Full chapter index

Children often begin with the numbers because numbers look actionable. Yet a word problem contains relationships before it contains operations. If 3/8 of a tank is filled, the filled amount cannot exceed the full capacity. If Ali has fewer marbles than Ben, Ali’s number cannot be larger. These facts can be written before any arithmetic.

A guardrail eliminates models that violate the story. It is not a substitute for solving. Knowing that an answer must be below 120 does not tell us whether it is 45 or 75. It simply prevents 180 from being accepted after a neat but inappropriate multiplication.

The question “What can it never be?” is deliberately concrete. Some children find “state the constraints” abstract. They can usually identify an impossible answer: 2.5 buses when whole buses are counted, a negative distance, or a 70 cm remainder from a 60 cm piece. From that impossibility, the relevant rule can be named.

Parents should keep the tone exploratory. An impossible answer is evidence about a method, not evidence that the child is “not a maths person”. Ask which line first crossed the guardrail. The error may be an operation, a reversed relationship, a unit conversion or an assumption. Each requires a different repair.

Present this invented problem: “A box holds 48 identical packets when full. It is 3/4 full. How many packets are inside?” Before calculation, ask for two things the answer cannot be. It cannot exceed 48, and it cannot be a non-whole packet under the stated identical-packet context. Then the child may calculate 3/4 × 48 = 36.

If the child calculates 64 by dividing 48 by 3/4 and accepts it, the whole-part relationship is reversed. If the child obtains 36 but cannot state why 64 is impossible, the method may be remembered without conceptual checking. If the child says “below 48” but chooses an unreasonable operation anyway, the issue is using the guardrail during solution, not identifying it.

Now change the question: “Thirty-six packets are 3/4 of a full box. How many packets fit when full?” The answer is now 48. The same numbers appear, but the known quantity changed from whole to part. Ask how the impossible-answer statement changes: the full capacity cannot be less than 36.

One pair of problems is enough for the first diagnosis. Do not infer a general PSLE weakness from a single slip. Repeat later with money or length. Record whether the child can identify the reference whole, write a boundary, calculate, and check the result against it. These four actions locate the next teaching step.

Words such as *total*, *remaining*, *more than*, *less than*, *at most*, *at least*, *each* and *altogether* constrain a model. “At most 40” permits 40 and values below it. “Fewer than 40” excludes 40. A child should paraphrase these boundaries before using numbers.

Create a small symbol bridge where appropriate: *less than* becomes <, *at most* becomes ≤, and *at least* becomes ≥. Symbols are useful only if their meaning remains connected to the story. Ask for a spoken version after writing one. This catches reversed signs that look mathematically tidy.

Some constraints are structural rather than stated through comparison words. A rectangle’s opposite sides are equal. A remainder is smaller than the divisor. An average lies between the smallest and largest values unless weights or definitions change the situation. A percentage of a fixed whole connects the part and reference amount.

Make a “must / cannot / unknown” box. For each problem, record one fact in each category. This prevents assumptions from becoming rules. If two pupils share stickers, it may be unknown whether they receive equal amounts unless the question says so. Good constraint reading distinguishes given structure from imagined fairness.

“A library displayed 60 new books. Two-fifths were fiction. How many fiction books were displayed?” The fiction group is a part of the 60 books, so it cannot exceed 60. Since books are counted individually, the answer should be a whole number. Two-fifths of 60 is 24.

An answer of 150 reveals a likely division by 2/5. The arithmetic 60 ÷ 2/5 is correct, but it answers a different structure: 60 is two-fifths of what whole? The guardrail exposes the mismatch without claiming the child cannot divide fractions.

Change the context: “Sixty fiction books are two-fifths of all the new books.” Now 150 is permitted because 60 is the part and the unknown is the whole. Ask the child to draw bars for both versions. The position of 60 in the model decides which answer can be larger.

