Classical designs of randomized experiments, going back to Fisher and Neyman in the 1930s still dominate practice even in online experimentation. However, such designs are of limited value for answering standard questions in settings, common in marketplaces, where multiple populations of agents interact strategically, leading to complex patterns of spillover effects. In this paper, we discuss new experimental designs and corresponding estimands to account for and capture these complex spillovers. We derive the finite-sample properties of tractable estimators for main effects, direct effects, and spillovers, and present associated central limit theorems.

翻译：自20世纪30年代Fisher和Neyman提出的经典随机实验设计至今仍在实践中占主导地位，甚至在线实验领域亦不例外。然而，在诸如市场平台等多智能体群体以策略性方式交互的常见场景中，此类设计对于回答标准问题的价值有限，因为交互会导致复杂的溢出效应模式。本文探讨了新的实验设计及相应的估计量，以解释并捕捉这些复杂的溢出效应。我们推导了主要效应、直接效应和溢出效应可处理估计量的有限样本性质，并给出了相应的中心极限定理。