lz说的是logit模型么？

独立变量要求只取0和1两个值；潜变量模型的误差项e服从logistic分布：G（z）=exp(z)/[1+exp(z)]。

Yes, You can use logistic regressions for 不是配对的样本. Just make sure your dependant variable 'y' is yes or no (1 or 0) on the left side of your equation, and there are no restictions for your predictors (dependant variables) on the right of your logistic model.

谢谢!

不错                     。