Short answer: correlation means two things vary together. Causation means one thing produces or contributes to a change in another. A correlation can suggest a possible cause, but it does not prove one by itself.
If ice cream sales and sunburn both rise in summer, they are correlated. Ice cream does not cause sunburn. A third factor, hot sunny weather, helps explain both. This is the classic problem: two variables can move together because of coincidence, reverse causation, shared causes, measurement bias, or a real causal relationship.
Why the distinction matters
Confusing correlation with causation leads to bad decisions. A headline might say that people who drink more coffee live longer. That does not automatically mean coffee caused longer life. Coffee drinkers may differ from non-coffee drinkers in income, health, habits, work patterns, or medical history. Good research tries to account for these alternative explanations.
In many fields, observational studies are essential. Researchers cannot randomly assign people to every possible exposure. But observational results need careful interpretation because real life contains many overlapping causes.
How scientists test causation
Randomized controlled trials are powerful because random assignment helps balance other factors between groups. If the groups differ mainly in the treatment they receive, a later difference in outcome is easier to interpret causally. When trials are not possible, scientists use methods such as natural experiments, longitudinal studies, instrumental variables, matching, and mechanistic evidence.
A causal claim becomes stronger when several things line up: the cause comes before the effect, alternative explanations are addressed, there is a plausible mechanism, and different studies point in the same direction. Correlation is often the start of investigation, not the end.