Statistical Power for Probing Mediation Effects in Partially Nested Regression Discontinuity Designs

• In Event: Research Methodologies in Education

Abstract

Studies examining mediation effects play a crucial role in the unveiling of mechanisms that underly program theories. Previous literature has developed a suite of design and analysis methods to support these inquires across a broad range of designs and structures. Within this context, regression discontinuity designs have gained prominence across disciplines because of their flexibility in terms of targeted treatment assignment and their retention of high quality of inference on local treatment effects. A notable gap concerns the application of regression discontinuity designs to partially nested structures. In this study, we develop models, principles of estimation, sampling variability, and expressions to track the statistical power to detect mediation effects in partially nested regression discontinuity designs.

Authors

• Fangxing Bai, University of Cincinnati
  Presenting Author

• Benjamin Kelcey, University of Cincinnati
  Non-Presenting Author