What individual factors matter for secondary students’ ICT self-efficacy?

ICT self-efficacy reflects students’ confidence in using digital technologies. Prior studies often employed traditional statistical methods to examine a limited set of individual factors influencing ICT self-efficacy and were commonly based on small, single-country samples. These design and analytical constraints have limited both the identification of the most influential predictors and the generalizability of findings across educational systems. To address these gaps, this study analyzes data from 311,522 secondary students across 52 countries who participated at Programme for International Student Assessment (PISA) 2022 assessment. As the Extensive Digital Competence (EDC) model suggests, over 90% of the variance in ICT self-efficacy is attributable to individual-level factors and less than 10% explained by classroom- and school-level variables; this study classified 37 individual-level variables into four EDC domains: student demographic attributes, ICT-related characteristics, ICT usage situations, and general educational factors. We applied three machine learning models — random forest regression (RFR), support vector machine (SVM), and XGBoost — to predict ICT self-efficacy, coupled with Shapley Additive Explanations (SHAP). Results show that RFR achieved the best performance and identified the top 15 predictors. These findings are consistent with individual-level assumptions of the EDC model and echo Bandura’s social cognitive theory, highlighting that the ICT-related characteristics domain contributes most significantly to student ICT self-efficacy. This study also provides practical insights for educators and policymakers aiming to enhance students’ digital confidence.

Digital divide · Information and Communications Technology · Digital literacy in education · Education and Technology Integration · Gender and Technology in Education