Effect sizes across the included studies are highly heterogeneous (I² = 91.37%), indicating moderating factors¶

Claim · All claims
Evidence · 1 study · quant-synthesis r2 · q3
Currency · newest study 2026 · not yet checked against its sources · not reviewed by a person

Subclaims¶

q3 i? Residual heterogeneity among the 19 studies is significant (Q = 132.358; df = 18; p < .001) with I² = 91.37%, far above the 75% high-heterogeneity criterion. → Sa’idah 2026

Evidence¶

Sa’idah 2026¶

Sa’idah, N., Jailani, Sudiyatno, & Retnawati, H. (2026). Innovative teaching models and critical thinking in mathematics and statistics education: A meta-analytical review. International Journal of Research in Education and Science (IJRES), 12(1), 140-157. https://doi.org/10.46328/ijres.5054

q3 · i? · quant-synthesis · r2

Heterogeneity test of the 19 coded studies using Q and I² statistics with a 95% CI and p < .05 threshold. The article reports "an I² value of 91.37%", exceeding its stated 75% high-heterogeneity benchmark, so a random-effects model was adopted. No standardized effect size is printed for this test.

"The residual heterogeneity value (Q = 132.358; df = 18; p < .001) indicates significant heterogeneity."

Discussion¶

Related Claims¶

• Effect size distributions across AR-in-education studies were heterogeneous (Q=35.692, I²=60.776%), requiring a random effects model — related
• Individual studies of digital story effects are highly heterogeneous, requiring a random effects model — related
• Included flipped classroom studies show substantial heterogeneity, requiring a random-effects model — related
• Effect sizes across the included Indonesian blended-learning studies are heterogeneous, so a random effects model is used — related
• Effect sizes across the 31 included studies are highly heterogeneous (I² = 81.957%), warranting a random effects model — related
• The four schemes of work showed replicable significant heterogeneity in effectiveness, with the predicted-least-effective teacher-centred/non-human scheme among the most effective — related
• The distribution of effect sizes across the 19 studies is heterogeneous (Q = 202.569, exceeding the critical value of 38.582), requiring a random effects model — related
• Effect sizes of technology-delivered literacy instruction are heterogeneous across studies (Q = 165.38, I2 = 68.56) — related
• Quality of technology integration varies substantially across lessons, with most variance unexplained by teacher traits or time course — related
• Effect sizes across the 38 comparisons are highly heterogeneous, requiring a random-effects model — related
• Effect size estimates for hands-on science interventions are highly heterogeneous across studies — related
• Innovative learning models produce a large pooled effect on mathematical and statistical critical thinking compared with conventional instruction — related
• Effect sizes of multiple intelligence practices are highly heterogeneous across studies — related
• Effect sizes across the modelling study pool were heterogeneous, requiring a random-effects model — related
• High heterogeneity was present among effect sizes for both academic achievement (I2≈85%) and learning retention (I2≈61%) — related
• Effect sizes of creative drama studies are heterogeneous (Q = 60.78, df = 39, p < .05), motivating moderator analyses — related
• Heterogeneity across flipped-classroom studies was consistently high (I² > 90%) and was largely not explained by prespecified subgroup analyses, indicating context-dependent effects — a narrower finding that bears on this claim
• The pooled effect-size distribution is highly heterogeneous but resistant to publication bias — related
• Effect sizes across included studies are highly heterogeneous (I² = 94.79%), requiring a random-effects model — related
• Evidence for bilingual immunology teaching rests on few, mostly moderate-quality non-randomized studies with substantial heterogeneity — a narrower finding that bears on this claim