PRIMARY 4 MATHEMATICS LEARNING GUIDE · BATCH 7 · GUIDE 25
A diagnostic assessment is useful only when it changes what happens next. A score can tell us that some answers were wrong. A capability profile asks a more useful question: where did the solution first become unstable? Was place value misread? Was the comparison relationship misunderstood? Was a fraction applied to the wrong whole? Did a learner know the concept but lose accuracy in written calculation? Did the method work on a familiar worksheet but fail when the surface changed?
This guide provides an instructional diagnostic framework for Primary 4 Mathematics. It is not a standardised test, not a psychometric instrument and not a substitute for school assessment. Its purpose is to help a teacher, parent or learner gather evidence, classify the first weak dependency and route practice to the most relevant guide in the existing Primary 4 Mathematics hub.
Series route: return to the Primary 4 Mathematics Learning Hub. For mixed-topic transfer, use Mixed-Topic Problems, Representation Choice and Error Recovery.
Official curriculum boundary: the content families are referenced against the MOE Primary Mathematics Syllabus, updated October 2025. The diagnostic sequence, routing labels and repair framework are independently written teaching tools.
Navigate: purpose · capability profile · diagnostic protocol · task set · repair routing · retest · parent use.
1. Diagnosis is not the same as marking
Marking asks whether an answer is correct. Diagnosis asks what the answer reveals about the learner’s mathematical process.
Suppose two students both write 3/4 + 1/6 = 4/10. One may have added numerators and denominators because the common-denominator idea is not understood. Another may know the method but copy the denominator incorrectly under time pressure. The visible wrong answer is similar, but the repair should not be identical.
A useful diagnosis records the first unsupported step. Later errors may simply be consequences. If a learner reads 2.4 m as 24 cm, every later subtraction can be internally consistent but built on the wrong conversion.
Do not diagnose personality from mathematics. Labels such as lazy, weak, careless or gifted are too broad to guide a precise repair. Record observable mathematical evidence instead.
The question is: what capability was required here, what did the learner actually do, and what changed when the task was rotated?
2. Use three evidence levels: direct, changed and transfer
Direct task: the familiar form. Example: 864 ÷ 8.
Changed task: the same relationship with a different unknown or representation. Example: 864 objects are placed equally into 8 groups; find one group.
Transfer task: the relationship appears inside a less familiar multi-step context. Example: eight identical boxes contain 864 items altogether; 17 items are later removed from each box. How many remain?
A learner who succeeds only on the direct task may have procedural recognition without flexible control. A learner who succeeds on all three provides stronger evidence that the relationship travels.
Do not make the transfer task harder merely by increasing the numbers. Change the surface or dependency structure while preserving the core mathematics.
3. Separate concept, representation, operation and execution
| Layer | Question | Example weak signal |
|---|---|---|
| Concept | Does the learner understand the relationship? | Thinks “four times as many” means add 4 |
| Representation | Can the learner show the relationship clearly? | Bar model labels the difference as a whole amount |
| Operation | Can the learner choose the needed calculation? | Finds a total when the question asks for a difference |
| Execution | Can the learner carry out the chosen method accurately? | Correct long-division structure, arithmetic slip in one step |
| Verification | Can the learner test the result independently? | Repeats the same calculation as the only check |
| Interpretation | Does the final answer satisfy the context? | Reports 20 vans when 161 people need capacity for 21 |
These layers can fail independently. Repair the earliest weak layer rather than assigning more of the final calculation automatically.
4. Build a capability profile, not one overall label
A Primary 4 profile can contain separate evidence for:
- whole-number place value, rounding and estimation;
- multiplication, division, factors and multiples;
- fractions, mixed numbers and reference wholes;
- decimal place value and arithmetic;
- measurement and unit control;
- area, perimeter, angles, symmetry and spatial reasoning;
- tables, line graphs and pie charts;
- word-problem representation and multi-step sequencing;
- checking, explanation and error recovery.
A learner can be strong in one family and fragile in another. “Good at mathematics” and “bad at mathematics” hide that variation.
Use the profile to decide where to begin, not to rank the child.
5. Record confidence separately from correctness
Ask the learner to mark each task as certain, unsure or guessed before seeing the answer.
A correct answer with low confidence may indicate fragile retrieval. A wrong answer with high confidence may reveal a stable misconception. A wrong answer with low confidence may simply show that the method is unavailable.
Confidence is not proof. It is another piece of evidence.
Do not reward overconfidence or punish uncertainty. The goal is to compare self-monitoring with actual performance.
Over time, better calibration matters: the learner should increasingly know when an answer deserves checking.
6. Diagnostic protocol: one clean attempt first
- Give the task without teaching immediately before it.
- Allow ordinary working.
- Ask the learner to explain the first decision.
- Record the first unsupported step.
- Give one changed version.
- If needed, reduce the numbers while preserving the relationship.
- Route the repair to the earliest unstable dependency.
A diagnostic attempt should not become a coached performance. If an adult supplies the operation, model and check, the resulting correct answer measures support received rather than independent control.
