Predicting students engagement in asynchronous online learning: a mixed-method approach

Educational Technology Research and Development. 2026. Vol. 74. P. 769–791.

Капуза А. В., Гетман А. В., Котликова А. Б., Герасимова Ю. О., Уртенова П. Р., Kharakhordina M.

Predicting the level of student learning engagement in online learning is crucial for student success, especially for asynchronous courses. While digital traces can track students’ activity on the platform and help to measure the engagement level, they could provide contradictory results, so it is crucial to incorporate complementary methods which can triangulate the findings obtained from digital traces. This study aimed to develop and validate a model to determine the level of learning engagement in adult learners on an asynchronous online platform using a mixed-method approach. Data from digital traces, surveys, and interviews were combined. The study involved 2234 students and employed Extreme Gradient Boosting and Logistic Regression with L2 regularisation models to predict the level of engagement. The Extreme Gradient Boosting model more accurately predicted students in the low engagement group, providing crucial support for potentially vulnerable students. The number of finished homework assignments and attempts were found to increase the probability of high engagement. The diversity of activities, such as access to text materials, played a pivotal role in sustaining engagement. Interviews corroborated these results, suggesting the model effectively reflects engagement levels. The article discusses implications for constructing similar models in future research.

Ключевые слова: mixed methods, online learning, digital traces, Learning engagement, predictive modelling

ПУБЛИКАЦИЯ ПОДГОТОВЛЕНА ПО РЕЗУЛЬТАТАМ ПРОЕКТА: Использование цифровых следов для изучения поведенческих стратегий обучающихся (2025)