A poster can be visually finished and factually unfinished
A school display about a gas-sensor trial is ready for printing. Its headline says that Condition B makes carbon dioxide climb faster. The two bars beneath it show a mean final reading of 720 parts per million for Condition A and 760 parts per million for Condition B. The proposed conclusion sounds plausible because the taller bar belongs to B. But a final reading cannot establish how fast a quantity rose. The baseline and elapsed time matter. In this guide, you will decide whether the display should be released, revised or withheld by putting English reading, Mathematics and Biology into one disciplined review.
The earlier Biology investigation of exhaled air explains what carbon dioxide and limewater evidence mean. The Mathematics guide to paired sensor readings explains baseline subtraction and percentages. The English visual-text guide examines the impact of headlines, captions and axes. This volume has a different job: acting as the final evidence desk. Instead of answering one isolated question, you will check whether the whole public message can survive scrutiny, correct it without hiding the useful observations, and justify an honest publication decision.
Quick route: identify the exact public claim; identify what was actually measured; recover the missing starting readings; check units, intervals, sampling and controls; separate observation from explanation; rewrite the headline and caption; declare uncertainty; decide what can be released; and practise short responses for the separate G3 SEC English, Mathematics and Science examinations. No combined official EMS paper exists. EMS is the name of this learning connection, not an examination subject.
Use the real examination framework, not an invented combined paper
Singapore’s first Secondary Education Certificate examinations are scheduled for 2027. In the SEAB 2027 G3 catalogue, English Language is K300, Mathematics is K310 and Biology is K325. The older corresponding subject codes appear in the catalogue for reference; a student preparing for the 2027 examination should follow the applicable 2027 subject syllabus and school guidance. The three papers do not form one examination called EMS. They assess different knowledge and skills even when a useful classroom situation connects them.
The 2027 G3 English syllabus includes comprehension of visual and verbal information in Paper 2. It also assesses purposeful writing and oral communication. The 2027 G3 Mathematics syllabus expects correct operations, units, representations, working and interpretation of real-world mathematical information. The 2027 G3 Biology syllabus covers gas exchange and respiration as scientific concepts. An invented school exhibition is therefore a legitimate way to practise transferable reasoning, but it is not a prediction of the examination question, experimental protocol or mark scheme.
Every number, quote, trial result and school notice below belongs to an original fictional teaching scenario. The parts-per-million unit describes a concentration reading, not an individual person’s medical condition. No measurement threshold is used as health advice. The goal is academic judgement: what the supplied evidence shows, what it cannot show, and which sentences a careful writer may publish.
Start with the unedited notice and the source sheet
The proposed headline reads: ‘CONDITION B MAKES CO2 CLIMB FASTER!’ Beneath it are two vertical bars labelled A: 720 ppm and B: 760 ppm. A caption tells readers to choose the smarter trial. A small note says the bars are mean final readings from three runs in each condition, without showing the initial readings or observation duration. The visual scale begins at 700 ppm rather than zero. Each of those details matters to a different reviewer: the Maths reader asks what the bars measure, the Science reader asks how the measurements were obtained, and the English reader asks what impression the presentation encourages.
A later source sheet supplies mean starting values: A begins at 440 ppm and finishes at 720 ppm; B begins at 510 ppm and finishes at 760 ppm. Subtracting gives a mean rise of 280 ppm in A and 250 ppm in B, provided the starting and final aggregates refer to the same runs. The mean final concentration in B is 40 ppm higher, yet the mean change in A is 30 ppm greater. Both statements can be true because the groups began at different levels. No interval length has been supplied, so the speed of increase cannot be established.
The source sheet still does not disclose individual trial values, the location of the sensor, calibration status, background fluctuations, the meaning of Conditions A and B, or the time of reading. We must not invent those details to rescue a headline. The correct final decision will depend not only on repairing arithmetic but on telling readers exactly which gaps remain.
Four questions govern the release decision
First ask whether the sentence concerns the same quantity as the table. A statement about final concentration needs final readings. A statement about increase needs starting and final readings. A statement about rate needs change and time. A statement about cause needs a design that supports causal inference. A school poster can get every displayed number right yet claim something different from what its numbers measure. This is the most important gate because later checks cannot repair a claim about the wrong variable.
