Spotlight on Statistics with Rutger Carelse

Thursday 23 May 2024, 12:30 pm - 1:15 am
Bayesian Networks for Psychopathology Research

In the last decade the network approach for studying psychopathology has seen a rise in studies that investigate mental disorders as interactions among constituent symptoms. Network analyses are the central statistical analysis within this framework, where variables are depicted as nodes and edges represent connections between them. A limitation of the network approach is the absence of directionality that is present in frequently used partial correlation networks (PCN). Bayesian networks (BN), a form of Directed Acyclic Graphs (DAG), on the other hand are especially suited for inferring directionality (and potential admissible causal relationships) in multivariate data. However, there are very few studies that have implemented BN. In the presentation I will provide a short overview of differences between these three main network analyses (PCNs, BNs, and DAGs) and highlight the use of BNs in cross-sectional data using a tutorial paper from Briganti, Scutari, and McNally (2023). I will provide R code and discuss potential ways for exploratory and confirmatory research questions.

Bio of the Speaker:
Rutger Carelse is a second year Research Master student in the Research Master Behavioural Science at Radboud University in Nijmegen. He is also a research assistant at Novadic-Kentron, a large addiction center in Noord-Brabant, and a teaching- and student-assistant at Radboud University for various research-related projects and courses. His main research interests are substance use, clinical interventions, and statistical methods.

Location: MM 01.620

Thursday 23 May 2024, 12:30 pm - 1:15 am

No need to sign up.