3
$\begingroup$

Using the 'do calculus' + DAGS framework for causal inference, is this

If $(Y\perp X)_{G_{\underline{x}}}$ then $P(Y|X=x)=P(Y|\text{do}(X)=x)$

an axiom? Or can it be proven from first principles (and without using any notation from an alternative framework, such as potential outcomes)?

$\endgroup$

1 Answer 1

2
$\begingroup$

Intuitively, do-calculus is indeed the first axiomatic formulation for causal inference, see do-calculus rules (Axioms) from Stanford Encyclopedia of Philosophy's causal inference entry.

The equality asked can be regarded as the special case of the axioms. This is so-called the back-door criterion $P(Y|do(X),Z) = P(Y|X, Z) $. But it requires more strict conditions than only independence.

Axioms should provide equivalency to potential outcomes, however they have philosophical differences in how they construct solutions of course. One could argue that, DAGs provides more insights in the mechanisms, i.e., explicit causal model.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.