Correcting for Interference in Experiments: A Case Study at Douyin

TL;DR

This paper addresses interference in experiments on content marketplaces like Douyin, proposing a novel Monte-Carlo estimator that significantly reduces bias and variance, improving treatment effect estimation in real-world settings.

Contribution

It introduces a new Monte-Carlo estimator based on DQ techniques for interference, extending policy evaluation theory to all major MDPs, and demonstrates its effectiveness on Douyin.

Findings

Reduces MSE by 99% compared to existing methods

Provides robust, low-bias, low-variance estimates

Enables computationally cheap uncertainty quantification

Abstract

Interference is a ubiquitous problem in experiments conducted on two-sided content marketplaces, such as Douyin (China's analog of TikTok). In many cases, creators are the natural unit of experimentation, but creators interfere with each other through competition for viewers' limited time and attention. "Naive" estimators currently used in practice simply ignore the interference, but in doing so incur bias on the order of the treatment effect. We formalize the problem of inference in such experiments as one of policy evaluation. Off-policy estimators, while unbiased, are impractically high variance. We introduce a novel Monte-Carlo estimator, based on "Differences-in-Qs" (DQ) techniques, which achieves bias that is second-order in the treatment effect, while remaining sample-efficient to estimate. On the theoretical side, our contribution is to develop a generalized theory of Taylor expansions for policy evaluation, which extends DQ theory to all major MDP formulations. On the practical side, we implement our estimator on Douyin's experimentation platform, and in the process develop DQ into a truly "plug-and-play" estimator for interference in real-world settings: one which provides robust, low-bias, low-variance treatment effect estimates; admits computationally cheap, asymptotically exact uncertainty quantification; and reduces MSE by 99\% compared to the best existing alternatives in our applications.