PRIN2017 (HiDiNet) Seminar: Yulong Wang (Syracuse University)

Title: "Testing Limited Overlap"

  • Date: 26 November 2021 from 15:00 to 16:00

  • Event location: Microsoft Teams

Abstract

Limited overlap, reflected either by a large discrepancy in covariate distributions between the treatment and the control group, or by the presence of extreme propensity scores, can be a threat to the estimation of and inference on treatment effect parameters. In this paper, we propose a formal statistical test which helps assess the degree of limited overlap. Rejecting the null hypothesis in our test indicates either no or very mild degree of limited overlap, and hence reassures that standard treatment effect estimators will be well-behaved. One distinguishing feature of our test is that it only requires the use of a few extreme propensity scores, which is in stark contrast to other methods that require consistent estimates of some tail index. Without the need to extrapolate using observations far away from the tail, our procedure is expected to exhibit excellent size properties, a result that is also borne out in our simulation study.