st: teffects psmatch matching control cases?

[Josh Klugman <josh.klugman@gmail.com>]

I am exploring propensity score matching using teffects psmatch.  It
looks to me like the command will exit if there are no matches for
control cases within the specified caliper.  This does not seem like
the way PSM should work.

Take the example in the Stata documentation (p. 137 of the TEFFECTS
documentation):

*************************
Here we repeat the previous example [effect of pregnant mothers
smoking on birthweight], first specifying that we only want to
consider a pair of observations a match if the propensity score is
less than 0.03:

. teffects psmatch (bweight) (mbsmoke mmarried c.mage##c.mage fbaby
medu), caliper(0.03)
no nearest-neighbor matches for observation 2209 within caliper 0.03;
this is not allowed
r(459);

One mother who SMOKED could not be matched with a nonsmoking mother
whose propensity score was within 0.03 of the smoking mother's.
*************************

If one replicates this example (use
http://www.stata-press.com/data/r13/cattaneo2), and looks up the
observation identified in the error message, it turns out this mother
was in fact a NONsmoker.

. list mbsmoke in 2209

      +-----------+
      |   mbsmoke |
      |-----------|
2209. | nonsmoker |
      +-----------+

When I do this example using psmatch2 [ psmatch2 mbsmoke  mmarried
mage magesq fbaby medu, outcome(bweight) n(1) cal(.03) logit ] I have
no problems getting results.

Shouldn't teffects psmatch be trying to find nonsmoker mothers to be
matches with smoker mothers, not the other way around?  I do not
understand why an analyses should be aborted simply because there are
control cases without matches.  (in fact, I am not sure why the
command must always abort when it cannot find matches for all
treatment cases; but that is a separate issue).

Thanks in advance for any enlightenment shed on this issue.

--Josh
*
*   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/