Causal Calculus in the Presence of Cycles, Latent Confounders and Selection Bias
This work addresses foundational challenges in causal inference for researchers and practitioners dealing with real-world data that includes feedback loops and biases, representing a significant theoretical extension rather than an incremental improvement.
The authors tackled the problem of causal inference in complex scenarios with cycles, latent confounders, and selection bias by proving rules for causal calculus and generalizing adjustment criteria to a new model class (ioSCMs), enabling estimation of causal effects from observational data.
We prove the main rules of causal calculus (also called do-calculus) for i/o structural causal models (ioSCMs), a generalization of a recently proposed general class of non-/linear structural causal models that allow for cycles, latent confounders and arbitrary probability distributions. We also generalize adjustment criteria and formulas from the acyclic setting to the general one (i.e. ioSCMs). Such criteria then allow to estimate (conditional) causal effects from observational data that was (partially) gathered under selection bias and cycles. This generalizes the backdoor criterion, the selection-backdoor criterion and extensions of these to arbitrary ioSCMs. Together, our results thus enable causal reasoning in the presence of cycles, latent confounders and selection bias. Finally, we extend the ID algorithm for the identification of causal effects to ioSCMs.