Syndicate content

Add new comment

Submitted by Berk Ozler on
You can look at Lin's paper that is linked, but, yes, fairly straightforward. When you demean, the coefficient estimate on the uninteracted T will be evaluated at the mean of all variables, rather than for a specific left-out sub-group had you not centered the covariates.

On the R/Stata code, the authors have two estimators they propose and I was referring to only one of them. It's possible that the other estimator, the RWE, requires a little more assistance from the authors...