Matching Strategies for Observational Data With Multilevel Structures

• In Event: Propensity Score Methodology

Abstract

Despite the popularity of propensity score techniques for estimating causal treatment effects for observational data, they are not yet well studied for multilevel data where selection into treatment takes place among level-one units within clusters. Using a simulation study, we investigate two strategies for matching level-one units (students): (i) within-cluster matching where matches are formed within clusters (schools) and (ii) across-cluster matching where units may be matched across clusters. We conclude that both strategies are able to produce consistent estimates of the average treatment effect. However, we demonstrate that across-cluster matching requires stronger assumptions than within-cluster matching and that switching to an across-cluster matching strategy cannot directly compensate for a lack of overlap between treated and control units within clusters.

Authors

• Courtney E. Hall, University of Wisconsin - Madison

• Jee-Seon Kim, University of Wisconsin - Madison

• Peter M. Steiner, University of Wisconsin - Madison