Skip to content

Universal Design For Learning

Principle · All principles
Evidence · 20 claims (13 for, 7 mixed) · 33 studies (13 review, 7 causal, 5 quant-synthesis, 3 qualitative, 3 theoretical, 2 design), q1–q4 · 6 of 33 report an effect size · 12 claims rest on one study

Conditional relationship

This page owns a planning move: before teaching, find the demands an activity makes that are not part of the learning goal (seeing a diagram, hearing a lecture, decoding dense print, holding many items in memory, writing by hand, sitting through an unexplained task) and build alternatives to those demands into the design from the start, while the goal and its success criteria stay the same for everyone. For a class whose learners vary in sensory access, language, prior knowledge, attention or motivation, a design planned this way (several routes into the content, several ways to engage, several ways to show the outcome, all held to one standard) is expected to let more learners reach the goal, and need fewer after-the-fact fixes, than a design with one route of each kind. The expected change is in how many learners meet the shared criterion (immediate-performance, conceptual-understanding) and in voluntary engagement (motivation-affect, persistence). Whether the gain lasts (delayed-retention) or carries to new tasks (near-transfer) is a separate claim.

No claim in the wiki tests Universal Design for Learning as a framework, or compares a UDL-planned course with the same course planned for one route. The page's earlier text cited a meta-analysis of UDL studies (Capp, 2017) reporting positive effects alongside few rigorous experiments and inconsistent implementation; it has no claim page here, so its findings are not used in the model. What the wiki holds are claims about the parts: coordinated representations, segmenting, autonomy support, feedback level, learner control, redundancy, and the limits of matching instruction to a learner's style. The design below is therefore a labelled default.

The relationship is conditional on four things a page cannot settle in advance: which demands belong to the goal (if the goal is reading a printed text, decoding is not a barrier to remove; if the goal is understanding a process, it may be); who the learners are (a barrier exists only for a particular learner meeting a particular demand); how many options the learners can manage (options cost attention, and novices pay more); and what the outcome is (choice evidence is stronger for motivation than for achievement). It is a relationship about demands and goals, not about learning styles: offering several routes so that a learner can get past a barrier is a different claim from assigning a "visual learner" visual material, and the evidence is against the second.

How the converted siblings sit inside this: Multimodal Instruction holds the model for representing content in coordinated modes and giving an alternative mode where one is inaccessible; Accessible Vocabulary and Syntax holds removing language the objective does not need; Autonomy holds choice within structure and non-controlling language; Chunking and Cognitive Load Management hold segmenting and load; Assessment for Learning holds feedback against shared criteria. This page does not repeat those models. It owns the audit that decides where each is needed, the rule that the goal stays fixed while routes vary, and the check that alternative routes are equal in rigour.

Default design, while the relationship is untested

The earlier page's guidance, kept as a concrete default a designer can act on and revise. Each step says its evidence status.

