Dear statalisters,
I would like to tackle an attrition problem in a fixed effect logit
model. Does anybody have an idea how to do this?
My first idea is to estimate a selection model (just a probit model as
in the first step of the Tobit II model), take the error terms from this
estimation and verify whether they are correlated with the error terms
(not the fixed effects) of the FE logit model.
This idea is based on the fact that non-random selection is
non-ignorable if the selection into the sample depends on the variable
of interest, even when allowing the selection probability to depend upon
the observed explanatory variables.
Does this sound reasonable?
Marijke
Disclaimer: http://www.kuleuven.be/cwis/email_disclaimer.htm
*
* 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/