After diagnosis, teaching is appropriate. Keep the evidence-gathering phase and repair phase distinct.
7. Reduce numerical load to test whether the relationship survives
Suppose a learner fails 3,456 ÷ 8. Replace it temporarily with 56 ÷ 8 or 24 ÷ 6.
If the learner still cannot explain equal sharing or grouping, the weakness is conceptual. If the smaller problem is secure but the larger one fails, inspect written division, place value and fact fluency.
Reducing the numbers does not make the diagnosis childish. It removes one source of load so the relationship can be inspected directly.
Then restore the original scale after the first weak layer is identified.
This is the mathematical equivalent of isolating a variable in an experiment: change one source of difficulty while holding the target relationship steady.
8. Rotate the unknown to test ownership
Original: six boxes contain 24 items each. Find the total.
Rotation A: 144 items are shared equally among six boxes. Find each box.
Rotation B: 144 items are placed 24 per box. Find the number of boxes.
All three questions use the same multiplication relationship. A learner who can solve only the first may recognise multiplication without controlling the inverse forms.
Unknown rotation is one of the fastest ways to distinguish a remembered pattern from a reusable structure.
9. Change representation without changing mathematics
Present the same information as a sentence, table, bar model and simple diagram.
If performance falls only when a table is used, the content knowledge may be intact while representation reading is weak.
If a learner can read a pie-chart quarter but cannot solve “one quarter of 64”, investigate whether the visual partition is supporting a fraction concept that has not yet transferred to symbolic form.
Do not respond by drilling more arithmetic automatically.
The representation itself can be the first dependency.
10. Diagnostic task set A: whole numbers and operation control
- Write 40,706 in words and identify the value of the 7.
- Round 47,362 to the nearest hundred.
- Estimate 398 × 21 before calculating exactly.
- Calculate 2,436 ÷ 7 and check by multiplication.
- A number divided by 8 gives quotient 24 remainder 5. Reconstruct the number.
What to observe: place-value language, midpoint reasoning, magnitude prediction, placeholder control in division, remainder meaning and inverse checking.
If the learner calculates accurately but cannot explain the estimate or check, record verification as a separate capability rather than marking the whole family secure.
11. Diagnostic task set B: factors, multiples and multiplicative structure
- List all factors of 24 systematically.
- Find the common factors of 18 and 30.
- Two lights flash every 4 and 6 seconds. They flash together now. When next?
- A has four times B. Together they have 160. Find both.
- A has four times B and 90 more than B. Find both.
What to observe: factor-pair organisation, grouping versus synchronisation, total-unit count versus difference-unit count, and whether “times” is treated multiplicatively.
Question 10 is a deeper diagnostic. If it exceeds current classroom pacing, omit it rather than treating unfamiliar enrichment as a weakness.
12. Diagnostic task set C: fractions and decimals
- Convert 11/4 to a mixed number and explain the conversion.
- Calculate 2/3 + 1/4.
- A child uses 1/3 of 72, then 1/4 of the remainder. How much remains?
- Order 3.08, 3.8 and 3.18 from smallest to largest.
- Round 4.449 directly to one decimal place.
What to observe: fraction-unit size, common denominators, changing reference wholes, decimal place value and direct rounding rather than repeated rounding.
A learner who says 3.08 > 3.8 because 308 > 38 is showing a place-value misconception, not merely a comparison slip.
13. Diagnostic task set D: geometry, data and representation
- A rectangle has area 96 cm² and width 8 cm. Find its length and perimeter.
- Explain why adding the perimeters of two rectangles used to make one composite figure can double-count an internal edge.
- A graph scale has labelled values 20 and 40 with four equal gaps. What does one gap represent?
- A pie chart represents 64 pupils. One quarter belongs to Group A. Find A.
- Write one conclusion a graph can support and one conclusion it cannot support without more evidence.
What to observe: reverse area reasoning, boundary tracing, interval counting, reference-whole control and evidence boundaries.
14. Do not convert the task set into one misleading total score
A total such as 14/20 can conceal where the four errors occurred. Four failures concentrated entirely in fractions suggest a different next step from four errors distributed across reading, geometry, division and graph scales.
Instead record a short profile such as:
- whole-number calculation: stable;
- division remainder interpretation: developing;
- fraction reference whole: unstable;
- decimal comparison: stable;
- geometry reconstruction: stable;
- graph scale reading: developing;
- independent checking: weak across topics.
This profile points toward instruction. A raw score mostly points toward ranking.
15. Route the first weak dependency to the relevant guide
| Observed first weak link | Primary repair route |
|---|---|
| Place value, rounding, whole-number operations | Whole Numbers, Factors, Multiples and Four Operations |
| Division meaning or remainder use | Division, Remainders and Quotient Interpretation |
| Fraction reference whole or mixed numbers | Fraction Word Problems, Reference Wholes and Mixed Numbers |
| Decimal magnitude or rounding | Decimal Measurement, Place Value and Rounding Accuracy |
| Area/perimeter boundary reasoning | Composite Figures, Missing Lengths and Boundary Tracing |
| Graph scale or data evidence | Tables, Line Graphs, Pie Charts and Scale Reading |
| Representation selection or mixed problems | Mixed-Topic Problems, Representation Choice and Error Recovery |
Route narrowly. Do not assign the whole library because one dependency failed.
