502 – Propensity Score Methods to Conduct Observational Studies Using Complex Survey Data
Causal Inference Using Propensity Score Methods With Clustered Survey Data
Hyunshik Lee
Westat
Ning Rui
Westat
Duck-He Yang
Westat
The propensity score methods have been widely used for causal inference. Majority of studies ignored the sample design feature even when complex survey data were used. In recent years, a number of authors showed that ignoring the sampling weight would lead to biased results. However, causal inference using the propensity score methods for clustered survey data has not been much studied. This paper tries to fill this gap by providing correct ways of incorporating the sample design feature in the calculation of the propensity score and outcome analysis to estimate the treatment effect. The proposed methods will be studied using simulated and empirical survey data.