arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 1201.6385stat.AP
收藏

SPSS中倾向得分匹配

Propensity score matching in SPSS

Felix Thoemmes

摘要:

倾向得分匹配是非随机研究中用于因果推断的工具,允许对大量协变量进行调节。倾向得分在社会科学中的使用目前正在经历一个巨大的增长,但它远远不是一个常用的工具。更广泛地使用倾向得分方法的一个障碍是对专业软件的依赖,因为许多社会科学家仍然使用SPSS作为他们的主要分析工具。本文介绍了各种倾向得分匹配方法在SPSS中的实现。具体来说,SPSS自定义对话框允许研究人员使用熟悉的点击界面指定倾向得分方法。该软件允许使用逻辑斯谛回归估计倾向得分,并指定具有许多选项的最近邻匹配,例如,卡钳、公共支撑区域、有无替换匹配、一对多匹配。详细的平衡统计和图表是由程序产生的。

英文摘要:

Propensity score matching is a tool for causal inference in non-randomized studies that allows for conditioning on large sets of covariates. The use of propensity scores in the social sciences is currently experiencing a tremendous increase; however it is far from a commonly used tool. One impediment towards a more wide-spread use of propensity score methods is the reliance on specialized software, because many social scientists still use SPSS as their main analysis tool. The current paper presents an implementation of various propensity score matching methods in SPSS. Specifically the presented SPSS custom dialog allows researchers to specify propensity score methods using the familiar point-and-click interface. The software allows estimation of the propensity score using logistic regression and specifying nearest-neighbor matching with many options, e.g., calipers, region of common support, matching with and without replacement, and matching one to many units. Detailed balance statistics and graphs are produced by the program.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