Second ask whether the comparison is fair. A common time period, similar measurement method, well-defined conditions and appropriate controls are central to a persuasive practical investigation. Different starting values do not automatically invalidate a comparison, but they make the choice of comparison crucial. We can compare these sample end readings and net changes; we cannot explain why they differ without information about the experimental design.
Third ask whether the writing clearly marks its limits. ‘These three sample runs suggest’ is narrower than ‘this always happens’. ‘Mean final reading’ is more precise than ‘carbon dioxide level’ when the time point is important. ‘No conclusion about rate is possible without elapsed time’ helps readers know why the most dramatic headline was rejected. The limitation is part of the information, not an apology at the end.
Fourth ask whether a reasonable reader might act on an implication the source sheet does not justify. The command to choose the smarter trial could imply safety, efficiency, environmental benefit or scientific superiority without defining any criterion. Either define and support the decision rule or remove the command. A conscientious editor does not use decorative confidence to fill a factual gap.
What the arithmetic establishes and what it leaves open
Write all values with labels: A mean initial 440 ppm, A mean final 720 ppm; B mean initial 510 ppm, B mean final 760 ppm. A change is final minus initial, so A changes by 720 − 440 = 280 ppm and B by 760 − 510 = 250 ppm. The difference between final readings is 760 − 720 = 40 ppm in favour of B. The difference between increases is 280 − 250 = 30 ppm in favour of A. The order flips because the starting point differs. This is not a contradiction. It is what happens when one question concerns final values and another concerns changes.
If someone wants percentages, require the denominator to be named. Relative to A’s starting mean, its increase is 280 ÷ 440 × 100, about 63.6%. Relative to B’s starting mean, its increase is 250 ÷ 510 × 100, about 49.0%. Those are within-group increases relative to different baselines. They are not the percentage by which Condition A is scientifically more effective or safer. Choosing a denominator is a meaningful decision; adding a percent sign does not make it automatically meaningful.
Do not average the two percentages and present that as a universal combined change. Aggregation needs a defined population or comparable weights. It also matters whether each mean came from the same number of valid trials. Here the note says three in each group, but does not show the raw readings. The arithmetic demonstrates what follows from the reported averages, not how reliable those averages are. A mathematical check makes the claim more precise while leaving the biological and experimental questions visible.
Why science needs more than a tidy pair of means
A gas-sensor reading is a measurement of a concentration at a place and time. Carbon dioxide in exhaled air is linked to cellular respiration and gas exchange, but the existence of carbon dioxide in a container does not show how a person breathes, how much oxygen reaches the blood, or whether a room is healthy. A school experiment can establish an observation about its apparatus without establishing an effect on a person. These are different levels of explanation.
Suppose the detector sat closer to a source in one condition, or one sensor was warmed differently. A final-value difference might reflect those differences rather than the intended variable. Suppose the reading had not stabilised at the start, or someone recorded values to different decimal places. Those details could influence the comparison. You do not need to accuse the experimenters of making those mistakes; identify what information would be required to rule them out.
A proper control gives an alternative explanation a chance to fail. If the intended variable were the presence of a living source of carbon dioxide, the control might be a comparable arrangement without that source, while maintaining the other conditions as closely as the question allows. If the intended variable were airflow, both setups would need consistent sensor location and observation time. The correct control depends on the scientific question. Never write ‘keep everything the same’ as if the tested variable should remain unchanged too.
Make a release sheet before making a beautiful sheet
The release sheet is a quiet working document. Its first column is the proposed claim, its second the supporting observation, its third the scientific or mathematical uncertainty, and its fourth the editorial action. For ‘B finishes at a higher mean concentration’, the evidence is the two reported final averages and the action may be ‘retain with sample description’. For ‘B climbs faster’, the evidence is insufficient because duration is missing and the action is ‘delete or replace’. For ‘choose the smarter condition’, the success criterion is missing and the action is ‘delete’. These decisions protect the reader better than merely changing a colour.