Extend carefully: a part can equal the whole when the fraction is 1, and can exceed a stated reference when the relationship is not part-of-that-whole. “Sales this year are 120% of last year’s sales” permits a larger amount. Avoid turning “part never exceeds whole” into a slogan used outside its defined relationship.

“A tank is 65% full and contains 130 litres. What is its capacity?” Capacity must exceed 130 litres because the current amount is less than 100% of capacity. Since 65% corresponds to 130 litres, 100% corresponds to 200 litres. The inequality was known before the exact answer.

If the child multiplies 130 by 0.65 and obtains 84.5 litres, the result violates the boundary. Ask which quantity represented 100%. The repair concerns reference, not decimal multiplication. Once the whole is identified, the operation becomes meaningful.

Now ask: “The amount increases from 130 litres by 65%.” The final amount can exceed 130 because 130 is the original reference, not a capacity of which the final amount is a part. It becomes 214.5 litres in a purely numerical context. Similar percentages can describe different relationships.

Use language deliberately: *65% of*, *increased by 65%*, and *is 65% more than* are not interchangeable. Ask what the final answer must be relative to the reference—less, equal or greater—before calculating. This relational prediction is more precise than a vague instruction to “check if it makes sense”.

“A teacher packs 157 cards equally into bundles of 12. How many full bundles and how many cards remain?” The remainder must be a whole number from 0 to 11. It can never be 12, because another full bundle could then be made. Dividing gives 13 full bundles with 1 card remaining.

A child might write 12 bundles remainder 13 after stopping too early. The total can be reconstructed—12 × 12 + 13 = 157—but the remainder violates its definition. Repack one additional bundle to reach the canonical form. The guardrail does more than verify the total; it verifies the representation.

Change the question to the number of boxes needed to transport all cards when each box holds at most 12. Thirteen full boxes are not enough for 157 cards; 14 boxes are needed, with the last not full. The quotient-and-remainder information must be interpreted in context. Rounding down would leave a card without a box.

Ask two checks: does divisor × quotient + remainder reconstruct the total, and is 0 ≤ remainder < divisor? Both matter. A reconstructed total alone can hide a nonstandard remainder. This is a good example of how a mathematical definition creates an impossibility boundary.

“Eight pupils share 30 identical oranges equally. How many oranges does each receive?” If oranges must remain whole, each receives 3 whole oranges and 6 remain, unless the context allows cutting. If oranges may be divided, each receives 3.75 oranges. The object and instruction determine the allowable answer type.

Students sometimes apply “answers must be whole numbers” too broadly. Length, mass, time and money can take fractional or decimal values within their units. Counts of indivisible objects normally require whole numbers. But people can share a pizza in fractions, and a rope can be cut. Context decides divisibility.

Add units to the guardrail. A calculated 2.5 cannot be evaluated until we know whether it means buses, kilometres, hours or dollars. If the question asks for buses needed, 2.5 buses is not a feasible count and may require rounding up. If it asks for travel time, 2.5 hours may be perfectly valid.

Have the child label the answer type before calculating: count, measure, money, time, proportion or rate. Then state whether decimals and fractions are permitted and how rounding would be justified. This reduces automatic rounding based solely on the appearance of a decimal.

“A group buys identical notebooks costing $3.80 each with $25. How many can they buy, and how much remains?” Seven notebooks would cost $26.60 and are impossible. Six cost $22.80, leaving $2.20. The spending boundary is at most $25, and the notebook count is whole.

An answer of 6.58 notebooks is a useful intermediate quotient but not the requested purchase count. Rounding to seven by the usual nearest-whole rule would exceed the budget. The context determines rounding direction. The child should explain why six is the greatest feasible whole count.

Now reverse the task: “How much money is needed to buy seven notebooks?” The answer is exactly $26.60. The previous budget no longer constrains the result unless it remains part of the question. Do not carry guardrails from one problem into another after the conditions change.

