nanyagao 发表于 2011-3-30 11:01 
最近在看一篇文章，其中一个模型是这样的 ln hin=B0+B1worse+B3Health+U，因变量是家庭收入，自变量为worse表示健康变差，health表示健康状况 ，实证健康对家庭收入的影响，作者对因变量取了对数，这样做的目的是什么呢，我觉得这么做没有必要，直接用原数据就行了，作者在文中也没有说明这么做的原因，欢迎大家指教
The analyst believes the error term is multiplicative rather than additive. And further more if the error is assumed as log-normal, then the above model would be a simple linear model appeared in most text book.
I am not sure about how the data is collected or how many factors are under controlled. Generally the model is underspecified, the factors like education, years of experience, sex are not included in the model.
扫码加好友，拉您进群
全部版块
我的主页
收藏
