Correlation Is Not Causation
Two things can rise and fall in perfect step without either one causing the other. Mistaking that pattern for cause is the most common error in reading evidence.
When A and B move together, three explanations survive: A causes B, B causes A, or some third thing C causes both. The correlation alone cannot tell you which. Ice-cream sales and drowning deaths rise together — because summer heat drives both, not because one causes the other.
The pattern is seductive because our minds are built to see causes. A story arrives instantly and feels like understanding, while the boring truth — a shared cause, or pure coincidence — takes work to uncover and offers no narrative.
This is why the controlled experiment exists: it is the machine built specifically to turn a correlation into a causal claim by changing one thing while holding everything else steady. Absent that, a correlation is a lead to investigate, never a conclusion to announce.
Interrogate the link
Name all three arrows
For any correlation, write out A→B, B→A, and C→both. Force yourself to state the reverse direction and the hidden third cause before accepting the obvious one.
Hunt the lurking variable
Ask what could drive both at once — season, wealth, age, who chose to take part. The best confounder is often something nobody measured.
- What causes both of these?
- Who selected into each group, and why?
Demand an experiment for a causal claim
If the stakes are causal — will this intervention work? — insist on a study that manipulated the cause. Observation alone cannot carry that weight.
The world is full of things that move together and have nothing to do with each other.