Guidance, immediate feedback, and system-directed control showed unadjusted moderator signals that did not survive multiplicity adjustment, making them hypothesis-generating only¶

Claim · All claims
Evidence · 1 study (3 entries) · quant-synthesis r3 · q3
Currency · newest study 2026 · checked against its sources 2026-10-20 (automated, 3 of 3 entries) · not reviewed by a person

Subclaims¶

q3 i? Guided interventions (g = 0.760) showed a larger unadjusted estimate than unguided interventions (g = 0.479), F(1, 31) = 4.84, p = .035; immediate-feedback interventions (g = 0.744) larger than delayed (g = 0.408), p = .025; and system-directed interventions (g = 0.780) larger than learner-directed (g = 0.499), p = .030. → Yijia Yuan & Yiran Du 2026 q3 i? After Benjamini–Hochberg correction across eight omnibus tests, adjusted p = .095 for guidance, feedback timing, and learner control; no moderator met the false-discovery-rate threshold of 0.05. → Yijia Yuan & Yiran Du 2026 (2) q3 i? The three design characteristics co-occurred substantially at study level (14 of 20 guided studies also provided immediate feedback), so their independent contributions cannot be isolated confidently from the univariable contrasts. → Yijia Yuan & Yiran Du 2026 (3)

Evidence¶

Yijia Yuan & Yiran Du 2026¶

Yijia Yuan & Yiran Du. (2026). Technology-Supported Interventions and Learners' Metacognition: A Three-Level Meta-Analysis. Educational Psychology Review. https://doi.org/10.1007/s10648-026-10227-3

q3 · i? · quant-synthesis · r3

Exploratory univariable three-level mixed-effects moderator models; guidance was one of three implementation variables meeting the conventional unadjusted 0.05 threshold, alongside feedback timing (p = .025) and learner control (p = .030).

"Guided interventions produced a larger subgroup estimate ( g = 0.760, 95% CI [0.598, 0.921]) than unguided interventions (g = 0.479, 95% CI [0.275, 0.683]), F(1, 31) = 4.84, p =.035."

Yijia Yuan & Yiran Du 2026 (2)¶

Yijia Yuan & Yiran Du. (2026). Technology-Supported Interventions and Learners' Metacognition: A Three-Level Meta-Analysis. Educational Psychology Review. https://doi.org/10.1007/s10648-026-10227-3

q3 · i? · quant-synthesis · r3

Multiplicity analysis across the eight moderator tests in the same meta-analysis; cluster-robust CR2 inference additionally showed feedback timing no longer below threshold (p = .057) while guidance (p = .028) and learner control (p = .031) remained below 0.05 before multiplicity adjustment.

"Benjamini–Hochberg correction across the eight omnibus moderator tests produced an adjusted p =.095 for guidance, feedback timing, and learner control; no moderator met the false-discovery-rate threshold of 0.05."

Yijia Yuan & Yiran Du 2026 (3)¶

Yijia Yuan & Yiran Du. (2026). Technology-Supported Interventions and Learners' Metacognition: A Three-Level Meta-Analysis. Educational Psychology Review. https://doi.org/10.1007/s10648-026-10227-3

q3 · i? · quant-synthesis · r3

Descriptive three-way cross-tabulation of guidance, feedback timing, and learner control (Table 3); the authors state the overlap means the patterns are hypothesis-generating rather than evidence of independent moderator effects.

"Fourteen of the 20 guided studies also provided immediate feedback."

Discussion¶

Related Claims¶

• Dictionary-based SEANCE features yielded 21 nominally significant within-person associations with wearable outcomes, but none survived Bonferroni or FDR correction — related
• Progressively larger treatment estimates among intervention students crossing engagement thresholds are hypothesis-generating only — related
• Immediate experiential advantages in knowledge of cognition and engagement converged with the instructional group by five weeks — related
• In an immediate post-test design, participants with an unrestricted chatbot achieved significantly higher learning deltas than participants with a Socratic chatbot (d = 0.80) or an EEG-adaptive exercise system (d = 0.85) — related
• Study design and post-test timing show only borderline moderating effects in between-subject analyses, not significant in within-subject analyses — related
• Feedback direction moderates effects: student-to-student feedback is most effective (d = 0.85), teacher-to-student d = 0.47, student-to-teacher d = 0.35 — related
• The pooled association is robust to dependence, baseline-adjustment, risk-of-bias, and effect-size-derivation sensitivity analyses — related
• Technology type did not significantly moderate effects on metacognitive outcomes, though generative-AI interventions showed a descriptively larger estimate — related
• Under bias-adjusted planning assumptions, approximately 14 independent moderator effect sizes per moderator level are needed for 80% power, and about 60.2% of corpus effect sizes fell below this threshold — related
• Merrill's instructional design principles and feedback features did not significantly moderate the effects of ICT-supported writing interventions — related