Hello,
We're using firthlogit for rare events as we have 50,000 observations, only about 1,000 of which are positive. We used the code suggested by Jeff Wooldridge (https://www.statalist.org/forums/for...-probit-models) to test our data for heteroskedasticity, and found that we do have an issue. But we can't use hetprobit, because hetprobit doesn't adjust for rare events. How do we handle such a case?
Thanks for all help!
Margaret
We're using firthlogit for rare events as we have 50,000 observations, only about 1,000 of which are positive. We used the code suggested by Jeff Wooldridge (https://www.statalist.org/forums/for...-probit-models) to test our data for heteroskedasticity, and found that we do have an issue. But we can't use hetprobit, because hetprobit doesn't adjust for rare events. How do we handle such a case?
Thanks for all help!
Margaret
