Based on the data from a 2013 study conducted by the Institute of Gerontology, Renmin University of China, this study reexamined the ifndings of CAO Yang and WANG Jiwen’s research concerning the impact of elders’ participation in public benefit activities on the level of their life satisfaction. CAO and WANG’s study was also based on the data of the above-mentioned 2013 survey on the situation and service demand of the elderly in Xicheng district of Beijing. Unlike CAO and WANG’s study, this research uses both ordinary binary logistic regression and propensity score n weighted regression. By comparing ifndings resulted from the two methods, this study ifnds that if sample selection bias is not controlled, ordinary binary logistic regression method can’t differentiate the influence of some factors (gender, living style and whether giving care services to others) on urban retirees’ level of life satisfaction. Moreover, ordinary binary logistic regression method will underestimate the effect of age, marital status, health status and number of children. All this obviously inhibits researchers from fully tapping the policy value of empirical research ifndings.%使用2013年中国人民大学老年学研究所在北京市西城区开展的老年人基本情况和服务需求调查数据，重新检验了曹杨和王记文使用同样数据所做的公益活动参与对退休老年人生活满意度的影响研究。与曹杨和王记文研究的不同之处在于，在方法上既使用普通二分类logistic回归也使用倾向值加权回归。通过比较这两种方法得出的研究结论发现，如果不对样本选择偏差加以控制，普通二分类logistic回归不仅无法识别性别、居住方式、为他人提供照料服务对城镇退休老年人生活满意度的影响显著性，还会低估年龄、婚姻状况、健康状况、子女数量的作用程度，这显然不利于研究者充分挖掘实证分析结果所内含的政策价值。
展开▼
