A learner sees one worked example and says, “I understand.” Then the numbers change, the diagram is rotated, the sentence looks different, or the context shifts—and the method disappears.
The learner may have remembered the example rather than learned the rule.
Contrasting cases are designed to make the rule harder to miss. Instead of showing one example in isolation, the teacher places two or more deliberately chosen cases close enough to compare. The cases share important structure and differ in one or two features that matter. The learner asks what changed, what stayed the same, which feature determines the category or method, and why one case should be treated differently from another.
This is not the same as simply giving “more examples.” Ten unrelated examples can increase exposure while leaving the deep relation invisible. Contrasting cases use systematic variation. The teacher chooses examples so the difference itself becomes information.
Institute of Education Sciences research has examined this mechanism directly through programmes on systematic variation and contrasting cases, including mathematics studies where learners compare solution methods side by side. The central promise is not that comparison is always superior to ordinary instruction. It is that carefully selected contrast can draw attention to deep structure, category boundaries, conditional strategy knowledge and invariants that one example often leaves hidden.
This article owns one narrow canonical job on eduKateSG: contrasting cases as deliberate example variation for detecting invariants, category boundaries, strategy conditions and transferable deep structure. It does not own worked examples generally, generic comparison tasks, interleaving, variation theory as a whole, or multiple representations broadly.
The useful question is not, “Did students see two examples?” It is: what relation becomes visible only because these cases are placed together, and can the learner use that relation when the surface changes again?
The 50-second answer
Contrasting cases work when examples are chosen so that comparison reveals the feature that controls meaning, category or method.
- Name the target distinction. Which feature should learners notice?
- Keep irrelevant differences small. If everything changes, learners cannot tell which difference matters.
- Place cases close together in time and space.
- Ask what is the same and what is different.
- Ask which difference changes the rule or strategy.
- Use non-examples and boundary cases deliberately.
- Require an explanation, not only a label.
- Add a fresh case after comparison.
- Vary surface features later to test transfer.
- Do not let comparison replace explicit teaching when learners lack the necessary knowledge.
The shortest principle is: one example can show what works; a good contrast can show what makes it work.
1. Comparison needs a learning target
“Compare these two examples” is too broad. Learners can notice colour, length, vocabulary, layout or difficulty while missing the intended relation.
The teacher should know what distinction the comparison is meant to expose: proportional versus additive change, valid versus invalid evidence, metaphor versus literal description, convergent versus divergent plate boundaries.
The cases are selected around that target.
2. Cases should be near enough to compare
If examples are separated by twenty minutes of instruction, the learner must reconstruct the first from memory before comparing.
Side-by-side display reduces this memory burden and lets attention move directly between relevant features.
Proximity is part of the mechanism.
3. One changed feature can be highly informative
Suppose two equations look almost identical, but in one case the variable appears on both sides. That single change may determine whether a familiar method needs an extra step.
Minimal contrasts make the controlling feature easier to notice.
They are especially useful early in concept formation.
4. Too many differences create noisy comparison
If one example uses different numbers, notation, context, colour, diagram orientation and method, learners may attribute the outcome to the wrong feature.
Reduce irrelevant variation when the distinction is new.
Later, increase surface diversity to test whether the rule survives.
5. Invariants are what stay true while examples change
Mathematics is full of invariants: equality preserved by equivalent operations, ratio maintained under scaling, area preserved under rearrangement.
Ask, “What stayed the same even though the appearance changed?”
Invariants organise knowledge around structure rather than picture memory.
6. Category boundaries need examples close to the edge
Students learn categories poorly if they see only obvious members.
To understand mammals, metaphor, democracy, renewable resources or acute angles, they need cases near the boundary.
Boundary cases reveal which feature is essential and which is merely typical.
7. Non-examples are powerful when they are plausible
A random wrong example teaches little. A near-miss teaches the boundary.
For a persuasive claim, use a statement that sounds confident but lacks evidence. For a proportional relation, use a case that looks similar but is additive.
The learner has to discriminate.
8. “Same or different?” should become “same in what way?”
Every two cases are both same and different at some level.
Ask students to name the dimension: same operation, different representation; same cause, different evidence; same surface, different structure.
This develops comparison language.
9. Explanation turns noticing into a rule
A student can point to the difference without understanding its significance.
Follow with “Why does that difference change the answer?”
The explanation connects feature to consequence.
