Statalist


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

st: Interpreting pstest outputs: Pseudo R2 from probit estimation


From   ippab ippab <ippab.statalist@gmail.com>
To   statalist@hsphsun2.harvard.edu, Stata Technical Support <tech-support@stata.com>
Subject   st: Interpreting pstest outputs: Pseudo R2 from probit estimation
Date   Sun, 22 Nov 2009 14:58:18 -0700

Hello Everyone,

I am struggling to understand the meaning of the final output from the
pstest.  The help file indicates that it reports the  Pseudo R2 from
probit estimation of the conditional treatment probability (propensity
score) on on all the variables in varlist before matching, and Pseudo
R2 from the same probit on the matched samples.  I am wondering how to
interpret these results.  What would be an indication of good matches.
 For example, is joint insignificance of all the regressors after
matching is desired?

Here is what I am getting with kernel k(normal) matching:

--------------------------------------------------------------
    Sample |    Pseudo R2      LR chi2        p>chi2
------------+-------------------------------------------------
 Unmatched |        0.083       771.26         0.000
   Matched |        0.002        13.83         0.054
--------------------------------------------------------------



When I use  kernel k(biweight), I get different results:



--------------------------------------------------------------
    Sample |    Pseudo R2      LR chi2        p>chi2
------------+-------------------------------------------------
 Unmatched |        0.083       771.26         0.000
   Matched |        0.000         1.44         0.984
--------------------------------------------------------------



In the second case, for the matched sample, we fail to reject the H0:
all regressors are jointly insignificant.  So, can I conclude that
kernel k(biweight) matching is better?  What would be a test to check
the fit (performance) of the matching method?

Thanks a lot!

Best regards,

ippab

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



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index