While learning causal models, I came across the below assertions:

to express the conditional independence of Y and X given Z: $$ Pr(Y|X, Z) = Pr(Y|Z) $$

we can write:

$$ E(Y|X, Z) = E(Y|Z) $$ If X, Y and Z are random variables with multivariate normal distributions.

My question is shouldn't this hold for all distributions?

$$ E(Y|X, Z) = \int_{z}\int_{x} y \, dPr(y|x, z) = \int_{x} y \, d Pr(y|x) = E(Y|Z) $$

A bit rusty on the math 12 years after graduation, but what did I miss?

Any place where I can read about a proof using the multinormal property to prove the expectation equation?


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.