很多人种学实地研究致力于此，但是依旧总是出现一些观测偏差。
Many ethnographic field studies attempt to do this, though there are always some observation biases.
计算结果表明，计算值与观测值吻合得比较好，计算值与观测值相对偏差的平均值小于2.5%。
The results show a good agreement between measured and computed va-lues and mean value of the bias is less than 2.5%.
非观测效应，以及样本选择偏差和测量误差所引发的内生性问题对结果的影响很大，导致普通最小二乘法低估了我国的教育收益率。
The study also finds that unobserved effects, sample selection bias and measurement errors play important role on results, and OLS really underestimates yield of schooling.
应用推荐