Bookmark and Share

Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.


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

Re: st: teffects psmatch matching control cases?


From   "David M. Drukker" <[email protected]>
To   [email protected]
Subject   Re: st: teffects psmatch matching control cases?
Date   Mon, 4 Nov 2013 15:25:09 -0600 (CST)

Josh Klugman <[email protected]> asked some questions about why -teffects psmatch- refuses to estimate when it cannot find at least one match.

In short, without any matches -teffects psmatch- cannot compute an estimate
of the observation-level conditional-on-covariates treatment effect.

What follows is a more detailed discussion of this point.

Josh typed

. use http://www.stata-press.com/data/r13/cattaneo2
. teffects psmatch (bweight) (mbsmoke mmarried c.mage##c.mage fbaby medu), ///
   caliper(0.03)

and -teffects psmatch- refused to estimate after providing the error message

no nearest-neighbor matches for observation 2209 within caliper 0.03;
this is not allowed
r(459);

Josh was surprised at this error because observation 2209 is a nonsmoker,
someone with the control level.

Josh did not specify the -atet- option, so -teffects psmatch- will estimate
the average treatment effect which needs to impute the other potential
outcome for individuals that received the treatment and for individuals that
received the control.

-teffects psmatch- correctly identified observation 2209 as an observation
for which no match could be found and refused to estimate the ATE.

My guess is that Josh would like to estimate the average treatment effect on
the treated (ATET).  Estimating the ATET only requires imputing the control
potential outcome for those that get the treatment.  In this case, one would
not care that no match could be found for the nonsmoker in observation 2209.

Unfortunately, typing

. teffects psmatch (bweight) (mbsmoke mmarried c.mage##c.mage fbaby medu), ///
   caliper(0.03) atet

produces the same error message.  We will loosen this check of the overlap
condition in a future update.

-- David                    -- Rich
   [email protected]          [email protected]

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


© Copyright 1996–2018 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   Site index