A0533
Title: Moving toward best practice when using propensity score weighting in survey observational studies
Authors: Guangyu Tong - Yale University (United States) [presenting]
Abstract: Propensity score weighting is a common method for estimating treatment effects with survey data. The method is applied to minimize confounding using measured covariates that are often different between individuals in treatment and control. However, existing literature does not reach a consensus on the optimal use of survey weights for population-level inference in propensity score weighting analysis. Under the balancing weights framework, a unified solution is provided to incorporate survey weights in both the propensity score estimation and the outcome regression model. Estimators are derived for different target populations, including combined, treated, controlled, and overlap populations. A unified expression of the sandwich variance estimator is provided, and it is demonstrated that the survey-weighted estimator is asymptotically normal, as established through the theory of M-estimators. Through an extensive series of simulation studies, the performance of the derived estimators is examined, and the results are compared with those of alternative methods. Two case studies are also carried out to illustrate the application of different propensity score weighting methods with complex survey data. The discussion is concluded with the findings, and practical guidelines are provided for propensity score weighting analysis of observational data from complex surveys.
