Participants of this one-week summer school will be introduced to the field of statistical analysis of network data, with an emphasis of model applications in health research.
After a brief review of traditional compartmental (SIR) models and the methodology for classical descriptive network analysis, (static) Exponential family Random Graph Models (ERGMs) and dynamic temporal ERGMS will be introduced. ERGMs represent the processes that drive the formation of links in networks and are a natural and flexible tool to model (sexual) network data. Participants will learn how to develop stochastic network models for epidemics, with a focus on empirical models of HIV transmission and control.
Stochastic Actor Oriented Models (SAOMs) offer an alternative approach to model the evolution of a network, and the changes in actor attributes. In this modelling framework, individuals change their network ties and adjust their attributes (such as health seeking behaviours) in response to the current network. The approach allows distinguishing processes of selection and influence in network contexts.
During computer lab sessions, participants will go through examples and tutorials with applications of both the ERGM and SAOM approach to modelling network data. The labs will develop programming skills, using the R packages statnet, EpiModel, and RSiena. These software packages provide state-of-the-art tools for statistical network analysis, simulation and visualisation.