Khan Buchwald

Talk: A Novel Method to Support the Network Theory of Psychological Disorders with Bayesian Networks and Structural Equation Modelling:
New Horizons in the Land of Psychometrics

The network theory of psychological disorders is a novel theory of psychopathology which posits that symptoms of psychological disorders (or responses to items of a psychometric test) influence, cause, exacerbate, or are associated with the expression of other symptoms. This is contrasted with the latent variable model (LVM) approach which posits that responses to items of an assessment are conditionally independent given one or more latent variables. No known research has compared a network model with latent variables (Combined model) to a nested LVM without dependencies between items, using a likelihood ratio test and model fit statistics. We aim to demonstrate the advantages of the network approach over the LVM by testing their model differences on a well-known data set. We sourced 796 individuals from the North American Prodromal Longitudinal Three Study who completed the Scale of Prodromal Symptoms. We implemented a network model and a latent variable model of the positive, negative, and disorganised items of the SOPS. We found that the combined model had a significantly better goodness of fit than the LVM. This indicates that responses to items of the SOPS should not be considered independent, given the latent variable. Hence, we found empirical support for the network theory of psychological disorders, for the first time, and supported the notion that items should have interdependencies, as prescribed by network theory. We conclude by providing suggestions on how to utilise the network model in clinical research settings to help researchers apply these models in common research designs.

Khan Buchwald is currently enrolled in a PhD in Health at the Auckland University of Technology. In 2022 he completed the Master of Applied Statistics degree at Massey University while working at the Northland District Health Board in New Zealand as a Public Health Statistician. Khan currently runs a statistical consulting company, providing advanced statistical solutions to meet the needs of clientele. His research interests focus on people diagnosed with, or are ultra-high risk of, schizophrenia, and modelling complex systems using network analysis.