1 article
Why a high p-value on a parallel-trends test can mislead, and how Rambachan-Roth sensitivity bounds reveal when a difference-in-differences claim is fragile.