Principle: Adjustment is a causal decision before it is a statistical decision. Confounder A pre-treatment common cause of treatment and outcome. Often a good control. Mediator A downstream step through which the treatment works. Controlling for it changes the estimand. Collider A variable caused by two other variables. Conditioning on it can open a false path. Why it works DAGs force assumptions into the open. The point is not perfect certainty; the point is to stop hiding causal assumptions inside a regression formula.
Read the full lesson
Sign up free: one personalized lesson every day, matched to your role and goals.
Already have an account? Sign in