lshzhh 发表于 2010-10-17 18:40 
请问:多重共线性是指解释变量之间的高度相关性。 如果现在并不知道解释变量之间的相关系数 (是因为缺乏数据而不能做出相关系数),但是只有模型,知道模型中解释变量的系数,那么可以通过模型来判断解释变量之间的多重共线性吗?
谢谢!!
The problem is that determinate of the moment matrix is close to zero. The inverse moment matrix is not quite stable. Remember that the error estimationin ols is s*(x'x)**(-1). The result is t-statistic is small for some coefs. and sensitive to some data points.