How to test difference in coefficients between two samples in Regression Discontinuity

[Hao Liang]

Dear Statalists,

Happy New Year!

I have a question about testing the difference in coefficients between two samples (say using variable Z to partition the whole sample into High-Z subsample and Low-Z subsample) in a Regression Discontinuity setting. I know in standard OLS regressions, this can be achieved via seemingly unrelated estimation ("suest") or testing the coefficient on the interaction term between X and the High-Z dummy. However, I was wondering how to do this in an RD setting. Does suest still work?

Thank you for your help in advance!

Kind regards,
Hao Liang

[Hao Liang]

Just to add: I am using the command "rdrobust" – bias corrected estimates from Calonico, Cattaneo and Titiunik (2014). I need to test the statistical difference between two bias-corrected coefficients.

[Rocio Valdebenito]

Hi Hao,
Did you find a solution to your question?
I'm trying to do exactly the same thing.
Thanks!

[Jeff Wooldridge]

If you have something like a random sample to begin with, and you split the sample on the basis of an exogenous variable, then the estimators will be independent. So the standard error of the difference is sqrt(se1^2 + se2^2). That is, square each standard error, add them together, and then take the square root at the end.