A worked example is one of the most powerful forms of instructional support because it can reveal a route that a novice does not yet know how to build. But that same strength creates its central risk: the example can carry so much of the reasoning that the learner appears competent without becoming independently capable.
The teaching problem is therefore not simply whether to show examples. It is how to design them, what to reveal, what to withhold, how to compare them, when to ask the learner to explain, when to convert them into completion problems, when to remove them and how to verify that the learner can now construct the route alone.
This article takes the teacher-side design job. The learner-side owner is How Worked Examples Work in Learning | Seeing the Route, Explaining the Decisions and Then Carrying It Alone. The two articles should be read together but they solve different problems. One asks how a learner studies an example. This one asks how a teacher builds, sequences and retires one.
It also extends How Teaching Works, How Diagnostic Teaching Works, How Explanation Works in Teaching, How Cognitive Load Works in Learning and How Self-Explanation Works in Learning.
The Direct Answer
Worked examples work in teaching when they reduce unnecessary search, make expert decisions visible, focus attention on underlying structure, create opportunities for learner explanation, and then progressively remove support until the learner can select and execute the route independently.
The example is not the learning outcome. The learner owning the route is the outcome.
The Teacher-Side Worked Example Cycle
- Diagnose: identify what the learner cannot yet do independently.
- Select: choose a problem that exposes the target structure without unnecessary noise.
- Represent: make the translation from problem to representation visible.
- Model: show the route and the high-value expert decisions.
- Annotate: explain why key steps fit, not every trivial action.
- Probe: ask the learner to predict, justify and self-explain.
- Compare: use a second example when comparison reveals a boundary or alternative strategy.
- Complete: remove selected steps and transfer responsibility to the learner.
- Fade: withdraw prompts, labels, representations and method cues.
- Close: remove the example entirely.
- Verify: use a fresh problem to test independent route construction.
- Vary: change surface features and later mix problem types.
- Revisit: return after delay to test durability.
1. Worked Examples Solve a Search Problem
When a learner is new to a task, unguided problem solving can require substantial search. The learner must decide what matters, how to represent the information, which method might fit, what step should come first and how to evaluate the result.
Some search is useful. Blind search is not automatically useful. If the learner has almost no basis for choosing among possible moves, effort may be spent on trial and error rather than on learning the structure that experts use.
A worked example can reduce that unnecessary search. It frees attention for a different question: why does this route work?
This is why worked examples are especially valuable when the task is new, the solution contains several interacting steps, and the learner does not yet possess a reliable schema for organising those steps.
2. The Example Must Be Chosen for a Teaching Job
A worked example should not be chosen merely because it appears next in the textbook. The teacher should know what the example is intended to reveal.
Is the purpose to demonstrate a representation? Introduce a procedure? Reveal a decision rule? Correct a misconception? Compare two strategies? Show how checking works? Expose the difference between a valid and invalid inference?
The stronger the instructional purpose, the easier it becomes to decide which details belong in the example and which details are noise.
3. Start With the Problem Before Showing the Route
Learners should usually see the problem before the completed solution. This creates an opportunity to orient attention and sample prior knowledge.
Ask: What is being asked? What information matters? What representation might help? What would you expect the answer to look like? What first step seems plausible?
The learner is not required to solve the whole problem. The purpose is to activate the problem structure before the solution makes every decision visible.
4. Reveal Representation as an Expert Decision
Teachers often move too quickly from problem statement to equation, diagram, table or paragraph structure because representation is automatic for them.
For novices, representation may be the hardest step. Translating a word problem into algebra, turning experimental information into a variable structure, or converting a reading question into a claim-evidence relationship can require more expertise than the later calculation or writing.
A worked example should therefore reveal why a particular representation was chosen and how the original information maps into it.
5. Show Expert Decisions, Not Just Finished Steps
A complete solution can hide the most important part of expertise: selection.
The learner sees that a method was used but may not know why it was chosen. The learner sees a diagram but may not know why that representation is useful. The learner sees a quotation selected but may not know why it is better evidence than another.
Teacher annotation should therefore concentrate on high-value decisions: What feature triggered this method? What alternative was rejected? What invariant is being preserved? What check will detect a mistake?
6. Do Not Narrate Every Trivial Step
More explanation is not always better. Excessive annotation can overload attention and hide the structure beneath commentary.
The teacher should distinguish between steps that are already fluent and steps that carry the learning objective. If the learner can add fractions reliably, a long explanation of every arithmetic operation may distract from the new algebraic relationship being taught.
