How to decide whether a school food-test report is ready to publish
A school food-test demonstration becomes useful only when its records, calculations and written claims agree. A positive chemical test is evidence about a particular substance in a particular sample under stated conditions. It is not a verdict about which food every student should eat. A correct fraction is evidence about the records counted, not a licence to generalise to every product in a canteen. An attractive report does not repair either mistake. In this guide, students carry out a complete, careful evidence hearing before releasing a fictional school-club summary.
This fourth volume in the food-test sequence puts three earlier skills together: Biology explains what the tests actually detect, Mathematics identifies valid results and denominators, and English communicates the conclusion to the intended reader. You will work from a single invented evidence file, check each stage, correct weak sentences, answer questions resembling the reasoning required in school assessments, and decide what can responsibly appear on a school display. The aim is independent judgement, not an extra subject called EMS.
Quick route: name the precise claim; identify the test and its control; sort valid, invalid and unresolved records; choose the matching denominator; calculate without rounding away meaning; separate observation from interpretation; revise the report for a specific reader; make one defensible recommendation; and check the completed message against its original evidence.
Know the official 2027 examination setting
The Singapore–Cambridge Secondary Education Certificate begins in 2027 and assesses subjects at G1, G2 or G3 according to the subject level taken. There is no separate combined English–Mathematics–Science paper named EMS. EMS is a learning method used here to practise the connections between three actual subjects. The SEAB SEC overview explains the qualification and timetable, while the 2027 G3 syllabus catalogue lists current codes. G3 English Language uses K300 (formerly 1184), G3 Mathematics K310 (formerly 4052), and pure G3 Biology K325 (formerly 6093). Students taking combined Science have different subject codes and assessment arrangements; follow their registered syllabus.
For orientation, K300 English has Writing, Comprehension, Listening and Oral Communication papers. Its writing paper includes editing, situational and continuous writing; a school report is one possible situational text type, not a guaranteed examination genre. K310 Mathematics has two 90-mark written papers, each 2 hours 15 minutes, and its syllabus expects interpretation of data in real-world contexts. K325 Biology has a 40-mark multiple-choice paper, an 80-mark structured/free-response paper and a 40-mark practical paper, weighted 30, 50 and 20 percent respectively. This article’s evidence hearing is original revision practice, not an official question or an additional examination component.
The fictional evidence file: forty-eight food-test records
A secondary-school science club has prepared a demonstration display. The record sheet contains 48 attempts at one stated nutrient test. Of these, 42 have valid, classified outcomes: 15 positive and 27 negative. Another six cannot be classified reliably because the control or observation record is missing or ambiguous. The club’s draft says: ‘Most tested samples contained the nutrient, so the school should recommend them as the healthiest choice.’ Your task is to decide which parts can be published and which must be corrected.
Notice what the file does not say. It does not identify a representative sample of all foods sold at the school. It does not show that every sample was prepared in the same way. It does not establish nutritional value, portion size, energy intake or medical suitability. It does not tell you that the 48 records came from 48 distinct foods. Some may be repeats. It does not tell you whether the positive outcomes were evenly distributed between sample types. Each absence matters because it limits the question that the evidence can answer.
A reliable audit begins by writing the claim in a smaller form: ‘In this set of 42 valid classified test records, 15 gave positive results for the substance investigated.’ That sentence is less dramatic, but it is supportable. The confirmed positive proportion among valid classified records is 15 ÷ 42, approximately 35.7 percent. Among all 48 attempts it is 15 ÷ 48, or 31.25 percent. Those percentages describe different populations of records. Neither justifies the word ‘most’ in the normal majority sense. The six unresolved attempts cannot be quietly counted as negative.
Build an evidence chain with five deliberate questions
First ask exactly what was tested. Iodine solution is used for starch, Benedict’s reagent with appropriate heating for reducing sugars, the Biuret test for proteins, and the ethanol emulsion test for lipids. These are school-level qualitative tests, not a complete chemical composition analysis. A positive result supports the presence of the tested class of substance under the conditions of that test. It is not a measurement of how much nutrient a person should consume.