Good publication review is not a hunt for errors that makes the experiment worthless. It preserves what can be defended. The group means offer a useful classroom comparison. Their different baselines produce an instructive reversal between rankings of final values and increases. That is worth communicating. The review removes the unsupported story around those facts rather than hiding the facts themselves.
Give each correction a traceable reason. ‘Replace faster with higher mean final reading because the poster contains no time data’ is specific. ‘Take out the exaggerated word’ is not enough, because it does not identify what was exaggerated. ‘Retain 760 ppm but label it a fictional mean of three classroom readings’ states both the usable value and the restriction. Traceable revisions create a way for another student to check your judgement.
Build a trustworthy headline and caption
A defensible revised headline is: ‘Two Class Experiments Ended at Different Mean CO2 Readings’. Its subheading might read: ‘In three example runs per condition, B ended at a higher reported mean concentration, while A showed the larger mean increase from its reported start.’ This is not as dramatic as ‘B climbs faster’, but it tells the reader the genuinely interesting result: comparing endpoints and comparing changes answer different questions.
Replace the two isolated bars with a small text table listing the mean initial reading, mean final reading and mean change for each condition. If a graph is used, label its axis clearly, state that the values are means, and avoid a misleading graphic scale. A nonzero axis is not automatically dishonest, but it must be legible and appropriate to the intended comparison. In a public setting, a simple paired table may be clearer than an axis that exaggerates a small part of the range.
The caption should distinguish the available results from the missing evidence: ‘Illustrative mean readings from three runs per condition; conditions, raw trials and elapsed observation periods not supplied in this example. Differences in reported means do not establish relative speed or cause.’ This wording is useful because it does not say ‘the study proves nothing’. It says exactly what the data do and do not support.
What to do if the source sheet changes at the last minute
If a revised sheet provides equal observation periods, recalculate rates explicitly rather than treating the old final values as the rates. An increase of 280 ppm over 20 minutes would be a mean rate of 14 ppm per minute; 250 ppm over 10 minutes would be 25 ppm per minute. These are invented alternative values, not missing facts about the original trial. They demonstrate that A can have the larger total increase while B has the larger mean rate when the durations differ. A new table can reverse the original conclusion again.
If a new sheet says that one of the three trials was excluded, find out why and whether the published means already exclude it. A replacement mean cannot be computed from the old mean alone unless enough information about the changed dataset is known. If an observation was dropped because of a genuine sensor fault, report the count and rule consistently; if it was dropped simply because it contradicted the desired headline, the resulting comparison is not trustworthy.
If a new source supplies raw data, recompute rather than trusting an unexplained percentage copied from a slide. Check each initial-final pair, units and missing-value rule. Save the unchanged source alongside the corrected display. A reader does not need the editorial history in the final poster, but the team needs to know which version supports the published claim.
English performance: answer the message, not the illustrator’s imagined motive
In the G3 English Paper 2 visual-text component, you may need to explain how language and visual presentation influence the reader. A strong answer connects a visible device to a probable effect. ‘The enlarged headline makes the faster-climb claim sound decisive before the reader notices that the graph displays only end readings’ is stronger than ‘large letters attract attention’. It names the specific misunderstanding at issue.
A comparison question may ask how two texts differ in purpose. The original poster urges a choice using a striking claim; the correction note asks the reader to distinguish final values from increases and to avoid unwarranted conclusions. That contrast is a communication contrast, not merely an arithmetic distinction. Quote only a short relevant phrase or point to a specific graphic feature. Do not spend time retelling the entire experiment if the question asks about tone, purpose or effect.
In a Writing task, the same evidence could support a report, email or speech only if the requested audience and purpose are respected. A laboratory note to a teacher should be exact about procedure and uncertainty; a school notice to families should be understandable without specialist jargon. Neither format gives permission to invent health conclusions. The best writing is both accessible and faithful.
Mathematics performance: show the method that makes a claim auditable
When marks depend on reasoning, write the subtraction before the comparative sentence. Display 720 − 440 = 280 ppm and 760 − 510 = 250 ppm. Then state, ‘A’s mean reported increase exceeds B’s by 30 ppm’. This allows the examiner to follow the calculation even if a subsequent conclusion needs correction. Writing ‘A is better’ would throw away the meaning of the result and may not answer the question.
