Interventions lasting one to two months show the largest effect, and effects do not increase consistently with longer duration¶

Claim · All claims
Evidence · 1 study (2 entries) · quant-synthesis r3 · q4 · i1–i3
Currency · newest study 2026 · checked against its sources 2026-10-20 (automated, 2 of 2 entries) · not reviewed by a person

Subclaims¶

q4 i3 The highest effect size by intervention duration occurred for interventions lasting more than 1 month and up to 2 months (g = 0.833). → Doğan Yunus 2026 q4 i1 Interventions lasting more than 1 and up to 7 days showed a weaker effect that was not statistically significant (g = 0.256, p = 0.070), and no consistent pattern of increasing effectiveness with longer durations was found. → Doğan Yunus 2026 (2)

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 · i3 · quant-synthesis · r3

Duration moderator analysis across the 35 studies (between-group Q = 16.95, df = 5, p = 0.004). Interventions of more than 1 up to 2 months yielded the largest pooled effect, g = 0.833, 95% CI [0.67, 0.99].

"Regarding intervention duration, the highest effect size was observed in interventions lasting more than 1 month and up to 2 months (g = 0.833, 95% CI [0.67, 0.99], Z = 10.30, p < 0.001)."

Doğan Yunus 2026 (2)¶

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 · i1 · quant-synthesis · r3

In the same duration moderator analysis, the 1–7 day group showed a small, non-significant pooled effect (g = 0.256, p = 0.070); equivalence was not tested. Short-term interventions of ≤5 h were significant (g = 0.621, p = 0.003).

"However, for interventions lasting more than 1 and up to 7 days, the effect was weaker and not statistically significant (g = 0.256, p = 0.070)."

Discussion¶

Related Claims¶

• AI-supported instruction has a moderate positive effect on student achievement in STEM education (g = 0.670 under a random effects model) — the general claim this finding bears on, with the studies behind it
• Intervention duration significantly moderates Gen-AI's effect on HOT, with the strongest effects at 8–16 weeks (ES = 0.759) and weaker effects for shorter (0.494) and longer (0.372) interventions — related
• Interventions longer than one month showed the largest effect size for robot-assisted STEM learning, though duration was not a significant moderator overall — related
• Scaffold effects peak at 2–4 weeks of treatment and decrease dramatically for treatments lasting 9–24 weeks — related
• Longer simulation duration is associated with higher effects, but evidence for simulations over one month is inconclusive — related
• Educational level, instructor support, instructional strategy, interactive type, intervention duration, robotic type, and control group condition were not significant moderators of robotics effects — related
• Programme characteristics — longer duration, higher frequency, teacher delivery and older students — were tentatively associated with more significant well-being effects — related
• Teacher-supported learning scaffolds show the largest effect (ES = 0.92), and longer intervention durations yield larger effects — reports the opposite
• Educational robotics has a moderate positive overall effect on K-16 students' STEM learning compared to non-robotics instruction — related
• mCSCL effects vary significantly by intervention duration, peaking at 1–4 weeks — related
• Intervention duration in weeks and study-quality characteristics did not significantly moderate extensive reading effects — related