sem or gsem or ...?

[ingrid spinowen]

Hi everyone,
I am trying to run two Tobit regressions with correlated error terms. I should use "full information maximum likelihood" estimation for solving endogeneity problem. How can I do that?
As I know "gsem" use for Tobit family but couldn't calculate correlation between error terms. On the other hand, in gsem couldn't use FIML. And sem estimation couldn't use for Tobit model. How can I solve this problem?

Thanks.

[Clyde Schechter]

So you want three things: tobit model, correlated errors, and FIML. I don't think there is any way you can get all three in Stata. Maybe MPlus can do this--I'm not sure.

If you are willing to forego FIML, you can do tobit with correlated errors using -cmp-, a user-written program by David Roodman. You can get it from SSC. The help file is quite extensive and includes many clearly explained and well-worked examples.