Parametric g-formula for Testing Time-Varying Causal Effects: What It Is, Why It Matters, and How to Implement It in...

Area of Science:

• Psychology
• Causal Inference
• Longitudinal Data Analysis

Background:

• Longitudinal designs are crucial for examining causal effects over time.
• Treatment-dependent confounding complicates causal inference with time-varying treatments.

Purpose of the Study:

• Introduce psychologists to the parametric g-formula for causal inference.
• Demonstrate the g-formula's utility in handling treatment-dependent confounding.
• Provide an accessible tool for psychological research using longitudinal data.

Main Methods:

• The parametric g-formula estimates the joint distribution of post-treatment variables.
• Statistical models are specified using standard multiple linear regression.
• Implementation is demonstrated using the lavaan R package for structural equation modeling.

Main Results:

• The parametric g-formula effectively handles treatment-dependent confounding.
• The method is conceptually intuitive and easy to implement for psychologists.
• It enables the estimation of marginal structural models for time-varying treatment effects.

Conclusions:

• The parametric g-formula offers a robust solution for causal inference in psychological research.
• This tutorial equips researchers with a valuable analytic tool for longitudinal data.
• It facilitates a deeper understanding of causal relationships in psychological phenomena.