Skip to content

Learning by Producing (multimedia production as learning)

Pattern · All patterns
Evidence · 20 claims (12 for, 8 mixed) · 31 studies (10 causal, 9 quant-synthesis, 4 design, 3 theoretical, 2 review, 2 qualitative, 1 associational), q1–q4 · 10 of 31 report an effect size · 12 claims rest on one study

Description and scope

A reusable policy for a unit in which learners make a media product for someone else to watch, hear or use (a short video, a narrated slide show, an audio programme, a web page or database, an exhibition), and learn the subject content and the media skills through making it. The default sequence, from Tanaka (1997), runs: study exemplar products → investigate the content → plan a scenario and a blueprint (storyboard, script, layout) → produce the parts in assigned roles → assemble → present or perform to an audience, and appreciate one another's work. The intended changes are in conceptual-understanding of the subject content the product carries, in the complex-skill of composing for an audience across media (media literacy), and in collaboration; the source also names self-actualization and talent development as aims.

No claim in this wiki tests learning by producing media, against another way of teaching the same content or against a simpler product. The only source for the pattern is a descriptive article with no comparison, and the claim the earlier page listed for it ("Learning By Producing Multimedia") does not exist as a claim page. What the wiki holds is evidence about neighbouring relationships: sustained projects that end in a product against traditional teaching, writing for a real audience against writing for the teacher, preparing to teach and teaching against studying, guidance for novices, rubric-referenced self-assessment, and one direct test in which building a representation did worse than studying a provided one. Each is carried to this pattern by extrapolation, and every step below says how far. The sequence is the earlier page's, kept as the default; the response-dependent policy is an untested design proposal.

How it sits beside converted pages. The problem-based learning pattern owns the general decisions for a problem- or project-centred unit (how much guidance novices get, whether a problem-first phase comes before instruction, consolidation); this page applies them and adds only what is specific to making a media product: studying exemplar products, a planning blueprint checked before production, production roles, and a presentation to an audience. The authentic audiences and purposes principle owns the model of what a real audience changes in a product; this page uses it for the final step and does not restate it. The anchored instruction pattern is the inverse arrangement: learners work inside a media scenario someone else made, where here they make one. The former Multimedia Projects and multimedia-literacy-through-production principle pages stated the same idea without a tested relationship; they were folded into this page on 2026-10-05, and their bodies are kept in comments below.

Inputs: establish the design brief

Field Record or elicit
Learner and starting observation Age band (child, adolescent, adult) and task-specific starting knowledge of the content (novice, intermediate, advanced, or ?), recorded apart from media and tool skill, which is a separate starting state. Before the project, ask each learner to explain the topic in a few sentences and to say what a good product of this kind does for its audience; record both. Preserve unknowns.
The product and its audience Format (video, audio, slides, web page, exhibition), length, who will receive it and what they will do with it, whether the audience is real or simulated, and whether the product will be public, shared within the school, or seen only by the class.
Context and activities Setting (classroom, online-instructor-led, online-self-paced, workplace-clinical, mixed), duration (single-session, days-weeks), number of sessions and their length, team size, equipment per team (shared devices, phones only), who facilitates and how much time they have per team, and which subjects or teachers the unit spans.
Learner-valued goal Ask what the learner wants from making this (to make something good to show family or a community, to learn the tools, to learn the topic, to pass). Record disagreement with the designer objective; a learner who values the media craft may not value the content the product is meant to carry.
Designer objective Which of these the assessment will weigh: the subject content, the media composition skill, collaboration, or the product's effect on its audience. Name which content the product must carry, and which criteria say a product carries it.
Outcome Separate instruments for individual understanding of the content (a short explanation or test, scored apart from the product, at immediate and delayed horizons), the product scored against content criteria and against media criteria separately, each learner's contribution, and any audience response recorded as an observation, not as the grade.

A request to "have students make a video about it" does not specify these. Ask what content the product must carry, for whom, in how many sessions and with what equipment, before choosing the format, the team structure or the presentation.

Sequence and conditional policy

The default sequence is the earlier page's (Tanaka 1997), with doses added as labelled proposals. Steps 1 and 4 are untested; steps 2, 3, 5 and 6 have a neighbouring claim, carried by extrapolation.

