Singapore School Mathematics Operating Manual · Chapter 1
A student can carry out the right calculation and still deliver the wrong answer. The question asks for a length; the student gives an area. It asks for the greatest possible integer; the student gives a decimal boundary. It asks for a reason; the student supplies another example. The difficulty is not always a missing formula. Sometimes the student has not identified what the finished response is supposed to be.
This chapter teaches that decision directly. A task contract is our teaching name for the combination of action, mathematical object, conditions and answer form required by a question. It is not an official examination term. Its purpose is practical: before starting a solution, know what you are trying to produce; before finishing, check that you have actually produced it.
Use the BTT Mathematics Curriculum Overview to locate a topic or school stage. Use this chapter when the topic is familiar but the wording changes the job. The examples move from arithmetic into algebra, graphs and calculus. They are original teaching examples, not past-paper questions, and their stage labels indicate prerequisites rather than compulsory placement in every school programme.
Read the task · Compare worked demands · Understand explanation and proof · Try the practice · Check the answers
1. Read the question as a request for a mathematical object
Consider the instruction “Find the coordinates of the minimum point.” It contains more information than “do something with the quadratic”. The action is to find. The object is a point. The point must be the minimum point, not an intercept or any convenient point on the curve. The answer therefore needs two coordinates in the correct order.
Now compare “Find the minimum value.” This asks for the smallest output of the function. A single value can answer it. The input at which that value occurs may support the explanation, but the input is not the requested minimum value. These two instructions may use almost identical working while requiring different final responses.
A useful reading sentence is: “I need to produce ___, under ___, in the form ___.” For a measurement question, this might become “a length, using the stated scale, in metres”. For an inequality, it might become “all positive integer values, satisfying the inequality, as a list”. For a proof, it might become “a general argument, using the given conditions, showing why the claim must hold”.
The sentence is a temporary scaffold. It should not become a compulsory paragraph before every straightforward calculation. Once a student recognises the task reliably, the preparation can be mental. The value of the scaffold is that it exposes a misunderstanding before several lines of correct but irrelevant work are written.
Four questions make the contract visible. What action is requested? What object is required? What restrictions apply? What will count as a complete answer? Notice that “restrictions” includes words, not just numbers: positive, distinct, integer, exact, at least, in terms of, using, and without drawing can all change what is acceptable.
2. A command word is a starting point, not a mechanical code
Words such as calculate, state, solve, explain and show guide the response, but no isolated word determines the entire solution. “Find” may ask for a number, an expression, an equation, a probability or a set of values. “Explain” may require a property of a diagram in one question and a limitation of a model in another.
Read the complete instruction, including any sentence after it. “Calculate the length” and “Calculate the length, giving your answer correct to two decimal places” have different reporting requirements. “Solve the equation” and “Solve the equation for positive integer values” have different answer sets. Underlining only the verb leaves the rest of the contract unread.
The meanings in this guide are teaching interpretations, not a universal official glossary or a promise about marking. The question paper, its instructions and the relevant syllabus remain authoritative. A teacher may also impose a method during practice to develop a particular skill. An efficient alternative can be mathematically valid while not meeting that specific practice instruction.
Do not attach a fixed assessment-objective label to a verb. A “find” question can require substantial reasoning, while a “state” question can ask for a feature already visible in a prepared expression. The mathematical demand depends on what has been supplied, what must be connected and what must be justified.
3. One algebraic expression, four different jobs
Keep the same expression in view: x² − 5x + 6. The topic has not changed, but the instruction can change the finished product completely.
Expand
“Expand (x − 2)(x − 3)” asks for an equivalent expression with the product multiplied out. The calculation is x² − 3x − 2x + 6, so the answer is x² − 5x + 6. It does not ask for values of x. Writing x = 2 or x = 3 would introduce an equation that the question never supplied.
Factorise
“Factorise x² − 5x + 6” asks for the expression written as a product. We seek two numbers with sum −5 and product 6: −2 and −3. The answer is (x − 2)(x − 3). Expanding this product is a useful check, but leaving the final answer expanded would undo the requested form.
Solve
“Solve x² − 5x + 6 = 0” asks for values that make an equation true. Factorising gives (x − 2)(x − 3) = 0. Hence x = 2 or x = 3. The factorised expression is an intermediate result, not the complete answer to the solving task.
