Reproduce “Federated Causal Inference from Multi-Site Observational Data via Propensity Score Aggregation” 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
Rémi Khellaf, Aurélien Bellet, Julie Josse
The authors propose federated estimators of the average treatment effect by blending site-specific treatment probabilities according to site-membership weights. Sites can fit different propensity models without sharing records. In simulations with no local overlap, federated estimates were reported unbiased when global overlap held, with burden about 6.22; this depends on causal assumptions and does not guarantee estimated-model results.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.