Help needed using nnmatch

[jacksnape]

Hello,

I'm using nnmatch with ATT on a series of datasets containing varying numbers of treated observations. Some of the datasets contain only 1 treated observation, which seems to cause nnmatch problems.

The first error that came up was "Insufficient observations". I worked out that switching from ATT to ATE fixed that bug in most cases. However, for some of the single obeservation datasets I now get the error "Insufficient memory to create more variables: Either increase memory or decrease m". These aren't big datasets (6 variables, 70,000 observations) so memory shouldn't be a problem at all. I suspect something else funny is going on (to do with only having a single treated observation). Any thoughts?

Grateful if anyone can help me with a solution to nearest neighbour matching on a dataset with only 1 treated observation.

Regards

Jack

[jacksnape]

I should add that I'm using Stata 12 with this version of nnmatch - http://ideas.repec.org/c/boc/bocode/s439701.html

[Roberto Ferrer]

You are asked to post exact input and output. Providing a complete minimum example that reproduces the error is even better.

The first error that came up was "Insufficient observations"

This does't seem like a direct error from nnmatch. It's not part of the .ado file.

Insufficient memory to create more variables: Either increase memory or decrease m

This one does.

A reference for that command is:
Implementing matching estimators for average treatment effects in Stata by Alberto Abadie, David Drukker, Jane Leber Herr, Guido W. Imbens,
The Stata Journal, Volume 4 Number 3: pp. 290-311.