21.1 The Ladder of Causation

Pearl’s Ladder of Causation describes three hierarchical levels of causal reasoning:

Level	Activity	Questions Answered	Examples
Association	Seeing	What is? How does seeing $X$ change my belief in $Y$?	What does a symptom tell me about a disease?
Intervention	Doing	What if? What happens if I intervene and change $X$?	If I study more, will my test score improve?
Counterfactuals	Imagining	Why? What would have happened if $X$ had been different?	If I had quit smoking a year ago, would I be healthier today?

(Adapted from (Pearl 2019), p. 57)

Each level requires more cognitive ability and data. Classical statistics operates at Level 1 (association), while causal inference enables us to reach Levels 2 and 3.

References

———. 2019. “The Seven Tools of Causal Inference, with Reflections on Machine Learning.” Communications of the ACM 62 (3): 54–60.