Abstract
Directed acyclic graphs (DAGs) offer a clear, structured approach to depicting causal relationships in observational research, helping to distinguish likely causation from mere association. By explicitly mapping assumptions about confounders, mediators, and colliders, DAGs guide appropriate variable selection for adjustment and help avoid common sources of bias and confounding. This transparent framework supports more rigorous causal inference and serves as a compelling complement or alternative to large randomised controlled trials. Here we introduce some foundational principles of DAGs in support of recent DAG-guided work published around the SNAP-3 project.