I'm digging through stats literature, and not really finding what I need. I'm looking for a difference in differences estimator that uses the within transformation (subtracting off the means of all variables to lose the fixed effects), that can accomodate more than one non-linearly specified variable, and that gives me a cluster-robust standard errors.
I've got four periods of panel data and about 300 individuals. I've got one treatment dummy, time period dummies, and three nonparametric controls (two of which are interactions). The covariates are all exogenous conditional on the fixed effects, which I lose with the demeaning.
Thanks in advance for any advice, lecture notes, papers, or books.
Advertise on Talk Stats