: Count Data again


From   "Nick Cox" <n.j.cox@durham.ac.uk>
To   <statalist@hsphsun2.harvard.edu>
Subject   st: RE: Count Data again
Date   Thu, 28 Feb 2008 17:52:33 -0000

On a international list, I advise against statements that don't apply uniformly 
to all latitudinal and longitudinal zones, namely references to time of year or time
of day. 

To the point: 

Poisson in some ways functions as a default or null model here. 

With this mean and variance you should certainly try it out and see what happens, 
But as always then look carefully at all the diagnostics. 

Nick 
n.j.cox@durham.ac.uk 

Markus Köller

-> I am trying to predict the total number of children a woman gives birth to. I want to use a Count data model with a Poisson-regression.
-> I am using crosssectional data which has been collected within three waves. Total number of observation is 1350.

-> The mean of my variable "amount of children" is 1,27 and my variance of this variable is 1,131.

-> I see that my variable seems to be underdispersed, since the mean exceeds the variance.

My question is: The mean is bigger than the variance, but is it really "too" much, so that one can speak about "underdispersion"?  To me, 1,27 and 1,131 seems not so very different. And if it is not underdispersed, does it mean, that normal Poisson-distribution is the correct approach?

Have a nice evening and thanks everyone,

*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   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