Dose for a first project (untested proposal). About 8 sessions of 45–60 minutes over 2–4 weeks: exemplars 1 session, content investigation 2, planning 1, production 2–3, assembly and rehearsal 1, presentation and appreciation 1. A 2–3 minute product per team of 3–4. Content is checked twice before production effort is spent: at the end of the investigation and on the blueprint.

  1. Study exemplar products against criteria (1 session). Show two or three products of the kind learners will make, at least one weak; ask what each product does for its audience and what makes it work; turn the answers into a short criteria list with separate content and media rows. Evidence status: untested for media production. The nearest claim is rubric-referenced self-assessment [+M]: in two non-randomized studies (162 middle-school writers who reviewed a model essay, generated criteria and self-scored drafts; 218 pre-service teachers self-assessing a concept map), work was rated higher than in the comparison conditions, and in the second study the rubric group also reported more stress and a performance-avoidance orientation. In a case study of four physics students writing computational essays [+W], both pairs said the provided example essays made the open-ended assignment understandable.
  2. Investigate and learn the content, with explicit teaching for novices (2 sessions). The product can only carry content the team understands. For learners who are novice in the content, teach the essential content explicitly (a short explanation, a worked example, sources selected for them) before or alongside their own research, rather than leaving them to find it while also learning the tools. At the end, check each learner individually: a few sentences explaining the core idea. Move on when most can; reteach the rest before planning. Evidence status: minimal guidance is less effective for novices than explicit instruction [~S]: across 580 comparisons explicit instruction beat unassisted discovery (d = 0.38), while discovery enhanced with feedback, worked examples, scaffolding or elicited explanations beat other instruction; not tested in production projects.
  3. Plan a scenario and blueprint, and check it for content before production (1 session). Each team writes what the audience should understand by the end, then a storyboard or script naming, for each scene or segment, the content point it carries and the evidence for it. The teacher checks each blueprint against the content criteria and returns it with one or two corrections; production starts only on an approved blueprint. Evidence status: untested; from the earlier page. A reason for the content check: in three experiments with college readers [~M], learners who built their own graphic organizers did worse on transfer than learners given author-provided ones (d = 0.24 ns, 0.43, 0.84), with no difference on retention; making a representation is not automatically generative, and its quality has to be checked.
  4. Produce the parts in assigned roles (2–3 sessions). Assign roles (for example writer, illustrator or camera, editor, sound, narrator), each responsible for named content in the blueprint, and rotate them on the next project. Keep tool instruction short and just in time, and fix the product's length so production cannot expand to fill the unit. Evidence status: roles from the earlier page and the role-assignment strategy; structured collaboration improves achievement [+M] (two meta-analyses, early adolescents and undergraduate STEM, group work against none) bears on group work with individual accountability, not on production roles.
  5. Assemble, rehearse and revise against the criteria (1 session). Teams assemble the product, self-assess it against the content and media rows separately, revise once, and rehearse the presentation. Each learner prepares to explain the part they made and answer a question about it. Evidence status: rubric self-assessment as in step 1, [+M]; preparing to teach and teaching [+M]: across 28 studies, preparing and then teaching raised the teacher's own domain knowledge more than studying without that expectation (g = 0.56; preparing alone g = 0.35), more when the teaching was interactive. A presentation is a weak form of teaching unless the audience asks questions.
  6. Present to an audience, take questions, and appreciate (1 session). Show the product to the audience it was made for, or the nearest real one; have the audience ask questions or respond; then have teams say what they would change. Afterwards, check each learner's understanding of the content individually, apart from the product. Evidence status: authentic audiences [+M]: seventh graders writing to distant peers and second graders writing for an external audience produced higher-rated work than when writing for the teacher (within-subject, no effect size); this is product quality on writing, not learning measured apart from the product. Interactive questioning is the configuration learning by teaching [+M] favours.

Across projects (untested proposal). First project: exemplars and a blueprint template provided, one medium, roles assigned, a short fixed length, an audience inside the school. Second: learners write their own criteria from exemplars, choose between two media, rotate roles. Third and later: learners propose the product and audience, plan with a blank blueprint, and run peer review of blueprints themselves. Move a learner or team on to more open planning when their blueprint passes the content check without correction and their individual content explanation is accurate; keep the template where it does not.

Whole-unit reference point. A meta-analysis of project-based learning [+M] (30 studies, 46 effect sizes, 12,585 students in 9 countries; read as an abstract) found sustained projects ending in a product or performance raised achievement over traditional instruction (d+ = 0.71), moderated by subject, school location, instructional hours and IT support. Media production is one kind of project; the claim does not isolate it, and a short end-of-unit "make a video" activity is outside its scope.

