Fwd: st: Standard errors for difference-in-differences

[Clyde Schechter <clyde.schechter@gmail.com>]

the two time periods are obtained on the same people (or firms, or
animals, or whatever) or whether these are two different cohorts.  The
solution I proposed will work in the latter case, but will be
incorrect in the former.

If the two time points provide longitudinal data on the same cohorts,
then he should use:

-mixed outcome i.predictor##i.time || id: -

where id is an identifier for individual cohort members.

Clyde Schechter
Dept. of Family & Social Medicine
Albert Einstein College of Medicine
Bronx, NY, USA



---------- Forwarded message ----------
From: Clyde Schechter <clyde.schechter@gmail.com>
Date: Wed, Mar 19, 2014 at 7:35 AM
Subject: Re: st: Standard errors for difference-in-differences
To: statalist@hsphsun2.harvard.edu


 Michael Thiombendia is asking how to calculate the standard error for
a difference in differences.  He has a normally distributed interval
outcome variable and a categorical predictor measured at two time
points.

The simplest way to do this is probably:

-regress outcome i.predictor##i.time-

The coefficient of the interaction term in the output will be the mean
difference in differences, and the standard error he seeks will appear
right next to it.

Hope this helps.

Clyde Schechter
Department of Family & Social Medicine
Albert Einstein College of Medicine
Bronx, NY, USA
*
*   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/