Ladder of causation (association/intervention/counterfactual)

Description

A three-level framework classifying questions and evidence as association (seeing), intervention (doing) or counterfactual (imagining), clarifying what each study can prove.

Pearl's ladder sorts causal claims by the information needed to support them. Rung one, association, answers what users who did X also did; correlations, funnels and segment comparisons live here. Rung two, intervention, answers what happens if we make users do X; randomised experiments and adjusted observational analyses reach it. Rung three, counterfactual, answers what would have happened to this specific user had we acted differently; it requires structural models and stronger assumptions. Data from a lower rung cannot answer a higher-rung question without assumptions. For UX, the ladder is a classification tool for research questions and claims, exposing reports that reach rung-three language on rung-one evidence.

Attributes

Loss type
Latent
Stage
Strategy
Effort
Hours

Steps

  1. List each research question and each claim in a plan or report.
  2. Classify each by rung: association, intervention or counterfactual, based on the wording of the claim.
  3. Classify the evidence supporting each claim by the rung it can reach (e.g. survey correlation is rung one; an A/B test is rung two).
  4. Flag mismatches where the claim sits higher than the evidence, and either reword the claim or plan a design to climb the ladder.
  5. For rung-two needs, choose an experiment or quasi-experiment; for rung-three, specify the structural model and assumptions required.

Sources

  • Causal Inference: What If (Miguel Hernán & James Robins)
  • DAGs+SEM Playbook (Vivek Gopinath)
  • Nash Equilibrium Playbook (Vivek Gopinath)
  • The Book of Why (Judea Pearl & Dana Mackenzie)
  • Causality (Judea Pearl)
  • Causal Inference (Hernán & Robins)
  • DAGs+SEM Playbook (Gopinath)
  • Nash Equilibrium Playbook (Gopinath)
  • The Inquiry Series, Ch 45 — The Causal Revolution — Pearl's Ladder & the Fundamental Problem

Also known as

  • Pearl's causal hierarchy
  • Three rungs of causation

Categories

  • Research Operations
  • Research Design & Rigor