10. Strategy comparison builds conditional knowledge
Two methods may both solve a problem. The useful question becomes: when is each method efficient, transparent or robust?
Students need more than procedural knowledge. They need conditions for use.
Contrasting methods can make those conditions visible.
11. Side-by-side solution comparison can expose structure
In Mathematics, compare two correct solutions to the same problem.
Which steps are equivalent? Which method uses fewer operations? Which makes the invariant clearer?
The goal is not to crown one universal winner but to build strategic flexibility.
12. Compare correct and incorrect solutions carefully
A plausible error can reveal a misconception.
Ask where the reasoning first becomes invalid and which rule was misapplied.
Do not flood novices with errors before they possess a stable correct model.
13. Contrast can reveal hidden assumptions
Two Science scenarios may differ only in whether a system is closed. Two economic examples may differ in whether other variables are held constant.
Students often overlook assumptions until a contrast makes them consequential.
Ask which condition changed the inference.
14. Contrast helps when vocabulary terms are near neighbours
Words such as accuracy and precision, weather and climate, mass and weight are often confused because they occupy nearby conceptual space.
Place paired examples together and require students to justify the label.
Definitions become usable distinctions.
15. Contrast can teach grammar through function
Compare “The experiment failed because the temperature changed” with “Although the temperature changed, the experiment continued.”
Students can see how conjunction choice changes the relation between clauses.
The goal is not only terminology but meaning.
16. Literature benefits from contrasting interpretations
Two interpretations can use the same passage differently.
Ask which evidence each privileges and what assumption connects evidence to claim.
Students learn that interpretation is constrained argument, not free opinion.
17. History benefits from contrasting sources
Two sources describing the same event can differ in perspective, purpose and detail.
Comparison reveals provenance and selection.
Students learn to ask why accounts differ rather than treating one source as “the truth” and another as “wrong.”
18. Science models can be contrasted by explanatory reach
Place two models beside the same observations.
Which observations can each explain? Where does one fail?
Contrast supports model evaluation rather than memorising the approved diagram.
19. Graphs can be contrasted to reveal scale effects
Two graphs can show identical data with different axis ranges.
Ask how visual impression changes and what remains numerically true.
This teaches representational judgement and guards against misleading visuals.
20. Worked examples and contrasting cases can work together
First show a well-designed worked example. Then place a near case beside it and ask what changes.
The example provides a model; the contrast reveals the boundary.
The mechanisms are complementary.
21. Contrast is especially useful after an initial model exists
Complete novices can compare random features if they do not know what the domain considers important.
Give enough instruction to establish a starting schema.
Then use contrast to refine it.
22. Discovery without guidance can produce false rules
Learners may notice a correlation in the cases and infer the wrong cause.
Teacher prompts and feedback should confirm the relevant feature.
Comparison generates hypotheses; instruction stabilises valid ones.
23. Ask learners to predict before the second case is explained
After studying Case A, show Case B and ask whether the same rule should apply.
The prediction reveals the learner’s current generalisation.
Then compare reasoning with the outcome.
24. Variation should be systematic, not decorative
Changing names, colours or contexts can create variety without changing the conceptual relation.
Systematic variation changes a feature because the teacher wants to reveal what that feature does.
Every change should have a reason.
25. Surface variation becomes important later
Once the learner identifies the structure, vary context, notation and appearance.
Can the rule be recognised in a new setting?
This tests abstraction and transfer.
26. Contrast can expose overgeneralisation
A learner may believe “multiply makes bigger.” Contrast positive fractions, negative numbers and zero.
The near counterexamples force revision of an overbroad rule.
Good contrasts are antidotes to slogans.
27. Contrast can expose undergeneralisation
A student may believe a method works only for the exact format seen in class.
Show the same structure with different notation.
The learner discovers the rule is broader than the original case.
28. Category learning needs both within-category and between-category variation
Show several members of the same category with different surface features, then near non-members.
Students learn both what can vary and what cannot.
This is essential for robust concept formation.
29. Contrast can teach problem representation before solution
Place two word problems with different stories but the same mathematical structure together.
Ask students to represent both before solving.
They begin to see deep structure beneath narrative surface.
30. Contrast supports transfer when learners name the shared structure
Do not stop at “these are similar.”
Ask, “What exactly is the same relation?”
The named relation becomes retrievable in future contexts.
31. Teachers can create a contrast matrix
Rows are cases; columns are critical features.
Students fill cells and inspect which feature predicts category or strategy.
