Re: st: Propensity Score Matching

[jean ries <ries@ires.ucl.ac.be>]

Monica Ospina wrote:

> Hi eveyone, 
> I am doing a propensity score matching by "psmatch". After matching
> I am doing a weighted Ordered-Probit Adjusted Estimator of ATT, but I
> have noticed that the standard errors increase slightly. I would like to
> know if it is possible to get the variance-covariance matrix after do
> the matching. I really don't know if I have to correct my dataset (or
> the standard errors) in some sense after matching and before to use it
> for other regressions.  What do you think??
>  
Monica,

1. Note that there is a module written by Edwin Leuven and Barbara 
Sianesi that supersedes -psmatch- : it is called -psmatch2- . You can 
get it by typing -ssc install psmatch2- .

2. If I correctly understand your posting, you are doing a kind of 
"regression adjusted matching". In this case your standard errors need 
to be corrected. Abadie and Imbens (2004) discuss such an estimator, and 
how to get the correct standard errors. However, they are using linear 
regression where you are using ordered probit. Abadie et al. (2004) 
discuss the implementation of these estimators in Stata. They have 
written a command called -nnmatch- that implements these estimators. 
Typing -findit nnmatch- in Stata will take you to it.

A. Abadie and G. Imbens, 2004, Simple and Bias-Corrected Matching 
Estimators for Average Treatment Effects, mimeo, 
http://elsa.berkeley.edu/~imbens/wp.shtml

A. Abadie, D. Drukker, J. L. Herr, and G. Imbens, 2004, Implementing 
matching estimators for average treatment effects in Stata, Stata 
Journal, 4(3), 290-311.


Hope this helps,

jean

*
*   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/