Evidence-appraisal glossary
Exchangeability
The assumption that treated and untreated groups are comparable, so that the untreated group's outcome fairly represents what would have happened to the treated group had it gone untreated. Randomization creates it by design; observational studies must assume it after adjustment.
Also called: no unmeasured confounding, ignorability.
Exchangeability means treatment assignment is unrelated to the potential outcomes, at least once you account for measured confounders, which is called conditional exchangeability. It is the formal statement of no unmeasured confounding. You cannot verify it directly from the data, which is why unmeasured confounding is the central worry in observational research and why sensitivity analyses and negative controls are used to gauge how badly a violation could distort the finding.
This is a plain-language methodology definition for reading research. It is general education, not medical advice.