Comparing Learning Analytics-Supported and Traditional Mathematics Instruction across Higher Education
Abstract
Learning analytics (LA)-supported instruction has shown promise for improving student engagement, yet empirical evidence from infrastructure-constrained higher education systems remains limited. This study addresses that gap by pursuing two objectives: first, to compare mathematics achievement between students taught using LA-supported technology-based instruction and students taught using traditional lecture-based instruction at Samangan University, Afghanistan; and second, to examine the extent to which learning analytics engagement indicators login frequency, video time-on-task, quiz attempts, and forum participation predict students' mathematics achievement. A quasi-experimental, nonequivalent control-group, pretest–posttest design was employed with 96 second-year mathematics education students across two intact sections (experimental n = 48; control n = 48) over a 14-week term. Data were analyzed using independent-samples t-tests, analysis of covariance (ANCOVA), Pearson correlations, and multiple regression. Students in the LA-supported condition achieved significantly higher posttest scores (M = 78.6, SD = 7.4) than those in the traditional condition (M = 65.2, SD = 9.0), t(94) = 7.42, p < .001, d = 1.52, an advantage that remained significant after adjusting for pretest differences, F(1, 93) = 54.87, p < .001, partial η² = .37. Among the engagement indicators, video time-on-task (β = .34, p = .004) and quiz attempts (β = .27, p = .024) emerged as significant independent predictors, with the full set of indicators jointly accounting for 46% of the variance in posttest scores (adjusted R² = .41). Although the nonrandomized design limits causal inference, these findings extend empirical evidence on LA-supported mathematics instruction to a resource-constrained, fragile-state higher education context and indicate that behaviorally substantive engagement is a more informative predictor of achievement than platform access alone.