Add a discount or voucher only after the basic model is clear. Ask whether the discount applies per notebook or to the total, and whether the voucher can create a negative payment. The aim is not to make the arithmetic messy. It is to show that rules governing a transaction define which numerical results are possible.

“A rectangle has perimeter 50 cm and length 16 cm. Find its width.” Since 2(length + width) = 50, length + width = 25 and width = 9 cm. Before calculating, width must be positive and, under the stated values, less than 25 cm. It need not be less than the length for every rectangle, though it is here.

An answer of 17 cm gives perimeter 66 cm, violating the given total. An answer of −9 cm may result from a reversed subtraction and violates physical length in this context. Substitute dimensions into the perimeter relation rather than merely checking whether the number “looks reasonable”.

For a triangle, each side must be shorter than the sum of the other two. If lengths are 4 cm, 7 cm and 12 cm, they cannot form a non-degenerate triangle because 4 + 7 is not greater than 12. This impossibility can be recognised before area or perimeter work.

Avoid inventing diagram properties. A sketch that looks like a square does not establish equal sides unless markings or statements do. Constraints come from definitions, labels and givens, not visual appearance alone. Ask, “Which fact authorises this equality?” That question protects against one of the most common diagram assumptions.

“A film starts at 10:45 p.m. on Friday and ends at 12:20 a.m. on Saturday.” The finish clock reading is smaller numerically, but the event ends later because the date changes. Elapsed time cannot be negative. From 10:45 p.m. to midnight is 1 hour 15 minutes, then 20 minutes more, totalling 1 hour 35 minutes.

Subtracting 10:45 from 12:20 as same-day times can create confusion. Draw a timeline and place the day boundary. The constraint is chronological: end follows start. A clock face repeats every 12 hours, so the date or sequence supplies information the clock reading alone lacks.

Now ask for a possible latest start time if an activity must finish by 6:00 p.m. and lasts 1 hour 40 minutes. The start cannot be after 4:20 p.m. if the full duration is required. Work backwards and then verify forwards.

Time questions may include waiting, travel and breaks. List which intervals count towards the requested total. “Journey time” might exclude a later unrelated activity, while “time from leaving home to arriving” includes waiting. The impossible-answer check depends on the defined interval, not a generic rule about all times in the story.

“The ratio of red to blue beads is 3:5. There are 40 beads altogether. How many are red?” Eight equal parts represent 40, so one part is 5 and red beads number 15. Red must be less than blue and less than the total under this ratio.

An answer of 24 red beads may come from treating 3/5 as red’s fraction of the total. But 3:5 means red is 3 out of 8 combined parts, not 3 out of 5. The answer 24 would also make red larger than blue if 16 remain, contradicting 3:5.

Now state “red beads are 3/5 as many as blue beads”. This expresses the same comparison: red:blue = 3:5. But “red beads are 3/5 of all beads” gives red:blue = 3:2. Ask the child to identify what the denominator refers to—blue or total.

Use the boundary before the model: when red:blue = 3:5, red cannot be half or more of the total. This does not supply the exact count, but it immediately rejects a wrong interpretation. Diagram and inequality should support each other.

“Four quiz scores are 12, 15, 16 and 17. What is the mean?” Their total is 60, so the mean is 15. The mean must lie between 12 and 17. An answer of 60 reports the total, while an answer of 20 cannot be the arithmetic mean of these four values.

The range guardrail is useful but incomplete. A number inside the range is not automatically correct. Fourteen is possible-looking yet wrong. Reconstruct the total: mean × number of values should equal the sum. This second check tests the exact relationship.

If an additional score is above the current mean, the new mean must increase; if below, it must decrease. The size of the change depends on the score and number of values. Ask for direction before calculation. A correct computation that moves the mean the wrong way signals a setup or arithmetic error.

Weighted averages require care because groups may have different sizes. A simple average of two group means is valid only when the groups have equal numbers or the weighting happens to justify it. The constraint comes from total quantities, not the visual symmetry of two numbers.

