MEPX
Chapter 4 of 10All chapters
  1. 1What correlation measures
  2. 2The five explanations
  3. 3Confounders
  4. 4Experiments
  5. 5Natural experiments
  6. 6Weighing causal evidence
  7. 7Selection effects
  8. 8Regression to the mean
  9. 9Language and reporting
  10. 10Acting under uncertainty

Chapter 4 of 10

Experiments

The strongest evidence.

Why randomisation works

Assigning treatment by chance balances every other factor on average, including ones nobody has thought of. That is what no amount of statistical adjustment can replicate.

  • A control group shows what would have happened anyway.
  • Blinding stops expectations shifting the outcome on either side.

When it is impossible

Ethics, cost and practicality rule out experiments for many important questions. That does not make weak evidence strong; it means conclusions stay provisional.