Ratio-of-mediator-probability weighting (RMPW) for causal mediation analysis under treatment-by-mediator interaction¶

Theory · All theories
Evidence · 2 claims (2 for) · 1 study (1 theoretical), q2 · 0 of 1 report an effect size · 2 claims rest on one study

Description¶

RMPW is a weighting-based approach to causal mediation analysis that decomposes a total effect into natural direct and indirect effects even when the mediator–outcome relationship depends on the treatment condition. The article gives "an intuitive explanation of the RMPW rationale, a mathematical proof, and simulation results for the parametric and nonparametric RMPW procedures." Unlike model-based alternatives, it relies on propensity score models for the mediator rather than a fully specified outcome model.

Design Implications¶

Context¶

Requirements¶

• Correct specification of the propensity score models for the mediator when parametric RMPW is applied

Constraints¶

• Causally valid results require that the sequential ignorability assumptions hold

Target Learners¶

• education and behavioral science researchers conducting mediation analysis

Target Learning Objectives¶

• decomposing total treatment effects into direct and indirect components in the presence of treatment-by-mediator interaction

Claims¶

• RMPW was applied to test whether employment mediated the effect of an experimental welfare-to-work program on maternal depression [+W]
• RMPW decomposes total effects into natural direct and indirect effects, with the indirect effect further split into a pure indirect effect and a natural treatment-by-mediator interaction effect [+W]

Related Theories¶

Examples¶

• RMPW software program and online Stata code

Key Sources¶

• Guanglei Hong, Jonah Deutsch, Heather D. Hill. (2015). Ratio-of-Mediator-Probability Weighting for Causal Mediation Analysis in the Presence of Treatment-by-Mediator Interaction. Journal of Educational and Behavioral Statistics, vol. 40, no. 3. https://www.mathematica.org/publications/ratioofmediatorprobability-weighting-for-causal-mediation-analysis-in-the-presence