Reproduce “On Multi-Cause Causal Inference with Unobserved Confounding: Counterexamples, Impossibility, and Alternatives” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Alexander D'Amour

The paper examines whether observational data with multiple causes and an unobserved confounder identify causal effects. It considers a latent factor that renders causes conditionally independent and blocks confounding paths, then shows that, without parametric restrictions on the outcome model, factorization does not identify intervention distributions. Either intervention and observational distributions coincide almost everywhere, or the intervention distribution is unidentified.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.