st: RE: psmatch2 and p-value

["Colette.Grey" <Colette.Grey@ul.ie>]

Hi Ariel,

I appreciate you taking time to help me with these queries, I am learning and making some progress I hope. I have two questions about the output from psmatch2.

I have data for three years so I want to estimate the propensity score annually and match the firms annually and am using psmatch2 for this. Then I use the results from the three years together to asses the ATT. Yes, I do have more treatment than control firms. I am starting by just looking at the matched firms.

(1) What I have is a result below where RAW is the outcome of interest and TARG is the treatment:


. psmatch2 TARG, outcome (RAW) pscore (ps2) logit neighbor(1) noreplacement
----------------------------------------------------------------------------------------
        Variable     Sample |    Treated     Controls   Difference         S.E.   T-stat
----------------------------+-----------------------------------------------------------
         RAWAWCA  Unmatched | .005222726   .015996705  -.010773979   .006727021    -1.60
                        ATT | .007248377   .015996705  -.008748328   .009040694    -0.97
----------------------------+-----------------------------------------------------------
Note: S.E. does not take into account that the propensity score is estimated.

What I don't know form the above is whether or not the T-stat is significant, I need a p-value. Do you know please how I can get a p-value?


(2) The below is also printed when I run the above command:


psmatch2: |   psmatch2: Common
 Treatment |        support
assignment | Off suppo  On suppor |     Total
-----------+----------------------+----------
 Untreated |         0        140 |       140 
   Treated |       293        140 |       433 
-----------+----------------------+----------
     Total |       293        280 |       573

My question is if the above is calculated on the 280 matched firms or the 573 firms?
How do I tell Stata which I want, assuming I am able to do so please?

Any help would be great, thanks,

Colette 




-----Original Message-----
From: owner-statalist@hsphsun2.harvard.edu [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of Ariel Linden, DrPH
Sent: 23 November 2010 19:52
To: statalist@hsphsun2.harvard.edu
Subject: st: RE: Pscores: assignment to Treated and Control?

Hi Colette,

You are now providing additional information that you didn't share last
time. For future postings I suggest that you provide as much detail as
possible for readers to speak intelligently (and succinctly) to your
problem. 

I gather from your question that (a) you have multiple panels, and (b) that
you have more treated than controls?

These are non-trivial issues and so generalized statements could be
misleading. I don't know if your intent is to treat these data as
longitudinal or roll them up to a point-treatment effect? 

Also, you need to seriously think about how you treat these data if your
treatment group is larger than your potential control group. You can match
with replacement, but you need to review how many times any control was
matched. Conversely, you could switch it around to match treated to controls
(as opposed to matching controls to treated). These are all acceptable
methods, but they could produce bias. You'll need to do some sensitivity
analyses and some common sense tests.

As for longitudinal vs point-treatment, that is not something I can comment
about without understanding your intent.

Ariel

Date: Mon, 22 Nov 2010 11:55:05 -0000
From: "Colette.Grey" <Colette.Grey@ul.ie>
Subject: st: RE: RE: Pscores: assignment to Treated and Control?

This is a multi-part message in MIME format.

- ------_=_NextPart_001_01CB8A3C.CAF8D289
Content-Type: text/plain;
      charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable

Thank you very much Ariel, I tried to use cem and got an error [<istmt>: =
 3499  cemStata() not found], I will keep trying.
=20
Is it OK to run psmatch2 for three different years, store the results in =
new variables (for example _treatedYEAR1) and then combine the new =
variables (_treatedYEAR1, _treated YEAR2 and _treatedYEAR2) and use =
these as the results of psmatch2 for three years would you know please?
=20
I read that optmatch is most suitable for samples which have more =
control that treated observations so didn't try this.
=20
Again, many thanks, Colette
=20

________________________________

From: owner-statalist@hsphsun2.harvard.edu on behalf of Ariel Linden, =
DrPH
Sent: Fri 19/11/2010 18:19
To: statalist@hsphsun2.harvard.edu
Subject: st: RE: Pscores: assignment to Treated and Control?



Hi Colette,

It dose not appear that pscore offers the option to generate a variable =
for
the matches.

I am more familiar with other user written programs that do =
automatically
generate the matches.  -psmatch2- generates all sorts of useful =
variables
for post further analytic purposes. -cem- (written by Gary King and =
found on
his website, as well as a descrption in a recent article in Stata =
Journal),
uses coarsened exact matcing and also leaves a couple of new variables =
in
the file for the match and the weight. -optmatch- is yet another user
written program that performs matching (optimal), written by Mark Lunt
(findit - optmatch).

There are others as well, but these are my top picks of programs for
matching (propensity score or otherwise)

I hope this helps

*
*   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/