|Saturday, February 17|
|CS24 Causal Inference||
Sat, Feb 17, 11:00 AM - 12:30 PM
Causal Inference with Multilevel Data Structures (303531)*Luke Keele, Georgetown
Lindsay Page, University of Pittsburgh
Sam Pimentel, UC-Berkeley
Keywords: causal inference, clustered data, observational studies
Many observational studies of causal effects occur in settings with clustered treatment assignment. In studies of this type, treatment is applied to entire clusters of units. For example, an educational intervention might be administered to all or a subset of the students in a school while being withheld from all students in another school. In this session, we outline both a conceptual framework and a method of statistical adjustment for observational studies with clustered treatments. First, we outline possible assignment mechanisms at the cluster level. We then derive different designs based on different possible assignment mechanisms. For each of these designs, we develop a matching algorithm for multilevel data based on a network flow algorithm. Our algorithm is fast and scales to large data sets. We also allow the algorithm to trim treated observations to increase balance and increase overlap in the covariate distributions. We apply our algorithm to an intervention in North Carolina.