Dear Statalist,
I am able to apply the different matching procedures with psmatch2 in a
single treatment framework. Now my problem is that I need a multiple
treatment framework and it is unclear to me from the 'help file' how this
works.
Concretely, I would like to apply the mahalanobis metric matching procedure
to investigate the joint impact of four types of funding, FUND1,FUND2,
FUND3, FUND4, on the level of R&D, RD (the treated), as well as their
individual impacts. I would like to do this with common support and both
with SE that do not take into account that the propensity score is estimated
and with SE based on bootstrap simulation.
If there would only be a single type of treatment, say FUND1, I would do:
psmatch2 FUND1 X1 X2 X3 X4 X5, outcome(RD) mahalanobis(Z1,Z2,Z3,Z4) common
and
bootstrap r(att),reps(100) seed(1): psmatch2 FUND1 X1 X2 X3 X4 X5,
outcome(RD) mahalanobis(Z1 Z2 Z3 Z4) common
where the Xi are exogenous variables in the probit equation explaining FUND1
and Zi are exogenous variables entering the mahalanobis metric besides the
propensity score obtained from the probit equation.
Now in case of four treatments, FUND1, FUND2, FUND3, FUND4, we need first to
estimate the multinomial probit model explaining the four treatments, derive
from this four propensity scores, and then use them in the mahanalobis
metric.
So my questions are:
1. how does one do this?
2. how does one obtain an ATT estimate on the joint impact of the four
treatments?
3. how does none obtain separate ATT estimates on the separate impacts of
the four treatments?
4. can this also be done with bootstrap SE?
Thanks in advance.
Dirk
*
* 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/
