Exploring Effect Heterogeneity in Multi-Centre Trials with First-Party Measurements of Mediators and Confounders

• Activity Number: 77 - Data, Linked Data, and Model-Based Analytics in Social Science
• Type: Contributed
• Date/Time: Sunday, July 30, 2017 : 4:00 PM to 5:50 PM
• Sponsor: Social Statistics Section
• Author(s): David Judkins*
• Companies: Abt Associates, Inc.
• Keywords: Multi-level analysis ; Cross-level interactions ; Causal analysis

Abstract

Several papers have analyzed collections of linked experiments to try to learn about sources of variation in the effectiveness of the interventions tested (Bloom, Hill, and Riccio, 2003; Bloom, Raudenbush, Weiss and Porter, forthcoming). To measure variation in program features and in control conditions, they have relied on third-party measurement systems. In this paper, methods are presented to use the aggregate experiences of members of the treatment groups as mediators of program effectiveness and to use the aggregate experiences of members of the control groups as confounders. Projection of power is also discussed.