Research map / Probabilistic and scientific modeling / Treatment effects
A/B Testing in Dense Large-Scale Networks: Design and Inference
Preetam Nandy, Kinjal Basu, Shaunak Chatterjee, Ye Tu
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use OASIS instead of cluster-based randomization | Treatment-effect estimation under network interference · Simulation, average degree 120, δ = 1/4 | 95% CI empirical coverage: 0.974 -> 0.956 | Improved |
| Use OASIS instead of cluster-based randomization | Treatment-effect estimation under network interference · Simulation, average degree 120, δ = 1/2 | 95% CI empirical coverage: 0.592 -> 0.938 | Improved |
| Use OASIS instead of cluster-based randomization | Treatment-effect estimation under network interference · Simulation, average degree 120, δ = 1 | 95% CI empirical coverage: 0.880 -> 0.932 | Improved |
| Use OASIS instead of cluster-based randomization | Treatment-effect estimation under network interference · Simulation, average degree 21, δ = 1/4 | 95% CI empirical coverage: 0.974 -> 0.958 | Hurt |
| Use OASIS instead of cluster-based randomization | Treatment-effect estimation under network interference · Simulation, average degree 21, δ = 1/2 | 95% CI empirical coverage: 0.976 -> 0.968 | Hurt |
| Use OASIS instead of cluster-based randomization | Treatment-effect estimation under network interference · Simulation, average degree 21, δ = 1 | 95% CI empirical coverage: 0.946 -> 0.928 | Hurt |
| Use OASIS instead of cluster-based randomization | Treatment-effect estimation under network interference · Simulation, average degree 55 | 95% CI empirical coverage: 0.948 -> 0.950; 0.946 -> 0.950; 0.938 -> 0.954 | No clear effect |
| Apply OASIS to the recommender-system test | LinkedIn newsfeed, M^(1) versus M^(0) · LinkedIn newsfeed experiment | Daily Unique Contributors lift: +0.52% | Improved |
| Apply OASIS to the recommender-system test | LinkedIn newsfeed, M^(1) versus M^(0) · LinkedIn newsfeed experiment | Members in Active Community lift: +1.14% | Improved |
| Apply OASIS to the recommender-system test | LinkedIn newsfeed, M^(1) versus M^(0) · LinkedIn newsfeed experiment | Feed Interaction received Uniques lift: +1.34% | Improved |
| Compare the shadow population rather than the population matched on producer-side exposure | OASIS real-world experiment · LinkedIn newsfeed experiment | Producer-side outcome significance | No clear effect |
Closest papers by meaning
- Just Ramp-Up: Debiasing Regression-based Estimator for A/B Tests under Network Interference (NeurIPS 2026 · similarity 0.89)
- Independent-Set Design of Experiments for Estimating Treatment and Spillover Effects under Network Interference (ICLR 2024 · similarity 0.88)
- Online Experimental Design With Estimation-Regret Trade-off Under Network Interference (NeurIPS 2025 · similarity 0.87)
- Higher-Order Causal Message Passing for Experimentation with Complex Interference (NeurIPS 2024 · similarity 0.87)
- Telling Peer Direct Effects from Indirect Effects in Observational Network Data (ICML 2025 · similarity 0.84)
- Doubly Robust Causal Effect Estimation under Networked Interference via Targeted Learning (ICML 2024 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