16. Distinguish knowledge unavailable from knowledge overloaded
A learner may know a fraction method in isolation but lose it inside a three-step word problem. That does not necessarily mean the fraction concept has disappeared.
Remove one source of load. Present the fraction step alone. If it becomes secure, inspect working-memory load, sequencing and representation instead of reteaching the concept from the beginning.
Conversely, if the fraction step is still wrong when isolated with small numbers, the concept itself needs repair.
This distinction prevents unnecessary repetition of already-known content.
It also prevents the opposite mistake: assuming that a learner “knows it” because one heavily scaffolded example was correct.
17. One successful retest is evidence, not proof of permanent mastery
After repair, use one close question and one changed question. If both are secure, wait and return later with another representation or mixed context.
Immediate success can reflect short-term support. Delayed retrieval gives stronger evidence that the capability remains available.
Do not demand a fixed number of repetitions for every learner. Use the evidence to decide whether the relationship is stable enough to move on.
A retest should not be identical to the taught example. Otherwise memory of the page can masquerade as mathematical transfer.
The goal is increasing independence, not merely increasing exposure.
18. A repair receipt should be short and specific
Record:
- first weak dependency;
- evidence that exposed it;
- repair guide used;
- one close retest result;
- one transfer retest result;
- next return date or trigger.
Example: “Reference whole changed after the first fraction step. Repaired using changing-whole examples. Close retest secure. Transfer question with money context still required a prompt.”
This record is more actionable than “Fractions 70%”.
19. Diagnostic errors can occur in the assessment itself
A badly worded question can make a capable learner appear weak. An unfamiliar vocabulary word can test reading more than mathematics. A question can accidentally require content not yet taught.
Before concluding that a child has a weakness, inspect the task. Is the condition clear? Is the representation legible? Is the content within the intended learning boundary?
If changing the wording while preserving the mathematics immediately restores performance, record the reading or wording dependency rather than a false concept deficit.
Diagnosis should be humble about its own measurement limits.
No single worksheet can capture the whole learner.
20. How parents can use this without becoming the answer key
Parents can ask neutral questions:
- What is the question asking you to find?
- What does this number represent?
- What was your first decision?
- Can you draw or restate the relationship?
- How could you check this differently?
Avoid announcing “This is division” or “Use a bar model” before the learner has made a decision. That removes the very capability being assessed.
When a weakness appears, reduce the task and route the repair. Do not turn diagnosis into a long interrogation.
The aim is to find the next useful teaching move with the least unnecessary testing.
21. Diagnostic mini-laboratory: classify the first weak link
For each case, identify the first layer to investigate.
- A learner says 3.8 < 3.18 because 8 < 18.
- A learner draws four equal units for “A has four more than B”.
- A learner correctly models 160 as five equal units but calculates 160 ÷ 5 = 35.
- A learner calculates 161 ÷ 8 = 20 remainder 1 and reports 20 vans.
- A learner reads a graph interval of five as one.
- A learner solves 1/3 of72 correctly, then applies 1/4 to72 instead of the remainder.
- A learner gets 936 ÷ 6 wrong but solves 60 ÷ 6 and explains equal sharing correctly.
- A learner gets a word problem wrong but solves the same numerical relationship when shown as a bar.
22. Suggested classifications
1. Decimal place value/comparison concept.
2. Representation of additive versus multiplicative comparison.
3. Execution/arithmetic after a correct unit model.
4. Interpretation of remainder and capacity condition.
5. Representation reading/scale reconstruction.
6. Reference-whole tracking in a multi-step fraction problem.
7. Written division/place-value load rather than basic division meaning.
8. Translation from words into a mathematical representation.
These are starting hypotheses, not permanent labels. Confirm them with a changed task before routing extended repair.
23. Batch 7 handover: from diagnosis to the error atlas
Diagnosis identifies where the mathematical route first becomes unstable. The next guide catalogues recurring error families and shows how to repair the first weak link without replacing the whole solution process.
Continue to Primary 4 Mathematics Error Atlas, Misconceptions, First Weak Link and Correction.
Final checkpoint: can the adult or learner distinguish a wrong answer from its cause, rotate the task to confirm the hypothesis and route the smallest effective repair?
Source and editorial note
The official content boundary is referenced against the MOE Primary Mathematics Syllabus, updated October 2025. The capability profile, diagnostic protocol and routing framework are independent eduKate teaching tools and are not represented as a validated standardised assessment.
Editorial approach: Wintour House V1.0 · CivDJ · eduKate Publishing. Observe the first divergence, test it with a rotation, repair narrowly and return for evidence.
