A Framework for Selecting Causal Estimands in Observational Studies with Limited Overlap

This paper proposes a new framework for selecting causal estimands in observational studies, particularly when facing limited overlap between treatment and control groups, aiming to balance statistical performance with preserving the intended target population.