A systematic error from how participants were chosen, not from the intervention itself.
Selection bias arises when the way individuals are recruited, allocated, or retained in a study makes the sample systematically unrepresentative of the target population, distorting the observed association between exposure and outcome. Common forms include non-random allocation, self-selection into treatment groups, and differential loss to follow-up. It can inflate, mask, or reverse a true effect, so results should be interpreted alongside how the sample was assembled.
This guide was auto-drafted and is pending editorial review.