Methods in causal inference. Part 1: causal diagrams and confounding.

Area of Science:

• Causal inference and statistical modeling.
• Methodology for observational data analysis.

Background:

• Causal inference necessitates comparing counterfactual scenarios under interventions.
• Deriving these comparisons from data relies on specific assumptions and complex workflows.
• Causal diagrams are vital for assessing the identifiability of counterfactual contrasts.

Purpose of the Study:

• To elucidate the application of causal directed acyclic graphs (DAGs) in causal inference.
• To demonstrate how to determine the identifiability of causal effects from non-experimental data.
• To provide practical guidance and strategies for avoiding common errors in causal analysis.

Main Methods:

• Utilizing causal directed acyclic graphs (DAGs) to represent causal relationships.
• Applying DAG-based criteria to assess the identifiability of causal effects.
• Developing a structured workflow for causal effect identification from observational data.

Main Results:

• A clear framework for using causal DAGs to ascertain identifiability of causal effects.
• Identification of key assumptions and potential pitfalls in observational causal inference.
• Practical reporting guidelines for causal analyses based on DAGs.

Conclusions:

• Causal DAGs are essential tools for determining the identifiability of causal effects.
• A systematic approach using DAGs enhances the rigor of causal inference from observational data.
• Adherence to best practices and awareness of pitfalls are crucial for valid causal conclusions.