Virtual Characters Help K–12 Students Learn and Improve Motivation: A Meta-Analysis

Pedagogical agents, conversational agents, motivational agents, and other virtual characters have long been used in educational technologies. We built and analyzed the most comprehensive dataset to date of studies examining how virtual characters influence K–12 students’ learning and learning-related outcomes using three-level meta-analytic procedures. The results from five three-level meta-analyses indicate that virtual characters helped K–12 students learn (g = 0.42, p < .001, k = 70) and improved their motivation ( g = 0.48, p = .001, k = 47) but did not have any significant effects on emotions ( g = 0.60, p = .20, k = 15), perceptions (g = 0.05, p = .88, k = 34), or cognitive load ( g = −0.09, p = .84, k = 5) compared to systems without a virtual character present. We conclude that virtual characters can provide a meaningful addition to learning environments for K–12 learners.