
Do-calculus is a set of mathematical rules devised by Judea Pearl in 1995 to determine whether causal effects can be identified from observational data under specific assumptions encoded in a causal graph. It provides a systematic method for transforming expressions involving the do-operator (representing interventions) into expressions involving only observable probabilities, enabling the identification of causal relationships.
Causal queries involving interventions (e.g.,
P(y\middo(x))
The rules[2] apply to a causal graph
l{G}
P(y\middo(x),z,w)=P(y\middo(x),w) ifY\perp\perpZ\midX,Winl{G}\overline{X
This rule allows the removal of irrelevant observations (
Z
Y
X
W
X
P(y\middo(x),do(z),w)=P(y\middo(x),z,w) ifY\perp\perpZ\midX,Winl{G}\overline{X\underline{Z}}
This rule permits replacing an intervention (
do(z)
z
Y
Z
Z
P(y\middo(x),do(z),w)=P(y\middo(x),w) ifY\perp\perpZ\midX,Winl{G}\overline{X\overline{Z(W)}}
This rule removes irrelevant interventions (
do(z)
Y
Z
Z
Do-calculus can be applied to various domains within causal inference such as mediation analysis in decomposing direct and indirect effects.[3] [4] It can be used for meta-synthesis to combine the results from heterogeneous studies.[5]
The do-calculus is considered complete: if repeated application of the rules cannot eliminate the do-operator, the causal effect is not identifiable. This result was formalized in 2006 by Huang, Valtorta, Shpitser, and Pearl.[6]
Critics have pointed out that other frameworks, such as structural equation modeling (SEM) or Bayesian networks, may offer more intuitive approaches to causal inference for certain applications. These methods often emphasize parameter estimation rather than identifiability, which can be more relevant for applied research.[7]