Statistical Power for Individual and Contextual Mediation in Cluster-Randomized Designs

• In Event: 66.052 - Estimation Accuracy and Conditions for Complex Multilevel Models

Abstract

We develop closed-form expressions to estimate the power to detect causally-defined individual, contextual, and cumulative indirect effects in two-level group-randomized studies examining individual-level mediators (i.e., 2-1-1 mediation). We formulate our approach within the purview of typical multilevel mediation models and anchor their interpretation in the potential outcomes framework. The results provide simple power analysis formulas for asymptotic- and resampling-based methods that reduce calculations to simple functions of the primary path coefficients and common summary statistics (e.g., intraclass correlation coefficients). Probing these formulas suggests that group-randomized designs will typically be well-powered to detect individual indirect effects and can be well-powered to detect contextual and cumulative indirect effects when carefully planned. The power formulas are implemented in freely available software.

Authors

• Benjamin Kelcey, University of Cincinnati

• Nianbo Dong, University of Missouri - Columbia

• Jessaca K. Spybrook, Western Michigan University

• Kyle T. Cox, University of Cincinnati