The Interplay between Inspectable Student Models and Didactics of Statistics

Statistics is a challenging subject for many university students. In addition to dedicated methods of didactics of statistics, adaptive educational technologies can also offer a promising approach to target this challenge. Inspectable student models provide students with information about their mastery of the domain, thus triggering reflection and supporting the planning of subsequent study steps. In this article, we investigate the question of whether insights from didactics of statistics can be combined with inspectable student models and examine if the two can reinforce each other. Five inspectable student models were implemented within five didactically grounded online statistics modules, which were offered to 160 Social Sciences students as part of their first-year university statistics course. The student models were evaluated using several methods. Learning curve analysis and predictive validity analysis examined the quality of the student models from the technical point of view, while a questionnaire and a task analysis provided a didactical perspective. The results suggest that students appreciated the overall design, but the learning curve analysis revealed several weaknesses in the implemented domain structure. The task analysis revealed four underlying problems that help to explain these weaknesses. Addressing these problems improved both the predictive validity of the adjusted student models and the quality of the instructional modules themselves. These results provide insight into how inspectable student models and didactics of statistics can augment each other in the design of rich instructional modules for statistics.