Correlation vs causation: funny examples with real data
A short guide with 37 examples you can show a class, a colleague or a news editor.
Correlation means two things move together. Causation means one makes the other move. The rest of this page is about the gap between the two, explained with charts that are statistically real and causally ridiculous.
Four reasons unrelated things correlate
- A shared trend over time. Anything that grows every year correlates with anything else that grows every year: streaming subscribers, ghost tours, the price of candy. Time is the hidden variable.
- The seasons. Ice cream truck complaints and campsite nights both peak in July. Neither causes the other; summer causes both.
- A common cause. More people means more of almost everything. Cities, incomes and smartphones push many numbers in the same direction.
- Chance and too many comparisons. We compare 292 datasets with each other: tens of thousands of pairs. Some will line up perfectly by luck, and those are the ones that make headlines.
How to spot a spurious correlation
- Ask for the mechanism. If nobody can explain how A changes B, be suspicious.
- Look at the time axis. Two lines that only rise steadily will always correlate.
- Check what else changed in the same period.
- Ask how many other pairs were tested before this one was found.
Examples
Every example below uses real public data and a real Pearson correlation. The explanations are jokes, and each page explains why the correlation means nothing.
For teachers
Pick two variables in our correlation generator and let students argue for a cause, then take it apart together. Or let them upload their own numbers and see what their data correlates with. Every figure names its sources, listed on the data sources page.
Questions
What is the difference between correlation and causation?
Correlation means two things change together. Causation means one of them makes the other change. A correlation can come from a cause, but also from a shared driver, a common trend over time, or pure chance.
What is a spurious correlation?
A spurious correlation is a strong statistical relationship between two things that have nothing to do with each other, such as cheese stockpiles and the number of new programming questions online. With enough datasets, you will always find some.
Why do unrelated things correlate?
Mostly because both grow or shrink over the same years (population, prices, technology), because both follow the seasons, or because so many pairs were compared that some line up by chance.
How do I know if a correlation is causal?
You need a plausible mechanism, a controlled experiment or a natural experiment, and results that hold up in other data. A high r value alone proves nothing.