Metacognitive Monitoring Accuracy and Self-Regulated Learning Outcomes in Undergraduate STEM Cohorts: An Empirical Analysis of Calibration Bias Under Distributed Practice Schedules

Abstract

Metacognitive calibration — the alignment between learners' confidence judgments and actual performance — remains insufficiently characterized within distributed practice contexts in tertiary STEM education. This study examined 318 undergraduate students across three research-intensive universities, employing pre/post judgment-of-learning (JOL) protocols, spaced retrieval tasks, and hierarchical linear modeling to isolate the effects of calibration bias on summative assessment outcomes. Findings indicate that over-confident learners (positive calibration bias ≥ 0.25) demonstrated a statistically significant 14.3% decrement in end-of-semester examination scores compared to well-calibrated peers (p < .001), independent of prior academic achievement. Instructional scaffolding targeting metacognitive awareness attenuated this bias by 38.6% across a 12-week intervention. These results underscore the urgency of embedding systematic calibration training within undergraduate STEM curricula to mitigate performance gaps attributable to inaccurate self-assessment.