How to Apply Network Models to Attitudinal Surveys: Mapping Psychiatric Disorders and Political Belief Systems (in R)
Speaker(s):
Bio: Todd Hartman is Professor of Quantitative Social Science in the Department of Social Statistics at the University of Manchester. His research explores the psychological underpinnings of public opinion and behaviour using cutting-edge research methods and statistical techniques. His work has been published in prestigious peer-reviewed academic journals such as Nature Communications, Nature: Scientific Reports, Psychological Medicine, Big Data & Society, British Journal of Political Science, The Journal of Quantitative Criminology, Social Psychological and Personality Science, Political Psychology, Political Communication, and The Geographical Journal. Professor Hartman has been working with an interdisciplinary team to study the impact of COVID-19 on the public. This project secured early funding from the UK's Economic and Social Research Council (ESRC) and has collected nationally representative panel data in multiple countries (e.g., UK, Ireland, Spain, and Italy) from multiple survey waves of respondents beginning when the first UK Lockdown was announced (on 23 March 2020). This unique collaboration is only one of two social science research teams to receive ESRC funding to collect new longitudinal survey data since the start of the pandemic to study the implications of COVID-19 on adults living in the UK (e.g., see this funding announcement). While this project has been immensely challenging, given the speed with which things have changed locally, nationally, and internationally, it has also been a once-in-a-lifetime opportunity (hopefully!) to study a global health crisis which has wrought about such societal upheaval.
Abstract:
Do you want to learn how to analyze the relationships among survey items in your data and use sophisticated, algorithms to visualize their structure? This hands-on session will introduce students to 'psychometric' network analysis, which assumes that survey questions typically used to measure things like personality traits, psychiatric disorders, and political beliefs are not necessarily caused by a common underlying factor (i.e., latent variable models). Instead, network models are agnostic and use a data-driven approach in which each item may causally influence others in the system. These new analytical techniques thus allow links among survey items to be mapped in a visual network, in which the 'nodes' correspond to elements within the network and 'edges' to the magnitude and direction of their connections similar to what you might find in a social network. When networks are estimated for different subpopulations, network comparison statistics are available to test each network's overall connectivity, as well as the strength of individual edges for different groups.