Second ask whether the control behaves as expected. A known positive reference helps show that the reagent and procedure can produce the intended positive indication. A suitable negative control helps reveal contamination or a misleading appearance. A failed control makes the experimental result less secure. The claim ‘we followed the steps’ is not a substitute for showing the control outcome. In an actual school laboratory, all chemical tests must follow teacher instructions and approved safety procedures; do not independently heat reagents or handle ethanol as a home activity.
Third ask how results were classified. A clear observation such as a characteristic colour change under the correct conditions is different from ‘looked a bit unusual’. Ambiguous colour, incomplete heating, missing code or improper preparation may prevent confident classification. A record can be useful for diagnosing a procedure while still being unsuitable for the positive-versus-negative table.
Fourth ask whether the calculation uses the stated group. Fifteen confirmed positives among 42 valid classified records means 15/42. Fifteen confirmed positives among all 48 attempts means 15/48. A comparison with the 27 negatives must use 42, because those are the valid positive and negative outcomes. Stating a denominator is part of the explanation, not an optional technical detail.
Fifth ask what conclusion reaches a reader. In English, the public message should distinguish the demonstration’s purpose, the result and the limit. A careful report can still recommend useful action: correct the display, label the six unresolved attempts, repeat uncertain measurements under supervision and explain what each test detects. It should not issue dietary advice based on this limited exercise.
Use the correct biological meaning for each result
Consider the iodine starch test. A blue-black change supports the presence of starch under the specified procedure. Without that change, an accurately performed negative test does not prove that a sample contains no carbohydrate of any kind. It is testing starch, not every carbohydrate. The same restraint applies to Benedict’s reagent: evidence of a reducing sugar is not a claim about all types of sugar. Treat ‘does not show the tested result’ as narrower than ‘none of this nutrient exists’.
The Biuret test answers a different question, concerning protein, while the ethanol emulsion test is used for lipids. The class result is also not the same as the quantity consumed, because a qualitative positive does not by itself tell us grams per serving. It is tempting to turn four test names into four columns marked ‘healthy’ and ‘unhealthy’. Resist that shortcut. The correct question is ‘What does this reagent establish about this prepared sample?’.
Some pupils confuse a control with a controlled variable. A control set-up or reference is a comparison whose expected result helps interpret the test. A controlled variable is a condition deliberately kept the same between cases, such as sample volume, preparation time or reagent amount. In an investigation, you may need both. The PSLE Science control-set-up guide revisits the foundation; this G3 volume expects you to use the distinction when defending a conclusion.
Calculate only what the records permit
Start with the basic identity: valid classified records plus unresolved attempts equals all attempts. Here 15 + 27 + 6 = 48, and 15 + 27 = 42. These simple checks expose many reporting errors before percentages are calculated. If a draft table adds to 46 but the report states 48, either two records are unaccounted for or the totals have been transcribed incorrectly.
The positive share of valid records is 15/42 × 100 = 35.714… percent, reported as about 35.7 percent if the purpose supports one decimal place. The negative share is 27/42 × 100 = 64.285… percent. They add to 100 percent, within rounding. Among all attempts, the shares are 15/48 = 31.25 percent positive, 27/48 = 56.25 percent confirmed negative, and 6/48 = 12.5 percent unresolved. The three percentages add to 100 percent without reassigning the unresolved group.
If someone removes the six unresolved attempts from the total but keeps the word ‘all’, the statement becomes inconsistent. If someone includes them as negatives, the apparent negative fraction rises to 33/48, but that describes an assumption, not an observation. If someone claims a food was tested once for every record, ask for the sample identifiers. A table of attempts is not necessarily a table of unique foods.
A useful check is to work backwards. If the report says that 35.7 percent of the 42 valid records were positive, multiply 0.357 × 42 to see whether the implied count is close to 15. When counts must be integers, a percentage rounded too early can produce a misleading reverse result. Preserve the original count in the master table and treat the percentage as its description.
Write a report that lets the coordinator act
The audience is a school science coordinator who must decide whether the display can be used. Begin with the task, not a general essay about nutrition. One strong opening is: ‘The club reviewed 48 trial records for a single nutrient test, of which 42 produced valid positive or negative outcomes.’ This establishes the purpose, scale and scope. The next sentence should say what the results actually were.
