Can I use Orderprobit model to regress with both dependent and independent variables are discrete,(e.g. there is only 1 explanatory variable)?
Economics Job Market Rumors » Econometrics Job Rumors
A small question
(2 posts)-
Posted 4 months ago #
-
- This is technically feasible: Stata will (probably) give you an answer.
- It is probably undesirable, because the model is not semiparametrically identified, so the results may rely too much on the distribution of the error terms.
- If you have only one explanatory variable, this seems really undesirable and there is probably a better way to observe this, e.g. a basic cross table (possibly with the cumulative distributions)Posted 4 months ago #