Confounding Equivalence in Causal Inference
Abstract
Abstract The paper provides a simple test for deciding, from a given causal diagram, whether two sets of variables have the same bias-reducing potential under adjustment. The test requires that one of the following two conditions holds: either (1) both sets are admissible (i.e. satisfy the back-door criterion) or (2) the Markov boundaries surrounding the treatment variable are identical in both sets. We further extend the test to include treatment-dependent covariates by broadening the back-door criterion and establishing equivalence of adjustment under selection bias conditions. Applications to covariate selection and model testing are discussed.
Cite
Text
Pearl and Paz. "Confounding Equivalence in Causal Inference." Conference on Uncertainty in Artificial Intelligence, 2010. doi:10.1515/jci-2013-0020Markdown
[Pearl and Paz. "Confounding Equivalence in Causal Inference." Conference on Uncertainty in Artificial Intelligence, 2010.](https://mlanthology.org/uai/2010/pearl2010uai-confounding/) doi:10.1515/jci-2013-0020BibTeX
@inproceedings{pearl2010uai-confounding,
title = {{Confounding Equivalence in Causal Inference}},
author = {Pearl, Judea and Paz, Azaria},
booktitle = {Conference on Uncertainty in Artificial Intelligence},
year = {2010},
pages = {433-441},
doi = {10.1515/jci-2013-0020},
url = {https://mlanthology.org/uai/2010/pearl2010uai-confounding/}
}