Jeroen Mulder is a postdoctoral researcher at the Methods and Statistics department of Utrecht University, and is part of the Data Analytics Support Core of Stress in Action (SiA). His postdoc project investigates how machine learning techniques can best be used to aid SiA projects with causal inference from longitudinal, nonexperimental data (both panel data, and intensive longitudinal data). A specific focus is on the estimation of propensity scores for time-varying exposures, which can be used to adjust for time-varying confounding. Furthermore, Jeroen is interested in how these formal causal inference techniques can be combined with structural equation modeling methods.
Abstract
The website of Stress in Action states that “(…) we aim to gain insight into the causes and consequences of daily life stress, and to provide a path towards more stress-resilient citizens.” The investigation of causes, and the development of effect treatments for stress requires robust causal inference methodology. This postdoc project investigates how machine learning techniques can best be used to aid causal inference from longitudinal, nonexperimental data (both panel data, and intensive longitudinal data). A specific focus is on the estimation of propensity scores for time-varying exposures, which can be used to adjust for time-varying confounding.
Areas of Expertise
Publications