If a question asks for percentage increase, identify each original value and its role. For A, the relevant fraction is 280/440, not 280/720. For B, it is 250/510, not 250/760. If the percentages are compared, they refer to separate starting bases. The difference between approximately 63.6% and 49.0% is about 14.6 percentage points, not 14.6% of one or another group unless a further denominator is specified.
A graph task rewards proper axes, a chosen scale, plotted values and a truthful interpretation. An attractive plot with incorrect units is mathematically weak. The earlier guide to misleading graphs owns the general technique; this chapter applies it to an editorial release decision, where a numerical statement has an actual reader-facing consequence.
Science performance: qualify explanations at the right level
In a structured Biology answer, a valid explanation connects cause, mechanism and observation where the question supports it. Human cells release carbon dioxide during aerobic respiration, and carbon dioxide diffuses from blood to alveolar air down its concentration gradient, before being exhaled. That account gives context for exhaled-air evidence; it does not establish the cause of a difference between two unspecified classroom gas-sensor conditions. A correct biological fact cannot substitute for a missing experimental detail.
The limewater test described in the previous Biology volume can show evidence of carbon dioxide when performed properly. Our poster, however, gives instrument readings. Do not pretend the same test was performed or infer that a clear limewater sample was measured. A teacher may ask you to distinguish the instrument reading from what a chemical indicator would reveal, but the evidence must remain attached to the actual described method.
An excellent explanation also states what would improve the investigation: the same observation period, a clearly identified manipulated variable, comparable apparatus, repeated independent runs, a recorded control and enough raw readings to judge variation. These suggestions follow from the intended claim. They are not a claim that the original experiment definitely failed; they describe what the source sheet would need to support a stronger conclusion.
Twenty worked release decisions: foundations first
Case 1 — Endpoint. The headline says ‘B ended higher.’ The sheet shows mean final readings of 760 ppm and 720 ppm. This narrow comparison is supported, provided we say it describes the reported sample means. Accept it with the numbers and correct units, rather than turning it into a universal claim about every run.
Case 2 — Increase. The line says ‘B rose more.’ Subtraction gives A an increase of 280 ppm and B 250 ppm. Reject or reverse the line. Show both calculations so the reader sees why a larger endpoint does not mean a larger change.
Case 3 — Speed. The line says ‘B rises faster.’ No elapsed periods appear in the source. Withhold the rate claim. A change divided by time cannot be calculated from end readings alone, and a larger increase would not by itself determine a faster rate if durations differed.
Case 4 — Percentage. The line says ‘A is 14.6% more effective.’ The figure 14.6 comes approximately from the difference between two within-condition percentage changes, measured in percentage points. It has no defined effectiveness denominator. Reject the sentence and state the two percentage increases separately if the audience needs them.
Case 5 — Causation. The poster says ‘Condition B causes higher carbon dioxide’. The reported conditions are not defined, and baseline values differ. Even a statistical association would not establish causation without suitable experimental controls. Replace the causal verb with a description of the observed means.
Case 6 — Raw data. A student argues that three trials make the result certain. Three is a count, not a guarantee. Without individual results, we cannot check spread or unusual readings. Retain ‘mean of three example runs’, but avoid ‘every run showed’ or ‘results were always consistent’.
Case 7 — Axis. The bars start at 700 ppm, visually stretching the gap between 720 and 760. The numbers may be accurate, but the chosen scale can create an exaggerated impression. Either use an appropriate full scale with a clearly stated purpose or present a plainly labelled table.
Case 8 — Source. A logo is added to make the exhibit seem official. A logo does not establish independent verification. If no approved institutional source is documented, do not use its visual authority as scientific evidence. Name the work as a fictional classroom exercise.
Twelve further worked decisions: harder evidence questions
Case 9 — Timing. A new note gives A twenty minutes and B ten minutes. Now mean rates would be 14 and 25 ppm per minute respectively, using the reported mean changes. The rate ranking differs from the rise ranking. Before publishing, confirm that both intervals are actually part of comparable experimental protocols.
