Playing a digital maths game versus a human leads to better outcomes than playing versus a computer player but only for boys and non-anxious students

This study investigates how two competitive game modes affect performance in an educational mathematics game and how these effects interact with gender and mathematics anxiety. A total of 598 students from 30 classes across eight Irish primary schools participated in a between-subjects design, playing either against a computer-controlled opponent (vs. computer) or a classmate (vs. human). Move optimality did not differ significantly between conditions; however, students in the vs. human condition made fewer errors. Significant interactions emerged between game mode and gender for both performance measures, and between game mode and mathematics anxiety for move optimality. Boys showed higher optimality than girls in both conditions, with a larger gender difference in the vs. human condition. Gender differences in error ratio also varied by mode: boys had a significantly lower error ratio than girls in the vs. human condition, whereas no statistically significant gender difference was found in the vs. computer condition. Higher mathematics anxiety was associated with lower move optimality only in the vs. human condition. Differences between modes cannot be attributed to reduced enjoyment, as participants reported higher enjoyment of the vs. human version regardless of gender or mathematics anxiety. Educational relevance . Our study suggests that competitive gameplay against human opponents tends to benefit boys, who experience it as both more enjoyable and more effective than competition against AI-driven computer opponents. In contrast, girls and students with high levels of mathematics anxiety may require alternative intervention strategies to ensure equitable and effective learning outcomes. Students reported greater enjoyment on average when playing against a human opponent, suggesting that enjoyment alone does not necessarily translate into improved learning outcomes. These findings have important implications for educational game designers. Games should incorporate customizable gameplay options to enhance both effectiveness and inclusiveness: players could choose between competitive and non-competitive modes, different types of competition, or collaborative modes to accommodate diverse learner profiles and reduce potential anxiety-related barriers to learning.