Then explain the correction. ‘Fifteen of the 42 valid records were positive, so the proposed claim that most samples tested positive is inaccurate.’ Next state the limit. ‘The test identifies evidence of a particular nutrient in prepared samples; it cannot determine the healthiest food for every student.’ End with a practicable recommendation. ‘Replace the headline, publish the result categories, and repeat the six uncertain records with proper controls under teacher supervision before adding them to the table.’
Note the difference between an absolute promise and a measured recommendation. ‘This proves that our science programme succeeds’ is an unsupported leap. ‘This display can teach how a qualitative result is interpreted and why ambiguous records are not forced into a category’ expresses an achievable educational purpose. Good public writing is not less confident because it acknowledges the boundary of evidence; it is more trustworthy.
A complete original evidence-hearing worked example
Imagine a student committee seated with a lab notebook, an electronic sheet and a report draft. Their role is to decide whether each line of the display survives review. The science reviewer checks reagent, observation and control. The mathematics reviewer checks the tally, the denominator and the proposed visual. The English reviewer checks whether the sentence says what the data mean to this reader. All three must agree before publication. This exercise has no official marking scheme; it is a disciplined way to join ordinary examination habits.
The notebook says ‘Sample R4: recorded positive; colour description missing; negative control unavailable.’ The spreadsheet nevertheless classifies R4 as a positive. The reviewer cannot repair the missing observation by making the spreadsheet look neat. They return R4 to unresolved until evidence is recovered or the trial is repeated. The report writer must then avoid giving an exact outcome for R4. Correcting the label may reduce a reported positive count, but consistency is more important than protecting a preferred result.
Another row says ‘Sample S2: negative under the starch test’. A proposed caption says ‘No carbohydrates found’. This is a biological overreach, not a mathematical error. The corrected caption is ‘No positive starch-test result under the recorded conditions.’ Meanwhile the chart labels 15 out of 48 as ‘35.7%’. That is a denominator error, because 35.7% comes from 15 out of 42. The reviewers must fix both errors independently.
Suppose the student author has room for only one number in the first paragraph. The most informative count might be ’42 valid classified records out of 48 attempts’. That tells the reader immediately why a later percentage is based on a smaller group. A longer report can add 15 positives and 27 negatives. A school poster might use a simple three-part tally. A well-chosen number serves the decision; it does not merely make writing seem scientific.
Twenty-five original hearing cases with answers and transfer
Hearing 1: The hidden denominator
Case. A poster says ‘31% of the valid tests were positive’ although the file has 15 positives and 42 valid results. Reasoned response. The 31.25% figure refers to all 48 attempts, not the valid group. Replace it with about 35.7% of valid classified results, or rewrite the denominator explicitly. No biological interpretation is possible until the arithmetic sentence names its population. Transfer. Before accepting the conclusion, ask which observation, classification or denominator would need to change for the statement to become true. This habit makes the correction reusable on a different examination question rather than dependent on memorising this particular example.
Hearing 2: The false majority
Case. A draft states that the largest category was positive because positive outcomes were ‘important’. Reasoned response. Importance is not a count. Among 42 valid outcomes, 27 negatives exceed 15 positives. The committee should ask what ‘largest’ measures and replace the unsupported phrase with the observed frequencies, not a persuasive adjective. Transfer. Before accepting the conclusion, ask which observation, classification or denominator would need to change for the statement to become true. This habit makes the correction reusable on a different examination question rather than dependent on memorising this particular example.
Hearing 3: The missing six
Case. A graph contains only 15 positive and 27 negative bars but its title claims to summarise every attempt. Reasoned response. The graph represents classified outcomes, not all attempts. Either label it ’42 valid classified outcomes’ or show the six unresolved attempts as their own category. Quiet omission makes the plotted population appear larger or more complete than it is. Transfer. Before accepting the conclusion, ask which observation, classification or denominator would need to change for the statement to become true. This habit makes the correction reusable on a different examination question rather than dependent on memorising this particular example.
Hearing 4: Negative becomes absent
Case. An iodine test did not turn blue-black, so the caption says the sample contains no carbohydrate. Reasoned response. A negative starch-test observation is not a complete carbohydrate analysis. State that the recorded procedure did not yield a positive starch result. If broader carbohydrate claims matter, a different valid investigation is needed. Transfer. Before accepting the conclusion, ask which observation, classification or denominator would need to change for the statement to become true. This habit makes the correction reusable on a different examination question rather than dependent on memorising this particular example.
