Learn Before
Concept icon
Concept

Causal Definition of Confounder

Pearl defines a confounder as anything that causes a discrepancy between P(Y∣X)P(Y | X) and P(Y∣do(X))P(Y | do(X)). That is, in the absence of confounders P(Y∣X)=P(Y∣do(X))P(Y | X) = P(Y | do(X)), and if confounders are present then P(Y∣X)≠P(Y∣do(X))P(Y | X) \neq P(Y | do(X))

0

5

Concept icon
Updated 2021-06-27

References


Tags

Data Science