Re: st: heckman

[<fjbradbury@cox.net>]

Maria et al,

Can you please explain how Heckman should be applied?  I am trying to use it to eliminate self-selection bias in a retrospective study in which the controls are self-selecting.  Do you run Heckman and then a separate regression model in which you apply the coefficients derive from Heckman?

Thanks, Felix

============================================================
From: "Santos, Maria Emma" <maria.e.santos@vanderbilt.edu>
Date: 2005/03/23 Wed PM 07:08:38 EST
To: statalist@hsphsun2.harvard.edu
Subject: st: heckman

Thanks for the advice and refernces provided so far about non-convergence in 
Heckman MLE.
Though I couldn't find a way to solve the non-convergence problem I've worked 
out a way to estimate my regressions with heckman two steps (reducing all the 
weights inequal proportion and creating already weighted variables). 

Now my question is this: I would like to make sure that I have no 
heteroskedasticity problems. Heckman MLE allowed for the the option "robust" 
and also 'cluster'. But with two steps I can't use these options.
I have tried to make a test after the regressions with hettest but it 
says "last estimates not found". 

This is what I am writting:

heckman LogW memPrim age agesq NE NW 
PAMPEANA CUYO PATAGONICA if complPrim==0, twostep select 
(employed memPrim age agesq NE NW PAMPEANA CUYO PATAGONICA 
assets married children6 children618);
hettest;

I have also tried to do the test by myself first creating a variable with the 
resids, but it says "option resid not allowed".

Do you have any suggestions or tricks to check for heteroscedasticity in 
Heckman 2 steps? (And clustering?...)

Thanks in advance,

Maria.

-----------------------------------------------------------------
Santos, Maria Emma
Vanderbilt University
Email: maria.e.santos@Vanderbilt.Edu
*
*   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/
============================================================


*
*   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/