Good worked examples are selectively explicit.
7. Use Signalling to Direct Attention
A worked example can contain many visual elements: symbols, diagrams, arrows, labels, written explanations and intermediate results. The teacher should make it easy to see which element corresponds to which decision.
Place explanations near the step they explain. Use consistent labels. Avoid decorative complexity. If a diagram carries the reasoning, connect labels directly to the relevant part rather than forcing the learner to search back and forth across the page.
The objective is not aesthetic minimalism for its own sake. It is reducing avoidable search so attention can be spent on the learning relation.
8. Worked Examples Should Expose the Why
A worked example that shows only what to do can teach imitation. A stronger example connects each major move to a principle.
Why is the same operation applied to both sides of an equation? Why does this variable need to be controlled? Why is this quotation evidence for the interpretation? Why does this graph support association but not necessarily causation?
These explanations help the learner build knowledge that can survive changes in surface form.
9. Ask the Learner to Predict the Next Step
Prediction converts a worked example from passive viewing into active reasoning.
Pause before a major step and ask what should happen next. The learner may not know the whole route, but even an imperfect prediction reveals the current model and forces retrieval of relevant knowledge.
The teacher can then compare the prediction with the expert route and explain the difference if necessary.
10. Self-Explanation Is the Learner’s Half of the Example
A worked example becomes much stronger when the learner explains relationships rather than merely follows them.
Ask: Why is this step allowed? What does this symbol represent? What stayed the same? What changed? Why would another method be less efficient here? What misconception would produce this wrong step?
The goal is not constant verbalisation. The goal is to sample whether the learner has constructed the underlying relations.
For the learner-side mechanism, see How Self-Explanation Works in Learning.
11. Example–Problem Alternation Keeps the Learner Acting
A long block of fully worked examples can create familiarity while requiring very little reconstruction. One useful teaching structure is to alternate an example with a related problem.
The example reveals the route. The problem tests what the learner can now carry. The next example can then be selected in response to that evidence.
This creates a tighter teaching loop than showing many examples first and discovering much later that the learner never learned how to begin independently.
12. Completion Problems Are a Bridge, Not a Compromise
Moving from a fully worked solution directly to a completely independent problem may be too large a jump. Completion problems let the teacher transfer responsibility in controlled steps.
The learner may complete the last two steps, supply a missing representation, choose the method, diagnose an incorrect line or write the explanation that connects two provided steps.
The missing part should be chosen deliberately. Remove the decision the learner is ready to own.
13. Fading Should Follow the Structure of the Task
Fading is often described as gradually removing steps. But the best fading is not random deletion. It follows the intellectual structure of the task.
If representation is the weak point, require the learner to construct the representation while keeping later calculations visible. If strategy selection is weak, hide the method name but preserve a reminder to classify the problem. If checking is weak, remove the final verification and ask the learner to create one.
The teacher is not merely reducing support. The teacher is transferring ownership of specific decisions.
14. The Example Must Eventually Close
The most important move in worked-example teaching may be physically or mentally removing the example.
As long as the route remains visible, recognition can masquerade as knowledge. Closing the example forces retrieval.
Ask the learner to reconstruct the first decision, recreate the representation, explain the central principle or reproduce the route in outline form before attempting a fresh problem.
The learner-side guide How Worked Examples Work in Learning develops this transition in detail.
15. A Fresh Problem Is the First Real Independence Check
Redoing the exact example can show route memory. A fresh but structurally related problem gives stronger evidence.
The teacher can vary numbers first, then context, then representation, and finally mix the problem with neighbouring types so the learner must select the method rather than reproduce it.
Success should increasingly depend on the learner’s own recognition and planning rather than on similarity to the model.
16. One Example Is Rarely Enough to Define a Concept Boundary
A single example can accidentally teach superficial features. Learners may infer that the colour, context, diagram shape, number range or wording is part of the concept when it is not.
Teachers should therefore use multiple examples when variation is needed to reveal what changes and what remains invariant.
The second example should not exist merely to provide more practice. It should have a specific comparative job.
17. Use Contrasting Examples When Method Selection Matters
If learners confuse two methods, teach the contrast directly.
Place two similar-looking problems side by side where different strategies are appropriate. Ask what feature changes the route. Or show two solution methods for the same problem and compare efficiency, transparency and generality.
Contrast teaches discrimination. It moves the learner from “I can follow this method” toward “I know when this method fits.”
18. Compare Strategies When the Comparison Teaches Something
Multiple solutions can be valuable, but showing alternatives without purpose can create confusion.
