G3 SEC English · Mathematics · Science | Learner’s Guide Vol 0101. Decide what the supplied evidence is sufficient to establish, with original exercises and worked answers.
A student has answered most of a problem correctly, then hesitates. Is there enough information to finish? In English, the passage may support a clear inference without naming a character’s motive outright. In Mathematics, the numbers may be sufficient for an exact calculation but not for a broader comparison. In Science, a graph may support a trend while leaving two explanations open. Knowing when an answer is justified is a skill in its own right.
This G3 SEC Learner’s Guide Vol 0101 teaches you to distinguish missing information from missing confidence. You will learn to identify the exact question, list the evidence required, separate conclusions that are already supported from claims requiring another check, and finish efficiently without inventing facts. Our examples connect English, Mathematics and Science through a shared reasoning habit. They do not create a separate EMS examination.
The workshop brings together Volume 0097 on branching outcomes, Volume 0099 on misleading graph scales and Volume 0100 on repeat measurements and drift. In each, a plausible answer depended on checking the right condition. Here we ask a further question: what is the minimum information needed to reach the particular conclusion being requested?
Chapter index
- 1. Sufficient for what?
- 2. The current SEC G3 formats, without mixing subjects
- 3. Four ways a question may have enough information
- 4. English: how much can we infer from a gesture?
- 5. Mathematics: the missing value test
- 6. Science: describe a trend without inventing its mechanism
- 7. Stop collecting irrelevant information
- 8. What one additional check should you ask for?
- 9. A complete EMS case: the study-room extension
- 10. When absence of data is itself part of the answer
- 11. Original practice: twelve sufficiency decisions
- 12. The answer clinic: stop at the right boundary
- 13. How to use this under examination pressure
- 14. Seven days to make sufficiency a habit
- 15. A complete answer has a clean edge
1. Sufficient for what?
The word “sufficient” is meaningless without a task. Two measured temperatures may be sufficient to calculate their difference, but not to determine the precise moment a sample began warming. A table may be sufficient to find an average, but not to establish why the average changed. A passage may be sufficient to say a character refused an invitation, but not to determine every private reason for that refusal.
Start by naming the exact output the question asks for. If it asks “Calculate the percentage increase”, look for the original value and new value. If it asks “Explain why the increase occurred”, the two values alone may not be enough. Both tasks concern the same data, yet they require different evidence.
Consider a shop whose recorded daily visits increase from 80 to 100. We can calculate the change as 20 visits, or 25% relative to 80. The arithmetic is determined. But if a poster says the increase was caused by a new signboard, the two totals do not isolate that cause. Other factors may have changed. It would be wrong to refuse the percentage calculation because causation is uncertain; it would also be wrong to treat the calculation as proof of causation.
The first important distinction is between a numerical answer, a descriptive answer, an inference and a causal explanation. Each carries a different burden. Do not demand a perfect experiment to calculate an ordinary fraction. Do not treat an ordinary fraction as sufficient evidence for a causal theory.
In an examination, you can practise a one-line pre-answer check: “I need [two values] to calculate [difference]”, or “I need [the character’s action and context] to infer [attitude]”. That statement reveals whether the evidence you have matches the job you must do.
2. The current SEC G3 formats, without mixing subjects
The 2027 Singapore-Cambridge SEC examines subjects at G1, G2 or G3 as appropriate to the candidate. The official SEAB 2027 G3 list identifies English Language K300, Mathematics K310 and combined Science K326, K327 and K328. There is no single EMS paper, so we use the shared habit within each subject’s own assessment framework.
The English K300 syllabus assigns 35% to Writing Paper 1, 35% to Comprehension Paper 2, 10% to Listening Paper 3 and 20% to Oral Communication Paper 4. The two written papers each last 1 hour 50 minutes. Comprehension includes literal, inferential and evaluative questions and language for effect. Answering with evidence rather than unsupported invention matters in those tasks.
The Mathematics K310 syllabus has two compulsory papers, each lasting 2 hours 15 minutes, worth 90 marks and weighted at 50%. Candidates must show essential working, and the syllabus generally calls for non-exact answers to three significant figures unless otherwise specified. Interpreting results in context and reasoning mathematically are assessed along with standard techniques.
