proc logistic data=temp descending;
model depvar=indepvar / link=probit;
run; quit;
我还尝试了下面的code：
proc probit data=temp; class depvar; model depvar=indepvar; run;
上面两组程序output出的结果，coefficient符号相反。
我同时用stata运行：
probit depvar indepvar
但是得到结果却与SAS的截然不同，请问这是什么问题？哪里参数用得不对？谢谢！
另外，有人做过logit order regression么？作者: shiyiming 时间: 2010-7-22 09:30 标题: Re: Probit Regression The SAS procedures, such as QLIM, LOGISTIC, and PROBIT can fit the ordered logit model.
For sure, the coefficients’ signs from logistic and probit are the opposite. The difference between STATA and SAS logistic on ordered probit model may be about the assumption on intercept, where SAS set it as 0 while STATA estimates it instead. On the other hand, you can get comparable results between STATA .oprobit and SAS QLIM.
In general, the QLIM procedure are the [quote:2b6f3z6a]most recommended[/quote:2b6f3z6a].作者: shiyiming 时间: 2010-7-22 21:39 标题: Re: Probit Regression 谢谢，又学习了一个，我再去钻研一下qlim语句。