The teacher should know what the comparison is meant to reveal: generality, efficiency, representation, robustness, elegance, error risk or the underlying mathematical or conceptual relationship.
Ask learners to explain not only how the methods differ but why one may be preferable under particular conditions.
19. Incorrect Worked Examples Can Teach Error Detection
A deliberately flawed example can be useful when the learner already has enough knowledge to evaluate it.
The learner can identify the first invalid step, explain the misconception, repair the route and predict what kind of wrong answer the error would produce.
But incorrect examples are risky for novices who cannot yet distinguish correct from incorrect reasoning. The teacher should not use error exposure as a puzzle when the learner lacks the knowledge required to resolve uncertainty.
20. Worked Examples Should Change as Expertise Grows
Support that helps a novice can become redundant for a more experienced learner.
Early examples may need detailed representation, annotated decisions and explicit checks. Later examples can be shorter, partially complete, comparative or focused only on subtle strategic choices.
The teacher should therefore continually ask: What does the learner still need the example to carry?
When the answer becomes “very little,” the example should recede.
21. This Is Where Cognitive Load Matters
Worked examples are closely connected with cognitive load because they can reduce demands that are not yet productive for a novice.
A teacher can simplify search, externalise intermediate steps, integrate representations and sequence complexity so limited working memory is spent on relevant relationships.
But the goal is not permanent reduction of difficulty. As knowledge becomes organised, the learner should carry more of the task internally.
See How Cognitive Load Works in Learning.
22. Worked Examples and Diagnostic Teaching
A worked example can also be a diagnostic tool.
If the learner cannot explain the first decision, the problem may be conceptual. If the learner understands the example but cannot start a fresh case, the issue may be representation or strategy selection. If the learner can complete the route once a cue is given, retrieval may be weak. If the learner succeeds only while every intermediate step is visible, working-memory coordination may still be unstable.
The teacher can therefore use partial example removal to identify what support is currently carrying performance.
See How Diagnostic Teaching Works.
23. Worked Examples and Explanation
A worked example is a specialised explanatory form. It makes a process visible across time.
But it should follow the same explanatory rule: reveal the structure, avoid unnecessary detail, test reconstruction and give the thinking back.
The worked example should therefore not become a transcript of everything the teacher knows. It should be a carefully designed representation of the decisions the learner needs to see now.
See How Explanation Works in Teaching.
24. Worked Examples and Scaffolding
A worked example is one kind of scaffold. Like any scaffold, it should support performance while capability is being built and then become unnecessary.
The critical question is not whether support exists but whether the support is being transferred into learner capability.
If the learner repeatedly succeeds only when the example is present, the teaching system has created dependence rather than independence.
See How Scaffolding Works in Learning.
25. The Expertise-Reversal Problem
An explanation that is useful to a novice can become redundant to an expert. The same is true of worked examples.
Once the learner already has a strong schema, detailed worked solutions may consume time without adding much learning. They may even interfere with the learner’s own more efficient strategy.
Teachers should therefore avoid turning a successful novice scaffold into a permanent classroom ritual.
26. Worked Examples in Mathematics
Mathematics is a natural domain for worked examples because symbolic procedures can hide the decisions that make them meaningful.
A strong mathematics example should reveal problem representation, method selection, transformations, invariants and checking. It should distinguish between a legal move and a merely familiar-looking one.
For example, instead of writing “move the 3 across,” the teacher can model that the same operation is applied to both sides to preserve equality. Instead of merely substituting into a formula, the teacher can identify what each quantity represents and why the formula fits the relationship.
Later examples should reduce commentary and require the learner to select the method independently.
27. Worked Examples in Science
Science worked examples should expose the route from evidence to inference.
A teacher can model how to identify variables, read a graph, distinguish observation from explanation, connect a change to a mechanism and avoid overclaiming causation.
For experimental questions, the example can show why a control is necessary and what comparison makes the inference possible.
Later completion tasks can remove the explanation but leave the data, asking the learner to reconstruct the causal chain independently.
28. Worked Examples in English
English teaching often shows polished final answers without making the composition decisions visible.
A worked paragraph can reveal how the writer identifies the task, chooses evidence, frames a claim, connects evidence to interpretation and revises wording for precision.
The teacher can annotate where the analysis actually happens, why one quotation is stronger than another, and what a weak paragraph would do differently.
As support fades, the learner should move from completing missing explanation sentences to planning and writing the entire argument independently.