The combined Science syllabus K326–K328 includes a multiple-choice Paper 1 (40 marks, 1 hour, 20%); two relevant structured-and-free-response disciplinary papers (65 marks and 1 hour 15 minutes each, 32.5% each); and a practical Paper 5 (30 marks, 1 hour 30 minutes, 15%). Practical evaluation may ask candidates to identify limitations or suggest suitable improvements.
These facts describe formats. They do not suggest that a named “evidence sufficiency” question will appear. The topic is a general discipline drawn from the official skills of interpretation, explanation, problem-solving and evaluation. Use the G3 SEC learner-guide route for the complete syllabus path.
3. Four ways a question may have enough information
A question may be fully determined. If a train travels 24 km in 30 minutes at constant speed, the average speed over the given interval is 24/(30/60)=48 km/h. You need not ask what colour the train was or which driver operated it. Those facts are irrelevant to the requested calculation.
A question may support a bounded inference. If a character says “I cannot stay” and immediately leaves when the bell rings, the passage supports that departure occurred and that the character said they could not stay. It may suggest urgency, but does not prove the character was angry. Your answer can still be useful and correct if it distinguishes observation from uncertain motive.
A question may permit several outcomes. A Science investigation shows one factor increasing and another decreasing, but supplies no magnitudes needed to find the net effect. The correct response may be that a unique net outcome cannot be determined. That is a reasoned conclusion about the evidence, not an excuse to avoid the problem.
Finally, a question may require an assumption expressly supplied by the problem. In a mathematical model, we often assume a constant rate, equal likelihood or a particular formula because the question says so. Within that model, you can solve correctly even if a real-life situation would contain additional complexity. Do not add imaginary complications that invalidate the stated task.
These four cases can look similar when a learner is anxious. The remedy is to identify the required information for this task, not the information that would be nice to have about the whole world.
A good answer has a boundary. It covers what is asked, explains what can be supported and stops before unsupported claims appear. This is not minimalism for its own sake; it is accuracy with purpose.
4. English: how much can we infer from a gesture?
Read this original passage. “When the winner was announced, Amir clapped twice, then lowered his hands. He had spent the afternoon helping the winner rehearse. As the room emptied, he stayed to collect the abandoned programmes.” What might the writer be showing? We know Amir helped the winner and then tidied the room. We also know he clapped briefly. These facts can support an impression of involvement and quiet helpfulness.
But they do not prove he was jealous, secretly delighted or disappointed. The short applause might be meaningful or merely how the event was described. A question asking “What did Amir do?” needs only the actions. A question asking “What impression of Amir is conveyed?” requires interpretation of those actions in context. A question asking “Why did he clap only twice?” may not have a uniquely supported answer unless more of the passage supplies a clue.
A strong impression answer might say, “Amir appears quietly supportive because he helps the winner rehearse and stays afterwards to collect programmes, although the passage does not directly state his feelings about the result.” The inference is grounded in two actions. It does not claim he is always selfless or has no mixed emotions.
Now imagine a later sentence: “He paused at the doorway, staring at the unused certificate bearing his own name.” This new evidence changes the possible reading of the scene. It suggests an unresolved personal connection with the award or event. But even this does not establish one exact emotion without further context. A reader should update the interpretation appropriately, not replace a cautious answer with a dramatic certainty.
This continues Volume 0094 on narrator knowledge and Volume 0098 on dialogue and action. Those volumes teach how evidence belongs to particular voices and actions. Here the aim is deciding how much can responsibly be concluded from the available clues.
Do not add unsupported alternative interpretations simply to sound thoughtful. If the question asks for the strongest supported impression, choose it and justify it. If the passage remains deliberately ambiguous, explain the ambiguity. The task wording determines the depth.
5. Mathematics: the missing value test
Imagine a question: “A coach travels 90 km. Find the travel time.” Without speed, timetable or another constraint, no unique duration follows. A learner could propose a time only by adding an assumption. If the question instead says the coach travels at a constant 45 km/h, time becomes 90/45=2 hours. The added rate completes the problem.
