Evidence for Learning with others: peer instruction and explaining to learn #

Every substantive claim on the Learning with others: peer instruction and explaining to learn page is checked against current research. Here is each claim, how well today’s evidence supports it, and the sources. The full, de-duplicated source list lives on the references page.

Supported · moderate evidence — Learning material in order to teach it, and especially actually teaching it to others, produces better understanding and retention for the teacher than studying the same material only for oneself (the protege effect).

Fiorella and Mayer found that students who actually taught material learned more than those who only studied it, and a broader learning-by-teaching literature supports the effect; the benefit is reliable but moderated by whether teaching involves genuine generative explanation rather than mere recitation.

Sources: Fiorella & Mayer (2013), The relative benefits of learning by teaching and teaching expectancy, Contemporary Educational Psychology — https://doi.org/10.1016/j.cedpsych.2013.06.001 · Kobayashi (2019), Learning by preparing-to-teach and teaching: A meta-analysis, Japanese Psychological Research — https://doi.org/10.1111/jpr.12221 · full reference ›

Mixed · moderate evidence — Actually teaching material yields larger and more durable learning gains for the teacher than merely expecting to teach it (teaching expectancy alone).

Fiorella and Mayer reported that the act of teaching outperformed expectancy alone, but the size and even existence of a pure expectancy benefit is inconsistent across studies; meta-analytic work finds the reliable gains come from the generative act of explaining, so the act-vs-expectancy contrast is supported in direction but contested in magnitude.

Supported · moderate evidence — Learning by teaching works because explaining to a learner requires retrieving the material from memory and reorganising it for an audience, which exposes and prompts repair of the teacher’s own knowledge gaps.

Roscoe and Chi’s review of tutoring found that tutors learn most when they engage in reflective knowledge-building—generating explanations and integrating ideas—rather than simply delivering information, consistent with retrieval and elaboration accounts of why teaching benefits the teacher.

Sources: Roscoe & Chi (2007), Understanding tutor learning: Knowledge-building and knowledge-telling in peer tutors’ explanations and questions, Review of Educational Research — https://doi.org/10.3102/0034654307309920 · full reference ›

Supported · moderate evidence — Peer instruction—posing a conceptual question, having students commit to an answer, discuss it with a peer who disagrees, then re-answer—improves conceptual understanding compared with traditional lecturing.

Crouch and Mazur’s decade of physics-course data showed peer instruction substantially improved scores on conceptual inventories versus traditional lecturing, and the broader active-learning evidence base reinforces the gain; magnitude varies with implementation quality and question difficulty.

Sources: Crouch & Mazur (2001), Peer Instruction: Ten years of experience and results, American Journal of Physics — https://doi.org/10.1119/1.1374249 · full reference ›

Supported · strong evidence — Active-learning methods in which students discuss and work problems, such as peer instruction, raise exam performance and lower failure rates relative to passive lecturing across STEM subjects.

Freeman et al.’s meta-analysis of 225 studies found active learning raised average exam scores by about half a standard deviation and cut failure rates from 34% to 22% compared with lecturing, making the superiority of active over passive instruction one of the better-established findings in education research.

Sources: Freeman, Eddy, McDonough, Smith, Okoroafor, Jordt & Wenderoth (2014), Active learning increases student performance in science, engineering, and mathematics, PNAS — https://doi.org/10.1073/pnas.1319030111 · full reference ›

Supported · moderate evidence — Reciprocal teaching—students taking turns to lead structured dialogue using summarising, questioning, clarifying and predicting—improves reading comprehension, particularly for weaker readers.

Palincsar and Brown’s foundational studies and later reviews show reciprocal teaching reliably improves comprehension over typical instruction, with notable benefits for struggling readers; effect sizes vary considerably across implementations and on standardised versus experimenter-designed measures.

Sources: Palincsar & Brown (1984), Reciprocal teaching of comprehension-fostering and comprehension-monitoring activities, Cognition and Instruction — https://doi.org/10.1207/s1532690xci0102_1 · full reference ›

Supported · moderate evidence — Cooperative group work produces better learning than individualistic work when it combines a shared goal with individual accountability so that each member must contribute (as in jigsaw structures).

Johnson and Johnson’s extensive meta-analyses report that well-structured cooperative learning, defined by positive interdependence plus individual accountability, outperforms competitive and individualistic arrangements on achievement; the conditional design requirements are widely accepted, though some effect-size estimates come from the approach’s own proponents.

Sources: Johnson & Johnson (1999), Making cooperative learning work, Theory Into Practice — https://doi.org/10.1080/00405849909543834 · Kyndt, Raes, Lismont, Timmers, Cascallar & Dochy (2013), A meta-analysis of the effects of face-to-face cooperative learning, Educational Research Review — https://doi.org/10.1016/j.edurev.2013.02.002 · full reference ›

Supported · strong evidence — Group work can reduce individual effort and learning through social loafing: when responsibility is diffuse and individual contributions are not identifiable, people exert less effort than when working alone.

Karau and Williams’s meta-analysis of 78 studies established social loafing as a robust effect that is reliably reduced when individual contributions are identifiable and evaluable, directly supporting the claim that unaccountable group work depresses individual effort.

Sources: Karau & Williams (1993), Social loafing: A meta-analytic review and theoretical integration, Journal of Personality and Social Psychology — https://doi.org/10.1037/0022-3514.65.4.681 · full reference ›

Supported · moderate evidence — Collaborative learning as a method is distinct from a social learning-style preference: the structure of the group task, not the learner’s sociability, determines whether collaboration improves learning.

Evidence that collaborative benefits depend on task design (individual accountability, interdependence) rather than learner disposition, combined with the broad failure of learning-styles matching to improve outcomes, supports treating collaboration as a technique separable from any social style preference.

Sources: Johnson & Johnson (1999), Making cooperative learning work, Theory Into Practice — https://doi.org/10.1080/00405849909543834 · Pashler, McDaniel, Rohrer & Bjork (2008), Learning styles: Concepts and evidence, Psychological Science in the Public Interest — https://doi.org/10.1111/j.1539-6053.2009.01038.x · full reference ›