Factors associated with 21st-century learning outcomes in higher education: A learning analytics model grounded in social cognitive theory

Learning Analytics (LA) has emerged as a powerful approach for assessing student achievement; however, much of the field continues to rely on indirect indicators of outcome attainment, such as course grades, which often lack standardization and privilege knowledge recall over the development of professional competencies. Moreover, many existing models overlook key explanatory variables and lack a solid theoretical foundation. To address these limitations, this study, grounded in Bandura's Social Cognitive Theory (SCT), employs Structural Equation Modeling (SEM) to examine how personal, instructional, and contextual factors influence professional learning outcomes (LOs) in higher education. Using validated and standardized instruments, we directly assessed three professional LOs (communication, ethics, and teamwork) across three independent student samples (n = 253–452). Each LO was modeled separately to capture domain-specific relationships among instructional design, learner characteristics, and contextual variables. Results indicate that the proposed structural model demonstrated an acceptable to good fit across all three LOs, with χ ² /df ratios within recommended thresholds, broadly supporting the SCT-based hypotheses. While the overall pattern of relationships was consistent, the strength of individual predictors varied across LOs, underscoring the domain-specific nature of professional competencies. These findings contribute methodologically and practically to the LA field by demonstrating the value of combining SEM with direct, standardized LO measures. Moving beyond grade-based analytics, this approach offers a more authentic account of student achievement and provides actionable evidence to support continuous improvement, program evaluation, and informed decision-making in higher education.