A useful missing-value check names the relation you plan to use. Distance = speed × time contains three quantities; if only distance is known, a unique calculation of time requires another independent piece of information. But the question might provide a graph rather than a written speed. Search the entire given material before declaring data insufficient.
Sometimes several equations together determine unknowns. Two simultaneous conditions can be sufficient even when neither alone is. For example, two numbers have sum 20 and difference 4. Let the larger be x and smaller y. Then x+y=20 and x−y=4; adding gives 2x=24, so x=12 and y=8. Neither condition alone determines both values, but together they do.
Do not confuse having many numbers with having enough information. Ten unrelated measurements cannot solve for an unknown that none constrains. Conversely, one carefully chosen relationship can unlock a complete solution. This is the practical meaning of identifying useful evidence.
A common examination mistake is to add a hidden assumption. If a problem says two pupils cover the same distance, do not assume they took the same time. If a bag contains different numbers of coloured counters, do not assume the colour outcomes are equally likely. If a chart starts at a non-zero axis, do not treat the displayed bar-height ratio as the underlying value ratio.
Volume 0099 on misleading graph scales and Volume 0095 on changing probability spaces show two different ways an apparently complete answer can depend on an unnoticed assumption.
6. Science: describe a trend without inventing its mechanism
Suppose a reaction’s cumulative gas volumes are 0, 12, 20 and 25 cm³ after successive one-minute intervals. We can say the total collected gas increased, and the increments across those intervals were 12, 8 and 5 cm³. Their average collection rates were 12, 8 and 5 cm³/min. The observations show smaller amounts being added in later intervals.
Can we conclude why the rate fell? Perhaps the reaction slowed as reactants were used, under a suitable chemical model. But the data alone may not establish that mechanism uniquely. The apparatus might affect gas collection, and the full experiment conditions matter. When the question provides relevant theory and conditions, use them. When it does not, describe the trend without claiming the specific cause is proved.
Another example: a thermometer rises from 25 °C to 30 °C after a heater is switched on. It is legitimate to describe the measured increase. Whether the sensor response lagged behind the sample, how uniformly the sample warmed and what temperature it will ultimately reach may remain unknown.
A third example: two students record different results from supposedly identical trials. Without knowing the instruments, sample preparation, time of measurement and variation, you cannot automatically choose which result is correct. An appropriate reference check or repeated comparable trial may help. Volume 0100 on drift provides a more focused diagnostic route.
The important distinction is between lack of evidence for a claim and evidence that the claim is false. If a system’s future value was not measured, it may be impossible to establish its final state from the available data. That is different from proving the value will never be reached.
In a Science paper, a well-formed response might say: “The measurements support a decreasing average rate over the recorded intervals. To determine whether reactant depletion explains it, the experiment would need suitable information about reactant conditions and measurement reliability.” The first sentence answers what is known. The second identifies the precise limitation.
7. Stop collecting irrelevant information
An inexperienced learner sometimes keeps reading or calculating after the decisive answer is already available. If an English question asks which sentence shows reluctance, choose the specific language that conveys reluctance rather than retell the entire passage. If a Mathematics task asks for a percentage change from two stated values, compute it and stop rather than introducing a new scenario. If Science asks for one appropriate control, name the relevant one and explain why.
But stopping too early is also possible. “Because the line went up” may not answer a Science question asking for a mechanism. “He was anxious” may not satisfy an English question requiring textual support. “40” may not be enough for a Mathematics answer if the unit, essential working or interpretation is required.
The correct stopping point depends on the requested output. A calculation answer is complete when the mathematical relation, substitution, result, accuracy and unit are adequate. An inferential comprehension answer is complete when it gives the claimed meaning and relevant textual evidence. An evaluation answer is complete when it identifies a weakness, its consequence and, where requested, a justified improvement.
Imagine an English question: “What effect does the writer create by placing the empty chair beside the unopened letter?” A simple identification of the objects is not enough. The effect may be absence, suspense or uncertainty depending on the surrounding scene. The learner must explain how the details influence the reader, not merely list them.