Case 10 — Missing baseline. One run lacks its initial value. Do not silently insert zero or use the last observed reading as a substitute. Label the run incomplete for paired-change analysis and determine whether the mean end value and mean change use the same number of valid observations.
Case 11 — Sensor position. One detector was closer to the air source. This could be a confounding difference. The remedy is not a speculative numerical correction; it is a controlled repeat with a defined, consistent detector position.
Case 12 — Technical units. A caption says ‘760 units of carbon dioxide’. The reported figure is a concentration of 760 ppm. Restore the unit and context. The number alone does not tell us the total mass of gas or a rate of release.
Case 13 — Question wording. An English question asks for the likely effect of the exclamation mark. ‘It increases urgency and reinforces a decisive claim’ responds to the communicative effect. ‘The mark proves the science is false’ is an unsupported leap.
Case 14 — Contradiction. The final-value ranking favours B and the increase ranking favours A. A learner calls the data contradictory. Explain that they compare different quantities; start level differs. The contradiction lies in the original wording, not in the arithmetic.
Case 15 — Calibration. The methods sheet never mentions calibration. State that calibration information is missing. Do not invent an offset value or claim the sensor was faulty. Request a calibration record before raising the conclusion’s confidence.
Case 16 — Outlier. One high final reading is reported but not shown. A reviewer wants to remove it immediately. Without a prior quality rule or demonstrated measurement error, exclusion may bias the result. Inspect the raw record and report a justified exclusion transparently.
Case 17 — Common baseline. A classmate suggests subtracting 440 ppm from both final values. That would falsely treat B as starting at A’s baseline. Use each condition’s own reported starting value. A fair comparison never manufactures a common initial observation.
Case 18 — Small sample. The means are labelled from three illustrative runs. That may be enough to practise arithmetic but not enough to generalise confidently to an entire population. Avoid extrapolating to all classrooms, people or outdoor conditions.
Case 19 — Audience. A notice intended for younger pupils uses the abbreviation ‘ppm’ without explanation. Define it as parts per million, a concentration unit, without converting it into a medical claim. Clarity is part of accuracy, not a departure from it.
Case 20 — Deadline. A team argues that the poster is due today, so the headline should stay. A deadline cannot turn unsupported wording into a valid scientific statement. Release the narrower, supported claim on time or withhold the poster until the source can be checked.
A model decision memo that a pupil can defend
The editor’s decision is ‘revise before display’. The sample readings may be retained, because the reported mean final values and reported starting values permit a straightforward comparison. The claim that B makes carbon dioxide climb faster must be removed because the displayed bars show only final levels and the duration of observation is not supplied. The claim that B is the smarter choice must also be removed because the poster provides no defined performance or safety criterion.
The revised public text should present the actual values: mean initial readings A 440 ppm and B 510 ppm; mean final readings A 720 ppm and B 760 ppm; reported mean rises A 280 ppm and B 250 ppm. It should state that each condition has three illustrative runs. The source note should acknowledge that raw results, methods and elapsed time are not supplied in the example. The display should avoid claims about health, room safety, effectiveness or causation.
This is a useful model because it does not begin with a personal judgement about which group is good. It starts with the claim, cites the numbers, describes the restriction and states a proportionate remedy. It shows the reader how to act responsibly when evidence is interesting but incomplete. A shorter examination response can preserve the same sequence in two or three well-constructed sentences.
Practice questions with model reasoning
Question A. Which condition has the higher mean final reading, and by how much? Answer: B, by 760 − 720 = 40 ppm. A mark-worthy response names the final reading, not just the condition. Do not confuse the 40 ppm endpoint difference with either group’s increase.
Question B. Which group has the larger reported mean increase? Answer: A, because 720 − 440 = 280 ppm, compared with 760 − 510 = 250 ppm for B. A’s increase exceeds B’s by 30 ppm. The calculation is central, and both quantities must be comparable.
Question C. Explain why the faster-climb headline is not supported. Answer: The poster shows mean end concentrations, not concentration change per unit time. Initial readings are omitted from the display, and no observation durations are supplied. We therefore cannot conclude which condition had the faster increase.