Evaluate
“Evaluate x² − 5x + 6 when x = 4” asks for a numerical output at one specified input. Substitution gives 16 − 20 + 6 = 2. The number 4 is given, not something to solve for. The final answer is 2, not the two roots of the related quadratic equation.
These distinctions prevent a common kind of topic-triggered response. Recognising “quadratic” is not enough. The same quadratic can be expanded, factorised, evaluated, solved, sketched or used to establish a bound. The command and requested object tell the student which relationship needs to become visible.
4. Primary arithmetic: the number must answer the named quantity
In an invented transport problem, 112 pupils must travel, and each identical bus can carry at most 26 pupils. Assume there are no other passengers and that every pupil needs a seat. The calculation 112 ÷ 26 = 4 remainder 8 is useful, but it does not by itself answer every possible question about this situation.
“What is the minimum number of buses needed?” asks for a whole-number capacity decision. Four buses hold only 104 pupils. Five buses hold 130. The answer is five buses. It is not 4.31 buses, because a fraction of a bus is not an available booking in the stated model.
“How many seats are unused when the minimum number of buses is used?” asks for a different quantity. Having established five buses, calculate 5 × 26 − 112 = 18. The answer is 18 seats, not five. The difficult intermediate decision must not replace the final target.
“What fraction of the available seats is occupied?” asks for a comparison against the capacity of the five buses. The fraction is 112/130 = 56/65. The denominator is not 112, and it is not 26. It is the total available seating in the arrangement being described.
A child does not need algebra to recognise these distinctions. Ask the child to finish the sentence “My answer counts…” before calculating. Buses, empty seats and occupied fractions are different objects. The arithmetic can be related while the answers remain different.
The lesson is not “always round up”. If a different question asks how many complete groups of 26 can be formed from 112 pupils, the answer is four complete groups, with eight pupils remaining. The context determines the whole-number decision. A memorised rounding direction cannot replace the requested quantity.
5. Solve, list, count and identify an extreme are not interchangeable
Suppose 3x − 4 < 11. Adding 4 gives 3x < 15, so x < 5. Over the real numbers, that inequality describes infinitely many possible values. It is the complete solution set when no additional domain restriction is imposed.
Now add “x is a positive integer”. The allowed values become 1, 2, 3 and 4. If the question asks to list them, write the list. If it asks how many there are, answer four. If it asks for the greatest value, answer 4. The same work supports three different outputs.
Boundary language matters. Replacing “less than 5” with “at most 5” includes 5. Replacing “positive integer” with “non-negative integer” admits 0. Replacing “integer” with “real number” means there is no greatest value strictly below 5: however close a chosen value is to 5, another larger allowed value can be found.
This last point is a useful extension. The answer to “What is the greatest real number less than 5?” is not 4.999, because 4.9995 is larger and still below 5. The absence of a greatest value is a property of the task, not a failure to calculate enough decimal places.
When a pupil confuses a boundary with an answer, ask for one allowed example and one forbidden example. For x < 5, 4.5 is allowed and 5 is forbidden. This quick test often reveals whether the inequality has been understood or merely copied.
6. Verify an example, disprove a claim, prove a claim
Consider the statement “The sum of two consecutive odd integers is divisible by 4.” Testing 3 + 5 = 8 verifies one instance. Testing 9 + 11 = 20 verifies another. These examples support a conjecture, but they do not yet establish the statement for every allowed pair.
A general proof names an arbitrary odd integer 2n + 1, where n is an integer. The next odd integer is 2n + 3. Their sum is 4n + 4 = 4(n + 1). Since n + 1 is an integer, the sum is divisible by 4. The argument covers all consecutive odd pairs, rather than only those selected for testing.
Now change the claim to “The sum of any two odd integers is divisible by 4.” This is false. The pair 1 and 5 gives 6, which is not divisible by 4. A single valid counterexample disproves a statement that claims to hold for every allowed case.
The counterexample has to satisfy the conditions. The pair 2 and 5 would not disprove a claim about two odd integers because 2 is not odd. Likewise, a negative value cannot disprove a claim restricted to positive values unless the claim actually includes negatives.
