Too Long; Didn't Read

This paper establishes a comprehensive framework for experimental design in network settings, specifically addressing spillover effects. It introduces key assumptions, such as first-order local interference, and restrictions on potential outcome models and exposure mappings. The concept of local asymptotics is explored, capturing scenarios where spillover effects converge to zero at varying rates. The study concentrates on cluster designs, offering a detailed analysis of estimators, including covariate adjustment. The focus is on optimizing experimental designs for global treatment effects while considering spillover effects in network experiments.

L O A D I N G
. . . comments & more!