Help with areg

[Ann Steele]

Hello everyone,
I have a panel data and I want to do DID. My dependent variable is discrete. Can I use areg?? I know that is for linear models, but I saw some papers that do that instead of using probit/logit.
Many thanks!

[Carlo Lazzaro]

Ann:
I've never heard about -areg- with non-continuous regressand (but it may well be my fault).
Do the papers you're referring to report on linear probability models?

[Ann Steele]

I read that linear probability model generally gives consistent results with logit/probit and the interpretation of the estimated coefficient is straightforward. Angrist and Pischke explain that in the Mostly Harmless Econometrics.
thanks!

[Carlo Lazzaro]

Ann:
correct.
You might also be interested in https://www.stata.com/meeting/us20/s...20_Allison.pdf and related references if dealing with liner probability model.
In addition, I would recommend you to take a look at -areg- Tecnical note (please see related Stata .pdf manual entry) just to be sure that this tool is what fits your research needs.