A New Bayesian Foundation for Psychometric Network Modelling

Network modelling is quickly gaining ground as a promising way to understand psychological phenomena.

The rise of network analysis can be observed throughout the psychological sciences but has been particularly influential in psychopathology. While the network modelling literature has been rapidly expanding, methodological innovations struggle to keep pace. Reviews taking stock of the field invariably zoom in on the methodological challenges that network research faces.

The absence of a confirmatory scheme, the replicability of network results, and the struggle with population heterogeneity rank firmly among the field's top priorities.

These methodological challenges critically impede our understanding of psychological phenomena and the design of effective interventions.

This proposal outlines a new research program for psychological network modelling that addresses current methodological challenges.

Based on the basic principles of Bayesian inference, I develop a new confirmatory network methodology that uses model-averaging to deliver robust, replicable network results.

The new model-averaging approach will be designed for an exhaustive collection of network models and for cross-sectional and longitudinal applications.

I will develop new models that are urgently needed--but missing from the current set of networks--and advance solutions for modelling heterogeneous psychological data to complete the new program. The proposed work puts psychological network modelling on a firm methodological foundation.

To boost the project's impact, the new methods and models are made available in JASP (jasp-stats.org), a user-friendly, free statistical software package that I co-developed.

Armed with an exhaustive set of network models, a confirmatory methodology that delivers replicable results, and their implementation in open-source software, applied researchers can leverage the full potential of psychological network modelling.