PAIR-C scaffolding shows mixed evidence for deep understanding and reduced misconceptions in emergent-phenomena instruction¶

Claim · All claims
Evidence · 1 study · causal r2 · q3 · n=50 (24 PAIR-C, 26 control)
Currency · newest study 2026 · not yet checked against its sources · not reviewed by a person

Explicitly scaffolding the contrast between emergent (decentralized, collective) and sequential (individualistic, cumulative) causal structure — the PAIR-C framework — produced a statistically significant advantage on near-transfer tasks, a non-significant trend on deep-understanding questions, and no far transfer, in one study of 50 undergraduates randomly assigned to two conditions. The authors' abstract states the stronger reading: that PAIR-C "significantly improves students' deep understanding of natural selection, more effectively reducing misconceptions compared to the traditional method."

Subclaims¶

q3 i? PAIR-C-scaffolded instruction produced a significant Condition × Time interaction on near-transfer tasks (F=4.64, p=.036), favoring the PAIR-C group over a traditional-instruction control group using the same simulations. → Su, Chi & Nagashima 2026

q3 i? On deep conceptual understanding questions, the PAIR-C group's pre-to-post gain trended larger than the control group's, but the between-condition interaction only approached conventional significance (p=.059); far transfer to an unrelated emergent phenomenon was not achieved by either group. → Su, Chi & Nagashima 2026

q3 i? Reduction in the number and diversity of misconceptions (teleological reasoning, anthropomorphism, directed variation, use-and-disuse) was larger in the PAIR-C group, but the between-condition comparisons only approached significance (p=.060 for misconception categories, p=.095 for frequency). → Su, Chi & Nagashima 2026

Evidence¶

Su, Chi & Nagashima 2026¶

Su, M., Chi, M. T. H., & Nagashima, T. (2026). Applying the PAIR-C Framework to foster deep understanding and address misconceptions in science education. Journal of the Learning Sciences, 35(1), 130-175. https://doi.org/10.1080/10508406.2025.2546471

q3 · randomized experiment with a control condition, pre/post design ("randomly assigned to two groups", per the abstract) · i? · no d-family effect size is recorded here; the entry reports F and p values only · n=50 (24 PAIR-C, 26 control) · causal · r2

Fifty undergraduate students, participating online, learned natural selection using NetLogo agent-based simulations (Peppered Moths, Rock Pocket Mice), either with explicit PAIR-C-dimension scaffolding (annotations, prompts, and feedback framed around the five dimensions) or with traditional instruction focused on differential survival and reproduction using the same simulations. Both groups were assessed pre/post on shallow questions, deep (inter-level causal) questions, near-transfer questions (same concept, different context), far-transfer questions (an unrelated emergent phenomenon), and open-ended misconception coding (nine misconception categories, two raters, Kappa=0.843). The PAIR-C group significantly outperformed control on near-transfer gains. On deep-understanding questions and misconception reduction, PAIR-C showed a larger improvement than control, but the group-by-time interaction in each case did not clear the conventional p<.05 threshold (p=.059, p=.060, p=.095 respectively). Neither group achieved far transfer to an unrelated emergent phenomenon (geese flight formation). The abstract, by contrast, reports that PAIR-C "significantly improves students' deep understanding of natural selection, more effectively reducing misconceptions compared to the traditional method", and that epistemic network analysis showed a more focused misconception pattern in the PAIR-C group; the p-values above have not been re-checked against the full text, which was not available when this entry was corrected (2026-10-15).

Discussion¶

This is a single-domain (natural selection), single-population (undergraduates, not K-12 students) randomized study with a modest sample (n=50). On this entry's reading, only the near-transfer result cleared conventional statistical significance; on that reading the deep-understanding and misconception-reduction results are trends. The authors' abstract reports them as significant, so the two readings disagree, and the full text has to settle which holds before either is relied on. The complete absence of far transfer in both groups suggests that naming and contrasting the five dimensions within one domain is not, by itself, sufficient for learners to recognize the same causal structure in an unfamiliar emergent phenomenon — the authors suggest explicit cross-domain analogical comparison (per Jacobson et al., 2020) may be needed for that. Persistence of gains beyond the immediate posttest was not measured.

Related Claims¶

• Guidance during inquiry learning improves post-inquiry learning outcomes (d = 0.50) — the general claim this finding bears on, with the studies behind it
• Contingent scaffolding improves learning more than fixed or absent support. — PAIR-C is a form of dimension-targeted, contingent scaffolding aimed at learners' specific misconceptions
• Analogical Reasoning Improves Transfer — related
• Science-classroom studies of cooperative learning show mixed results, with several null or negative comparisons — related
• Inquiry Based Teaching Improves Science Achievement — a broader claim this one bears on
• Teacher Guided Inquiry Outperforms Student Led — a broader claim this one bears on
• Human tutoring has a medium effect over no tutoring, and in one small study students learned as well when tutors only prompted them as when tutors also explained and gave feedback — related
• The review reports that simulation games support near transfer of knowledge but evidence for long-term or far transfer is less robust — related
• AI-assisted inquiry produced larger decision-making gains than inquiry-only and traditional instruction in secondary climate-change education — related
• AI-section students showed tool-linked self-efficacy and attenuated metacognitive calibration, with the human-pair contrast group showing larger concept-map gains in cross-links and hierarchy — related
• AI-driven methodologies in chemistry laboratory classes foster deeper understanding of chemical phenomena while raising questions about human intuition and creativity — related
• Openness to material feedback is proposed as the single organizing dimension behind the FRA differences between science and generative models — related
• Scaffolding effects varied by education population, with traditional learners significantly outperforming underperforming learners — related
• Models can only be understood with respect to the aims and intentions of a cognitive agent, not solely in terms of representation — related
• Misconceptions decreased over time in all three instructional methods, with no difference between methods
• AI-supported inquiry-based learning significantly improved secondary students' creative mathematical performance compared with traditional instruction — related
• The learning cycle is reported as an effective means for learning science concepts and processes — related
• Nature-based instruction improves academic achievement over traditional instruction, with larger gains at larger doses — related
• Well-controlled comparison studies show clear advantages of inquiry-based instruction over direct instruction for science conceptual learning — related
• Trained students outperformed controls at selecting the more reliable of two contradictory texts, at both post-tests — related