The method of difference-in-differences (DID) is widely used to study the causal effect of policy interventions in observational studies. DID employs a before and after comparison of the treated and c
Valid causal inference from observational studies requires controlling for confounders. When time-dependent confounders are present that serve as mediators of treatment effects and affect future treat
Assumptions that are sufficient to identify local average treatment effects (LATEs) generate necessary conditions that allow instrument validity to be refuted. The degree to which instrument validity