This externalises comparison for complex domains.
32. Tables are useful when several cases must be compared
Side-by-side prose can become unwieldy.
A structured table reduces memory load and highlights dimensions.
Keep columns tied to the learning target.
33. Students can generate contrasting cases
Ask for one example that fits the rule and one near example that does not.
Generation demonstrates understanding of the boundary.
Teacher review is important because learners can create invalid pairs.
34. Student-generated non-examples are diagnostically rich
A bad non-example can reveal what the learner thinks the defining feature is.
Discuss and refine it.
Conceptual errors become visible through case construction.
35. Contrast can help feedback become specific
Instead of “your paragraph is weak,” place it beside a stronger version and ask what changed: evidence, cohesion, precision, sentence control?
Comparison turns vague quality into observable features.
The stronger example should not become a template copied mechanically.
36. Contrast can help learners judge quality
Experts recognise quality partly through experience with many examples.
Students can accelerate that learning by comparing carefully selected work of different quality.
Use criteria after initial noticing to stabilise judgement.
37. Contrast should not humiliate student work
Do not publicly place one named student’s weak work beside another’s strong work without consent.
Use anonymised or teacher-created examples.
Psychological safety matters.
38. Digital interfaces can support rapid comparison
Split-screen documents, interactive sliders and toggles can let learners alternate between cases.
Technology is useful only if the relation remains clear.
Do not hide the cases behind unnecessary clicks.
39. Animation can make contrast difficult if states disappear
For dynamic processes, freeze critical moments side by side.
Learners compare better when both states remain inspectable.
Temporal proximity is not enough if visual memory carries the whole burden.
40. Contrast and interleaving are related but distinct
Interleaving mixes problem types over practice to strengthen discrimination and selection.
Contrasting cases deliberately juxtapose examples to expose the distinction.
A curriculum can use both.
41. Contrast and variation need pacing
Too many new cases at once can overwhelm novices.
Start with one discriminating pair, then expand.
Complexity should grow as the learner’s schema grows.
42. Teachers should anticipate likely wrong features
If students tend to attend to colour instead of shape, or wording instead of logical form, design cases that hold the distracting feature constant while varying the critical one.
Example selection becomes a response to misconception.
43. Contrast should be followed by independent classification
After comparing cases, show a fresh example without the paired support.
Can the learner classify or choose a method and justify it?
This is the immediate transfer test.
44. Later transfer should change the surface more substantially
Days later, present the same deep relation in a different context.
Recognition there is stronger evidence than success on a near copy.
Spacing and surface change test durable abstraction.
45. Explanations matter after classification
A correct label could still be guessed.
Ask which feature determined the decision.
The explanation reveals whether the intended distinction was learned.
46. Contrasting cases can reduce memorisation burden
Instead of memorising ten separate rules, learners may discover one general rule plus the conditions that modify it.
Structure compresses knowledge.
This is one reason comparison can feel clarifying.
47. The teacher’s hardest work happens before the lesson
Good contrasts are designed, not improvised randomly.
Select cases that differ in the right dimension, sequence them and plan the prompt.
The classroom activity can then be simple.
48. The endpoint is discrimination and transfer
Students should be able to see a new case and decide what matters, what category it belongs to and which strategy fits.
The comparison has succeeded when the original pair is no longer needed.
Worked case 1 — Additive versus proportional
Students see two tables. In Table A, every output is input + 3. In Table B, every output is input × 3. Both increase as input increases.
The teacher asks what stays constant in each relation. Students identify constant difference versus constant ratio, then classify fresh tables.
The contrast reveals why “both go up” is too shallow a rule.
Worked case 2 — Strong versus weak evidence
Two history paragraphs make the same claim. One cites a source directly connected to the event; the other cites a later opinion with vague relevance.
Students compare evidence quality, not writing fluency.
They then revise a third paragraph by choosing stronger evidence.
Worked case 3 — Two correct algebra methods
Students compare substitution and elimination on two systems of equations. On one system, a variable is already isolated; on another, coefficients align conveniently.
They identify conditions that make each method efficient.
Strategy choice becomes conditional knowledge rather than teacher preference.
Worked case 4 — The non-example that teaches the boundary
Students have learned examples of renewable energy but assume anything “natural” is renewable. The teacher contrasts solar energy with natural gas.
Both are natural resources; only one replenishes on a human timescale.
The near contrast corrects the overgeneralisation.
