Thanks Peggy
I was avoiding doing this, but I think you're correct: the only solution to
my problem is to introduce interaction terms and / or powers. It is the only
path that can be followed to achieve the balancing. But I also need to know
whether it is necessary to use the propensity obtained through weighted
regression in the estimation of atts. It seems that there is a indication
that the nearest neighbor method is insensitive to weighing (it is easy to
test this) but in other methods, such as the kernel and local linear
regression results change with weighting. There is also a proposition that
weight the treated observations with the inverse of the propensity and with
the inverse of 1 minus the propensity for untreated cases. I am not sure
about this procedure.
Henrique
-----Mensagem original-----
De: [email protected]
[mailto:[email protected]] Em nome de P Chrisman
Enviada em: quinta-feira, 24 de abril de 2008 22:38
Para: [email protected]
Assunto: st: RE: using pscore and pstest
Hello Henrique,
I am not an expert at pscore by any means, but I have been using propensity
scores for a current project. I had difficulty with the balancing tests and
found that introducing higher order terms for the variable(s) that did not
balance was the easiest way to correct this problem. I think it was the
Stata documents by the authors of the pscore procedure that suggested trying
a less parsimonious model to fix balancing errors.
Peggy
*
* 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/
No virus found in this incoming message.
Checked by AVG.
Version: 7.5.524 / Virus Database: 269.23.2/1389 - Release Date: 21/4/2008
08:34
No virus found in this outgoing message.
Checked by AVG.
Version: 7.5.524 / Virus Database: 269.23.2/1389 - Release Date: 21/4/2008
08:34
*
* 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/
