Dr. Johannes Bleher assists researchers from TU Dortmund University by conducting machine learning-based analysis of German PIRLS data, specifically using post-selection regressions on principal components derived from multiple imputed and bootstrapped datasets.
The study examines how vital social resources, more precisely social interactions with peers and teachers, are for two central aspects of school success, namely academic achievement and well-being. To this end, the representative German PIRLS 2016 data of 3959 fourth-grade students (MAge = 10.34 years; N = 1,940 girls, 71% white) were analyzed. Social interactions were operationalized using factors indicating whether students experienced bullying from peers, and how much teacher support they perceived. We found that fewer bullying experiences and more perceived teacher support were positively related to academic achievement and enjoyment of school as a prominent aspect of school-related well-being. Applying machine-learning methods to avoid overfitting while including important control variables, only the effects of bullying experiences and perceived teacher support on well-being remained robust. The results underlined that positive relationship experiences were particularly important for students’ well-being but not necessarily incremental to students’ academic achievement.