Design and Analysis of Clustered Regression Discontinuity Studies Probing Mediation Effects

• In Event: Marriott Roundtable Session Five
  In Roundtable Session: Educational Research Methodology (Table 3)

Abstract

Clustered regression discontinuity (CRD) designs draw on cluster-level assignment of a treatment on the basis of a cut score variable. Prior research has suggested that CRD designs are a formidable alternative to cluster randomized designs because they provide flexibility in treatment assignment while maintaining a high-quality basis for. However, CRD designs have not been fully developed to address the array of core effects (e.g., main, moderation, and mediation) typically examined in common structures and problems in education studies. We develop closed-form expressions to estimate the statistical power to detect indirect/mediation effects in CRD designs. We formulate models, develop principles of estimation, sampling variability, and inference as well as expressions to estimate the statistical power to detect the main effects.

Authors

• Fangxing Bai, University of Cincinnati
  Presenting Author

• Benjamin Kelcey, University of Cincinnati
  Non-Presenting Author

• Yanli Xie, Florida State University
  Non-Presenting Author

• Kyle T. Cox, University of North Carolina - Charlotte
  Non-Presenting Author