Practical route for teachers
Start with the distinction you want learners to notice. Choose two cases that share most irrelevant features and differ on the critical feature. Place them together. Ask what changed, what stayed the same and why the difference matters.
Follow with a fresh case, then later vary context and representation. Use student explanations to check whether the intended feature—not a superficial cue—drives the decision.
Practical route for learners
When studying examples, do not ask only “How do I do this one?” Ask: What would have to change for this method not to work? Which feature makes this an example of the rule? What stays the same if the numbers or context change?
Build pairs of examples and near non-examples. The contrast can reveal what your notes leave implicit.
Practical route for parents and families
When helping, ask the child to compare two similar problems instead of giving another explanation immediately. “What is different here?” “Which difference changes what you do?”
Keep the contrast small enough to inspect. The goal is to help the learner notice structure, not to create a puzzle with hidden tricks.
Common failure modes
- Showing multiple examples with no target distinction.
- Changing too many features at once.
- Using obvious examples but no boundary cases.
- Using random wrong answers instead of plausible near-misses.
- Asking only for labels without explanations.
- Comparing cases before learners have enough domain knowledge.
- Assuming students notice the intended feature automatically.
- Allowing false rules to persist after comparison.
- Using one canonical diagram or context forever.
- Confusing contrast with simple variety.
- Comparing student work in humiliating ways.
- Overloading novices with too many cases.
- Failing to test a fresh case afterward.
- Never changing surface features later.
- Using comparison as a substitute for explicit teaching.
- Ignoring strategy conditions.
- Failing to name invariants.
- Letting digital interfaces separate cases so comparison becomes harder.
- Assuming side-by-side placement alone causes learning.
- Stopping before transfer.
Frequently asked questions
What are contrasting cases?
They are deliberately selected examples placed together so learners can compare features and identify what determines category, relation or strategy.
How are contrasting cases different from examples?
A single example shows one instance. Contrasting cases use systematic variation so the differences between instances reveal the underlying rule.
Do cases have to include a wrong example?
No. Two correct methods, two valid categories or two representations can be contrasted. Near non-examples are useful when category boundaries matter.
When should teachers use contrasting cases?
After enough initial instruction for learners to recognise relevant features, especially when misconceptions involve confusing near categories or memorising surface procedures.
Can novices discover the rule alone?
Sometimes, but unguided comparison can produce false generalisations. Specific prompts and feedback are often valuable.
Does contrast improve transfer?
It can support transfer by making deep structure and conditional rules more visible. Transfer should still be tested with fresh, surface-varied examples.
Is this the same as interleaving?
No. Interleaving mixes problem types across practice. Contrasting cases deliberately juxtapose examples to reveal distinctions. They can complement each other.
How many cases are needed?
Often one discriminating pair is enough to begin. Add more cases when the concept has multiple dimensions or when students need to see within-category variation.
Should students explain comparisons?
Yes. Explanations help reveal whether the learner noticed the intended feature and understands why it matters.
What is the success condition?
The learner can classify, choose a strategy or explain a new case based on deep structure without needing the original comparison present.
Evidence boundary and caveats
Contrasting-case studies differ in domain, learner expertise, prompt design and comparison condition. Some benefits arise when comparison is tightly guided; in other contexts, students may focus on irrelevant features or become overloaded. The mechanism depends heavily on case selection.
IES research on systematic variation and side-by-side solution comparison supports the approach as a promising instructional design, not as a universal replacement for direct explanation or practice.
Sources and further reading
- Institute of Education Sciences research on systematic variation and contrasting cases: https://ies.ed.gov/
- Institute of Education Sciences research on comparing multiple solution methods in mathematics: https://ies.ed.gov/
- What Works Clearinghouse, Organizing Instruction and Study to Improve Student Learning: https://ies.ed.gov/ncee/wwc/PracticeGuide/1
Continue exploring on eduKateSG
- How X Works Hub
- How Self-Explanation Prompts Work
- How Text-and-Diagram Integration Works
- How Transfer of Learning Works
- How Guided Practice Works
- How Checks for Understanding Work
The final idea
An example can become a photograph in memory: this is what the problem looked like when the teacher solved it.
A contrast asks a more useful question: what can change while the rule remains true, and what change makes the rule stop applying?
That is how examples become categories, methods become conditional knowledge, and procedures become transferable structure.
The educational power is not in seeing two cases. It is in discovering the relation between them.