The three jobs therefore have different stopping conditions. To verify a stated example, check that example. To disprove a universal claim, supply an allowed case where it fails. To prove a universal claim, provide an argument that reaches all allowed cases. More arithmetic is not automatically stronger reasoning; the type of evidence has to match the type of claim.
For younger learners, this distinction can be explored without formal symbols. “Every rectangle with perimeter 20 cm has the same area” can be challenged using a 4 cm by 6 cm rectangle and a 3 cm by 7 cm rectangle. Both perimeters are 20 cm, but the areas are 24 cm² and 21 cm². Two carefully chosen drawings do more work than twenty unrelated examples.
7. “Show that” gives a destination, not permission to assume it
A rectangle has perimeter 50 cm and width x cm. Show that its area is 25x − x² cm². The required expression is already printed, but the task is to derive it from the stated geometry.
Let the length be L cm. Since 2(L + x) = 50, L + x = 25 and L = 25 − x. The area is width multiplied by length, so A = x(25 − x) = 25x − x². The missing bridge between perimeter and area has now been supplied.
Beginning with “A = 25x − x² because the question says so” does not show the result. Neither does substituting one convenient width and observing that the formula gives the right area in that case. The task concerns the relationship, not a single numerical example.
Working backwards can be a legitimate planning method. A student may inspect 25x − x², factor it to x(25 − x), and recognise the two side lengths. But the finished explanation still needs to establish why those side lengths follow from the perimeter. Exploration and justification are related activities with different roles.
The domain also matters. A non-degenerate rectangle requires 0 < x < 25 in this model. The area expression is algebraically defined outside that interval, but those values do not describe the stated rectangle. A good response does not confuse an expression’s mathematical availability with its physical interpretation.
When an instruction asks the student to show a numerical result correct to a stated accuracy, copying the printed rounded number is not a derivation either. Show the relevant calculation and enough numerical detail to support the requested rounding. The specific G3 examination guidance on this appears in the source note below.
8. Explain the mathematical reason, not the sequence of buttons
“I divided because I used the calculator” describes an action. “I divided the total cost by the number of identical items to obtain the cost per item” identifies a relationship. The second explanation can be checked against the context.
A useful explanation often has three parts: the relevant fact, the rule connecting it to the result, and the conclusion. For example: “These two angles lie on a straight line, so their sum is 180°. Therefore the missing angle is 180° − 68° = 112°.” Every sentence has a mathematical role.
Not every explanation requires a long paragraph. “The denominator cannot be zero, so x ≠ 3” may be sufficient for a local domain restriction. Conversely, “because of the formula” is not automatically sufficient when the task asks why that formula applies to this situation.
In a model, explanation may require a limitation rather than another calculation. If a journey time is found using a constant speed, a sensible limitation is that stops or changes of speed are not represented. This is not the same as declaring the arithmetic wrong. The calculation can be correct within a model whose assumptions have a restricted reach.
Avoid collecting impressive mathematical words around a weak argument. “Therefore”, “obviously” and “by logic” do not create a missing connection. Ask whether another reader can identify the actual relationship that makes the next line follow.
9. Sketch, draw accurately, plot and describe
A sketch communicates important shape and features. An accurate graph drawn on specified axes also has scale and plotting requirements. A verbal description explains behaviour in words. These forms support one another, but they are not interchangeable responses.
Take y = (x − 2)² − 9. A useful sketch shows an upward-opening parabola with minimum point (2, −9), x-intercepts (−1, 0) and (5, 0), y-intercept (0, −5), and symmetry about x = 2. The curve should connect those features consistently. The intercepts and turning point are not optional decorations when they are the features the task requires.
If the instruction instead asks for an accurately plotted graph over a specified interval, follow its axes, scale and plotting directions. A rough curve with approximately placed features may show conceptual understanding without meeting the accurate-drawing instruction.
If the instruction asks to describe the minimum point, drawing an unlabelled curve is incomplete. If it asks to sketch the curve, writing only “it is a parabola” is incomplete. The mathematical content and the communication form both matter.
A student can rehearse the sketch without being given a finished picture: mark the two intercepts, place the minimum point below them on the symmetry line, mark the y-intercept, and draw a smooth curve consistent with all four features. Then compare the picture with the algebra. This is a construction check, not an invitation to estimate exact coordinates from an arbitrary drawing.
