Post-treatment bias arises when a regression model conditions on (includes as a covariate) a variable that is itself a consequence of the treatment, rather than a pre-treatment confounder. Because the post-treatment variable lies on the causal pathway from treatment to outcome, conditioning on it blocks part of the treatment's true effect and can open a collider-style spurious path, producing a biased estimate of the treatment's total effect. It is a specific case of included-variable bias (see Statistical Rethinking, 2nd ed., p. 171).