st: problems with heckman twostep procedure and manual computation

["Ozemela, Ekelechi Nneoma" <e_ozemela@abdn.ac.uk>]

I am doing out a heckman correction as I have a sample of children who have birth weight and majority of the other children do not have as most were born at home. My question is this. When I run the command heckman with two step, my results are very different from when I run the manual two stage equation using probit. I have read the difference is usually in the standard errors but I am getting results are entirely different from the heckman.

A short version of my model is:
heckman Birthweight class age twin bord momBMI, select (BWinfo clinic class age twin bord momBMI) twostep mills(imr)

Where BWinfo is a dummy to represent whether a child has birth weight information or not and clinic is the identifier(instrument) depicting place of delivery.

When I run the selection equation alone,

Probit BWinfo clinic class age twin bord momBMI

The results are entirely different from the previous command and by the time I compute the inverse mills ratio and run the regress command including it, my results are all different.

Kind Regards,

Ekelechi Ozemela
PhD candidate
University of Aberdeen Business School






The University of Aberdeen is a charity registered in Scotland, No SC013683.


*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/