You might try -linktest-. With such a skewed distribution an obvious
possibility is complementary-log-log (-cloglog). But I agree with David
Greenberg that the difference between logit and probit takes very very
very large datasets to be discerned.
HTH
Sam
On Mon, 23 Apr 2007, Yun Liu wrote:
> Dear statalist,
>
> I have a model with binary dependent variable. Only 16% of the
> observations are ones, the rest are zeros. How to determine whether to
> use logit, probit or other models? Thank you!
>
> Yun
> *
> * 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/
