All Times EDT
Keywords: Causal inference, Double robustness, Prognostic score, Propensity score
Unlike in randomized clinical trials (RCTs), confounding control is critical for estimating the causal effects from observational studies due to the lack of treatment randomization. Under the unconfoundedness assumption, matching methods are popular because they can be used to emulate an RCT that is hidden in the observational study. To ensure the key assumption hold, the effort is often made to collect a large number of possible confounders, rendering dimension reduction imperative in matching. Propensity score matching has been a long-standing tradition for handling high-dimensional confounding, however requiring stringent model assumptions. In this talk, I will introduce novel double score matching (DSM) utilizing both the propensity score and prognostic score. To gain protection from possible model misspecification, we posit multiple candidate models for each score. Theory shows that the de-biasing DSM estimator achieves the multiple robustness property in that it is consistent if any one of the score models is correctly specified. Recommendations will also be given regarding balance measures, variable selection, matching bolts and nuts.