Question D. Identify an experimental improvement. Answer: Record the same defined observation interval and comparable sensor conditions for both groups, then report time-stamped paired readings. This addresses rate inference; merely adding more coloured bars does not.
Question E. How could the headline influence readers? Answer: The emphatic capitals and exclamation mark make the faster-climb statement sound certain and urgent, encouraging acceptance before readers notice the small print about final readings. This explains effect without speculating about deliberate dishonesty.
Question F. Draft one accurate headline. Answer: ‘B Has the Higher Mean Final CO2 Reading in a Small Classroom Example.’ This sentence preserves the endpoint result and the sample context. It does not imply causation, rate or safety.
Question G. Does the evidence show the class should choose B? Answer: No. ‘Better’ is undefined and the source sheet gives no decision criterion. Higher mean final concentration might be relevant for one purpose and irrelevant or undesirable for another. A recommendation requires an explicit goal.
Question H. Why should the team preserve the raw trials? Answer: The raw records would allow checking means, spread, sensor anomalies and whether initial and final readings were paired. A single average can conceal patterns. The exercise does not supply raw trials, so the team must not pretend to have checked them.
Question I. What does a nonzero graph origin change? Answer: It changes the visual impression of the gap, often making a small absolute difference appear larger on the printed bars. It does not change the numerical values. A good graph makes the scale readable and appropriate to the question.
Question J. Can the poster say the carbon dioxide concentration caused illness? Answer: No. The fictional concentration measurements supply no evidence of a person’s exposure, symptoms or medical cause. The educational question is about evidence interpretation, not diagnosis.
An examination routine for calm, repeatable performance
At the start of a cross-subject evidence problem, underline the noun after the comparison: final reading, rise, rate, percentage, result, cause or effect. Many errors happen because the candidate silently changes that noun midway through an answer. Write a tiny quantity map in the margin: initial, final, difference, time, units. Then answer the requested comparison rather than the one that seems more impressive.
For English, identify the sender, intended audience, main claim and design or word choice that shapes interpretation. For Mathematics, calculate only after choosing the correct baseline and operation, and finish with units. For Science, name the observation and explain its mechanism only as far as the experimental design permits. When you revise, compare your final sentence against the question’s verb: identify, calculate, describe, explain, evaluate or suggest.
A practical study routine is to attempt three passes. On the first, use the source sheet without looking at the answer. On the second, mark each sentence with a supported number or an explicit limit. On the third, change one condition, such as observation duration, and see whether the conclusion still holds. This transfer test comes from the PSLE guide on refusing to generalise from one example and remains valuable when preparing for G3 examinations.
What this volume adds to the sequence
The earlier subject guides already teach their own owners: biology for the interpretation of a carbon dioxide demonstration, mathematics for the behaviour of baselines and graph claims, and English for visual-text comprehension. This volume does not replace them. It asks what happens when a real reader will encounter all three together as a single public message. The transferable skill is evidence stewardship: the ability to retain useful facts without elevating them into unsupported recommendations.
The next time a report says something ‘works faster’, ask where the time data are. When a bar looks taller, ask whether it shows an ending, a change or a rate. When a conclusion claims a cause, ask which alternative explanation was controlled. When a caption asks an audience to act, ask what the evidence actually permits that audience to decide. These are not cynical habits. They are the plain reading and reasoning skills that make examination answers more accurate and public communication more trustworthy.
Official references and related learner routes
The official subject sources are the SEAB 2027 G3 syllabus catalogue, G3 English Language K300, G3 Mathematics K310 and G3 Biology K325. For a subject’s exact examination conditions, assessment structure, syllabus changes or allowed equipment, use its applicable official document and school instructions. All invented numbers and posters here are exercises rather than official sample questions.
Continue with Vol 0563: Biology exhaled-air and carbon dioxide, Vol 0564: Mathematics paired gas-sensor data and Vol 0565: English visual-text reading. For graph-scale foundations, consult the Mathematics misleading-graphs guide. The purpose of these links is to give each subject its own home while this integrated volume concentrates on the decision to publish a claim.
