Dear _all,
After checking with the user who reported the bug, it appears that
-triprobit- is NOT bugged, and the model will be correctly estimated in
any case, even with endogenous binary rhs variables.
My apologies to the users who might have been troubled by this, they need
not to re-run all the estimations...
Best,
Antoine.
Antoine Terracol wrote:
> Dear statalisters,
>
> A quite serious flaw has been found by a -triprobit- user. The
> problem occurs when one has endogenous dichotomous rhs variables in an
> equation.
> For example, the following model
>
> y1=c1+x1*b1+a1*y2+a2*y3+eps1
> y2=c2+x2*b2+eps2
> y3=c3+x3*b3+eps3
>
> will be incorrectly estimated by -triprobit-
>
> However, a model with no endogenous rhs variables, such as
>
> y1=c1+x1*b1+eps1
> y2=c2+x2*b2+eps2
> y3=c3+x3*b3+eps3
>
> will be correctly estimated.
>
> The problem comes from an incorect use of the symmetry property of the
> normal CDF in the presence of endogenous rhs variables.
>
> A new version of -triprobit- will be released as soon as possible, my
> apologies to the users who might have suffered from the bug.
>
> Feel free to mail me if you have further questions.
>
> Best,
> Antoine.
>
> *
> * For searches and help try:
> * http://www.stata.com/support/faqs/res/findit.html
> * http://www.stata.com/support/statalist/faq
> * http://www.ats.ucla.edu/stat/stata/
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
