How Learning Analytics Can Help Orchestration of Formative Assessment? Data-driven recommendations for Technology-Enhanced Learning
Abstract
Formative assessment provides teachers with feedback to help them adapt their behavior. To manage the increasing number of students in higher education, technology-enhanced formative assessment tools can be used to maintain and hopefully improve teaching and learning quality thanks to the high amount of data that are generated by their usage. Based on literature and on authentic usages of the formative assessment system called Elaastic, we use learning analytics to provide evidence-based knowledge about formative assessment practices. An Elaastic sequence consists in asking learners to answer a choice question with a vote, a confidence degree and a rationale, then to confront their viewpoints through a peer grading activity, and finally to answer the same question a second time. Benefits of such sequences are measured through the increase of correct answers between the two votes. Our results suggest that: (A) Benefits of sequences increase when close to 50% of learners' first votes are correct; (B) Benefits of sequences increase when peers provide better grades to rationales related to correct answers than others; (C) Sequences benefits do not significantly increase when learners who provided correct answers are more confident than learners who did not; (D) Grades attributed by peers depend on such peer's confidence degree; (E) Self-grading is inaccurate in peer grading context; (F) The amount of evaluations each learner performs makes no significant difference in terms of sequences benefits. These results lead to recommendations regarding formative assessment and to a new data-informed formative assessment process which is discussed at the end of the paper.