  1. Fix the goal and its criteria first, and name what is not the goal. Write the capability, the instrument and the criterion; then list the demands of each planned activity and mark each as part of the goal or not. Only the second kind gets an alternative. Untested proposal; from the earlier page ("without a fixed goal, 'multiple means' collapses into unfocused choice").
  2. Run a barrier audit before the first session. For each activity, ask what a learner must be able to see, hear, read, remember, physically do and want to do, and which learners in this cohort (including those with recorded accommodations) would be stopped by it. Allow one to two hours per unit (a proposal). Untested proposal.
  3. Give core content at least two coordinated routes, each carrying part of the meaning. For example a diagram with a short explanation, a worked example with a spoken walk-through, and ask learners to translate between them. Provide the same text as editable text (for text-to-speech, enlargement and contrast), and captions and a transcript for audio and video. Basis: Multiple representations improve learning [+M] and Adding relevant pictures or graphics to text improves learning from it, with a small pooled effect on reading comprehension, though whether dual coding explains the benefit is disputed [+S], both conditional on the representations being relevant and connected; the representation model itself is on Multimodal Instruction. Captions alongside narrated animation are an access layer, not a learning gain for hearing learners: redundant on-screen text can impair learning when it competes with a visualization [~M], so let learners turn them on and off.
  4. Segment content and give learners the pause. One idea per segment, roughly 5–10 minutes of video or one page of text (a proposal), with pause and replay always available. Basis: Chunking reduces working memory load [+S] (laboratory memory research, carried to lessons); learner pacing helped on complex material in one experiment and learner control in general showed almost no effect (learner pacing and learner control [~M]).
  5. Offer one bounded choice per session, with a reason for the work that is not negotiable. Two or three vetted options (which example to work on, which text on the same concept, whether to work alone or in a pair), never a choice of whether to meet the goal. Basis: Autonomy support increases intrinsic motivation, engagement, and persistence [+M], for motivation outcomes; the choice model is on Autonomy. Learner control does not, in general, raise achievement (learner control [~M]).
  6. Where the format of a product is not the goal, offer two or three formats against one rubric. For example a written explanation, a recorded explanation with a sketch, or a labelled diagram with a short oral check. Score every format with the same criteria, and before the first use, mark one anchor sample in each format to check the criteria travel. Untested proposal; the argument that a test in one verbal code may underestimate some students' knowledge is an argument, not a tested result [+W].
  7. Give feedback on the task and the process, against the shared criteria, whatever the format. Say what the work does and does not yet meet and what strategy would close the gap; avoid praise or criticism of the person. Basis: Feedback Most Effective At Task And Process Levels [+S] (meta-analyses across education and work; not tested across expression formats).
  8. Keep individual accommodations alongside the universal options. After the first week, check who still meets a barrier the options did not remove, and arrange the targeted support (extended time, a scribe, assistive technology). Basis: a practitioner's judgment that some learners cannot access instruction without modifications [~W] (adult older-beginner ESL; expert opinion, not tested).
  9. Plan the arc across sessions, not only one lesson. A default for a unit of four to eight weeks: weeks 1–2, options curated by the designer, every learner tries each support and each product format once with the teacher modelling how to switch on captions, use a glossary or choose a format; weeks 3–5, learners choose their supports and format, write one sentence on why, and self-check against the rubric before submitting; from week 6, learners plan their own supports for a new task, and the designer removes options nobody used and adds one where a barrier persisted. Move a learner from curated to open choice when they meet the criteria on two consecutive products in their chosen format and can say why they chose it. Fade extra representations and step-by-step supports for learners who succeed without them (expertise reversal [~M]). The arc, its durations and its thresholds are untested proposals.
  10. Review the design after each unit, by barrier, not by learner label. For each learner who did not reach the criterion, record which demand stopped them and change the design for the next cohort, not only for that learner. Untested proposal.

When a condition is not met: if no one knows the cohort yet (a course built before enrolment), audit against the commonest barriers (print, audio, language, memory load, fine motor) and plan step 8's check for week one; if there is no time for parallel formats, keep one format and add the cheapest access layers first (editable text, captions, a glossary, pause control).

Fitting the design to a situation

The three facts that most change the decision:

  • Which demands belong to the goal. An alternative to a demand that is the goal (spelling in a spelling course, reading the drug label as printed) removes the learning, not a barrier. If unstated, ask: what must the learner be able to do at the end, with what aids, and which parts of each activity are only the way in?
  • Who is in the cohort and what stops them. Learners with a sensory or motor impairment, learners in a second language, learners with little prior knowledge and learners who have stopped engaging meet different barriers, and the options that help one can burden another. If unstated, ask: who has struggled with this unit before, what accommodations are on record, and which languages and devices do learners use?
  • What the learners can manage choosing. Novices pay for every option in attention; experienced learners can choose well and are slowed by supports they no longer need. If unstated, ask: can learners already do one task of this kind unaided, and have they chosen supports for themselves before?
If the brief says… Then change… Basis
Learners are novice on this content Curate options to two or three; integrate representations rather than offering many in parallel; model each choice before asking for it Instructional guidance that helps novices can become redundant or counterproductive as expertise grows. [~M] and Multiple representations improve learning [~M] (benefit depends on skill at connecting representations); bounding the number of options is untested
Learners are advanced, or succeed on the first task unaided Drop the extra representations and step-by-step supports by default and leave them available on request; widen choice of task and format Instructional guidance that helps novices can become redundant or counterproductive as expertise grows. [~M] (cognitive-load studies; no crossover point given)
Learners are child (primary age) Fewer, concrete choices (pick one of two); the teacher demonstrates each support; read-aloud and picture supports by default untested proposal; the learning-styles experiment with fifth graders shows only that matching to a reported style did not help (Learning Styles Matching Does Not Improve Learning [~S])
Learners are adult or professionals, including older beginners or learners with disabilities Ask learners which barriers they meet; keep individual accommodations; offer formats drawn from their own work modifications may be necessary for older beginners and learners with disabilities [~W] (one practitioner's judgment, adult ESL)
A colleague proposes assigning materials by each learner's "learning style" Do not; offer the routes to everyone and let content decide the primary mode (a diagram for spatial structure, speech for pronunciation) Learning Styles Matching Does Not Improve Learning [~S] (systematic review and a randomized experiment with fifth graders)
Goal is a concept or principle Pair a visual and a verbal representation of the same relation and ask learners to translate between them; feedback on the process, not only the answer Multiple representations improve learning [+M] (undergraduate physics, one field study) and Adding relevant pictures or graphics to text improves learning from it, with a small pooled effect on reading comprehension, though whether dual coding explains the benefit is disputed [+S] (comprehension from text with graphics)
Goal is a procedure or complex-skill whose format is the goal (handwriting, oral fluency, a clinical technique) Keep the expression format fixed; vary only the routes into the content and the practice supports untested; from the earlier page (options in how learners engage and access, not in what they are expected to learn)
Setting is online-self-paced, no teacher present Build the access layers into the media (editable text, captions and transcripts, pause and replay, glossary links); give automatic, criterion-referenced feedback on each product format Letting learners pace instruction helped on complex material in one experiment [~M] (primary-school animation and a meta-analysis of learner control; achievement gain not general); the rest untested
Narrated video or animation is the main medium Captions on request rather than always on; put essential words in the narration; give a transcript separately Redundant on-screen text impairs learning when it competes with a visualization [~M] (lab multimedia lessons; captions remain a necessary accommodation for deaf and hard-of-hearing learners)
classroom with one teacher and a large group Offer choice at the level of the whole task (two versions of one activity) rather than per learner; one shared rubric displayed for all formats; check-ins on who is stuck untested proposal
Time is a single session Do only steps 1, 3 and 4 (goal, two routes into the content, segmenting and pause); skip format choice untested proposal
Stakes are high (certification, grading) Validate that each expression format measures the same capability before allowing it; mark anchor samples in every format; record accommodations given untested proposal; the concern that a single verbal format underestimates some learners is an argument [+W]
Learners have stopped engaging, or work only for marks Add a stated reason for the work and one meaningful choice per session; replace controlling incentives and language with informational feedback Autonomy support increases intrinsic motivation, engagement, and persistence [+M] (motivation outcomes; learning outcomes weaker, see Autonomy)

