Undergraduate Mathematical Sciences Seminar
Wednesday, 29 November 2006, 12:30--1:50pm
An Overview of Treating Social Networks with Sampled Data
Krista Gile
Abstract: Social network analysis is used to characterize the paterns of relations between sampled units. Often, this analysis must be carried out on a sample where not all relations are observed. This talk
introduces social networks and mecahnisms of sampling from social networks. It then describes design-based and model-based inference for partially observed social networks, including the Horvitz-Thompson estimator and the concept of ignorability. It concludes with an example of a complex model-based analysis using an exponential random graph model for partially observed data.