Hello!
I wanted to use the diff command to produce kernel propensity score diff-in-diff estimates.
However, the dataset I am using is an unbalanced panel and therefore I am not sure how to go about it.
I tried the following two versions of code:
1) diff outcome, t(treatment) p(time) cov(`controls') kernel id(ID)
2) diff outcome, t(treatment) p(time) cov(`controls') kernel rcs id(ID) where rcs is repeated cross section
I am getting a sensible output for 2) but for 1) the output looks like this:
any idea why this is happening?
My larger question is when we combine propensity score matching with diff-in-diff, is the panel nature of the data preserved while matching? What difference does the rcs option make in the diff command?
Will be grateful for any sort of help - thanks in advance!
I wanted to use the diff command to produce kernel propensity score diff-in-diff estimates.
However, the dataset I am using is an unbalanced panel and therefore I am not sure how to go about it.
I tried the following two versions of code:
1) diff outcome, t(treatment) p(time) cov(`controls') kernel id(ID)
2) diff outcome, t(treatment) p(time) cov(`controls') kernel rcs id(ID) where rcs is repeated cross section
I am getting a sensible output for 2) but for 1) the output looks like this:
any idea why this is happening?
My larger question is when we combine propensity score matching with diff-in-diff, is the panel nature of the data preserved while matching? What difference does the rcs option make in the diff command?
Will be grateful for any sort of help - thanks in advance!
