Hello,
I manually try to estimate a Heckman sample selection model using following code:
It works without a problem (at least I think so).
If we manually estimate the Heckman model, we need to correct the standard error in the second step. Using bootstrap command seems to be a choice in normal case.
However, my data contains survey weights, what I need to do in the second step is something like that:
As you probably guess, the code gives an error: weights not allowed
The question is that how can I use bootstrap with survey weights?
I hope I have given enough details for the problem.
Thank you
Alper
I manually try to estimate a Heckman sample selection model using following code:
Code:
probit ydummy x predict ydummyhat, xb gen IMR = normalden(ydummyhat)/normal(ydummyhat) regress y x IMR
If we manually estimate the Heckman model, we need to correct the standard error in the second step. Using bootstrap command seems to be a choice in normal case.
However, my data contains survey weights, what I need to do in the second step is something like that:
Code:
bootstrap, reps(100) seed(1): regress y x IMR [pweights=mysurveyweights]
The question is that how can I use bootstrap with survey weights?
I hope I have given enough details for the problem.
Thank you
Alper
