Learning Analytics for Assessment Design and Pedagogical Decision-Making in Online Computer Science Higher Education

Olga Pishchukhina, Daria Gordieieva, Maria Angela Ferrario, Neil Anderson

This study explores how learning analytics (LA) can support evidence-based assessment design and pedagogical decision-making in online computer science higher education. As online learning environments continue to expand, educators require effective ways to use student engagement data to improve assessment alignment and support learning outcomes. The aim of this study is to examine how analytics derived from formative assessment activities can inform the design and refinement of summative assessments and teaching strategies. The research draws on data collected from a fully online postgraduate computing module delivered over five academic years. LA from the institutional online learning platform was used to analyse student engagement with formative multiple-choice assessments. Item-level analytics, including difficulty indices, discrimination indices, and distractor effectiveness, were applied to evaluate question performance and identify areas, where students experienced learning difficulties. The analysis enabled evidence-based refinements to assessment design and informed pedagogical adjustments within the module. Findings indicate that LA derived from formative assessment engagement can effectively guide improvements to summative assessment tasks and support the development of a more robust and aligned assessment approach. By linking patterns in formative and summative performance with targeted pedagogical interventions – such as revising underperforming questions, aligning tasks with identified learning challenges, and enhancing targeted student support – the study demonstrates how LA can be used to strengthen assessment alignment and improve student outcomes in online computer science education. These findings provide practical evidence on how data-informed approaches can enhance assessment design and curriculum delivery in online learning environments. Overall, the study illustrates how LA can move beyond monitoring student activity to become a strategic tool for shaping assessment design, guiding pedagogical decisions, and advancing data-informed innovation in online computer science higher education.