http://www.businessanalysis.cn/x/html/46/t-14946.html

这个之前也搜到了，感觉没太看懂……
哎~不过还是多谢好心人帮助，呵呵！

因为我现在在看事件史分析的东西，明明在说logit，但是用在spss里面就用logistics来做，这两个一样吗？很困惑啊

Logit model and logistics model are the same thing here.

Suppose that y* is a person's utility which a modeler does not observe.

                y*=x*beta + err;

if              Y* >0 then y=1;
else                           y=0;

Here 1=buy
         0 not buy.

A Modeler observes this action instead.  

If err is a logistic distribution, then it is a logistic model(regression).

p(y=1) = cdf(x*beta)  and logit=log(p/(1-p))=x*beta         The math is easy to work with. <----- this is nice one, looks like a linear regression.


BTW, if err is normally distributed, then it is probit model. this is not true in probit (log(p/(1-p))=x*beta) in general.

Hope this helps.

谢谢讲解  貌似清楚了一点点  \(^o^)/~