B1016
Title: Independence weights for causal inference with continuous treatments
Authors: Jared Huling - University of Minnesota (United States) [presenting]
Abstract: Studying causal effects of continuous treatments is important for gaining a deeper understanding of many interventions, policies, or medications, yet researchers are often left with observational studies for doing so. In the observational setting, confounding is a barrier to the estimation of causal effects. Weighting approaches seek to control for confounding by reweighting samples so that confounders are comparable across different values of the treatment. Treatments with continuous or otherwise non-categorical values are often present in medical studies involving Electronic Health Record Data. For example, in studying the causal effect of mechanical power of ventilation in those with acute respiratory disease, mechanical power is continuous. Yet, for continuous treatments, weighting methods are highly sensitive to model misspecification. We elucidate the key property that makes weights effective in estimating causal quantities involving continuous treatments. We show that to eliminate confounding, weights should make treatment and confounders independent on the weighted scale. We develop a measure that characterizes the degree to which a set of weights induces such independence and propose a new model-free method for weight estimation by optimizing our measure. The empirical effectiveness of our approach is demonstrated in a suite of challenging numerical experiments, where we find that our weights are quite robust and work well under a broad range of settings.