A useful revision habit is to underline the command word, then write a short checklist beside the task. The checklist should be specific. For “calculate”, you may need numbers, formula, unit. For “explain”, you need evidence and mechanism. For “evaluate”, you need a criterion and its implications. Finish when that list is satisfied.
This approach is calmer than assuming longer answers always earn more marks. More writing can help when it develops evidence, but irrelevant writing may waste time and obscure the conclusion.
8. What one additional check should you ask for?
When evidence is insufficient, identify the smallest useful missing piece rather than writing “more information is needed” without explanation. In Mathematics, it might be the total number of students before calculating a percentage. In Science, it might be a control reading or instrument calibration. In English, it might be the subsequent paragraph that clarifies a character’s motive.
Consider a school report saying “Most students prefer weekend revision” after surveying only one club. To infer a schoolwide preference, we need information about representativeness, the sampling procedure and the wider student population. Asking simply for “more opinions” may not solve the bias if the new opinions come from the same club.
Suppose a graph shows temperature at 0 and 10 minutes, but the question asks whether the temperature remained constant between those times. The most useful additional evidence is a series of measurements within the interval, taken by a suitable method. Another endpoint at 20 minutes would not necessarily reveal what happened between 0 and 10.
For a probability problem, an important missing fact might be whether a chosen object is replaced before a second draw. Knowing that single procedural detail changes the sample space. More decimal places in the first probability would not address the issue.
For a passage, the question may ask whether a speaker’s claim was sincere. If the narration gives actions but not enough information to decide intent, a later disclosure, internal thought or strong contextual clue could help. The reader cannot request more text in an examination, but can state the supported inference without overstating intent.
The goal is to choose a check that distinguishes competing explanations. This continues the reasoning in Volume 0097 on branching outcomes. Instead of listing every conceivable unknown, ask which one would actually change your conclusion.
9. A complete EMS case: the study-room extension
A school opens a study room for an extra hour each afternoon. Its announcement says: “Students who register before noon may use the extension, subject to available seating.” Records show 60 visits on the first Monday and 75 on the following Monday. A staff member suggests the new hours caused the increase. A separate note says examination revision classes began that week.
The English task is to interpret the announcement. Does registering before noon guarantee a seat? No. The phrase “subject to available seating” limits the promise. The given words are sufficient to answer that interpretation question. We do not need to know the room’s exact number of chairs to recognise the condition.
The Mathematics task is to calculate the change. Visits increased by 15, or 15/60 ×100%=25%. The two totals are sufficient for that arithmetic under the stated definitions. However, they do not tell us how many distinct students attended if some visited more than once. Do not silently convert visits into people.
The Science-style evidence task is to assess causation. The new hours could contribute to the increase, but the new revision classes are a competing explanation. With only two Mondays and no appropriate comparison, the effect of extended hours is not isolated. We may ask for visit times, reasons for attendance and comparable observations before the change or at another suitable period.
Now add a measurement complication: an electronic counter missed some visitors on the first Monday. That information affects confidence even in the numerical increase as a true comparison of attendance. The arithmetic of the recorded values remains 25%, but the records may underestimate the first day. A careful conclusion distinguishes “recorded visits increased by 25%” from “true attendance increased by exactly 25%”.
This exercise shows why evidence sufficiency is task-relative. The same dataset is sufficient to calculate a recorded percentage, insufficient to count unique visitors and insufficient to isolate a particular causal effect. The right answer depends on the question being asked.
A student who gives all three answers with the same certainty has missed the distinction. A stronger student can be definite about the calculation, precise about the condition and measured in the causal evaluation.
10. When absence of data is itself part of the answer
Sometimes a question deliberately tests whether you will overclaim. A graph is shown only until 10 minutes, then asks whether a quantity remains constant forever. No finite observation window normally establishes that unlimited future claim without additional governing conditions. The correct response identifies the gap between recorded interval and proposed conclusion.
In English, an extract may end before a character answers a difficult question. If asked whether the character agrees, the passage may leave the answer unresolved. You can describe signs of hesitation or the effect of the unresolved ending, but should not invent a later decision.
