Correlation vs. Causation

Why two things moving together does not prove that one causes the other, and the questions to ask before accepting a causal claim.

When two things tend to move together, they are correlated. When one thing actually makes the other happen, that is causation. Economic news is full of claims that link two numbers - “countries that spend more on education grow faster,” “cities with more cafés have higher incomes” - and it is easy to hear a correlation and quietly assume a cause. Learning to separate the two is one of the most valuable skills for listening to economic arguments.

Why things move together without causing each other

There are three common reasons two things can be correlated without one causing the other.

The first is a confounding variable: a third factor that drives both. Ice cream sales and drowning accidents both rise in hot weather, but ice cream does not cause drowning. Summer causes both.

The second is reverse causation: the cause runs the opposite way from what you assumed. Richer countries tend to have more hospitals, but that does not simply mean hospitals make countries rich; being rich also allows a country to build hospitals.

The third is plain coincidence. With enough data, you can find two unrelated series that happen to rise and fall together purely by chance.

Tutoring and exam scores

Suppose a report finds that students who pay for private tutoring score, on average, 15 marks higher on exams than students who don't. It is tempting to conclude that tutoring adds 15 marks. But families who can afford tutoring often also have more books at home, quieter places to study, and parents with more education. Those factors could explain much of the gap. To learn how much tutoring itself helps, researchers would need to compare similar students where the only real difference is the tutoring - for example, by randomly offering free tutoring to some students and not others, then comparing results.

How economists test for causes

Economists take causation seriously and use several methods to get closer to it. Randomized controlled trials assign a policy or program to some people at random and compare them with a similar group that did not receive it. Because the choice is random, other differences tend to even out. Economists Abhijit Banerjee, Esther Duflo, and Michael Kremer shared the 2019 Nobel memorial prize in economics partly for using such experiments to study poverty.

When experiments aren’t possible, researchers look for natural experiments - situations where something like a new rule affected one area but not a very similar neighbouring area, allowing a fair comparison.

Questions to ask when you hear a claim

When a speaker says one thing “leads to,” “drives,” or “boosts” another, ask: Could a third factor explain both? Could the cause run the other way? Was there a fair comparison group? Words like “linked to,” “associated with,” and “tied to” usually signal a correlation only, and a careful reporter uses them deliberately.

Hearing "linked to" as "caused by"

A common mistake is treating a report that two things are "linked" as proof that changing one will change the other. A correlation is a starting point for questions, not an answer. Look for evidence from fair comparisons before accepting that one thing causes another.

Key takeaways

Correlation means two things move together; causation means one makes the other happen.
A confounding variable can drive both things at once.
Reverse causation means the cause may run the opposite way.
Randomized trials and natural experiments help economists identify causes.
"Linked to" and "associated with" usually describe correlation only.