Utilizing Technology to Support Collaborative Problem-Solving of K-12 Students in STEM Education: A Multivariate Multi-level Meta-analysis

As classrooms in the 21st -century increasingly incorporate digital tools, educators and researchers have turned their attention to the promising potential of technology in enhancing essential competencies such as collaborative problem-solving (CPS) skills (OECD, 2017). CPS skills, including joint decision-making and coordinated problem resolution, are necessary for success in both academic and real-world contexts. While many studies have investigated individual problem-solving in technology-enhanced learning environments, the collaborative dimension remains underexplored (Graesser et al. 2018b). This dissertation addresses that gap by conducting a multivariate multi-level meta-analysis to examine the overall effects of technology use on K–12 students’ CPS skills in STEM education and to identify the instructional and learner-related variables that moderate these effects. A total of 23 empirical studies met the inclusion criteria, spanning a variety of educational technologies (e.g., simulations, games, collaborative platforms), grade levels, geographic contexts, and study designs. The analysis revealed a statistically significant and positive overall effect of technology use on CPS outcomes, indicating that digital interventions can support the development of collaborative competencies when implemented with intentional pedagogical design. Moderator analyses uncovered key trends. Structured and guided learning environments, including scaffolding, prompts, or scripts, tended to yield larger gains than minimally guided approaches, although differences across treatment types were not statistically significant. Interventions targeting cognitive outcomes showed the most reliable effects, while motivational and social outcomes demonstrated positive but less consistent gains. Studies incorporating teacher feedback yielded statistically significant gains, although the difference from interventions without feedback was insignificant. Learner- and context-related moderators also influenced outcomes: triads (3-member groups) demonstrated the most consistent benefits, and students with intermediate technology proficiency experienced statistically significant gains. Middle school learners showed the most reliable improvements, though all grade bands benefited. The study also applied the (M)UTOS framework to organize moderator variables across methodological, learner, treatment, outcome, and setting dimensions. This theory-driven approach enhanced the interpretability and conceptual clarity of the findings. While the study offers robust insights, limitations include the relatively small number of studies, variability in measurement instruments, and inconsistent reporting across primary sources. This dissertation contributes to the growing research on technology-enhanced collaborative learning by offering a quantitative synthesis and practical recommendations. It highlights the need for intentional instructional design, developmentally appropriate supports, and ongoing evaluation of technology integration. Recommendations for future research include examining emerging technologies such as AI-powered platforms, conducting longitudinal studies, and incorporating qualitative meta-syntheses to better understand students’ lived experiences in CPS contexts. Ultimately, this work aims to inform educators, instructional designers, and policymakers as they seek to foster meaningful collaboration through technology in K–12 STEM education.