Hearing 5: A control fails
Case. The known positive reference does not develop the expected colour, but all unknown samples are recorded as negative. Reasoned response. A failed positive reference raises doubt about reagent, timing or method. It is unsafe to infer that the unknown samples lack the substance. Label the series inconclusive, investigate the procedure and repeat under supervision. Transfer. Before accepting the conclusion, ask which observation, classification or denominator would need to change for the statement to become true. This habit makes the correction reusable on a different examination question rather than dependent on memorising this particular example.
Hearing 6: Two tests in one column
Case. A spreadsheet mixes Benedict’s outcomes with iodine outcomes under a heading marked ‘nutrient present’. Reasoned response. The results concern different substances and should not be pooled as though each were the same test. Make separate columns with test names, expected observations and valid-status flags before any calculation. A combined count needs a clearly defined scientific question. Transfer. Before accepting the conclusion, ask which observation, classification or denominator would need to change for the statement to become true. This habit makes the correction reusable on a different examination question rather than dependent on memorising this particular example.
Hearing 7: The mystery sample
Case. One tube was marked ‘perhaps positive’ after a faint colour change. Reasoned response. ‘Perhaps’ is not a confirmed positive. Record the actual appearance and classify the result as unresolved until the relevant observation standard and controls support a decision. This protects the integrity of the category totals. Transfer. Before accepting the conclusion, ask which observation, classification or denominator would need to change for the statement to become true. This habit makes the correction reusable on a different examination question rather than dependent on memorising this particular example.
Hearing 8: Repeat counts
Case. Three records came from repeated attempts on the same prepared sample, and the writer calls them three different foods. Reasoned response. The units are attempts, not unique foods. The record count may still be 48, but the report must not claim 48 different foods. A responsible revision identifies repeated sample codes or states that uniqueness cannot be established. Transfer. Before accepting the conclusion, ask which observation, classification or denominator would need to change for the statement to become true. This habit makes the correction reusable on a different examination question rather than dependent on memorising this particular example.
Hearing 9: Rounding too early
Case. A student rounds 15/42 to 0.4, then reports that 40% of results were positive. Reasoned response. Rounding to one decimal as a proportion loses important precision. Calculate with the original fraction and round once at the end to about 35.7%. Check the requested accuracy in the question; do not invent a universal rule. Transfer. Before accepting the conclusion, ask which observation, classification or denominator would need to change for the statement to become true. This habit makes the correction reusable on a different examination question rather than dependent on memorising this particular example.
Hearing 10: An impossible percentage
Case. A draft gives 35.7% positives and 65.3% negatives, a total of 101%. Reasoned response. The negative share is approximately 64.3%, not 65.3%. Recalculate both from 42, then verify their sum. Slight rounding differences can be acceptable, but a whole percentage-point mismatch signals an error worth repairing. Transfer. Before accepting the conclusion, ask which observation, classification or denominator would need to change for the statement to become true. This habit makes the correction reusable on a different examination question rather than dependent on memorising this particular example.
Hearing 11: A convenient label
Case. A board calls all unresolved attempts ‘not detected’ to simplify the chart. Reasoned response. Missing or uncertain evidence is not a test result. Keep an unresolved category and explain why the attempts cannot be classified. Convenience is not a scientific criterion for turning a missing result into a negative result. Transfer. Before accepting the conclusion, ask which observation, classification or denominator would need to change for the statement to become true. This habit makes the correction reusable on a different examination question rather than dependent on memorising this particular example.
Hearing 12: A comparison without conditions
Case. Two groups use different sample volumes, and the report says one food contains more starch because its colour seemed darker. Reasoned response. The procedures are not controlled well enough for the proposed comparison, and colour appearance need not give a reliable quantity in a qualitative test. Compare only what the validated method supports; redesign the investigation if quantities are needed. Transfer. Before accepting the conclusion, ask which observation, classification or denominator would need to change for the statement to become true. This habit makes the correction reusable on a different examination question rather than dependent on memorising this particular example.