Ten of the thirteen rows cite a claim, though two of them (primary-age learners, high stakes) are untested proposals that cite a claim only for a limit or an argument; three cite none. Most cited claims are carried from laboratory or single-setting studies (multimedia lessons, physics recitations, memory experiments, feedback meta-analyses) to course design, and none tests a UDL-planned design as such.

Observation, state and explanation

What can be observed: which supports and formats each learner used, whether their product met each criterion, how long it took, what they said about their choice, and which activity they stopped at. "Faces a barrier", "is overloaded by options" and "is avoiding difficulty" are inferred states.

Response Plausible explanations Observation that could change the interpretation
A learner uses none of the supports offered and does not reach the criterion Did not know the supports existed or how to use them; using them is visible to peers and embarrassing; the barrier is one the supports do not address Show the support privately and ask the learner to try one item with it; ask what stopped them on the activity.
A learner always chooses the same, easiest-looking format Avoiding a demand that is part of the goal; the format genuinely removes a barrier; the other formats were not modelled Check whether the chosen format meets every criterion; ask the learner to try one other format once with support, and compare.
Work in an alternative format meets fewer criteria than written work The format makes some criteria hard to show; markers apply the criteria less consistently across formats; the learner understood less Mark the same learner's explanation in two formats against the same rubric; have a second marker score anchor samples.
Many options offered, little work completed Time spent choosing; too many options for novices; options not tied to the goal Halve the options for a week and compare completion; time how long choosing takes.
A learner with recorded accommodations still fails the same activity The universal options do not cover their barrier; the barrier is in the goal itself; an unrelated difficulty with the content Ask the learner and any specialist which demand stops them; try the targeted accommodation on one activity.
Engagement rises, scores do not Choice served motivation but not learning; the assessment does not match what learners chose to do; too short a horizon Compare criterion attainment for learners who did and did not take up options; check the rubric against each format.

These rows are proposals, untested as diagnostics. Asking a learner about a barrier, or offering a support, changes what you then observe, so record each probe and keep the explanations that remain open.

Evidence and qualifications

No claim in the wiki tests UDL as a framework, or compares a design with several routes against the same design with one. The claims below test parts of it, mostly in other settings.

Learners, tasks and outcomes differ across these claims (memory experiments, physics undergraduates, fifth graders, adult literacy classes), so do not rank them by effect label. None measures how many learners reach a shared criterion under a UDL-planned design.

The evidence by question

Further evidence, not yet read against this model

Objective and learner-valued goal

The designer's objective is that every learner reaches the same criterion on the same capability, by whatever route works. Elicit separately what each learner wants: to keep up without being singled out, to use the tool they already rely on (a screen reader, a phone, their first language), to get the course finished, or to show what they can do in a format that does them justice. Agreement and divergence are both observations. A learner who values not standing out may refuse a support that is visibly different, which is an argument for making supports available to everyone rather than issued to a few; a learner who values speed may choose the format that meets fewest criteria.

