The Future Potential of Combining Educational Data Mining With Stealth Assessment

Abstract

As a field, educational data mining (EDM) develops methodologies and practices for exploring large-scale datasets generated by modern educational technologies, such as educational games. These methods leverage on the nature of the data generated by these systems, able to explore and examine differences between and generate novel predictions about potential student outcomes that can themselves be fed into new models. This chapter explores the affordances that EDM and stealth assessment can have in terms of fine-grained, non-intrusive analysis and assessment of student behaviors in educational spaces. It surveys existing work within the field of EDM, and profiles an example of a merger of these two fields in the context of identifying when students are at risk of quitting during gameplay of the educational physics simulation game Physics Playground. It ends with a discussion on the limitations of these fields, as well as a suggestion of how future research may continue to synthesize these domains to better identify, assess, and improve student learning outcomes.