0
$\begingroup$

I'm having some trouble fitting a model for differences-in-differences regression in Stata. Specifically, I'm confused as to the B3 coefficient in the below:

y = B0 + B1p + B2trt + B3(p*trt) + u

Here, p and trt are dummy variables where p = 0 for pre-treatment and 1 for post-treatment, and trt = 0 for control group and 1 for treatment group. p * trt is an interaction term indicating the marginal impact of the treatment group over the experimental period.

My question is whether this should be a simple dummy variable equal to zero if either p / trt = 0 and 1 if both p / trt = 1?

Improve this question
$\endgroup$
3
  • 5
    $\begingroup$ By construction the interaction of $p\cdot trt=1$ if both $p=1$ and $trt=1$. The coefficient $B3$ gives you the difference in differences estimate. $\endgroup$
    – Andy
    Commented Jan 24, 2014 at 20:57
  • $\begingroup$ Thanks, Andy. And then if either $p=0$ or $t=0$ then $p*t=0$, correct? $\endgroup$
    – Robb
    Commented Jan 24, 2014 at 22:18
  • $\begingroup$ That is correct $\endgroup$
    – Andy
    Commented Jan 24, 2014 at 22:22

0

Reset to default

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.