29. Worked Examples in Writing
Writing examples are most useful when they reveal decisions rather than encourage imitation of surface phrases.
The teacher can show why a sentence is moved, why an example is cut, why a paragraph needs a stronger logical link or why a particular word creates ambiguity.
Side-by-side drafts can be especially powerful because they reveal revision as a process rather than presenting the final text as if it appeared fully formed.
30. Worked Examples in Study Skills
Study strategies can also be modelled as worked examples.
A teacher might show how to turn a chapter into retrieval questions, how to analyse a wrong answer, how to plan spaced review or how to decide whether a learner needs more explanation, more retrieval or more mixed practice.
The example should reveal the decision rule so the learner can later choose the strategy without being told what to do.
31. Worked Examples in Small-Group Teaching
Small groups allow the same example to generate different levels of support.
One learner may need the complete route. Another may be asked to predict every second step. A third may receive only the problem and compare an independent route with the example afterward.
The shared object remains the same, but the teacher varies responsibility according to diagnostic evidence.
This connects with How Can Three Students Share a Class Without Receiving the Same Teaching?
32. Worked Examples and Homework
Homework can reveal whether worked-example learning survives without the teacher nearby.
A useful structure is to provide one example as reference, one completion problem and one fresh problem without support. Later homework can remove the reference and introduce variation.
The teacher should examine which parts fail after the example disappears. That failure is diagnostic evidence, not merely a reason to assign more questions.
See How Homework Works in Learning.
33. Worked Examples and Retrieval
An example can be understood today and inaccessible tomorrow. Retrieval must therefore be built into the sequence.
After studying a route, ask the learner to reconstruct it without looking. Later, return to a structurally related problem after time has passed.
That delayed reconstruction distinguishes durable learning from temporary fluency.
34. Worked Examples and Transfer
Transfer requires the learner to recognise the same deep structure beneath changed surface features.
Teachers should therefore vary context, representation and wording after the route is first learned. Ask what remains the same. Ask which feature still makes the method appropriate.
If the learner succeeds only when the fresh problem visually resembles the example, the example may have taught a template rather than a transferable structure.
35. Worked Examples and Metacognition
Worked examples can model not only task execution but self-monitoring.
The teacher can make checks visible: Does this answer have the right sign? Does the size make sense? Have I answered the actual question? Does my evidence support the claim? What assumption am I making?
Over time, these external checks should become questions the learner asks internally.
See How Metacognition Works in Learning.
36. Worked Examples and AI
AI can generate worked solutions instantly, which makes the design problem more important, not less.
If AI supplies a complete route before the learner has attempted to represent the problem, the tool may erase the most valuable diagnostic evidence. If every difficulty is solved by requesting another explanation, prompt dependence can grow quickly.
A stronger AI-supported sequence asks the learner to predict first, study only the missing portion, explain the key decision, close the solution and solve a fresh problem without AI support.
See How AI-Assisted Study Works | Help That Must Leave the Learner Stronger.
37. Do Not Confuse Recognition With Learning
The learner may say, “Yes, that makes sense,” while looking at the completed route. That statement can be sincere and still provide weak evidence of independent capability.
Recognition is easier because the important decisions are already present. Independent performance requires the learner to generate those decisions.
This is why the example must close and why fresh problems matter.
38. Do Not Confuse Copying With Reconstruction
Copying a worked example can create a clean record without creating much learning.
If copying is used, add intellectual work: predict, annotate the reason for a step, cover the solution, reconstruct the route and solve a related problem.
The question is always what the learner had to generate.
39. Do Not Confuse Volume With Coverage
Ten examples that all share the same surface structure may add less value than three examples deliberately chosen to expose different boundaries.
Worked-example design is not about accumulating pages. It is about selecting cases that reveal the conceptual space the learner must later navigate independently.
40. Do Not Show the Method Name Forever
Textbooks and worksheets often organise practice by chapter or method. This is useful early because it reduces search. But later it can hide a critical part of expertise: deciding which method applies.
Eventually the teacher should remove the method label and mix neighbouring problem types.
Independent performance includes selection.
41. The Worked Example Should Include a Checking Strategy
Many examples end when the answer appears. Expert problem solving often continues with verification.
Show how the result is checked: substitute back, estimate magnitude, inspect units, compare with constraints, test against the text, or ask whether the conclusion is supported by the evidence.
Then fade the check so the learner must generate it independently.
42. The Worked Example Should Reveal Common Error Routes
A good example can include a brief warning at the point where a common misconception becomes tempting.
