Abstract While frameworks like Illustration-Interaction-Intervention-Induction often guide data-driven learning (DDL), few studies examine whether they reflect actual learner behaviour or a prescriptive ideal. This paper analysed field recordings and worksheets from 61 participants using grounded theory, identifying ‘learning events’ (i.e., what learners do) and ‘learning sequences’ (i.e., when and why they do so), grounded in constructivism, sociocultural theory, and the Noticing Hypothesis. These macro- and micro-level insights reveal DDL as a non-linear, iterative process shaped by data observation, repeated rule extraction and testing. The findings inform DDL studies by highlighting the importance of considering non-linearity in task design and providing regular scaffolding. Beyond DDL, the paper contributes to a broader understanding of collaborative, inductive learning and offers implications for task design across educational contexts.
Jarvis Looi (Fri,) studied this question.