Title
Hierarchical Network Models
Speaker
Tracy Sweet, Carnegie Mellon University
Abstract
Current methods of social network analysis are ill-suited to modeling the multiple partially-exchangeable networks that arise in randomized field trials and observational studies in which multiple networks or organizations are involved. To address these needs and drawing from the statistics literature on single network statistical models, I present a new organizational framework, Hierarchical Network Models (HNM). HNMs can be used to extend single-network statistical network models to multiple networks, using a hierarchical modeling approach and allow one to not only model ensembles of networks from observational data but also fit network level interventions. The HNM framework is quite flexible in that any statistical network model can be used; I introduce the Hierarchical Latent Space Model (HLSM) as an example and illustrate its use with both observational and intervention data. I will also briefly discuss other work involving the HNM framework as well as future directions for this research.
