I'm trying to work out the effect of treatment on a group using a counterfactual approach. However, I don't know how I can infer the treatment start date for the comparison group.

For both groups I have a time point from which on they are eligible for treatment. I'm using kernel matching to match both groups on pre-treatment characteristics (i.e. from the time they became eligible). I don't want to use 1:1 matching because then I can't bootstrap the standard errors of the impact estimates.

I want to measure outcomes relative to the treatment start. For that I will have to estimate a pseudo treatment start for the counterfactual. How can it be done?

I thought of some sort of reverse matching (matching the treatment to the control as a weight on the treatment start date distribution), but not sure whether this could work.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.