Dynamic Inferential Network Analysis for Public Health

Social Science Statistics/Public Health 

Keywords - network, network science, network analysis, network dynamics, longitudinal analysis, data science, statistics, social sciences, complexity, complex system, inference, health behaviour, public health, health outcome, substance use, research methods, statistical model, peer pressure, peer influence, social influence, social epidemiology, alcohol use, drug use, mental health

Project Summary - The PhD candidate will work on the development and implementation of statistical techniques for inference on longitudinal networks and apply these techniques to datasets from the field of public health. Many public health outcomes, such as mental health or substance abuse, are network phenomena because they do not develop in isolation. It is vital for public health to understand not just the prevalence, but more importantly, also the causes of these "networked" health outcomes. Research has suggested that people's health risk behaviour and outcomes (e.g., smoking and obesity) may spread through networks ("contagion"); or may provoke network or friendship ties among people with similar behaviour ("homophily"). In recent years, research in network science, public health, the social sciences, and statistics has therefore developed inferential network analysis techniques for modelling the structure and dynamics of friendship or contact networks (e.g., the Temporal Exponential Random Graph Model; TERGM) and the diffusion of behaviour through networks over time (e.g., the Temporal Network Autocorrelation Model; TNAM). The innovative aspect of this project is the extension of the TERGM and the TNAM model to networks with multiple groups, multiple relations, and/or in a co-evolutionary joint process. This is of particular importance in public health, where behaviour like smoking and mental health can otherwise not be explained properly.

Project Team - The PhD student will have the opportunity to make use of a variety of datasets in public health and work in an exciting interdisciplinary environment. The candidate will work with supervisors from social science research methods, public health, and statistics, whose joint interest is the study of network dynamics. Dr Philip Leifeld is a Senior Lecturer in Research Methods in the School of Social and Political Sciences. His research focuses on network dynamics and applications in the social and political sciences. Dr Mark McCann is a Research Fellow in the Social and Public Health Sciences Unit and is an expert in substance use, mental health, adolescent health behaviour, and complex systems. Dr Nema Dean is a Lecturer in the School of Mathematics and Statistics. Her research focuses on model-based inference for dependent and high-dimensional data. The PhD project will be hosted by the Graduate School of Social Sciences, and a stronger involvement in any of the three disciplinary homes and institutions of the supervisors is possible, depending on the interests of the candidate. The PhD project will also offer the option of a research stay abroad in the wider network of the three supervisors.

PhD candidate - Sebastián Martínez