Observation Discriminating probe Proposed action, with its uncertainty
Polished product, thin or wrong content Ask each team member, separately, to explain the core idea without the product; score the product's content and media rows apart. If individuals cannot explain it, the effort went to production: reteach the content, re-check the blueprint, and shorten production next time. If they can but the product does not show it, the blueprint lost the content: add a content point per scene. Neither branch is tested.
Team stalls at planning; blueprint stays empty Ask what the audience should understand by the end; check whether they can state the content at all. If they cannot state the content, the stall is content, not planning: teach it explicitly. If they can, give a blueprint template with one scene filled in. Proposed.
One or two members do everything; others idle Ask each member to name their part and the content it carries; compare individual content checks. Assign roles with named content per role and an individual check; rotate roles next project. A finished team product does not show each member learned.
Tool problems take most of the time Time a production session: how much went to the tool and how much to the content? Simplify the medium (slides with narration rather than edited video), give a template, or cut the length; teach the tool in a short separate session. Proposed; no recorded claim measures production overhead.
Learner avoids presenting, or freezes Ask privately what worries them; offer a smaller or recorded audience. Allow a recorded narration or a presentation to a small group; keep the reader-facing task. The audience claim measured product quality, not anxiety, so this is untested.
Content correct in the product but not on the later individual check Same content in a new question, at a delay. The product may have been assembled from sources without understanding, or one member held the content. Add the step-5 explain-your-part rehearsal and an individual check before presentation.

Fitting the design to a situation

The three facts that most change the decision:

  • Whether learners already know the content the product must carry. A product built while the content is still being discovered by novices leaves them learning the content and the tools at once, and unassisted discovery learned less than explicit instruction. If unstated, ask: before the project, can learners explain the core idea in a few sentences?
  • Whether the target is the content or the media skill. If media literacy is the objective, production time is the learning; if the content is, production time is overhead to be kept short. If unstated, ask: what will the assessment weigh, and is the product scored for content apart from its polish?
  • How much facilitator time and equipment there is per team. The content checks on the blueprint and the individual checks are where the evidence-backed parts of the sequence live, and they need someone's time. If unstated, ask: how many teams per facilitator, how many minutes per team per session, and how many devices per team?
If the brief says… Then change… Basis
Learners are novice in the content Teach the essential content explicitly first (step 2), provide the sources, use a blueprint template with one scene filled in, fix the product to 2 minutes, and check content individually before production. Minimal guidance and novices [~S]; tested on discovery and inquiry, carried to production
Learners are advanced in the content but new to the media Skip the content investigation to a short review; spend the time on exemplars and media criteria; let learners choose the audience and write their own criteria. Untested proposal
Learners are child (primary age) One medium (narrated pictures, a poster with audio), roles named with pictures, the teacher scribes the storyboard, sessions of 30–40 minutes, and an audience they know (another class, families). Untested; the audience comparison with second graders (authentic audiences [+M]) concerned writing, not media
Learners are adult in a work setting (workplace-clinical, onboarding, professional development) Make the product something colleagues will use (a 2-minute procedure video, a job aid), with a subject-matter expert reviewing the blueprint for accuracy before recording; individual check on the procedure itself. Untested proposal; learning by teaching [+M] supports making each learner explain their part, not tested with adults at work
The goal is subject content; media literacy is incidental Cap production at a third of the unit's time, use the simplest medium that carries the content, score content apart from polish, and run the individual content check at a delay. Learner-constructed organizers did worse on transfer [~M], college readers of a text, carried by analogy
The goal includes cross-curricular or interdisciplinary learning Plan the product with two subject teachers, each owning content criteria on the blueprint; ask for one scene that connects the two subjects explicitly. Interdisciplinary humanities units [+M]: one non-randomized middle-school study; units, not media production
single-session only Drop production of new media: teams annotate or re-sequence provided clips or slides into a 1-minute explanation, present to another team, and answer one question each. Untested proposal; the project-based meta-analysis (project-based learning [+M]) excludes short single activities
Mixed mode: some learners remote, some in the room, or phone-only Give remote members roles that work on a phone (script, narration, research, still photos); keep the blueprint in one shared document; present by screen share with questions in chat, and run the individual check in writing for everyone. Untested proposal
One teacher for many teams, or scarce facilitator time (volunteers, a busy shift, a tutor checking in briefly) Spend the facilitator's minutes on two checks only: the blueprint's content (5 minutes per team) and the individual explanation before presentation. Teams peer-review blueprints against the criteria list first; use templates so tool questions do not reach the facilitator. Untested proposal; the blueprint check is the step most directly tied to the content-quality concern [~M]
Learners with low confidence, low literacy, or learning in a second language Allow pictures and speech instead of written text in the product and blueprint; give sentence frames for narration; let a learner present in a small group or by recording; keep the individual content check oral. Untested proposal
The product goes to a real audience (a community partner, families, a public site) Agree with the audience what they need and whether they will respond; plan time to revise after their response; score the content against criteria and record the audience's reaction separately; get consent for anything public. Authentic audiences [+M] (product quality on writing); the authentic audiences principle holds the conditions
High stakes: the product is graded for a qualification Grade content understanding individually as well as the team product; publish the criteria from the start; consider weighting process evidence (blueprint, revisions). Rubric self-assessment [~M], with its stress and avoidance finding; the grading argument for maker education [~W], an argument, not a test
The brief names its own routine (its own storyboard template, rubric, production checklist or presentation format) Keep it. Change only what the evidence says matters: add a content point per scene to the template, separate content from media rows on the rubric, and add one individual content check before presentation. Untested proposal
Little equipment (one device per class, no editing software) Make the product live rather than recorded (a performance, a gallery walk, a radio-style reading), or record on one device in turn while other teams plan and rehearse. Untested; from the earlier page's "handmade TV program" example