For example, an adult with dyslexia on a health-and-safety course may want to pass without reading long manuals, while the designer's objective is that they can identify hazards on site. Audio and diagram versions of the manual serve both; replacing the hazard-identification task with a quiz on the manual would serve neither. Record both goals, so that a completed quiz is not read as the capability the learner and the employer need.

What would revise this model?

The expectation should weaken if comparisons of a design with several routes against the same design with one route, with comparable learners and the same criterion-referenced outcome, show no increase in the share of learners reaching the criterion, or show that the options cost novices more than they help. If gains appear only in engagement and not in attainment, keep that split rather than calling both "learning". If alternative expression formats consistently yield different judgments of the same capability, the format options should be narrowed to those shown to measure the same thing. If targeted accommodations alone match a universal design for the learners who need them, the case for universal options rests on the stigma and timeliness arguments, not on learning. Do not protect the model by calling every failure "weak implementation" after the fact.

Claims

  • Multimodal Instruction — holds the conditional model for coordinated modes and alternative modes where one is inaccessible.
  • Accessible Vocabulary and Syntax — holds removing language the objective does not require.
  • Autonomy — holds choice within structure and non-controlling language.
  • Cognitive Load Management — UDL's multiple options must be curated to avoid imposing navigation load on novices
  • Chunking — segmented, structured representation is a core "multiple means of representation" tactic
  • Assessment For Learning — flexible expression formats work only when assessment criteria are transparent and goal-referenced
  • Active Learning — multiple means of action and expression typically manifest as varied, participatory performance formats
  • Adaptive Learning — technology-based adaptation is one systematic way to deliver differentiated representation and challenge

Examples

Illustrative

Choice Boards — A grid of equivalent tasks through which learners demonstrate the same objective (e.g., write, record, or diagram an explanation). A well-designed board enforces equal rigor across options and requires all learners to hit the same success criteria.

CAST UDL Guidelines — The canonical framework resource, maintained by CAST, organizing dozens of design checkpoints under the three principles. Widely used as a design-review rubric for curriculum and digital learning products.

Bookshare — An accessible ebook library providing alternate formats (audio, braille, large print, dyslexia-friendly) of the same texts, exemplifying multiple means of representation at scale for learners with print disabilities.

Khan Academy — Offers the same content as video, transcript, worked examples, and practice with hints, letting learners choose their entry point and re-represent material on demand — a large-scale implementation of multiple means of representation.

Accommodations — Individualized supports (extended time, scribing, assistive technology) that remain necessary alongside universal design; UDL reduces but does not eliminate the need for targeted accommodations.

Advance Organizers — Pre-reading structures and graphic organizers that give learners with weaker background knowledge an access route into the same content as their peers.

Key Sources

  • Meyer, A., Rose, D. H., & Gordon, D. (2014). Universal design for learning: Theory and practice. CAST Professional Publishing. https://udlguidelines.cast.org
  • Capp, M. J. (2017). The effectiveness of universal design for learning: A meta-analysis of literature between 2013 and 2016. International Journal of Inclusive Education, 21(8), 791–807. doi:10.1080/13603116.2017.1325074
  • Rose, D. H., & Meyer, A. (2002). Teaching every student in the digital age: Universal design for learning. ASCD. https://www.cast.org/products-teaching-every-student
  • U.S. Office of Educational Technology. (2016). United States National Educational Technology Plan. U.S. Department of Education. Republished in R. West (Ed.), Foundations of Learning and Instructional Design Technology. EdTech Books. https://edtechbooks.org/lidtfoundations/us_national_ed_tech_plan
  • Robinson, Peter J. (1988). Procedural and Declarative Knowledge in Vocabulary Learning: Communication and the Language Learner's Lexicon. Paper presented at the 22nd Annual TESOL Convention, Chicago. https://eric.ed.gov/?id=ED296585
  • Dyer, Penny. (1992). Training a Multicultural Workforce. Hobart Institute of Technical and Further Education. https://eric.ed.gov/?id=ED378844
Page metadata
Field Value
type principle
id universal-design-for-learning
aliases diverse-exercise-types-learning-styles, comprehensible-training-benefits-all-workers
title Universal Design For Learning
description For a class whose learners vary in access, language, prior knowledge and motivation, planning several routes into content, engagement and expression while the goal and criteria stay fixed is expected to let more learners reach the goal than a single-route design; no claim here tests UDL as a framework, only its parts.
status review
canonical True
generated by claude/unspecified on 2026-10-05
trust unverified — no one has confirmed this page's content
sources 5 — meyer-2014, capp-2017, rose-meyer-2002, robinson-1988, dyer-1992

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