Social network analysis: practical and statistical challenges

Abstract

Many thanks are owed to O’Malley and Marsden for taking on the challenging task of compiling this practical tutorial on the statistical analysis of social network data. Advancements in social network modeling have come over the years from a number of sources—including graph theorists, physicists, computer scientists, and sociologists. Only in the last few decades have statisticians really begun making their own contributions to the modeling and inference of network data. It is particularly important that we now communicate these methods and models for network analysis to a health-oriented audience, since many of the most powerful and exciting future developments are likely to occur within this arena. As highlighted in this tutorial, there are some fundamentally different ways that network models can be utilized: as predictors for individual outcome models, as dependent variables in relational models, or as substrates for dynamic processes that spread with personto-person contact (such as infectious diseases). It is important to emphasize the unique practical and statistical challenges that arise in each setting.