Marjolein Fokkema is an expert on machine learning and latent variable modeling. Since 2015, she is an assistant professor at Leiden University. She completed her PhD thesis at the Vrije Universiteit Amsterdam, on how we can improve measurement and prediction in psychology. For example, how can we assess whether pre- and post-assessments in clinical trials measure the same construct in the same way? Or how can we best predict which individuals have a high or low risk of developing a psychological disorder?
In addition to researching and teaching at Leiden University, Fokkema is an associate editor at The European Journal of Psychological Assessment and a member of the Dutch Comittee on Testing Matters (COTAN) of the Dutch Intitute for Psychologists (NIP).
Fokkema's research centers around developing statistical and machine learning algorithms that provide transparent and easy-to-understand results, which can be easily applied in clinical decision making. Many statistical methods require a lot of information and calculation, which makes it difficult to apply these methods for doctors or psychologists in clinical practice. In her research, Fokkema develops so-called decision-tree methods which, like their name already suggests, yield decision trees instead of mathematical formulas. Often these methods provide equally or more accurate results than other statistical methods, but are easier to understand and require less information for making a decision in clinical practice.
Fokkema teaches courses in the bachelor's programmes Psychology (e.g., Psychometrics) and in the (research) master's programme Psychology (e.g., Latent Variable Modeling, Statistical Learning and Prediction).
- 2017: Swiss National Science Foundation (SNF) - International Visit Grant (main applicant)
- 2017: ZonMW - Onderzoeksprogramma GGZ Middellange termijn (co-applicant)
- 2014: Psychometric Society - Student Travel Award (main application)