Causal Inference With Clustered Data

In this chapterpotential outcomes with unit-level and cluster-level treatments, interference, and estimands; causal diagrams with cluster nodes and their adjustment sets; identification from within clusters by fixed effects and the within-between model; propensity scores with clustered data; a simulation of estimators under cluster confounding; cluster-randomized and multisite trials with the Tennessee STAR experiment; difference-in-differences, regression discontinuity, and interrupted time series with clusters; mediation and moderation of causal effects; reporting.