Data-Informed instruction: pedagogical responses and obstacles in using learning analytics

E-learning resources and educational technology are increasingly used in STEM education, generating vast amounts of student-level data. Learning analytics tools can utilize this data, enabling instructors to adjust their pedagogy to support student success. Despite the potential benefits, the implementation of learning analytics does not always lead to improvements in teaching practices. This paper, through two case studies, investigates challenges instructors may face in adopting learning analytics. In Case Study 1, we examined how online activity data reflects student engagement by analyzing historical data from a learning management system (LMS) alongside observations of class schedules. Online activity was compared to semester timelines and qualitative codes to identify patterns of alignment. The findings suggest that accurate measurement of engagement requires the integration of both LMS data and contextual classroom information. In Case Study 2, we explored how learning analytics influences pedagogical change through surveys and interviews with instructors. Instructors generally found static data related to enrollment and academic standing more useful than dynamic data capturing students’ online behaviors. The difficulty in translating data into actionable pedagogical strategies rendered the learning analytics less effective for long-term course-level improvements. The case studies highlight both the challenges and potential of learning analytics in STEM education. Effective learning analytics intervention requires integrating qualitative insights, fostering collaboration, and providing targeted professional development to support evidence-based teaching and learning strategies.