Statistical/ psychometrical techniques for digital learning applications

Digital learning environments have become popular tools for practicing skills in students. As an important advantage, these learning environments can personalize the learning experience. For instance, Computerized Adaptive Practice (CAP) environments continuously adjust the difficulty level of the exercises to the current skill level of the student, thereby fostering the individual learning process. This adaptivity, however, requires skills to be reliably assessed on-the-fly, while considering that learning happens moment by moment. Rating systems – with the Elo Rating System (ERS) as a famous example – have proven to be powerful tools for tracking student skills in CAP environments. Recent research, however, demonstrated that current rating systems also have significant shortcomings. This project will propose several psychometric extensions to the state-of-the-art rating systems that mitigate these issues.