10. “In terms of” asks for a relationship with selected variables
Suppose the perimeter of a rectangle is P and its width is w. “Express its length in terms of P and w” asks for a formula whose right-hand side uses those quantities. Starting from P = 2L + 2w gives L = P/2 − w.
There is no need to invent values for P and w. The symbolic relationship is the answer. A student who insists on a final decimal has changed the job from expressing a relationship to evaluating a special case.
Restrictions on the permitted symbols must be read carefully. An answer containing a new undefined letter has not necessarily completed the task. If an intermediate quantity r was introduced, but the answer must be in terms of d, use the relationship between r and d before finishing.
For example, the area of a circle expressed in terms of diameter d is πd²/4. The route uses r = d/2 and A = πr², but the requested final expression should not still depend on an unexplained radius. The mathematics is not finished merely because a familiar formula has appeared.
11. Exact form, units and precision belong to the answer contract
If a positive quantity x satisfies x² = 20, an exact expression is x = 2√5. A decimal such as 4.47 is an approximation. When an exact answer is requested, replacing the exact expression with a rounded decimal does not meet the form requirement.
Conversely, a question can ask for a numerical result to a specified accuracy. In that case, provide the requested rounded result while retaining sufficient precision in the working. Do not treat exact form as an automatic exemption from a numerical reporting instruction.
Units are another part of the contract. A length of 8 and an area of 8 are not fully identified by the same bare numeral. Distinguish metres from square metres, hours from minutes, and a percentage from a decimal proportion. For coordinates, distinguish an x-value from an ordered pair.
Within the 2027 G3 Mathematics and G3 Additional Mathematics syllabuses, SEAB specifies default accuracy for non-exact numerical answers and additional requirements when demonstrating a stated rounded result. Those instructions are not a blanket rule for every PSLE, G1, G2, school or JC task. Read the actual paper and any question-specific instruction. See the G3 Mathematics syllabus, page 4 and G3 Additional Mathematics syllabus, page 5.
For the separate skill of protecting exact values through a calculation, continue to Why I Sometimes Stop a Student from Turning the Answer into a Decimal. Here the narrower question is whether the form delivered matches the form requested.
12. The same curve can ask for a value, a point, an equation or an interval
For learners who have studied differentiation, consider f(x) = x² − 6x + 13. Its derivative is f′(x) = 2x − 6. Completing the square also gives f(x) = (x − 3)² + 4. These relationships allow several tasks to be answered, but they do not make those tasks identical.
“Find the minimum value of f(x)” asks for 4. “Find the coordinates of the minimum point” asks for (3, 4). “Find the value of x at the minimum” asks for 3. Giving the right number from the wrong column of this relationship is still a target error.
“Find the gradient when x = 1” asks for f′(1) = −4. “Find the equation of the tangent when x = 1” requires both that gradient and the point (1, 8). The equation is y − 8 = −4(x − 1), or y = −4x + 12. Stopping at −4 leaves the equation task unfinished.
“Find where the function is decreasing” asks for the input interval on which the derivative is negative. Solving 2x − 6 < 0 gives x < 3. It does not ask for the decreasing values of y or for a single negative gradient.
This example is particularly useful because the calculations are short enough to expose the language. A learner can understand differentiation yet confuse gradient, point, equation and interval. Repeating harder differentiation exercises would not directly repair that distinction.
Calculator permissions also depend on the assessment. The 2026 H2 Mathematics syllabus, page 4 explains the role of graphing calculators and distinguishes situations in which supporting mathematical steps are required. Do not replace those specific instructions with a universal slogan that every answer must be obtained without technology.
13. Audit an answer before auditing every arithmetic line
Consider this response to “Find the minimum number of 26-seat buses for 112 pupils”: “112 ÷ 26 = 4.307692…, so 4.31.” The division is not the first problem. The answer is the wrong kind of object. The repair begins by interpreting the capacity decision, not by doing the same division more carefully.
Consider “Solve (x − 2)(x − 3) = 0”, followed by “The expression is already factorised.” That observation is true but does not supply the requested values. The missing operation is to use the zero-product relationship and state the solutions.
Consider “Show that all consecutive odd pairs have a sum divisible by 4”, followed by five successful examples. The arithmetic may be flawless. The evidence still has the wrong scope. Replace repeated verification with the general representation 2n + 1 and 2n + 3.