In Mathematics, an unknown may be underdetermined. Suppose x+y=10. Many pairs satisfy this relation: 4 and 6, 3 and 7, or 5 and 5. Without another independent condition, x and y are not individually fixed. The correct conclusion is that more information is needed to determine a unique pair, even though the sum is known exactly.
A common error is to treat missing data as a zero value. If a student’s test score is blank in a table, that may mean absent, not assessed, unrecorded or zero, depending on the definitions. Do not replace a blank with 0 unless the question says that is the coding rule. This matters for averages and comparisons.
Another mistake is treating lack of observed change as proof that no process occurred. A sensor with limited resolution may not record a small change. The statement “no measurable change within the instrument’s resolution” is different from “nothing changed”. The apparatus defines what can be seen.
The discipline is not pessimistic. It is a way to preserve trustworthy conclusions. Clear limits make the supported part of an answer more credible.
11. Original practice: twelve sufficiency decisions
A. English. A character whispers “Don’t tell anyone” and hides an envelope. Can the reader conclude what the envelope contains? What can be inferred about the character’s wish?
B. Mathematics. A runner completes 6 km. What additional value is needed to calculate average speed over the run?
C. Mathematics. A class rises from 24 to 30 pupils. Find the absolute and relative increase. Is this sufficient to prove the class has better learning outcomes?
D. Science. An instrument shows the same reading at minutes 3 and 4. Can you conclude the process stopped exactly at minute 3? What additional observation or apparatus information might help?
E. English. A narrator says “I expected her to be angry, but she laughed.” What is established about the narrator’s expectation and her actual response?
F. Mathematics. Two numbers add to 18. Give two different pairs that satisfy the condition, then state one additional independent fact that would determine a unique pair.
G. Science. A measured value rises after a lamp is moved closer, but the observation intervals differ between trials. What can the data establish and what comparability problem remains?
H. English. A passage reports that a student missed the bus, then failed to arrive on time. Does the sequence alone prove the missed bus caused the lateness?
I. Mathematics. A bag has two red and three blue counters. Two draws are proposed but the question does not state whether the first is replaced. Why does this matter for the probability of two reds?
J. Science. Five readings agree closely, but a known standard reveals a constant offset. Which quality is supported by the repeated values, and what remains in question?
K. Integrated. A school reports 70 library visits one week and 84 the next, then claims reading attainment improved. Calculate the recorded change and explain the mismatch in variables.
L. Decision. A question asks for one practical improvement to test whether a rising sensor reading is a real sample change or instrument drift. Propose a targeted check rather than a generic request for more data.
Answer each case with three headings on scrap paper if needed: known, required and unknown. You will often find that the final response can be much shorter than your planning notes.
12. The answer clinic: stop at the right boundary
A. The content of the envelope is unknown. The words and act of hiding support an inference that the character wants the information kept private. They do not reveal the exact contents or the full motive.
B. You need elapsed time for the 6 km to calculate average speed, or another relevant quantity from which that time can be determined. Distance alone is insufficient for a unique speed.
C. Increase is 6 pupils, or 6/24 ×100%=25%. The enrolment comparison says nothing directly about learning outcomes. The arithmetic is sufficient; the educational claim requires different evidence.
D. No. Equal readings at two sampled times do not prove the exact stopping instant. Information about sampling frequency, instrument resolution and additional measurements could help distinguish a real plateau from an unobserved or undetectable change.
E. The narrator expected anger, but the character laughed. The contrast is explicit. The laughter’s precise meaning—amusement, relief, surprise or something else—may still depend on context.
F. Pairs such as (8,10) and (7,11) both add to 18. If their difference is 2 and the first is the smaller, the pair is determined as (8,10). A suitable second independent equation resolves the ambiguity.
G. The readings may show an association between a changed lamp distance and the observed outcome, but the unequal intervals compromise a direct comparison of counts. Match intervals or use a justified rate with suitable controls before claiming the specific effect.
H. No. Missing the bus may contribute, but the sequence alone does not establish all relevant causes of lateness. The passage may support a plausible explanation without proving it uniquely.
I. Without replacement, the chance of two reds is (2/5)(1/4)=1/10. With replacement, it is (2/5)(2/5)=4/25. The procedure changes the sample space and therefore the answer.
