Illustrating How to Simulate Data From Directed Acyclic Graphs to Understand Epidemiologic Concepts.

Area of Science:

• Epidemiology
• Biostatistics
• Computational Biology

Background:

• Simulation methods are crucial for characterizing real-world phenomena and self-teaching epidemiologic concepts.
• Epidemiology curricula often include causal directed acyclic graphs (DAGs) but may not cover simulation methods.
• A gap exists in practical simulation skills among epidemiologists.

Purpose of the Study:

• To provide a structured approach for building data simulations directly from causal DAGs.
• To equip epidemiologists with practical simulation skills for understanding complex concepts.
• To facilitate self-directed learning and enhance the application of simulation methods in epidemiologic research.

Main Methods:

• Detailing a step-by-step methodology for constructing simulations based on causal DAGs.
• Providing illustrative examples and executable code for simulation procedures.
• Recommending the use of simple DAGs to learn fundamental simulation techniques.

Main Results:

• Demonstrated a feasible method for generating data simulations from causal DAGs.
• Illustrated how simulations can elucidate key epidemiologic concepts such as mediation, confounding bias, and M bias.
• Provided practical code examples for implementation.

Conclusions:

• Building simulations from causal DAGs is an accessible and effective skill for epidemiologists.
• This approach can bolster understanding of core epidemiologic principles and improve research capabilities.
• Empowering epidemiologists with simulation skills can positively influence the conduct of future studies.