Tutorial on causal mediation analysis with binary variables: An application to health psychology research

Area of Science:

• Health Psychology
• Causal Inference
• Statistical Methods

Background:

• Mediation analysis is crucial for understanding treatment effects in health psychology.
• Identifying mediators and their impact is a key research area.
• Causal mediation analysis provides a framework for these investigations.

Purpose of the Study:

• To introduce causal mediation analysis for binary exposure, mediator, and outcome variables.
• To focus on resampling and weighting methods within the potential outcomes framework.
• To provide practical R code examples for analysis.

Main Methods:

• Utilizes the potential outcomes framework for estimating causal effects.
• Employs resampling and weighting methods for analysis.
• Defines causal effects in a hypothesized mediation chain with binary variables.

Main Results:

• Demonstrates the application of causal mediation analysis with binary data.
• Highlights the importance of temporal order and confounding control.
• Provides functional R code examples using `mediation` and `medflex` packages.

Conclusions:

• Causal mediation analysis with binary variables can be effectively implemented using resampling and weighting methods.
• The tutorial provides practical guidance for researchers in health psychology.
• Understanding direct and indirect effects is essential for interpreting complex relationships.