Confounding
When a third factor explains the link you thought you'd found between two things.
A confounder is something associated with both the exposure and the outcome, which can create an apparent relationship between them, exaggerate a real one, or hide it. The textbook example: people who take supplements have better health outcomes, largely because people who take supplements also tend to exercise, smoke less and see doctors more.
It matters because confounding is the single most common reason a plausible health claim turns out to be wrong. Randomisation deals with it by design; observational studies deal with it by adjustment, which only works for confounders that were measured, measured accurately, and thought of in advance.
In practice, ask what else is true of the people in the exposed group. If the answer is "a great deal", treat the finding as a hypothesis. Residual confounding — what's left after adjustment — is the reason large observational effects sometimes vanish entirely when trialled.
Sources
Related terms
Written and reviewed by Mathew Beale, MSc Biotechnology, University of Reading.
Last reviewed: