Skip to content
Causal DAG templates

Biology — M-bias

A collider can bias even pre-exposure adjustment.

Template previewCausal DAG
M-biasXExposureYOutcomeACause of XBCause of YMCollider (measured)ExposureOutcomeConfounderMediatorUnobserved`Outcome` is not a descendant of `Exposure`, so P(Outcome | do(Exposure)) = P(Outcome): this diagram says the effect is NULL, and identified w…No arrow path leads from `Exposure` to `Outcome`. This diagram states that the exposure has NO effect on the outcome — which may be exactly the null hypothesis be…

Make it your own.

title "M-bias"
exposure X "Exposure"
outcome Y "Outcome"
node A "Cause of X"
node B "Cause of Y"
node M "Collider (measured)"
A -> X
A -> M
B -> M
B -> Y