Options often reveal different misconceptions. Before calculating fully, eliminate answers that violate units, part-whole order, parity or size. If 7 identical boxes contain an equal whole number of objects, the total must be divisible by 7. An option that is not divisible cannot fit the stated arrangement.

Elimination should be explained, not treated as pattern spotting. “This option is too large because it exceeds the original whole” is stronger than “It looks wrong”. If two options remain, perform the necessary calculation. Constraints reduce the search; they do not license guessing among survivors.

Be alert to options that are intermediate values. A question may ask for the amount remaining, while one option is the amount used. Label the target before solving. A value can satisfy many mathematical boundaries and still answer the wrong question.

Create a practice item with four options: one wrong unit, one reversed ratio, one arithmetic slip and one correct answer. Ask the child to identify the likely error behind each. Understanding distractors can reveal model choices without teaching suspicion of every option.

Before equations, write a short boundary such as “0 < width < 25”, “part < whole”, “remainder 0–11”, or “boxes are whole and round up”. Then draw the bar model, table, number line or diagram. The written constraint becomes a test throughout the solution.

When a line produces a violation, stop at that line. Do not finish three pages of algebra hoping the final answer recovers. Check whether the equation represented the story. This habit saves time after it is learned, though it may initially feel slower than immediate calculation.

Some problems have several simultaneous constraints. A number may need to be even, between 30 and 50, and divisible by 3. List all three. Their intersection narrows possibilities to 30, 36, 42 and 48 if the lower boundary includes 30. Read wording carefully to decide inclusion.

Teach the child to update constraints when new information appears. If a later statement says the number is greater than 40, remove 30 and 36. Mathematics problem solving is a controlled revision of the possibility set, not a commitment to the first guess.

Problem A: “A jug holds 2 litres when full and currently contains 750 millilitres.” The amount cannot exceed capacity, and units must be aligned. The empty space is 2000 − 750 = 1250 millilitres. An answer of 2750 millilitres would describe adding rather than remaining capacity.

Problem B: “A number leaves remainder 4 when divided by 7.” The number can never have remainder 7 or a negative remainder in the usual whole-number division context. Candidate values take the form 7k + 4. If the number lies between 30 and 50, possibilities are 32, 39 and 46.

Problem C: “A shop sold fewer than 3/5 of 200 tickets.” The number sold must be less than 120 and whole. Without more information there is no unique exact answer. Recognising insufficiency is part of constraint reasoning. Do not invent equality because the numbers invite multiplication.

Problem D: “The average of five positive whole numbers is 8.” Their total must be 40, but individual numbers are not uniquely determined. If one number is 30, the remaining four positive whole numbers total 10, which is possible. Ask what is fixed, what is bounded and what remains unknown.

For reversed part-whole errors, use paired questions with the same numbers and switch which quantity is known. Draw bars and state whether the unknown must be larger or smaller. Delay symbolic shortcuts until the relationship is stable.

For unit errors, require an answer-type line before arithmetic and a unit on each meaningful intermediate value. Convert only when quantities must be combined or compared. Ask whether the final unit matches the question, not whether the page contains a conversion somewhere.

For impossible decimals in count contexts, discuss divisibility and rounding purpose. Sometimes a decimal shows that the proposed equal distribution cannot occur without splitting objects. Sometimes it is an intermediate quotient that must be rounded up for containers. The context decides the response.

For children who know constraints but ignore them under time pressure, place the boundary beside the answer box. Practise a three-second final comparison: sign, size, type and unit. Reduce scaffolding gradually. The target is an internal habit, not dependence on a printed checklist.

Choose one problem and spend two minutes listing must, cannot and unknown facts. Spend five minutes solving. Spend two minutes testing the answer against every boundary. Use the final three minutes to change one condition and predict how the possibility set changes.

