A Tutorial on Bayesian Modeling of Change Across Time, Individuals, and Groups

• Published in: Computational brain & behavior 2023-12, Vol.6 (4), p.697-718
• Main Authors: Knight, Emma, Neal, Andrew, Palada, Hector, Ballard, Timothy
• Format: Article
• Language: English
• Subjects: Behavioral Science and Psychology; Cognitive Psychology; Mathematical Models of Cognitive Processes and Neural Networks; Original Paper; Psychological Methods/Evaluation; Psychology
• ISSN: 2522-0861; 2522-087X
• DOI: 10.1007/s42113-023-00173-6

Psychological theories often incorporate dynamic processes, but it can be difficult to accurately represent these processes with standard statistical tests. As such, there tends to be a misalignment between theory and statistical analysis. We provide a tutorial on a flexible Bayesian approach to developing and analyzing discrete dynamic models that overcomes many challenges associated with conventional methods. This approach can be used to analyze models of virtually any functional form, including models with feedback loops and dynamic (i.e., stock or level) variables. It allows one to quantify uncertainty in components of a dynamic process. This approach also provides a natural way to examine variation in a process between individuals, known groups, or latent subgroups. This framework has the flexibility to capture the dynamism inherent in many theories, which we believe will facilitate theory testing, and ultimately, cumulative theoretical progress.