J. Close agreement supports repeatability under the checked conditions. The offset relative to the standard calls accuracy into question. Repetition does not automatically remove systematic error.
K. Recorded visits increased by 14, or 14/70 ×100%=20%. Visits do not directly measure attainment. An appropriate learning assessment would be needed to evaluate that separate claim.
L. Measure a suitable stable reference at the same stages, check calibration and instrument range, or use an appropriate independent sensor. The explanation should say how the chosen check can distinguish apparatus drift from genuine sample change.
The answers are not all “yes” or “no”. Some require a calculation, some a supported interpretation, and some a specific limitation. That diversity is the point: evidence sufficiency depends on the job.
13. How to use this under examination pressure
Under timed conditions, begin with the command word. Circle calculate, describe, explain, infer, suggest or evaluate. Then identify what a complete response to that verb contains. A calculation needs the right relation and values. An inference needs passage evidence. An evaluation needs a relevant judgement and reason.
If you are stuck, ask whether the obstacle is a missing fact or a weakly understood skill. If the information is present but you do not recognise the model, revisit the relevant formula or concept. If the required fact genuinely is absent, do not invent it. State what can and cannot be determined where appropriate.
Check whether the examiner has already supplied a simplifying assumption. Mathematics questions may state that motion is constant, that objects are selected randomly or that a relationship is linear. Those assumptions allow a precise model answer within the given problem. Introducing unrequested real-world exceptions can distract from the solution.
When multiple readings or passages disagree, choose the relevant source and timeframe. One value may belong to a later condition. One quotation may belong to a character, not the writer. The solution may depend on a specific group or interval. Volume 0085 on reference auditing develops this foundation.
Do not turn caution into a universal refusal. A conclusion supported by appropriate evidence should be stated clearly. “The probability is 1/10” is right under the specified without-replacement model. “It might be 1/10” would be weaker and less precise when all assumptions are supplied.
A final answer earns trust when it is both complete for the question and limited to what the evidence warrants. That balance is a useful definition of readiness for unfamiliar problems.
14. Seven days to make sufficiency a habit
Day 1: task words. Sort a dozen short prompts into calculate, describe, infer, explain and evaluate. For each, name the minimum type of evidence required.
Day 2: known versus unknown. Read passages and small tables. Write what is explicit, what can be derived and what remains unsupported. Practise without solving full papers.
Day 3: Mathematics completion. Try problems with missing values, then add one new condition and solve them. Distinguish a missing fact from an unfamiliar method.
Day 4: English calibration. Write two-sentence inference answers that cite a specific gesture or phrase. Remove unsupported psychological labels.
Day 5: Science evaluation. Look at a time series with an apparent trend. State what the numbers establish and one targeted measurement that would distinguish competing explanations.
Day 6: integrated case. Return to the study-room example. Produce three answers: one precise interpretation, one calculation and one evidence-limited causal judgement. Do not make them sound equally certain.
Day 7: independent check. Attempt Questions A–L and classify errors as unsupported assumption, missed value, excessive caution, overextended claim or irrelevant extra working. Revisit only the weakest category with a new task.
For parents, a good question is, “What do we know for certain, and what else would we need for the stronger answer?” It helps students explain their reasoning without treating uncertainty as incompetence. Over time they should learn to answer confidently where the evidence is sufficient and carefully where it is not.
15. A complete answer has a clean edge
Examination performance is not a competition to write the most or to sound absolutely certain. It is the ability to meet a task with the information that supports it. A correct calculation need not pretend to explain why the world changed. A good inference need not reveal thoughts the story keeps hidden. A sound Science explanation need not claim that one experiment proved more than it measured.
Continue from this integrated volume through the G3 SEC learner-guide route and the subject owners: English, Mathematics and Science. If your difficulty lies in what is missing rather than what is present, a focused return to the correct foundation is more useful than repeating a whole revision chapter.
Exam details were checked against the SEAB 2027 G3 syllabus index, K300 English, K310 Mathematics and combined Science K326/K327/K328. All practice cases here are original educational exercises rather than official SEC examination questions.