On the next day, use a different domain—money after fractions, geometry after money. This checks whether the question “What can it never be?” transfers beyond a memorised topic. Keep arithmetic moderate while the reasoning habit is new.

Ask for explanations in ordinary language: “It cannot be 64 because 48 is the full capacity in this version.” Clear relational speech often precedes clear equations. Correct the model before polishing mathematical vocabulary.

Stop after the selected problem. Keep the original and corrected line. At the end of a week, ask the child to sort mistakes by relationship, unit, answer type or arithmetic. This small error map is more useful than counting pages completed.

The child begins to predict direction and range before calculation. They reject a larger part, a remainder equal to the divisor, a fractional bus and a negative length without waiting for an answer key. More importantly, they can say which condition is violated.

Calculation becomes connected to verification. The child substitutes dimensions into a perimeter, reconstructs a quotient, checks mean × count, or tests total spending. A result is not accepted merely because the calculator or written method produced it.

Progress may first appear as self-correction rather than fewer initial errors. Catching a reversed model independently is valuable. Over time, the boundary should influence setup earlier, reducing wasted work. Track that movement from after-the-fact check to before-solving design.

Use an observable goal: “On three unfamiliar word problems, state one boundary, solve, and explain whether the answer fits.” Do not equate success with a particular score from a tiny practice set. Stable reasoning across new contexts is the stronger sign.

If a vehicle travels at a positive speed for a positive duration, the distance must be positive. When two objects move towards each other, their separation should decrease until they meet. When they move apart, it should increase. State the expected direction before selecting a speed relationship.

Example: two walkers start 18 km apart and approach each other at 4 km/h and 5 km/h. Their combined closing speed is 9 km/h, so they meet after 2 hours. An answer of 18 hours ignores the rate relationship; an answer of −2 hours violates the time context.

If one walker starts an hour later, do not add speeds immediately across the whole interval. First account for the early walker’s movement and the changed separation. The start-time constraint divides the journey into phases. A single formula used without a timeline can combine quantities that do not operate simultaneously.

Ask the child to sketch positions at the start, after any delay and at the meeting. The diagram should agree with the calculation. A meeting point outside the original interval is impossible when both walkers travel towards each other from its ends without turning.

A person’s age increases by the same number of years as another person’s over the same time. Therefore, the age difference stays constant. If a parent is 28 years older than a child now, the difference remains 28 years in ten years, although the ratio changes.

Consider: Mei is 10 and her aunt is 34. In how many years will the aunt be twice Mei’s age? Let the time be x: 34 + x = 2(10 + x), giving x = 14. The future ages 24 and 48 satisfy the ratio and preserve the difference of 24 years.

A negative time can be meaningful if a question asks how many years ago, but it must be interpreted. If solving a future-tense question produces −3, the stated event occurred in the past and no future solution fits. The sign is evidence about the model and wording.

Use two exact checks: preserve the age difference and test the requested ratio. A result can satisfy one but not the other after an arithmetic mistake. Avoid invented assumptions about birthdays unless the question requires precise dates; standard school problems usually state enough to use the intended whole-year model.

Increasing a rectangle’s length while keeping width positive increases both area and perimeter, but not by the same rule. Area changes by width × change in length. Perimeter changes by twice the change in length. Predicting both directions prevents swapping formulas.

Take a 12 cm by 7 cm rectangle. Increase length by 3 cm. Area rises from 84 cm² to 105 cm², an increase of 21 cm². Perimeter rises from 38 cm to 44 cm, an increase of 6 cm. Units distinguish the quantities as well as the values.

If a question fixes perimeter, increasing length forces width to decrease. Area may increase or decrease depending on the starting dimensions. Do not carry the earlier “area must increase” guardrail into a changed condition. Constraints belong to the exact scenario.

Ask the child to draw two rectangles and label unchanged and changed quantities. Then substitute dimensions into both formulas. Visual appearance alone is insufficient, but a labelled diagram makes dependencies easier to track and exposes impossible negative widths.

