Saltar al contenido principal

Follow the Backdoor Path

Choose adjustment moves that block backdoor paths without opening new bias.

causal-inference5 min read

The tempting claim Mentorship participants promote faster than non-participants. If nomination favors already-rising employees, the apparent effect may be confounding. Backdoor adjustment Block non-causal paths, keep causal paths open A good adjustment set blocks common-cause paths without conditioning on mediators or colliders. Adjust for role, not correlation. 01 Treatment 02 Backdoor 03 Transfer Define treatment Step 1 Mentorship can mean invitation, attendance, or completion. Which treatment definition is most consistent? Choose controls

Read the full lesson

Sign up free: one personalized lesson every day, matched to your role and goals.

← Back to library

Contáctanos