Networks for AB Testing
Data Skeptic
00:00
Cluster-Level Experimentation in A/B Testing
This chapter explores the intricacies of cluster-level experimentation in A/B testing, focusing on the effective division of social networks into treatment and control groups via content creators. It highlights the importance of using engagement data to assign viewers accurately, the transition to more efficient data-processing technologies, and the significance of combating spillover effects during experimentation. Additionally, the chapter discusses ongoing monitoring of clusters and the procedural challenges faced when running large-scale experiments across vast networks.
Transcript
Play full episode
