Course 5, lesson 46 of 100, Ages 11+
Correlation vs causation
When two things move together
Like I’m 5
In summer, more people eat ice cream and more people swim. Ice cream doesn't make people swim. They both happen because it's hot.
The big idea
Correlation means two things tend to change together. Causation means one actually makes the other happen. AI is brilliant at finding correlations, but a correlation alone doesn't prove a cause.
Often a hidden third factor explains both, like hot weather explaining ice cream and swimming. To test causes, scientists run careful experiments, such as giving a medicine to one group and a dummy pill to another.
Examples
- Ice cream and swimming: Both rise in summer because of heat, not each other.
- Shoe size and reading: Older children have bigger feet and read better: age explains both.
- Fair tests: Medicine trials compare similar groups to find real causes.
How it works
- Notice that two things change together.
- Ask whether a hidden factor could explain both.
- Run a fair experiment to test for a real cause.
Check your understanding
- Ice cream sales and sunburn both rise in summer. What's the likely cause?
- Options: Hot, sunny weather; Ice cream causes sunburn; Sunburn causes ice cream.
Answer: Hot, sunny weather. Sunny weather drives both: a hidden third factor. - How do scientists test whether one thing causes another?
- Options: With fair experiments comparing similar groups; By finding any correlation; By asking a chatbot.
Answer: With fair experiments comparing similar groups. Controlled experiments separate causes from coincidences.
Remember
Things that move together aren't necessarily cause and effect. Look for hidden factors and test fairly.
Talk about it
Can you think of two things that happen together without one causing the other?
Go deeper
Confounding variables create spurious correlations. Randomised controlled trials, A/B tests and causal-inference methods are used to estimate real effects.