Promoting social and emotional learning in K-12 students through digital-based interventions: a systematic review and meta-analysis

Digital-based interventions that promote students’ social and emotional learning (SEL) development have expanded, but their effectiveness and impact remain debated. This study aimed to estimate effects of digital-based interventions on three SEL outcomes: social emotional skills (SES), affect and attitudes (AA), and prosocial and antisocial behaviors (PB), and to examine moderators, including social impairment status, school level, device type, intervention duration, study design, intervention gamification design features, control condition, reporter type, and setting of the intervention. Following PRISMA guidelines, we searched Web of Science, ASSIA, PsycINFO, MEDLINE, Embase, and ERIC through August 1, 2025, and IEEE Xplore and ACM Digital Library through January 5, 2026; included experimental and quasi-experimental studies; used random-effects models (Hedges’ g); assessed heterogeneity using Q and I²; and evaluated publication bias using Egger’s test and the trim-and-fill method. Thirty-seven studies (N = 4742) were included. In the primary meta-analysis, digital-based interventions showed positive effects across all three outcomes: SES (SMD = 0.631, 95% CI: 0.461 to 0.800), AA (SMD = 0.267, 95% CI: 0.087 to 0.448), and PB (SMD = 0.285, 95% CI: 0.106 to 0.465). After trim-and-fill, SES (trim-and-fill SMD = 0.369) and PB (trim-and-fill SMD = 0.353) remained significant, whereas AA became non-significant (trim-and-fill SMD = 0.078). In moderator analyses for SES, social impairment status and reporter type were significant. Samples with social impairments benefited more. Effects were largest in clinician reports, followed by parent and self reports (similar magnitude), and smallest in teacher reports. For AA, setting of the intervention was significant (home > controlled research > mixed school-and-home/school). For PB, device type was significant (computer-based > tablet/phone). Other moderators were not significant. Digital-based interventions significantly enhance K–12 students’ SES and modestly improve PB. For AA, the initially significant effect became non-significant after correcting for publication bias (trim-and-fill), indicating limited robustness and warranting cautious interpretation. These effects are moderated by social impairment status, reporter type, device type, and setting of the intervention. Future research should use larger samples and long-term follow-up to study digital-based interventions. PROSPERO (CRD420251069260).