Causal Calculus in the Presence of Cycles, Latent Confounders and Selection Bias
arXiv:1901.00433
Abstract
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.
Accepted for publication in Conference on Uncertainty in Artificial Intelligence 2019 (UAI-2019)
References in corpus (9)
- Pearl's Calculus of Intervention Is Complete
- Joint Causal Inference from Multiple Contexts
- On the Validity of Covariate Adjustment for Estimating Causal Effects
- Bayesian structure learning using dynamic programming and MCMC
- Nested Markov Properties for Acyclic Directed Mixed Graphs
- Testing Identifiability of Causal Effects
- A Complete Generalized Adjustment Criterion
- From Ordinary Differential Equations to Structural Causal Models: the deterministic case
- Markov Properties for Graphical Models with Cycles and Latent Variables