Illusory Correlation
Seeing a relationship between two things — a group and a behavior, a remedy and a recovery — that the actual numbers don't support.
Examples
Lena: “Every time I wash my car, it rains.”
It doesn’t. But wash-then-rain is a memorable pair, and wash-then-sunshine isn’t, so memory keeps score unevenly.
The harmless version ends there; the harmful one shapes how we see people:
Comment: “Another rude tourist from that country. It’s always them.” Reply: “Is it? Or do you remember the rude ones because they stand out twice — foreign and rude?”
In the lab, people shown two groups with identical rates of bad behavior still judge the smaller group more negatively: rare person plus rare act makes a double-distinctive pair that memory over-files. That’s a stereotype forming in real time, from data that contains no difference at all.
Why it happens
Your mind is a pattern detector that never shows its arithmetic. Vivid, unusual, or expected pairings get stored and recalled easily; the disconfirming cases — the thousand times nothing happened — leave no trace to count. When you then “consult your experience”, the tally is rigged before you start. Expectation makes it worse: once you believe in the link, confirmation bias files each fitting case as further proof.
Superstitions, miracle cures, and group stereotypes all ride this mechanism. Its deliberate, argumentative cousin is the Texas sharpshooter fallacy — drawing the target around the hits after the fact.
How to counter it
- Draw the full table. A real correlation needs four numbers: happened/didn’t for both cases. “Every time I X, Y” cites one cell of four.
- Ask what you’re not seeing: the quiet cases that never make memory — or the news.
- Distrust vividness as evidence. Memorable is not the same as frequent; it’s usually the opposite.
- For claims about groups of people, demand base rates before accepting the pattern — and offer them charitably when someone else has been fooled: “Standouts stick in memory. What do the overall numbers look like?”