Test to tool: Creating Value from Educational Assessment Data for Learners and Educators

Data are everywhere. In the field of educational assessment, data from tests are used to estimate the ability of students and characteristics of test questions, such as their difficulty. Tests are also increasingly often administered digitally, allowing more and different kinds of data to be captured. Besides capturing student answers, we can now also capture indicators of a student’s answering process, such as the time the student spent answering an item, or how often a student rewrote a response before submitting their answer. This increased availability of richer data, combined with recent improvements in computational processing power and scientific modeling breakthroughs, offers new opportunities for supporting learners and educators (e.g., more insights into learning processes, or higher automation of tasks), but new challenges and dangers simultaneously arise (e.g., model biases, ethical issues around learners’ privacy). This dissertation explores how added value from educational assessment data for learners and educators can be created responsibly.

The four studies in this dissertation address the main research question by exploring different methods through which different types of educational assessment data can generate added value. Collectively, the four studies in this dissertation show that educational assessment data create value when translated – through transparent, user-centered educational technology – into personalized insights, pedagogical feedback, and teacher support that strengthen both learning and teaching. However, such gains do not follow automatically when educational technology is applied. In order to achieve these gains, it is necessary to ground educational technology in theory and take an interdisciplinary and integrated approach, evaluating results from research and development collaboratively and thoroughly.