It should be Ordered Probit model, not like what the guy on floor 3 said, Multinomial Probit, in fact, there is no such a term defined as "Multinomial Probit", all I have heard about is just Multinomial Logit.
Ordered Probit model is applied to the circustance where the alternative outcomes are more than two, which cannot be solved by Binary Probit model. For example, we can use this in investgating the labor market. In the labor market, people may have at least three status, thus, unemployed, part-time, or full-time. So, if we set up a model to estimate one's probability of being each status, an ordered probit model should be used, and a binary probit model is inappropriate here.
Hope this might help!
[此贴子已经被作者于2005-10-12 9:36:52编辑过]