Hello all,

I am working on a Fuzzy RDD, and since my first and second stages come from different samples, I cannot use the command rdrobust with the Fuzzy option. So I want to estimate manually using the Indirect Least Squares (or the Wald estimator since my outcome is binary). My approach is to use two sharp RDD and use the ratio as my coefficient.

For the first equation, I have

Code:
rdrobust bvp psu_score if education==1, c(600)
Where bvp is is the student was selected for the scholarship (binary outcome), psu_score is a test score. If the student scored above 600, they get the offer for a tuition waiver, hence the c(600) option. And education is equal to 1 if the student enrolled in an education major.

and I am after this coefficient:

e(tau_cl) = .7257820300327013

The second equation would be

Code:
bvp education  psu_score, c(600)
and I am interested in this coefficient:

e(tau_cl) = .0313048533101998

How can I estimate the standard error of the ratio .0313048533101998/.7257820300327013?
I could get the same coefficients with the commands rd or lpoly if needed.

I was trying to use nlcom but I couldn't figure it out.

Any help is much appreciated!