Differentiated matching for individual and average treatment effect estimation

One fundamental problem of causal inference is estimating treatment eect with observational data where variables are confounded. The traditional way of controlling the confounding bias is to match units with different treatments but similar variables. However, traditional matching methods fail on selection and differentiation among the pool of numerous potential confounders, leading to possible under-performance. In this paper, we give a theoretical analysis of confounder differentiation and propose a novel Differentiated Matching (DM) algorithm for both individual and average treatment effect estimation by learning confounder weights for variable differentiation and unit matching. To address the distribution shift in confounder weights learning, we further propose a Propensity Score based DM (PSDM) algorithm by weighted regression with the inverse of the propensity score. Extensive experiments on both synthetic and real-world datasets demonstrate that the proposed algorithms achieve better performance than other matching methods on treatment effect estimation.