Statalist


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

st: AW: psmatch2 and the options ATE and COM


From   "Martin Weiss" <martin.weiss1@gmx.de>
To   <statalist@hsphsun2.harvard.edu>
Subject   st: AW: psmatch2 and the options ATE and COM
Date   Tue, 10 Mar 2009 13:51:09 +0100

<> 
Information about -pscore- is easily obtained from
http://www.stata-journal.com/sjpdf.html?articlenum=st0026

Both commands leave variables behind that let you check whether their
propensity scores coincide. Using the famous Lalonde data usually employed
in this kind of exercise -and described in the article above-, I found that
the common support is enforced inconsistenly between the two commands. Then
again, there is no reason to suspect that authors of competing commands
should be consistent among themselves...


*************
use lalonde.dta, clear
qui pscore treat age age2, pscore(mypscore) blockid(myblock) comsup
qui psmatch2 treat age age2, outcome(re78) neighbor(1)
compare _support comsup


                                        ---------- difference ----------
                            count       minimum      average     maximum
------------------------------------------------------------------------
_support=comsup               585
_support>comsup                29             1            1           1
                       ----------
jointly defined               614             0     .0472313           1
                       ----------
total   
*************



HTH
Martin


-----Ursprüngliche Nachricht-----
Von: owner-statalist@hsphsun2.harvard.edu
[mailto:owner-statalist@hsphsun2.harvard.edu] Im Auftrag von Sarah Wichmann
Gesendet: Dienstag, 10. März 2009 11:51
An: statalist@hsphsun2.harvard.edu
Betreff: st: psmatch2 and the options ATE and COM

Hello,
I am working with Propensity Score Matching and came across the
following problems (code below).

1) When comparing the results from psmatch2 with pscore (and
subsequently att*), I only get the same results if I do not use the
common support options. If I do that, the results differ. I guess the
common support is selected differently with the two commands, but I
could not find more information. Can somebody explain the difference?

2)  When I use psmatch2 again with the common support option, I get
different results for ATT depending on if I also use the ATE option or
not. As I understood the help file, the ATE option should only add
further details in the output instead of changing the calculation,
therefore I do not understand the changing result.

Thanks for your time and consideration,
Sarah


psmatch2 emsbinary logvossd, radius cal(0.1) out(perfbinary) logit com ate

Variable     Sample |    Treated     Controls   Difference         S.E.
T-stat
----------------------------+-----------------------------------------------
------------
      perfbinary  Unmatched |       .625   .582417582   .042582418
.045186234     0.94
                        ATT | .612903226   .620169943  -.007266717
.048401842    -0.15
                        ATU | .581497797    .55933642  -.022161378
                        ATE |                          -.018370454


psmatch2 emsbinary logvossd, radius cal(0.1) out(perfbinary) logit com

Variable     Sample |    Treated     Controls   Difference         S.E.
T-stat
----------------------------+-----------------------------------------------
------------
      perfbinary  Unmatched |       .625   .582417582   .042582418
.045186234     0.94
                        ATT | .612903226   .620364171  -.007460945
.048395422    -0.15


attr perfbinary emsbinary logvossd, pscore(pscore) logit comsup

n. treat.   n. contr.         ATT   Std. Err.           t
---------------------------------------------------------

      159         454       0.027       0.046       0.580





-- 
Sarah Wichmann
Eichelhäherkamp 94b
22397 Hamburg

Handy: 0046 73 683 57 91
           0049 173 612 44 42

*
*   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/


*
*   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