Jonas Haslbeck, University of Amsterdam

2021 Dissertation Prize

Date & Time: Friday, July 23 at 9:00am EST

In this talk, I will retrace the path of my PhD dissertation. I started out by developing methodology to estimate a variety of statistical network models for cross-sectional and time series data including Mixed Graphical Models, mixed Vector Autoregressive models, Moderated Network Models, and time-varying variants of these models. These and other network models have become extremely popular and are being used in applied research to better understand the systems underlying psychological phenomena. In the second part of my dissertation, I investigated how this should actually work. I show that statistical models are largely ill-suited as theories for psychological phenomena, and that making inferences from statistical models to unspecified systems is highly problematic. As an alternative, I put forward an abductive approach that uses different data types and statistical models to construct formal theories of psychological phenomena. I conclude by discussing open questions that need to be addressed to create a more formalized psychological science.

About the Speaker

Jonas Haslbeck is a postdoc in Denny Borsboom’s lab at the Psychological Methods Group at the University of Amsterdam. He recently completed his PhD with distinction, during which he created various extensions of statistical network models and methodology to construct formal theories in clinical psychology. The statistical methods Jonas developed during his PhD, which are implemented in the R-package mgm and allow the estimation of network models that are time-varying, contain different types of variables and higher-order interactions, have already been widely adopted in the field of network psychometrics. Next to his main research lines on statistics and psychological theory he is also engaged in interdisciplinary research on topics such as decision making, movement tracking, opinion dynamics and international affairs. In his postdoc, Jonas is working on further extending the toolbox of network models and creating tools to make theory construction more approachable for psychological researchers.