Reducing bias through directed acyclic graphs
2008

Reducing Bias with Directed Acyclic Graphs

publication Evidence: moderate

Author Information

Author(s): Ian Shrier, Robert W. Platt

Primary Institution: McGill University

Hypothesis

Can a simple 6-step approach using directed acyclic graphs (DAGs) effectively reduce bias in causal inference?

Conclusion

Using a simple 6-step DAG approach is likely to reduce bias in effect estimates in statistical models.

Supporting Evidence

  • The traditional approach to confounding may create bias rather than eliminate it.
  • Many researchers continue to use outdated methods due to a lack of understanding of DAGs.
  • The 6-step process helps identify appropriate covariates to minimize bias.

Takeaway

This study shows a way to use diagrams to help researchers avoid mistakes when figuring out if one thing causes another.

Methodology

The manuscript outlines a 6-step process for using directed acyclic graphs to assess and minimize bias in causal inference.

Potential Biases

The traditional methods of adjusting for confounding may introduce new biases instead of minimizing them.

Limitations

The study is limited to a conceptual discussion and does not provide empirical validation of the 6-step process.

Digital Object Identifier (DOI)

10.1186/1471-2288-8-70

Want to read the original?

Access the complete publication on the publisher's website

View Original Publication