Group-level scaffolding training increases teacher process support and student participation¶

Claim · All claims
Evidence · 1 study · causal r1 · q3 · n=4 seventh-grade mathematics teachers (109 students, 14 groups per condition), 5 lessons each on Early Algebra, Amsterdam
Currency · newest study 2022 · not yet checked against its sources · not reviewed by a person

Mathematics teachers coached to use the Small-Group Scaffolding Tool diagnosed the group's collective understanding and handed responsibility back to the group significantly more often than untrained teachers, gave content support less often, and drew substantially more students into each small-group interaction — but did not improve the contingency (accuracy of calibration) of the content support they did give.

Subclaims¶

q3 i? Teachers trained on the Small-Group Scaffolding Tool used the tool's diagnostic step (determining the group's collective maximum understanding) significantly more than untrained teachers (9.2%/7.3% of utterances vs. 0%/0%; Fisher's exact p=.001). → Calor et al. 2022

q3 i? Trained teachers checked whether the group could continue independently significantly more often than untrained teachers (6.9%/15.6% vs. 3.2%/7.4% of utterances; p=.045), and provided process support (encouraging peer explanation) significantly more often (18.5%/14.7% vs. 5.3%/7.4%; p=.004). → Calor et al. 2022

q3 i? Trained teachers gave direct mathematical content support in a substantially smaller share of utterances than untrained teachers (50.2% vs. 79.3%; p<.001), consistent with a shift toward more varied, less content-dominant scaffolding. → Calor et al. 2022

q3 i? The proportion of content support that was well-calibrated ("contingent") to the group's actual understanding did not differ significantly between trained and untrained teachers (61% vs. 49%; p=.134) — the tool changed which scaffolding moves teachers used, not the accuracy of the content support itself. → Calor et al. 2022

q3 i? The average number of students actively participating in a teacher–small-group interaction was substantially higher for trained teachers (M=3.48, SD=0.71) than untrained teachers (M=1.96, SD=0.73); t(56)=7.96, p<.001. → Calor et al. 2022

Evidence¶

Calor et al. 2022¶

Calor, S. M., Dekker, R., van Drie, J. P., & Volman, M. L. L. (2022). Scaffolding small groups at the group level: Improving the scaffolding behavior of mathematics teachers during mathematical discussions. Journal of the Learning Sciences, 31(3), 369-407. https://doi.org/10.1080/10508406.2021.2024834

q3 · quasi-experimental comparison (2 trained teachers vs. 2 matched untrained teachers, video-coded across 45 interaction fragments and 969 turns) with statistical hypothesis testing (Fisher's exact tests, t-test); not pre-registered, not randomized at the teacher level · i? · no standardized effect size printed (tests and percentages only; a d≈2.08 for student participation was derived by the page, not printed); one explicit null result reported · n=4 seventh-grade mathematics teachers (109 students, 14 groups per condition), 5 lessons each on Early Algebra, Amsterdam · causal · r1

Four teachers (two coached on the Small-Group Scaffolding Tool via individual training, on-the-job coaching, and post-lesson debriefs; two taught as usual) were video-recorded across five mathematics lessons designed to provoke small-group discussion. Coded teacher-turn frequencies show trained teachers used the tool's diagnostic and process-support steps significantly more often and gave direct content support significantly less often, while student-participation counts show trained teachers' interactions involved roughly 1.5 more students on average — a very large effect (t(56)=7.96, p<.001). The one explicit null result — no significant difference in how well-calibrated the content support was — is reported plainly by the authors rather than minimized.

Discussion¶

This is a genuine, well-instrumented quasi-experiment with interrater reliability reported (Cohen's kappa .82–.95) and honest reporting of a null finding alongside the positive results, which supports trusting the positive findings as not cherry-picked. The sample is nonetheless small at the teacher level (n=4, 2 per condition) — not randomized, and with teachers matched only on experience/age — so unmeasured teacher-level confounds cannot be ruled out. The behavioral changes documented are in teacher scaffolding moves and student participation counts; the study does not measure student mathematical learning outcomes directly, so the link from "more group-level scaffolding" to "better mathematics learning" is plausible but untested here.

Related Claims¶

• Teacher coaching programs improve teachers' instructional practice by a pooled 0.49 SD across 43 causal studies — the general claim this finding bears on, with the studies behind it
• The quality of feedback students give peers predicts gains in their own work, but no recorded study isolates a causal benefit of the assessor role — related
• Summarization Effective With Training — related
• Contrasting workshop practices (random vs. purposeful grouping) framed the problem of labeling students around local context and teacher agency — related
• Teachers receiving 11 or more hours of training on integrating digital content report almost double the reliance on software and the Internet compared with untrained teachers — related
• Coursemojo's color-coded engagement dots served as a real-time diagnostic that coaches reported enabled in-the-moment scaffolding and small group pulls — related
• Lessons highly rated for both the mathematics and access to mathematics used varied participation structures, but none of the 14 was also highly rated on agency, authority, and identity — related
• The scaffolding intervention unintentionally increased small groups' independent working time, roughly doubling it from premeasurement to postmeasurement — related
• Teachers who completed a scaffolding intervention program increased the contingency of their support far more than non-participating teachers — related