Definition
When two things rise and fall together like a pair of partners, that's called correlation. When one thing truly causes the other to happen, that's called causation. Many things seem to happen together, but that doesn't mean one is the cause of the other.
Moving together doesn't mean one causes the other
At the beach in summer, ice cream sales jump, and the number of drownings climbs right alongside them. Put both numbers on a graph, and you can clearly see them rise and fall together.
So does eating ice cream make people drown? Of course not. Behind both is a real shared cause: hot summer weather. When the temperature soars, more people reach for ice cream. At the same time, crowds head to the water, and accidents go up too.
When two things move together in time or direction like this, we call it correlation. When one event actually sets off another, we call it causation. Mix them up, and you might come up with a silly fix like banning ice cream to stop drownings.
The hidden third factor
Every morning in a country village, the rooster crows at the top of its lungs, and soon after, the sun always rises. Seen hundreds of times, these two events look like a perfect pair.
But the rooster's crow isn't what hauls the huge sun into the sky. Behind both events is a giant law of nature: the Earth's rotation. A factor that hides between two events and affects both is called a confounding variable (a hidden factor that distorts the result).
This mistake shows up in daily life and the news all the time. Say a study finds that people who drink lots of coffee live longer. We have to be careful to sort out whether coffee itself added years, or whether it's a lifestyle with enough money and free time to enjoy coffee.
A bit more precisely: how to prove causation
Correlation is relatively easy to find. Just collect lots of data on a computer and calculate whether two values change together. But proving real cause and effect is much harder and demands thorough testing.
To do that, scientists run a randomized controlled trial. They split participants into two groups by lottery, so other conditions spread evenly across both groups. Then they change just one possible cause and compare. For example, one group gets a real new drug and the other gets a look-alike fake pill, to see whether the drug's ingredient alone changes the result.
Time order matters a lot too. The cause must come before the effect, and we also check whether removing the cause shrinks the effect. The more of these checks a claim passes, the more confidently we can call it causation, not just correlation. For questions that can't be tested in an experiment, researchers sometimes weigh evidence from many observational studies together.
Common misconceptions
If two sets of data move in the same shape on a graph, one must cause the other.
It could be pure coincidence, or a hidden third factor (a confounding variable) affecting both. Moving together doesn't automatically mean cause and effect.
Where you meet it
A survey finds that elementary school kids with bigger shoe sizes have bigger vocabularies. But it's not the shoes. The kids are simply older.
The more firefighters sent to a fire, the greater the damage. But firefighters didn't make the damage worse. The fire was just big.
In one sentence
Happening together doesn't mean cause and effect, so check carefully for a hidden shared cause.