Thanks to Kit Baum, a new component, mahaselectunique, has been added
to the mahapick module on SSC.
Every once-in-a-while, someone asks for this, so I have finally made
it into a program for general use.
mahaselectunique enables you to take the output of mahapick (a set of
"near" observations, according to a Mahalanobis distance measure) and
selects unique matches based on a randomized priority.
That is, mahaselectunique would typically follow the use of mahapick
-- to reduce the resulting data to a unique set -- unique in that a
given control observation appears only once. It is not unique in that
a randomization process is used. And there is no optimization either;
that's another story for another day.
Thanks go to those of you who have been requesting this; most
recently this includes David Greenwood-Sanches and Paula Arce.
To install:
ssc install mahapick
(There are also some minor updates to some of the existing help files
therein -- for other components.)
