Abstract: This paper proposes an approach to estimate quantiles of the distribution of treatment effects under the identifying assumption that treatment assignment is based on  observed characteristics. We use a matching approach to derive the distribution of treatment effects from differences in outcomes between matched treatment and control units.  Our parameters of interest may be interpreted as generalized versions of the quantile treatment effect (QTE) and the quantile treatment effect on the treated (QTT), which can be identified without imposing a rank...

Abstract: Whether police discriminate on the basis of race and ethnicity when making stops is a topic of frequent debate among academics, in courts, and beyond. However, due to implausible assumptions about police behavior, the most commonly used tests are quite susceptible to indicating discrimination when it is not present or to indicating a lack of discrimination when it is present. This is true of research on the New York Police Department’s (NYPD) practice of Stop, Question, and Frisk (SQF), a particularly contentious case where findings have been mixed. Using data...

Abstract: Systematic discriminatory biases present in our society influence the way data is collected and stored, the way variables are defined, and the way scientific findings are put into practice as policy. Automated decision procedures and learning algorithms applied to such data may serve to perpetuate existing injustice or unfairness in our society.  We consider how to solve prediction and policy learning problems in a way which ``breaks the cycle of injustice'' by correcting for the unfair dependence of outcomes, decisions, or both, on sensitive features (e.g...

Abstract: This work introduces permutation weighting: a weighting estimator for observational causal inference under general treatment regimes which preserves arbitrary measures of covariate balance. We show that estimating weights which obey balance constraints is equivalent to a simple two-class classification problem between the observed data and a permuted dataset (no matter the cardinality of treatment). Arbitrary probabilistic classifiers may be used in this method; the hypothesis space of the classifier corresponds to the nature...