Effect sizes are highest in medium-sized classes (50–99 students), with class-size differences only marginally significant¶

Claim · All claims
Evidence · 1 study · quant-synthesis r3 · q4 · i2 medium
Currency · newest study 2026 · not yet checked against its sources · not reviewed by a person

Subclaims¶

q4 i2 The class-size moderator analysis found the highest effect size in medium-sized classes (g = 0.687), lower in small classes (g = 0.448), and moderate in large classes (g = 0.545), with between-group heterogeneity marginally significant (Q = 4.61, df = 2, p = 0.099). → Doğan Yunus 2026

Evidence¶

Doğan Yunus 2026¶

Doğan Yunus, Kılıç Zeynep, Kalınkara Yusuf, Talan Tarık. (2026). The Impact of Artificial Intelligence-Supported Instruction on Student Learning in STEM: A Systematic Review and Meta-Analysis. Journal of Intelligence. https://doi.org/10.3390/jintelligence14060109

q4 · i2 · quant-synthesis · r3

Class-size moderator analysis across the 35 studies, with class size coded as small (≤49), medium (50–99), large (≥100). Medium classes showed g = 0.687; the article reports the between-group test as "marginally significant (Q = 4.61, df = 2, p = 0.099)".

"Analyses based on class size revealed that the effect size was highest in medium-sized classes (g = 0.687). In small classes, the effect was lower (g = 0.448), while in large classes, it was moderate (g = 0.545)."

Discussion¶

Related Claims¶

• Six of thirteen coded moderators significantly influence the effect of CALL feedback: educational level, intervention provider, mother tongue, research context, subject domain, and target language — related
• Differences in AI-supported instruction effects across STEM disciplines are not statistically significant
• Active learning benefits students in all class sizes, with the greatest effects in classes of 50 or fewer students — reports the opposite
• Tutorial group size moderates SDL gains in PBL: small and medium groups gained SDL, large groups declined — related