Zastosowanie analizy wrażliwości do skorygowania obciążenia efektów oddziaływań oszacowanych za pomocą Propensity Score Matching
The application of sensitivity analysis to adjust bias of treatment effects estimated by Propensity Score Matching

Summary/Abstract: In observational studies we cannot be sure that we have taken into account all variables influencing the selection process to the treatment group and the outcome variable. Such an unmeasured confounder may result in the bias of estimated treatment effect. Rosenbaum [2005] recommends to routinely complement studies based on matching, including Propensity Score Matching, with sensitivity analysis for such an unobserved potential confounder. The author of the paper proposes the application of the sensitivity analysis method developed by VanderWeele and Arah [2011], which is an interesting alternative to Rosenbaum’s sensitivity analysis [2002], since it can be used to estimate the bias caused by an unobserved confounder and to adjust estimated treatment effect. The paper presents the application of VanderWeele and Arah’s method to the results of the study of the net effect of internships (estimated with PSM) organised by District Employment Office in Tarnów [Denkowska 2015].

• Details
• Contents