Dear Stata users,
I am using nnmatch to estimate the treatment effect. I found very
strange matching results: the estimates using one neighbor are
identical to the matching estimates using 4 neighbors. Please check
the following stata output. However, that can not be right. Does
anyone know why? Or is this a program bug?
By the way, there is no such problem when I do not use sampling weights.
Thanks,
David
. nnmatch mdBMI fs LBMI black hispanic [w=pw], tc(att) m(4) bias(bias)
robust(4)
(sampling weights assumed)
530 observations dropped due to treatment variable missing
Matching estimator: Average Treatment Effect for the Treated
Weighting matrix : inverse variance Number of obs =
469
Weight variable: pw Number of matches (m) =
4
Number of matches,
robust std. err. (h) =
4
------------------------------------------------------------------------------
mdBMI | Coef. Std. Err. z P>|z| [95% Conf.
Interval]
-------------+----------------------------------------------------------------
SATT | -.0273805 .0229841 -1.19 0.234 -.0724285
.0176676
------------------------------------------------------------------------------
Matching variables: LBMI black hispanic
Bias-adj variables: LBMI black hispanic
. nnmatch mdBMI fs LBMI black hispanic [w=pw], tc(att) bias(bias)
robust(4)
(sampling weights assumed)
530 observations dropped due to treatment variable missing
Matching estimator: Average Treatment Effect for the Treated
Weighting matrix : inverse variance Number of obs =
469
Weight variable: pw Number of matches (m) =
1
Number of matches,
robust std. err. (h) =
4
------------------------------------------------------------------------------
mdBMI | Coef. Std. Err. z P>|z| [95% Conf.
Interval]
-------------+----------------------------------------------------------------
SATT | -.0273805 .0229841 -1.19 0.234 -.0724285
.0176676
------------------------------------------------------------------------------
Matching variables: LBMI black hispanic
Bias-adj variables: LBMI black hispanic
. nnmatch mdBMI fs LBMI black hispanic [pweight=pw], tc(ate) m(4)
bias(bias) robust(4)
530 observations dropped due to treatment variable missing
Matching estimator: Average Treatment Effect ate
Weighting matrix : inverse variance Number of obs =
469
Weight variable: pw Number of matches (m) =
4
Number of matches,
robust std. err. (h) =
4
------------------------------------------------------------------------------
mdBMI | Coef. Std. Err. z P>|z| [95% Conf.
Interval]
-------------+----------------------------------------------------------------
SATE | -.0467686 .0160686 -2.91 0.004 -.0782626
-.0152747
------------------------------------------------------------------------------
Matching variables: LBMI black hispanic
Bias-adj variables: LBMI black hispanic
. nnmatch mdBMI fs LBMI black hispanic [pweight=pw], tc(ate)
bias(bias) robust(4)
530 observations dropped due to treatment variable missing
Matching estimator: Average Treatment Effect ate
Weighting matrix : inverse variance Number of obs =
469
Weight variable: pw Number of matches (m) =
1
Number of matches,
robust std. err. (h) =
4
------------------------------------------------------------------------------
mdBMI | Coef. Std. Err. z P>|z| [95% Conf.
Interval]
-------------+----------------------------------------------------------------
SATE | -.0467686 .0160686 -2.91 0.004 -.0782626
-.0152747
------------------------------------------------------------------------------
Matching variables: LBMI black hispanic
Bias-adj variables: LBMI black hispanic
. nnmatch mdBMI fs LBMI black hispanic [pweight=pw], tc(atc) m(4)
bias(bias) robust(4)
530 observations dropped due to treatment variable missing
Matching estimator: Average Treatment Effect for the Controls
Weighting matrix : inverse variance Number of obs =
469
Weight variable: pw Number of matches (m) =
4
Number of matches,
robust std. err. (h) =
4
------------------------------------------------------------------------------
mdBMI | Coef. Std. Err. z P>|z| [95% Conf.
Interval]
-------------+----------------------------------------------------------------
SATC | -.0476791 .0162158 -2.94 0.003 -.0794614
-.0158967
------------------------------------------------------------------------------
Matching variables: LBMI black hispanic
Bias-adj variables: LBMI black hispanic
. nnmatch mdBMI fs LBMI black hispanic [pweight=pw], tc(atc)
bias(bias) robust(4)
530 observations dropped due to treatment variable missing
Matching estimator: Average Treatment Effect for the Controls
Weighting matrix : inverse variance Number of obs =
469
Weight variable: pw Number of matches (m) =
1
Number of matches,
robust std. err. (h) =
4
------------------------------------------------------------------------------
mdBMI | Coef. Std. Err. z P>|z| [95% Conf.
Interval]
-------------+----------------------------------------------------------------
SATC | -.0476791 .0162158 -2.94 0.003 -.0794614
-.0158967
------------------------------------------------------------------------------
Matching variables: LBMI black hispanic
Bias-adj variables: LBMI black hispanic
*
* 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/
