Don van den Bergh

I’m Don van den Bergh, a postdoctoral researcher at the University of Amsterdam and part of the Bayesian graphical modeling lab. My work focuses on developing new models for psychological data, applying Bayesian inference, and advancing computational statistics. The modeling aspect of my work revolves around designing new models that accurately describe psychological processes. For example, in this paper, we developed a Bayesian hierarchical model to account for differences across patients, staff members, and items, applying it to data from a Dutch maximum-security forensic psychiatric center. In another, we introduced a multilevel Gaussian graphical model for fMRI data.

Bayesian inference underpins nearly all of my work, and I am a strong advocate of Bayes factor hypothesis testing. For example, in this paper, we developed a default Bayes factor to compare variances across multiple groups. As the number of competing models grows, I prefer Bayesian model averaging. In this paper, we compare all possible sets of equality constraints across groups and devise priors to accommodate multiplicity corrections.

I believe that an important part of research is not only developing new methods but also ensuring their application in empirical research. To this end, I develop open-source software, including R and Julia packages, and contribute to the statistical software program JASP.