A pattern rule must fit every stated term, not just the first pair. If a sequence is 5, 8, 11, 14, a constant increase of 3 fits all shown transitions. A proposed “add 3, then add 4” rule fails at the third term unless the pattern description supplies alternating changes.

Several rules can fit a short finite list. The expected rule in a school task usually follows the represented structure or simplest consistent pattern, but the child should use all available diagrams and words. Do not claim uniqueness from too little information.

For a growing arrangement of squares, count what is added at each stage. If stage 1 has 4 tiles and each new stage adds 3, stage n has 4 + 3(n − 1). The count cannot be negative, and stage number is normally a positive whole number.

Test a formula at an early stage and one later stage. A formula that fits stage 1 automatically because of construction may still fail elsewhere. Substitution is the exact constraint check. It turns a guessed general rule into one tested against the pattern.

“A snack costs either $2 or $3, and exactly five snacks cost $12.” Let x be the number of $3 snacks. Then 3x + 2(5 − x) = 12, so x = 2. Counts must be whole and between 0 and 5. These boundaries matter before solving.

An algebraic answer x = 7 would be impossible even if produced by a calculation because there are only five snacks. A decimal x = 2.5 cannot represent a count of whole snacks under this setup. The context gives both lower and upper bounds and an answer type.

For small problems, make a table of possibilities. Zero $3 snacks costs $10; one costs $11; two cost $12. A table is not an inferior method when it is systematic, complete and connected to the constraints. It can reveal structure before algebra is introduced.

Ask whether another solution exists. Since replacing a $2 snack with a $3 snack raises the total by exactly $1, each count creates a different total. Therefore, only two $3 snacks fit $12. Explaining uniqueness deepens the solution beyond finding a number.

Layer one checks representation: did the bar, table, diagram or equation match the wording? Layer two checks arithmetic. Layer three checks context—sign, size, type and unit. Layer four checks the defining relationship by substitution or reconstruction. A single tick cannot cover all four.

Suppose a child solves a ratio problem and gets 18. Arithmetic may be flawless, but if 18 represents the larger group where the ratio makes it smaller, layer three fails. If group totals do not reconstruct the stated whole, layer four fails. The routine identifies what must be repaired.

Teach the child to choose the cheapest decisive check. Recalculate only when arithmetic is suspect. Substitute when a formula is used. Rebuild the total when parts are found. Compare with the boundary when the answer type is the issue. Efficient checking is targeted, not repetition of every line.

During timed practice, write four tiny initials beside the answer—R, A, C, D—and use only those relevant. Gradually remove the marks. The goal is flexible verification that becomes internal, not a checklist so long that it competes with solving.

Extended practice: one problem, several impossible answers

Use this original problem: “A rectangular noticeboard has area 96 cm². Its length and width are whole numbers, and the length is greater than the width. The perimeter is less than 50 cm. Find possible dimensions.” The answer is not obtained by one immediate operation. Constraints define a search.

List factor pairs of 96: 1 by 96, 2 by 48, 3 by 32, 4 by 24, 6 by 16 and 8 by 12. Reverse orders are unnecessary because length is greater. The first pairs have very large perimeters. Calculate only as needed: 2(4 + 24) = 56 fails, while 2(6 + 16) = 44 and 2(8 + 12) = 40 fit.

The board therefore has more than one possible dimension pair under the given information. Recognising multiple solutions is not a failure. A child who reports only 8 by 12 may have stopped after finding a fit. Ask whether another factor pair survives all constraints.

Now add “the length is 10 cm more than the width”. The pair 6 by 16 fits the new difference, while 8 by 12 does not. Additional information narrows the possibility set to a unique solution. This demonstrates why every condition matters.

