Applications of learning analytics in the study of academic performance in higher education: A meta-review with an educational perspective

Abstract

This meta-review synthesizes evidence from systematic reviews (SRs) examining the application of Learning Analytics (LA) to the study of academic performance in higher education from an educational perspective. Despite growing interest in this area, existing reviews vary considerably in scope and exhibit notable substantive and methodological shortcomings. A comprehensive search strategy, up until May 2025, was conducted across four major databases (Scopus, Web of Science, ERIC, and PsycINFO), and gray literature. A narrative synthesis was employed to analyze findings from 19 SRs that met the inclusion criteria. The reviewed SRs paid limited attention to key educational aspects, such as instructional contexts, pedagogical aims, disciplinary domains, underlying educational theories, and student learning stages. They were also characterized by critically low methodological quality. Considerable variability was found in how academic performance was conceptualized and operationalized, ranging from broad constructs to clearly defined indicators. Demographic and socioeconomic variables, followed by academic background and performance indicators, were the most frequently examined factors, although their predictive power was rarely reported. Supervised quantitative approaches, particularly regression and classification-based techniques. However, the predominantly narrative and heterogeneous reporting of non-comparable performance metrics precluded robust conclusions. The educational implications of these patterns are discussed, emphasizing the need for future research to incorporate contextually relevant educational dimensions, multi-source data, and multivariate and multilevel analyses. Overall, the results underscore the importance of aligning LA research with pedagogical objectives and enhancing its practical relevance in higher education.