Six rows rest on a claim (novices, adults at work, content goals, interdisciplinary goals, a real audience, high stakes), each carried beyond the population or design its claim page reports, as its Basis cell says; scarce facilitator time and the child row lean on a claim only in part; the other rows are untested proposals.

Choosing configurations from evidence

  • Whether to expect a gain from production itself. No claim compares producing media with another way of teaching the same content. The nearest, project-based learning improves achievement [+M], covers sustained projects culminating in a product or performance against traditional instruction (one meta-analysis, d+ = 0.71, moderated by subject, location, hours and IT support), not media production specifically, and was read as an abstract. Do not promise a gain from the label.
  • Guidance. Minimal guidance is less effective for novices than explicit instruction [~S]: explicit instruction beat unassisted discovery (580 comparisons, d = 0.38), guided discovery beat other instruction (360 comparisons); with 112 third- and fourth-graders, more learned an experimental procedure by direct instruction. It argues for teaching the content explicitly within the project, not for abandoning the project.
  • Making a representation is not automatically better than studying one. Learner-constructed graphic organizers are not shown to outperform provided ones [~M]: three experiments with college readers found provided organizers gave better transfer, the gap growing as the construction task got simpler, and a meta-analysis found constructing concept maps had a larger effect than studying them, each against its own comparison. This is the limiting claim for this pattern: a product is a representation learners construct, and its benefit depends on its content being right. Generative accounts are recorded as generative learning improves retention [~M], whose only evidence is a self-explanation meta-analysis (g = 0.55), so it does not test production.
  • Audience. Authentic audiences improve student work [+M]: two small within-subject writing comparisons, outside audience against teacher-as-grader, rated product quality higher; no effect size; not learning measured apart from the product.
  • Explaining one's part. Learning by teaching improves tutor learning [+M]: preparing to teach and teaching raised the tutor's domain knowledge (g = 0.56 over studying; preparing alone g = 0.35), more when interactive; both meta-analyses read as abstracts. A presentation without questions is closer to preparing alone.
  • Criteria. Rubric-referenced self-assessment [~M]: higher-rated work in two non-randomized studies, with more stress and performance-avoidance in one.
  • Teams. Collaborative learning improves outcomes [+M]: group work against none, two meta-analyses; the claim page says it applies to structured collaboration with individual accountability, which is why roles carry named content and each learner is checked.

Do not rank these against one another: their comparators (traditional instruction, unassisted discovery, provided organizers, teacher-as-audience, studying without teaching, self-assessment without a rubric, no group work) and outcomes differ, and none was measured in a media-production unit. A precise abstention names the missing comparison (a production unit against the same content taught with a simpler product or none, on individual content understanding at a delay, with the time cost recorded) and the local observation that stands in for it.