Ask which proposed answers are impossible and why. A 5 by 19.2 board violates the whole-number-dimension condition. A 12 by 8 ordering violates the named length-greater-than-width convention if the labels are assigned incorrectly. A 3 by 32 board fits area but violates perimeter.

Then change “perimeter less than 50” to “perimeter at most 50”. In this case the same two listed pairs still fit, but the boundary language differs. Create a new area where equality matters so the child sees that less than and at most are not interchangeable.

Have the child design one extra condition that selects 8 by 12 without directly stating either dimension. “The difference between length and width is 4 cm” works. Designing a constraint requires understanding how information eliminates alternatives.

Finish with a verification table containing area, order, perimeter and added condition. This is a complete proof of fit for a small possibility problem. It models systematic search rather than random trial and shows exactly why rejected pairs cannot be answers.

Repeat the method with a number puzzle: a two-digit whole number is greater than 40, divisible by 6, and has digits that add to 9. Multiples of 6 above 40 include 42, 48, 54, 60 and so on. Testing the digit condition leaves 54 among the early candidates; continue only within any stated upper bound.

The example shows why bounds must be complete. If no upper limit is given, 54 is not automatically the only possible answer; 72 and 90 also have digit sum 9 and are divisible by 6. A question claiming uniqueness needs enough information. Children should not stop because the first satisfying value looks tidy.

Add “less than 70” to select 54. Then ask the child to explain each rejection rather than list crossed-out numbers. Forty-eight fails the digit sum, 60 fails it, and 72 exceeds the bound. Systematic elimination is mathematical reasoning, not guessing.

Design a different final condition that also selects 54, such as “the tens digit is three more than the ones digit”. Comparing conditions shows that information can be redundant or decisive. The child begins to see a problem as an intersection of allowable sets.

Bring the complete problem, working and answer. Ask the tutor to identify whether the breakdown is in reading the relationship, representing it, calculating, interpreting the quotient, or checking constraints. Correct arithmetic with a wrong model needs different teaching from a multiplication slip.

Ask how practice will move from explicit guardrails to independent use. A sensible route might include paired problems, diagrams, inequality statements, mixed domains and unfamiliar transfer. Repeating many near-identical questions can create fluency without showing whether the child knows when the method applies.

Confirm level, materials and current arrangements directly with any provider. A search for PSLE Mathematics tuition in Punggol does not verify fees, schedules, vacancies, class size or a particular programme. This guide makes none of those claims and offers no result guarantee.

Review new work for boundary statements, appropriate models and exact checks. Ask the child why an attractive wrong answer cannot fit. Confidence should rest on explained relationships, not on the speed with which a familiar operation was selected.

Is this just estimation?

No. Estimation predicts approximate size. Constraint reasoning identifies what is permitted or impossible by definitions and relationships. They work well together, but a remainder rule or whole-number count can reject an answer even without estimating its approximate value.

Will writing constraints waste examination time?

At first it may add seconds. With practice, a short note can prevent a long wrong method and speed option elimination. The note should be concise and used where it clarifies the problem, not turned into an elaborate ritual for every basic calculation.

Can a part ever be greater than a whole?

Not when it is literally a part of that specified whole. But quantities described as 120% of an earlier reference can exceed that reference, and an unknown whole can exceed a known part. Define the relationship before applying the rule.

Are decimal answers wrong in PSLE Mathematics?

No. Decimals are appropriate for many measures and money amounts. They may be infeasible for counts of indivisible objects or may need context-specific rounding. The answer type and question wording decide.

What if several answers satisfy the boundaries?

Then the constraints are not enough to determine a unique value, or further information has not been used. Boundaries filter possibilities; the complete relationships and calculation select the answer. Sometimes recognising that information is insufficient is itself correct reasoning.

What is the best final check?

Ask whether the answer has the right sign, size, type and unit, then verify the defining relationship. Reconstruct the total, substitute into the formula, or compare part with whole. A verbal “looks okay” is weaker than an exact check.