When Do Learning Habits Matter? Linking Behavioural Stages to Academic Performance Using Longitudinal Log Data

ABSTRACT Background Developing effective learning habits—such as studying at regular times—can significantly influence academic performance. While prior research has shown that structured routines benefit learners, most studies rely on self‐reported surveys and provide only a snapshot of behaviour change. Little is known about how learning habits evolve over time or how these shifts affect learning outcomes in technology‐enhanced environments. Addressing this gap is critical for designing personalised, timely support that adapts to learners' behavioural states. Objectives This study investigates how learners progress through different stages of learning habit formation and how these stages relate to academic performance over time. Methods This study analysed 3 years of daily learning log data from 120 Japanese junior high school students. The data were collected through a school‐wide digital learning environment that integrates a learning management system, a digital e‐book system, and a self‐regulated/directed learning support tool. Grounded in the transtheoretical model of behaviour change, this study extracted learners' monthly habit stages from the log data and examined their relationship with exam performance across nine mathematics assessments. Results and Conclusions Academic performance improved significantly as students reached the ‘preparation’ stage, marking a transition from irregular exploratory behaviours to regular weekly study within a single month. However, reaching the ‘maintenance’ stage—the highest stage in the transtheoretical model, representing the culmination of habit‐building through consistent weekly study for at least 4 months—led to performance stabilisation rather than further gains, suggesting a plateau effect once behaviours become automatic. Notably, some learners maintained their academic levels even after regressing to lower habit stages, providing evidence of a residual effect where the cognitive benefits of previously established habits persist despite behavioural fluctuations. This study contributes a novel method for identifying when to offer timely support and suggests how digital learning tools can adapt based on students' habit stages.