Inference for Social Network Models from Egocentrically-Sampled Data

Friday August 17, 2pm, Carslaw 173

Pavel Krivitsky
University of Wollongong, School of Mathematics and Applied Statistics

Inference for Social Network Models from Egocentrically-Sampled Data

Egocentric network sampling observes the network of interest from the point of view of a set of sampled actors, who provide information about themselves and anonymised information on their network neighbours. In survey research, this is often the most practical, and sometimes the only, way to observe certain classes of networks, with the sexual networks that underlie HIV transmission being the archetypal case. Although methods exist for recovering some descriptive network features, there is no rigorous and practical statistical foundation for estimation and inference for network models from such data. We identify a subclass of exponential-family random graph models (ERGMs) amenable to being estimated from egocentrically sampled network data, and apply pseudo-maximum-likelihood estimation to do so and to rigorously quantify the uncertainty of the estimates. For ERGMs parametrised to be invariant to network size, we describe a computationally tractable approach to this problem. We use this methodology to help understand persistent racial disparities in HIV prevalence in the US. Lastly, we discuss how questionnaire design affects what questions can and cannot be answered with this analysis. This work is joint with Prof Martina Morris (University of Washington).

Dr Pavel N. Krivitsky received his PhD in Statistics in 2009 from University of Washington, and has been a Lecturer in Statistics at the University of Wollongong since 2013. His research interests include statistical modelling of social network data and processes for applications in epidemiology and the social sciences, statistical computing, and data privacy. He has contributed to theory and practice of latent variable and of exponential-family random graph models for networks, particularly models for network evolution and for valued relations, understanding effects of changing network size and composition, and estimation of complex network models from difficult (perturbed, egocentrically-sampled, etc.) data. He develops and maintains a number of popular R packages for social network analysis.