最近写一篇小论文，用的是N为8，T为13的面板数据，模型经hausman检验p值大于0.05，那就使要用随机效应模型吗？我在一个老外的讲义上看到这样一句话：Random and fixed are not that much different if you have panels with lots of years. Random is usually preferred when you have large number of entities. 由于我的模型中N是8，那么最好使用固定效应吗？疑惑！！！到底是选固定效应还是随机效应呢？？？是不是我的数据太少了？
希望各位高手能够给些建议，感激不尽！！！
In a short panel, I advise you try random effect in order to minimize bias. Of course, the p value in your test is very small, which might be another reason you could use random effect
扫码加好友，拉您进群
全部版块
我的主页
收藏
