Causal Mediation Analysis for Effect Heterogeneity.

Area of Science:

• Epidemiology
• Biostatistics
• Social Sciences

Background:

• Understanding causal pathways from exposure to outcome is vital.
• Effect heterogeneity across subgroups complicates analysis and intervention.
• Existing methods may not fully capture subgroup-specific causal mechanisms.

Purpose of the Study:

• To extend causal mediation analysis for heterogeneous effect decomposition.
• To provide nonparametric definitions and identification assumptions for effect heterogeneity.
• To offer analytical formulas for direct and indirect effect heterogeneity measures.

Main Methods:

• Utilizing a counterfactual framework extended for heterogeneous effects.
• Developing nonparametric definitions and identification strategies.
• Applying causal mediation analysis to decompose effect heterogeneity.

Main Results:

• Introduced novel measures for direct and indirect effect heterogeneity.
• Provided a framework for understanding subgroup-specific causal effects.
• Demonstrated application using neighborhood poverty, mental health, and gender.

Conclusions:

• Effect heterogeneity decomposition offers deeper insights into causal mechanisms.
• The methodology clarifies for whom and in what context effects operate.
• Findings have implications for targeted interventions in public health and social sciences.