Finally, consider “Find the equation of the tangent at x = 1”, followed by “The gradient is −4.” This is useful progress, not a complete response. The repair is to obtain the point on the curve and form the line equation, not to discard the correct derivative.
These distinctions support precise feedback. Instead of writing only “careless”, identify the first mismatch: wrong object, missing condition, incomplete output, unsupported generalisation, or incorrect answer form. The existing Secondary 1 guide to communication, working and checking develops the wider presentation and verification habits.
14. Independent practice: decide the output before calculating
For each task, first name the required output in a few words. Then answer it. The set deliberately repeats mathematical structures with different instructions. That repetition is the exercise: notice which part of the job has changed.
1. Simplify 4(2x − 3) + 5x.
2. Factorise 12y + 18 fully.
3. Solve 12y + 18 = 0.
4. Evaluate 2t² − 3 when t = −4.
5. List all positive integers n satisfying 2n + 1 ≤ 12.
6. State the number of values in your answer to Question 5.
7. Find the greatest integer k satisfying 4k < 29.
8. State the gradient of y = 3x − 8.
9. Find the coordinates of the point where y = 3x − 8 meets the x-axis.
10. Prove that the product of two consecutive integers is even.
11. Disprove the claim that all rectangles with perimeter 20 cm have the same area.
12. Show that x² − 4x + 7 = (x − 2)² + 3. Hence state the minimum value of x² − 4x + 7 for real x.
13. An invented activity charges a fixed booking fee of $12 and $7.50 per participant. With a total budget of $74, find the maximum whole number of participants. There are no other charges.
14. A quantity increases from 80 to 100. Find the percentage increase. Then find the percentage decrease needed to return from 100 to 80.
15. Solve x² = 9 over the real numbers. Then state only the positive solution.
16. A student claims that 194 × 49 is approximately 1,000. Explain, using estimation rather than a full multiplication, why this is unreasonable.
15. Worked answers and the contract each answer satisfies
1. An expression: 13x − 12. Expand to obtain 8x − 12 + 5x, then collect the x-terms. There is no equation to solve. A statement such as x = 12/13 would answer a different, invented problem.
2. A product: 6(2y + 3). The greatest common numerical factor is 6. Writing 2(6y + 9) is a factorisation, but it is not fully factorised over integer coefficients because the bracket still has a common factor of 3.
3. A value of y: y = −3/2. Subtract 18 to get 12y = −18, then divide by 12. The expression 6(2y + 3) can help, but the solving task ends with the value satisfying the equation.
4. A number: 29. Substitute with brackets: 2(−4)² − 3 = 2(16) − 3. The negative input is squared before the multiplication and subtraction. This is evaluation at a supplied input.
5. A list: 1, 2, 3, 4, 5. The inequality gives n ≤ 5.5. The words “positive integers” select the five listed values. Neither 0 nor 5.5 belongs in the answer.
6. A count: 5. This answer counts the allowed values from Question 5. The greatest listed value also happens to be 5, but that coincidence does not make “count” and “greatest value” the same operation.
7. An extreme integer: 7. Dividing by 4 gives k < 7.25. The integer 7 satisfies the condition, while 8 does not. The boundary 7.25 is not the requested integer.
8. A gradient: 3. In y = mx + c, the coefficient m gives the gradient. The intercept −8 is a different feature, and a coordinate pair has not been requested.
9. An ordered pair: (8/3, 0). On the x-axis, y = 0. Therefore 0 = 3x − 8 and x = 8/3. Supplying only 8/3 omits the coordinate form explicitly requested.
10. A general argument. Of any two consecutive integers, one is even. Their product therefore has an even factor and is divisible by 2. Alternatively, write the pair as 2m and 2m + 1, or as 2m + 1 and 2m + 2, and show the factor of 2 in either case. One numerical example would not establish the universal claim.
11. A valid counterexample pair. A 4 cm by 6 cm rectangle has perimeter 20 cm and area 24 cm². A 3 cm by 7 cm rectangle has the same perimeter and area 21 cm². Both satisfy the premise, but their areas differ, so the claim is false.
12. An identity followed by a minimum value. Expanding (x − 2)² + 3 gives x² − 4x + 4 + 3 = x² − 4x + 7. Since (x − 2)² ≥ 0 for real x, the minimum value is 3, attained when x = 2. The point (2, 3) is consistent with the result, but the requested minimum value is 3.
