A Directed Acyclic Graph represents your causal assumptions about how variables relate to each other.
Key Concepts
- Nodes represent variables in your analysis (treatment, outcome, confounders, etc.)
- Edges represent direct causal relationships - an arrow from A to B means A directly causes B
- The graph must be acyclic - no variable can cause itself through any path
- Include all variables that affect both treatment and outcome (confounders)
The DAG encodes your causal assumptions. It's the foundation for determining whether and how causal effects can be identified from observational data.
Draw edges by dragging from one node handle to another