Explain not merely “do not do this” but why the incorrect move looks plausible and what principle rules it out.
This helps learners build error detection rather than memorise prohibitions.
43. The Worked Example Should Preserve the Learner’s Need to Think
The best example does not answer every possible question. It reveals enough structure for productive learning while leaving decisions the learner can reasonably make.
This balance changes with expertise. The same learner may need a complete example today and only a strategic hint next week.
Teaching quality lies in choosing what to reveal now and what to leave for the learner.
44. A Strong Worked Example Can Be Short
Length does not determine instructional quality. A compact example can be powerful when it reveals the critical decision and removes irrelevant detail.
Conversely, an elaborate solution can obscure the key relation.
The standard is not maximum explanation. It is sufficient explanation for the learner’s current state.
45. The Teacher Should Track What the Example Is Carrying
At any moment, a worked example may be carrying representation, method selection, procedural memory, intermediate calculations, checking, vocabulary or confidence.
The teacher should know which of those supports remain necessary.
That question turns fading from a routine into a diagnostic process.
46. A Worked Example Is Successful When It Becomes Unnecessary
This is the deepest design principle.
A successful worked example is not one that students keep beside them forever. It is one that helps build internal structure so the learner no longer needs the external route.
The example may remain useful later as a reference, but independent performance should not depend on consulting it.
47. The Minimum Worked-Example Teaching Test
After teaching with one example, close it and ask the learner to:
- state what kind of problem this is;
- identify the first important decision;
- reconstruct the representation;
- explain why the method fits;
- rebuild the next major step;
- name one likely error;
- describe a checking strategy;
- solve a fresh related problem.
This test reveals whether the learner merely followed the route or has begun to own it.
48. Common Teacher Failure Modes
- Showing the solution too early: diagnostic evidence disappears.
- Showing only steps: the learner cannot infer why the route works.
- Annotating everything: explanation overload hides the structure.
- Using examples with no contrast: surface features become mistaken for the concept.
- Never asking for prediction: the learner becomes a spectator.
- Never requiring self-explanation: recognition is mistaken for understanding.
- Using too many complete examples in a row: familiarity grows faster than independence.
- Removing all support at once: the transition becomes unnecessarily large.
- Fading random steps: responsibility is not transferred strategically.
- Never closing the example: the model continues to carry retrieval.
- Testing only identical problems: transfer remains unknown.
- Keeping detailed examples after expertise grows: support becomes redundant.
- Using incorrect examples too early: novices encode errors they cannot yet evaluate.
- Omitting checking: learners imitate production but not expert monitoring.
49. A Practical Worked-Example Sequence for Teachers
- Show the problem before the solution.
- Ask for a prediction or first-step hypothesis.
- Reveal the representation and explain why it fits.
- Model the first major decision.
- Pause before the next decision and ask the learner to predict.
- Annotate only the high-value steps.
- Ask the learner to self-explain one or two critical transitions.
- Show a contrast case if discrimination matters.
- Give a completion problem.
- Remove the example.
- Ask for route reconstruction.
- Give a fresh related problem.
- Vary the surface.
- Mix neighbouring problem types.
- Return after delay.
50. Evidence Boundary
Worked examples have a substantial instructional evidence base, especially for novice learning in structured problem-solving domains. The strongest general claim supported by this evidence is not that worked examples should replace problem solving, but that guidance can reduce unnecessary search while learners build useful schemas, and that examples should be paired with explanation, practice and eventual independent performance.
The U.S. Institute of Education Sciences guide Organizing Instruction and Study to Improve Student Learning includes worked-example guidance, and Improving Mathematical Problem Solving in Grades 4 Through 8 recommends using worked examples to compare strategies and asking students to explain why steps work while also providing opportunities for independent problem solving.
These sources do not imply that every worked solution in every subject produces the same effect, that more explanation is always better, or that worked examples should remain permanent. Domain, learner expertise, task complexity and the quality of the example design all matter.
The Core Principle: Show the Route, Then Transfer the Route
A worked example is successful when it changes what the learner can do without it.
The teacher uses the example to expose expert representation, decisions, steps and checking. The learner predicts, explains and reconstructs. The teacher removes selected support. The learner takes over more of the route. The example closes. A fresh problem tests whether the route can now be generated independently.
That is the teacher-side logic of worked examples: reduce unnecessary search at the beginning, increase learner responsibility through the middle, and remove the example before mistaking supported performance for independent learning.
Continue through the wider teaching architecture with How Teaching Works, or use the learner-side companion How Worked Examples Work in Learning.