Hearing 13: Recommendation from appearance
Case. One sample tested positive for protein, so the report recommends it as the best meal choice. Reasoned response. The Biuret result alone does not determine a balanced meal or the needs of particular students. Revise to a limited statement about a protein-test indication. Dietary guidance requires broader evidence and suitable expertise. Transfer. Before accepting the conclusion, ask which observation, classification or denominator would need to change for the statement to become true. This habit makes the correction reusable on a different examination question rather than dependent on memorising this particular example.
Hearing 14: The missing date
Case. One control sheet is dated a different day from the unknown-sample trials. Reasoned response. The committee should check whether reagent and conditions were sufficiently comparable. Do not automatically combine separate sessions as one controlled run. Where provenance is uncertain, mark the record for review rather than inventing consistency. Transfer. Before accepting the conclusion, ask which observation, classification or denominator would need to change for the statement to become true. This habit makes the correction reusable on a different examination question rather than dependent on memorising this particular example.
Hearing 15: A screenshot changes the count
Case. A graph exported earlier shows 14 positives, while the final notebook shows 15. Reasoned response. Identify which version is authoritative and reconcile the changed record with its underlying observation. Re-export the graph after the data are confirmed. A polished poster with an obsolete count remains inaccurate. Transfer. Before accepting the conclusion, ask which observation, classification or denominator would need to change for the statement to become true. This habit makes the correction reusable on a different examination question rather than dependent on memorising this particular example.
Hearing 16: The persuasive headline
Case. The heading ‘School tests prove food labels wrong’ is proposed after a small classroom demonstration. Reasoned response. No label accuracy study has been established by the provided file. A truthful heading is ‘What our nutrient-test demonstration showed’. Preserve the difference between teaching a chemical test and auditing commercial food labels. Transfer. Before accepting the conclusion, ask which observation, classification or denominator would need to change for the statement to become true. This habit makes the correction reusable on a different examination question rather than dependent on memorising this particular example.
Hearing 17: The careless ‘always’
Case. A reviewer writes ‘The result is always blue-black when starch is present’. Reasoned response. Describe the expected observation under appropriate test conditions without claiming every preparation always behaves perfectly. Experimental performance depends on factors such as reagent, procedure and sufficient sample. Precision protects a teaching statement from needless exceptions. Transfer. Before accepting the conclusion, ask which observation, classification or denominator would need to change for the statement to become true. This habit makes the correction reusable on a different examination question rather than dependent on memorising this particular example.
Hearing 18: The wrong recipient
Case. A report written to a coordinator begins with instructions for primary pupils and never states the decision requested. Reasoned response. The coordinator needs findings, limitations and an actionable recommendation. The writing should name the decision, then organise only the information needed to make it. Audience adaptation changes emphasis, not the truth of the evidence. Transfer. Before accepting the conclusion, ask which observation, classification or denominator would need to change for the statement to become true. This habit makes the correction reusable on a different examination question rather than dependent on memorising this particular example.
Hearing 19: Unverified practice
Case. The club asks students to reproduce heated Benedict’s tests at home. Reasoned response. The recommendation is unsuitable. Chemical reagents, heating and ethanol require appropriate laboratory oversight and safety practice. The public text can explain concepts or direct students to supervised school demonstrations; it must not invite unsupervised replication. Transfer. Before accepting the conclusion, ask which observation, classification or denominator would need to change for the statement to become true. This habit makes the correction reusable on a different examination question rather than dependent on memorising this particular example.
Hearing 20: Claims hidden in a pie chart
Case. A pie chart has 15, 27 and 6 records but its caption calls every sector a nutrient result. Reasoned response. Six attempts are unresolved process records, not confirmed chemical outcomes. A pie chart can show audit categories, but its caption must say so. Without that distinction, a reader may mistake a missing record for a nutrient state. Transfer. Before accepting the conclusion, ask which observation, classification or denominator would need to change for the statement to become true. This habit makes the correction reusable on a different examination question rather than dependent on memorising this particular example.
Hearing 21: Not all negative results mean failure
Case. A club member says that a scientifically sound negative test is ‘useless’. Reasoned response. A valid negative is a meaningful observation about the tested substance under the specified conditions. The aim of a controlled investigation is to find out what happens, not to guarantee positives. Separate an unexciting result from an invalid experiment. Transfer. Before accepting the conclusion, ask which observation, classification or denominator would need to change for the statement to become true. This habit makes the correction reusable on a different examination question rather than dependent on memorising this particular example.
