Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.
From | Inga van den Bongard <ivdb@uni-mannheim.de> |
To | statalist@hsphsun2.harvard.edu |
Subject | st: standard errors for conditional predicted probabilities after biprobit |
Date | Thu, 02 Dec 2010 19:43:02 +0100 |
Dear all,I am estimating a recursive bivariate probit model and calculate the conditional predicted probabilities. As prvalue or prgen does not work after biprobit, I tried predictnl estimate=(binormal(xb(#1),xb(#2),$rho)/normal(xb(#2)), se(se) in order to get standard errors of the conditional predicted probabilities. Unfortunately, the delta method standard errors seem to be extremely (too?) high (compared to an ordinary probit). Is it possible that the delta method does not perform very well in this context?
Can you recommend a way to calculate standard errors and/or confidence intervals for conditional predicted probabilities?
Many thanks! Best Inga * * 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/