Prof. Dr. Lilly Shanahan: Risk and Resilience from Adolescence to Young Adulthood
Prof. Dr. Lilly Shanahan leitet die Forschungsgruppe Risiko und Resilienz am Jacobs Center for Productive Youth Development der Universität Zürich (UZH) und ist Professorin für Klinische Entwicklungs-psychologie am Psychologischen Institut der UZH. Bevor sie an die UZH wechselte, hatte sie Professuren an mehreren US-amerikanischen Universitäten, unter anderem an der University of North Carolina in Chapel Hill. Als Co-Direktorin des Zürcher Projekts zur sozialen Entwicklung von der Kindheit bis ins Erwachsenenalter (z-proso) begleitet sie eine der bedeutendsten Längsschnittstudien zur sozialen Entwicklung über die Lebensspanne hinweg. Zudem leitet sie gemeinsam das Bevölkerungsforschungs-zentrum der UZH. Ihre Forschung untersucht, wie biologische, psychologische und soziale Faktoren zusammenspielen und die psychische Gesundheit beeinflussen, mit besonderem Fokus auf Depressionen, Angststörungen, selbstverletzendes Verhalten und Substanzkonsum im frühen Lebensverlauf.
Prof. Dr. Mirka Henninger: Enhancing the interpretability of machine learning models in psychological research
In the last years, machine learning has gained substantial popularity in psychological research. However, many machine learning models were not specifically designed for the data structures commonly encountered in psychology. This presentation explores early and recent developments in machine learning and interpretable machine learning, highlighting their potential benefits as well as possible challenges when applied in psychological research. First, I will present two popular algorithms for decision trees —CART (Classification and Regression Tree) and ctree (conditional inference tree) — and examine the implications of their software implementations for practical applications. I will also discuss the integration of decision trees with parametric models, such as those from psychometrics or for multilevel data. I will show how we can enhance the interpretability of these models, but also identify situations where researchers may risk drawing incorrect conclusions. Finally, I will explore the opportunities and challenges of interpreting predictions from machine learning models in psychological data. I will conclude by discussing how machine learning can benefit from the statistical expertise that we have in the field of psychology.
