Please note that
"For logistic regression models, it is well known that estimation of fixed-effects models by the inclusion of dummy variables yields inconsistent estimates of b (Hsiao 1986) due to the �incidental parameters� problem (Kalbfleisch and Sprott 1970), while conditional estimation does not suffer from this problem."
Allison and Waterman, FIXED-EFFECTS NEGATIVE BINOMIAL REGRESSION MODELS, Sociological Methodology, 32, page 249
Nicola
At 02.33 06/03/2008 -0500, you wrote:
>For FE you can include dummies. Here you pick up
>unobserved characteristics that do not vary over time.
> For clustering use: cluster (varname) this will pick
>up the within correlation. You can also include
>dummies instead of clustering which means that you
>allow for differences in the average level of across
>villages in addition to adjusting for the SE
>(intra-group and between group correlation).
>
>see:
>http://www.gseis.ucla.edu/courses/ed230bc1/notes3/cluster.html
>
>[...]
*
* 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/