← Evaluate statistically based reports
9 Causation vs correlation
The classic trap: a report finds two things move together and claims one causes the other. Association is not causation.Alternative explanations to raise
- Lurking (confounding) variable, a third factor drives both (e.g. ice-cream sales & drownings ← hot weather).
- Reverse causation. B might cause A instead.
- Coincidence / chance, especially with small samples or many comparisons.
- Causation needs a controlled experiment, a plausible mechanism, and consistency.