Hearing 22: The changed denominator
Case. A second round resolves two of six cases as positive, but the report retains the earlier 35.7% calculation. Reasoned response. Update the record: there would now be 17 positives and 44 valid classified results, with four unresolved, assuming the other counts stay unchanged. The new valid-positive fraction is 17/44, about 38.6%. State that this is a revised dataset, not a hidden correction. Transfer. Before accepting the conclusion, ask which observation, classification or denominator would need to change for the statement to become true. This habit makes the correction reusable on a different examination question rather than dependent on memorising this particular example.
Hearing 23: Reporting a confident count
Case. The writer cannot retrieve two observation sheets but wants a clean statement that 48 samples were fully tested. Reasoned response. The accurate statement is that 48 attempts were recorded, with 42 valid classified outcomes and six unresolved. ‘Fully tested’ implies completed usable evidence for every attempt, which the file does not support. Keep uncertainty visible. Transfer. Before accepting the conclusion, ask which observation, classification or denominator would need to change for the statement to become true. This habit makes the correction reusable on a different examination question rather than dependent on memorising this particular example.
Hearing 24: The overlong introduction
Case. A 300-word situational report devotes half its length to the history of nutrition lessons. Reasoned response. Give the coordinator the trial’s purpose and result early. Save words for the incorrect claims, why they are wrong, and what should change before publication. Good writing is complete without becoming an encyclopedia. Transfer. Before accepting the conclusion, ask which observation, classification or denominator would need to change for the statement to become true. This habit makes the correction reusable on a different examination question rather than dependent on memorising this particular example.
Hearing 25: The final sign-off
Case. The tally is correct and the wording is cautious, but no one has confirmed whether the sample codes were entered correctly. Reasoned response. The committee should not equate arithmetic accuracy with source accuracy. Reconcile the display with the original coded records, note who checked the figures, and release only the version whose claims can be traced to evidence. Transfer. Before accepting the conclusion, ask which observation, classification or denominator would need to change for the statement to become true. This habit makes the correction reusable on a different examination question rather than dependent on memorising this particular example.
The independent five-minute quality gate
Give yourself five minutes with no answer key. First, draw three columns labelled ‘Observation’, ‘Calculation’ and ‘Claim’. Put each factual sentence from a short school report in its correct column. Next underline any number that lacks a stated denominator. Circle words such as all, most, every, proves and safest. Then check whether the relevant data actually justify those words. Finally write one two-sentence recommendation that a school coordinator could act on without misrepresenting the trial.
A strong response will preserve five boundaries: a qualitative test is not a nutritional prescription; a valid record is not an attempt merely because it was logged; a percentage is tied to its denominator; a trend is not universal proof; and a recommendation must follow the evidence rather than precede it. These boundaries are important in Science, Mathematics and English, but each subject tests them differently. Learn the exact expectations in each official syllabus.
Model closing decision for the fictional committee
The committee should release a corrected learning display, not the original majority headline. The revised panel can state that forty-eight attempts were recorded, forty-two yielded valid classified outcomes, fifteen of those were positive and twenty-seven negative for the specified nutrient test, and six remained unresolved. It should explain that the result indicates the substance tested under recorded conditions. It should avoid ranking meals, claiming universal health effects or hiding failed controls. The coordinator may approve a simple educational account after checking the primary records and safety language.
This judgement is the end of the food-test sequence. The next learning move is to transfer the same discipline to an unfamiliar question: when a human-gas-exchange demonstration produces a visible change, ask what the observation shows, what process might explain it, and what else the experiment would need before a larger claim is justified. That is how foundations become advanced performance.
Read the exact owners rather than re-learning the same lesson
For the scientific owner, continue with Vol 0559 Biology nutrient-test evidence and the Advanced Science Tutorials food-tests guide. For the numerical owner, use Vol 0560 Mathematics valid denominators. For a report written to a reader, use Vol 0561 English situational report. The PSLE decision-first route remains useful when the page contains too much information. Check the official links above for 2027 format updates. All committee data and questions here are invented teaching examples.
