A Meta-Analysis of Intelligent Homework Systems (IHS): Effects on Academic Performance and the Role of Functional Features

Research on Intelligent Homework Systems (IHS) has expanded with the increasing integration of automated assessment, feedback, and adaptive support into homework environments. However, existing studies have reported mixed effects on academic performance, and it remains unclear whether variations in learner, instructional, and functional characteristics contribute to this heterogeneity. This study examined the overall effect of IHS on learners’ academic performance and the moderating roles of learner, instructional, and IHS functional characteristics. Thirty-seven experimental and quasi-experimental studies published between 1995 and 2025 were included, yielding 38 effect sizes from 17,682 learners across 15 countries or regions. Using a random-effects model, we found a statistically significant small-to-moderate positive effect of IHS on academic performance (Hedges’ g = .419, 95% CI [.298, .541], p < .001). Moderator analyses showed that scoring mechanism, granularity of diagnosis, and level of feedback elaboration significantly moderated the effects of IHS, whereas learning domain, cultural background, grade level, degree of adaptivity, and mastery criterion did not. The medium level of granularity of diagnosis and the medium level of feedback elaboration showed the largest pooled effects. The findings suggest that IHS effectiveness depends not merely on the presence of intelligent functions, but on how instructional support is configured.