I am using Stata's heckprob to estimate a heckit probit model. However, it
seems there is little available out there re explaining Rho, for example,
what's the meaning of Rho being positive or negative. I estimate four nested
heckprob models, and as I add more variables in, the sign of Rho changes,
and its significance level varies too. Any input?
Jun Xu
Department of Sociology
Indiana University at Bloomington