由于我们不能通过随机试验解决遗漏变量偏差和内生性问题，因此以下这些方式成为次优中的最优选择：二阶段最小二乘法，结构方程模型，广义矩估计，控制方程方法。这四种方法都能够得到相同的估计结果，因为都是在通过寻找工具变量来控制诸如内生性等问题，当然这里面最具一般性的方法就是广义矩估计，因为其他方法都可以通过广义矩估计得到。
注：点击图片，看清晰文字（PDF下载看文末）。
In the first example, I used an estimator that exists in Stata. In the last two examples, I used estimation tools that allow us to obtain estimators for a large class of models.
Concluding remarks
Estimating the parameters of a model in the presence of endogeneity or related problems http://blog.stata.com/tag/endogeneity/ is daunting. Above I illustrated how to estimate the parameters of such models using commands in Stata that were created for such a purpose and also illustrated how you can use gmm and sem to estimate these models.
需要PDF版本的，请看文末最后一句话。