Regression commands in Stata typically allow to account for survey weights (e.g. aweights, fweights, iweights or pweights) but I don't think rdrobust allow them. Is there a reason for that? Should not be useful if it allow them?
In my setting, I have around 20 million observations and rdrobust takes a considerable amount of time. In order to reduce the amount of time these regressions take, I thought it could be useful to run the regressions for unique observations in terms of the outcome and the running variable using frequency weights; but given that rdrobust does not allow survey weights, this is not possible. Any other suggestion on how I could speed up the process?
In my setting, I have around 20 million observations and rdrobust takes a considerable amount of time. In order to reduce the amount of time these regressions take, I thought it could be useful to run the regressions for unique observations in terms of the outcome and the running variable using frequency weights; but given that rdrobust does not allow survey weights, this is not possible. Any other suggestion on how I could speed up the process?
