Abstract

The blopmatching estimator, proposed by Díaz et.al (2015), is a nonparametric estimator of treatment effects in observational studies under strong ignorability. To impute the missing potential outcome of each unit, the blopmatching estimator solves an optimization problem that directly determines the weights and the matches used while optimizing the covariate balance. This work describes the blopmatch command that implements these estimators. Our routines also implement the corresponding estimator of the marginal variances of both the average treatment effect and the average treatment effect on the treated, providing confidence intervals for them.