You’ve seen the headlines: people who do X live longer. Drink red wine. Take this vitamin. Run marathons. Most of these come from observational studies, and most of them are quietly undone by one problem — the healthy-user effect.

People who take a daily supplement, or exercise, or eat their vegetables, tend to also be the kind of people who sleep enough, smoke less, and see a doctor. When they live longer, you can’t tell whether it was the supplement or everything else about them.

Good researchers try to adjust for this. But you can only adjust for things you measured, and human lives are full of variables nobody recorded. That’s why an RCT — where a coin flip decides who gets the treatment — is worth so much more than a correlation, however large the dataset.

So when the next “linked to longevity” headline lands: check whether anyone actually randomised the treatment. Usually no one did.