Reproduce “Qini curve estimation under clustered network interference” 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
Rickard K.A. Karlsson, Bram van den Akker, Felipe Moraes, Hugo M. Proença, Jesse H. Krijthe
Researchers introduced and evaluated methods for estimating Qini curves under clustered network interference, a setting where unit treatments affect outcomes of other units within a cluster. The study found that traditional estimation methods yield biased results when interference is present, while proposed beta-Additive Inverse Probability Weighting (β-IPW) strategies offer a practical balance of reduced variance and good policy ranking.
Key insights
- Ignoring clustered network interference leads to systematically biased Qini curve estimates, with bias increasing as cluster size grows.
- There is an inherent bias-variance trade-off in interference-aware estimators; unbiased methods like Standard IPW suffer from prohibitive variance for larger clusters.
- β-Additive IPW estimators provide a practical solution by balancing bias and variance, offering more stable estimates and strong discriminatory power for ranking policies, especially for larger cluster sizes.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.