Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.
From | "Andrea Morescalchi" <morescalchi@ec.unipi.it> |
To | statalist@hsphsun2.harvard.edu |
Subject | st: biprobit vs treatreg2 |
Date | Sun, 27 Feb 2011 12:59:31 +0100 |
Dear all,I am estimating a model with a binary outcome variable and an endogenous discrete regressor, which can be either binary or multinomial. The multinomial case, i.e. when my regressor has more than two states, is covered by the command - mtreatreg - which uses maximum simulated likelihood methods. Anyway, this command does not seem to support the case of a binary treatment which apparently was supported by the "obsolete" command - treatreg2 -. Unfortunately, it seems that treatreg2, while allowing a binary-treatment, does not allow yet a probit for the outcome. Moreover, the basic - treatreg - command is thought to be applied only for continuous outcome. In the end, it seems to me that the only command I can use for the binary-treatment case is - biprobit - (unless of course I assume exogeneity of the treatment). Is my understanding correct or there is a way to use a treatment-effect estimator for the probit/binary-treatment case?
I Thank you very much for your attention Andrea * * For searches and help try: * http://www.stata.com/help.cgi?search * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/