|
Abstract:
|
The propensity score method is being increasingly used as a less parametric alternative to the traditional regression methods in many fields . However, propensity score was developed and have been applied in cross-sectional settings with unstructured data. Data collected in the health policy studies are typically clustered or hierarchically structured, in the sense that patients are grouped together in one or more ways. In this talk, we present results in the following aspects: (1) develop and apply valid propensity score methods for causal inference in the context of hierarchically structured data; (2) elucidate assumptions underlying methods for causal inference in the context of hierarchically structured data; and (3) develop diagnostics, which make use of the clustered structure, for testing the sensitivity of causal inferences to important violations of assumptions.
|
