Dear statalist users,
This is a non-technical question on difference-in-difference estimation, or rather how to obtain a
relative estimate. The data consists of a pre-treatment and a post-treatment measure for
observations with and without treatment. However, the DiD-estimate has a difficult interpretation
since the scale of the outcome is, in this specific case, obscure. The idea is to divide the
(absolute) DiD-estimate with the level of the outcome to obtain a relative estimate. But what should
I use as denominator? The data has a v e r y strong trend, and the relative estimates differ
strongly if I use the average level of pre-treatment values or the average of post-treatment values.
(Taking logs of the outcome is unfortunately not an option due to the nature of the outcome
measure). Is there a strong theory for guiding this decision, or is this just ad hoc (i.e., like a
form of scaling indeterminacy)?
Many thanks in advance,
Martin
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
