SHARP A03

Personalized Support for Acquiring Task-Based Diagnostic and Intervention Skills in Mathematics and Physics Based on Eye Movements and Log Data

Goals/target groups

Good teachers understand how their students think: They identify where mistakes occur and select tasks that help students continue learning. This ability is called diagnostic competence. It can be learned, but so far it has rarely been specifically and individually fostered in teacher education programs.

The project aims to identify the knowledge and skills necessary for effective diagnosis and to determine how support tailored to the individual can promote the acquisition of this competency. The target group consists of teacher education students majoring in mathematics and physics. In the long term, the findings are intended to improve teacher education at universities.

Contents

In a digital learning environment (simulation), student teachers take on the role of a teacher. They analyze real student solutions on the topic of linear functions, identify typical errors and misconceptions, and select appropriate exercises to help students continue their learning.


While working on the tasks, an eye tracker records where the students are looking. It also logs what they click on and select within the simulation (log data). In three experimental studies, typical diagnostic approaches are derived from this data. Machine learning algorithms are designed to recognize these approaches even as the students are working. This allows the learning environment to provide appropriate guidance at the right moment—for example, regarding subject-specific pedagogical knowledge. The study investigates whether such personalized assistance actually improves the acquisition of competencies.


The subproject is part of the Transregional Collaborative Research Center (TRR) 419 SHARP at LMU Munich and the Technical University of Munich (TUM), funded by the German Research Foundation (DFG). SHARP investigates how simulation-based learning at universities can be personalized through the analysis of learning processes.

Persons and partners involved

Shayla Giap (TUM), Yichen Teresa Wang (LMU), Sonja Doppelfeld (TUM)

Prof. Dr. Andreas Obersteiner (TUM), Dr. Stefan Küchemann(LMU), Prof. Dr. Matthias Stadler (LMU Klinikum)

Prof. Dr. Jochen Kuhn(LMU), Dr. Michael Nickl (TUM)