Estimating Treatment Effects under Recommender Interference: A Structured Neural Networks Approach – MSI

Working Paper

Estimating Treatment Effects under Recommender Interference: A Structured Neural Networks Approach

Ruohan Zhan, Shichao Han, Yuchen Hu and Zhenling Jiang

Dec 16, 2025

Download

Combines a structural choice framework with neural networks to correct biases that can result from A/B tests of recommender algorithms on content sharing platforms.

Topics

marketing analytics marketing experiments Social Media & Social Influencers

research recap

How Content-Sharing Platforms Can Avoid the “Recommender Interference” Penalty

Content-sharing platforms use A/B tests to improve algorithms matching creators and audiences. Selection bias can result from highly personalized recommendation feeds or when tests boost treated content. This cost-effective alternative combines a structural choice framework with neural networks to account for rich viewer-content heterogeneity, improving results and business decisions.

By using MSI.org you agree to our use of cookies as identifiers and for other features of the site as described in our Privacy Policy.