|DATE:||November 17, 2016|
|TIME:||1:30 - 2:50 pm|
|LOCATION:||Medical School Office Building, Rm x303|
|TITLE:||Graph clustering for network experiments
Assistant Professor, Management Science & Engineering, Stanford
A/B testing is a standard approach for evaluating the effect of online experiments; the goal is to estimate the Average Treatment Effect (ATE) of a new feature or condition by exposing a sample of the overall population to it. A drawback with A/B testing is that it is poorly suited for experiments involving social interference, when the treatment of individuals spills over to neighboring individuals along an underlying social network. This talk will describe an experimental framework based on graph cluster randomization for identifying average treatment effects under social interference. Given this framework, we analyze the variance of the average treatment effect as a property of the graph cluster design, and bias/variance trade-offs under exposure model misspecifications. This talk will discuss joint work with Lars Backstrom, Dean Eckles, Brian Karrer, Jon Kleinberg, and Joel Nishimura.
Suggeted reading:J. Ugander, B. Karrer, L. Backstrom, J. Kleinberg. (2013) "Graph cluster randomization: network exposure to multiple universes" Proceedings of KDD.
D. Eckles, B. Karrer, J. Ugander (2014) "Design and analysis of experiments in networks: Reducing bias from interference", arXiv.
D. Walker, L. Muchnik (2014) "Design of Randomized Experiments in Networks", Proceedings of IEEE.