MALTS: Matching After Learning to Stretch

TL;DR

This paper presents MALTS, a flexible and interpretable framework that learns a covariate distance metric to improve the quality of matches in causal inference, especially when irrelevant covariates are present.

Contribution

It introduces a novel method to learn a covariate stretching metric for matching, enhancing match quality and interpretability in causal effect estimation.

Findings

Higher quality matches compared to traditional methods

Improved estimation of conditional average treatment effects

Enhanced interpretability of matching process

Abstract

We introduce a flexible framework that produces high-quality almost-exact matches for causal inference. Most prior work in matching uses ad-hoc distance metrics, often leading to poor quality matches, particularly when there are irrelevant covariates. In this work, we learn an interpretable distance metric for matching, which leads to substantially higher quality matches. The learned distance metric stretches the covariate space according to each covariate's contribution to outcome prediction: this stretching means that mismatches on important covariates carry a larger penalty than mismatches on irrelevant covariates. Our ability to learn flexible distance metrics leads to matches that are interpretable and useful for the estimation of conditional average treatment effects.