The evidence by question

Further evidence, not yet read against this model

Illustrative design instance and observation record

Invented illustration, not a tested unit. A lower-secondary class (adolescent, classroom, days-weeks) studying the water cycle makes 2-minute explainer videos for a primary class in the same school. The science teacher's objective is that each learner can explain evaporation, condensation and precipitation and why they happen; one learner says she wants to learn video editing. Agree that both the explanation and a well-edited video count, scored separately. The pre-check shows most learners can name the stages but not explain condensation, so the teacher teaches condensation explicitly with one demonstration before planning. Teams of four study two exemplar explainers (one confusing), write a six-row criteria list (three content, three media), and draft storyboards with a content point per scene; the teacher returns two storyboards in which condensation is shown but not explained. Production runs two sessions with fixed roles and a phone per team; each learner rehearses explaining their scene. The primary class watches and asks questions; a week later each learner explains condensation in writing.

Record: content and audience → each learner's starting explanation and what they say a good product does → content taught explicitly, and when → criteria list → blueprint and the corrections given → roles and the content each role carries → time spent on tools against content → product scored on content and media separately → each learner's explanation of their part → audience questions and responses → individual content check at the stated horizon. Preserve the learner's stated purpose and any change in it. An agent may summarize this record but must not fabricate missing inputs or treat audience praise as evidence of content learning.

Elements and limits

Multimedia production project, exemplars, explicit teaching, rubrics, self-assessment, group roles, revision, peer feedback and feedback. Strategies: assign diverse production roles and act as a learning supporter during production.

This pattern is scoped to units in which a team or individual makes a media product that carries named content to an audience, with the content taught and checked before production. Open-ended maker or studio work with no content target, a single "make a poster" activity at the end of a unit, and products judged only on polish or engagement need other configurations and evidence. The wiki records no comparison of learning by producing with another approach; the diagnostic branches above remain to be tested with learners.

Claims

Examples

Key Sources

  • Tanaka. (1997). Multimedia learning and cross-curricular learning. https://eric.ed.gov/?id=EJ545678
  • Chen, C.-H., & Yang, Y.-C. (2019). Revisiting the effects of project-based learning on students' academic achievement: A meta-analysis investigating moderators. Educational Research Review, 26, 71-81. doi:10.1016/j.edurev.2018.11.001
  • Kobayashi, K. (2019). Learning by preparing-to-teach and teaching: A meta-analysis. Japanese Psychological Research, 61(3), 192–203. doi:10.1111/jpr.12221
  • Stull, A. T., & Mayer, R. E. (2007). Learning by doing versus learning by viewing: Three experimental comparisons of learner-generated versus author-provided graphic organizers. Journal of Educational Psychology, 99(4), 808–820. doi:10.1037/0022-0663.99.4.808
  • Yang, Y. T. C., & Wu, W. C. I. (2012). Digital storytelling for enhancing student academic achievement, critical thinking, and learning motivation: A year-long experimental study. Computers & Education, 59(2), 339-352. https://doi.org/10.1016/j.compedu.2011.12.012
  • Kitalong, K. S., & Miner, R. L. (2017). Multimodal composition pedagogy designed to enhance authors' personal agency: Lessons from non-academic and academic composing environments. Computers and Composition, 46, 39-55. https://doi.org/10.1016/j.compcom.2017.09.007
  • Lim, J., & Polio, C. (2020). Multimodal assignments in higher education: Implications for multimodal writing tasks for L2 writers. Journal of Second Language Writing, 47, 100713. https://doi.org/10.1016/j.jslw.2020.100713
Page metadata
Field Value
type pattern
id learning-by-producing-pattern
title Learning by Producing (multimedia production as learning)
description A reusable policy in which learners make a media product for a real audience to learn its content and media skills: study exemplars, learn the content (explicitly, for novices), plan a blueprint checked for content, produce in roles, revise against criteria and present; no claim tests learning by producing itself, and the sequence rests on neighbouring claims about projects, guidance, audiences, teaching others and the quality of learner-made representations.
status review
moved_from principle/cross-curricular-multimedia-units, principle/multimedia-literacy-through-production, principle/multimedia-projects
canonical True
generated by claude/unspecified on 2026-10-05
trust unverified — no one has confirmed this page's content
sources 4 — tanaka-1997, yang-2012, kitalong-2017, lim-2020

Frontmatter is the page's machine-readable record — see the Schema & Guide.