When Clicks Reveal Student Frustration: A Sensor-Free Analytics Approach with Adaptive Thresholds

Abstract

The process of learning is deeply intertwined with students’ emotional states, with frustration recognized as a particularly disruptive affective condition that undermines motivation, engagement, and achievement. This study investigates frustration detection in an online postgraduate course delivered through a Moodle-based learning environment. Building on prior work in affect-aware tutoring, we propose a sensor-free approach that relies on anonymized logfiles that capture student interactions. An adaptive thresholding method was applied to identify anomalous bursts of activity, such as repeated clicking, that signal frustration. Results revealed 4,763 frustration incidents, with navigation and forum activities generating the majority of signals, while submissions and quizzes were secondary contributors. Peaks in frustration closely aligned with assignment deadlines, quizzes, and examinations, though sustained navigation and forum engagement also proved highly indicative of stress. The findings highlight the importance of contextual and category-specific analysis in capturing frustration dynamics, and they underscore the need for inclusive support strategies that extend beyond high-stakes assessment moments.