13. A whole-number capacity decision: 8 participants. After the fixed fee, $62 remains. Dividing by $7.50 gives 8.2666…. Eight participants cost $12 + $60 = $72. Nine cost $79.50 and exceed the budget. The neighbouring checks establish the maximum under the stated pricing model.
14. Two directed percentages: 25% increase and 20% decrease. The increase is 20 relative to the original 80, giving 20/80 × 100% = 25%. The return decrease is 20 relative to 100, giving 20%. The two instructions use different reference quantities.
15. A solution set, then a selected solution. Over the real numbers, x = −3 or x = 3. The positive solution alone is x = 3. The condition in the second request changes what should be reported, not the algebraic fact that both numbers square to 9.
16. An estimation argument. Since 194 is close to 200 and 49 is close to 50, the product should be close to 200 × 50 = 10,000. A value near 1,000 is about an order of magnitude too small. The explanation meets the request without replacing estimation with the exact multiplication.
16. A teaching routine that separates reading from calculation
Begin with a matched pair of questions rather than a large mixed worksheet. Ask the learner to identify the requested output without solving either question. “Find the gradient” versus “find the tangent equation” is useful for a calculus learner; “how many full groups?” versus “how many containers are needed?” is useful much earlier.
Next, let the learner solve one question and adapt the result to the other. This reveals which working can be shared and which final step must change. The aim is not to teach a longer written ritual. It is to make the dependence between instruction and response visible.
Finally, ask the learner to write a new instruction that would make an incomplete answer complete. A gradient of −4 is incomplete for a tangent-equation question, but complete for “Find the gradient at this point.” This reversal tests whether the learner understands the mathematical object rather than merely recognising an answer format.
For parent review, one question is often enough: “What kind of thing did this question ask you to produce?” Allow the child to point to the wording. Avoid supplying the missing interpretation immediately; the useful evidence is whether the child can recover it from the question.
When the issue is arithmetic, use the appropriate topic guide. When the arithmetic is sound but the output is wrong, practise task distinctions. When a long question changes what may be used from part to part, move to the linked-parts chapter. Different errors deserve different repairs.
A two-part instruction needs a two-part finish
A task may say, “Find the value and explain whether it is reasonable.” A numerical result alone answers only the first clause. The second asks for a comparison with the conditions of the problem. Equally, an explanation that the answer should be positive does not replace the requested calculation.
For an original example, suppose a model predicts that 7 identical packets cost $31.50. A student divides correctly and finds $4.50 per packet. If the question also asks whether $40 is enough for 9 packets, the response must continue: 9 × $4.50 = $40.50, so $40 is not enough in the stated fixed-price model. The unit price was an intermediate answer, not the whole task.
Read conjunctions carefully. “Find and explain”, “state and justify”, and “calculate and compare” each request more than one output. Tick the requested clauses mentally before stopping. This is not an invitation to add unrelated commentary; it is a way to avoid leaving a required mathematical job unfinished.
A useful practice variation is to hide the final instruction after showing the data. Ask a learner to propose three different questions the data could answer, then write a suitable final response for each. The exercise separates the mathematical situation from the particular object the examiner or teacher requests.
17. Continue through the School Mathematics manual
The next chapter, Data Sufficiency, Unique Answers and Counterexamples, asks whether the information actually determines the target. Linked Question Parts, Hence and Result Handoffs explains how instructions connect earlier and later results. Rounded Data, Thresholds and Guaranteed Conclusions examines what approximate information can establish.
Return to the BTT Mathematics Hub: School Mathematics Operating Manual for the four-chapter route, or use the Mathematics Article Directory to find an existing topic, examination or stage guide.
18. Sources, assessment boundaries and use
The 2026 PSLE formats, 2027 SEC G3 syllabuses and 2026 A-Level syllabuses are official reference doorways, checked for this edition on 6 September 2026. They concern different examinations and cohorts; one paper’s instructions should not be transferred automatically to another.
All worked cases, practice questions and teaching routines here are original BTT material. The command-word explanations are a learning framework, not an official marking glossary. No mark allocation, examination prediction or endorsement by MOE, SEAB or Cambridge is claimed. Select the examples appropriate to the learner’s actual course and prerequisite knowledge.