中文摘要：

通过文献回顾发现,目前老龄健康研究最大的问题是研究结论的不一致,甚至相互矛盾,其根本原因在于未能正确处理内生性、生态学谬误等问题。建议通过工具变量法解决联立性偏误、随机分配法解决自选择偏误、Heckman二阶段法解决样本选择偏误、面板数据解决变量遗漏、利用滞后期数据解决反向因果关系问题,利用HLM模型解决生态学谬误。

英文摘要：

Through literature review,the most serious current problem is that many of the test conclusions of health effects are inconsistent,and even contradictory.The basic reasons of inconsistent,including endogenous,ecological fallacy.The proposal is by IV to solve simultaneity bias,random distribution method to solve the selection bias, Heckman two-stage method to solve the sample selection bias,Panel data to solve omitted variable,lag data t solve the problem of reverse causality,using HLM model to solve ecological fallacy.