Quick answer. Compare corrosion-trial data by defining what a percentage measures, finding centre and spread, and checking whether groups have matched conditions. In an invented fourteen-day record, group P has a mean affected area of 4%, Z 2% and U 14%. A two-percentage-point difference is not the same as a two-percent relative change.
This is the statistics owner for a trial-data question, not a second guide to chemical rusting or four-year maintenance costs. It covers mean, median, quartiles, IQR, standard deviation, chart accuracy, missing values and conditional decisions using original worked questions.
Follow Volume 0776 on Chemistry rusting evidence and Volume 0774 on inspection cost schedules. For advanced applied mathematics consult the G3 Paper 2 application guide, the PSLE maths bridge, and the PSLE transition learning plan.
Contents
- What this mathematics chapter actually owns
- 2027 K310 SEC examination format
- The fictional six-specimen dataset
- First sort, then read the question
- Calculate the three means
- Percentage points and percentage change differ
- Range identifies a simple spread
- Quartiles and the middle half
- Standard deviation with the official formula
- The lowest mean is not a safety certificate
- Why matching times and areas matters
- The calculation-to-claim ladder
- How to spot a misleading graph
- Outliers must be described, not erased
- Missing data and selective reporting
- Comparing consistency is a defined job
- A conditional budget extension
- Probability is not the same as frequency in six trials
- Real-world Paper 2 answer structure
- What changes from PSLE mathematics
- A one-minute checking routine
- Thirty-two worked questions
- One complete extended problem and worked solution
- Common errors and exact repairs
- Four-week practice programme
- FAQ for learners
- Verified syllabus sources
What this mathematics chapter actually owns
A maintenance committee wants a short answer: which protective treatment appears most effective in a fictional 14-day observation record? Chemistry describes why materials may rust and how protective systems work. Mathematics must interpret the numbers carefully. The critical transition is from noticing one attractive result to comparing groups fairly. This chapter therefore owns numerical descriptions of results, not chemical mechanisms or real infrastructure certification.
The easiest mistake is to choose a single minimum value. A treatment that produces one excellent specimen may still produce uneven outcomes. Another mistake is to compare average percentages without noticing that groups contained different numbers of specimens or were inspected after unequal times. We will move from ordering six observations to calculating measures of centre and spread, then to a conclusion that gives both a finding and its limitations. Every number below is invented for teaching, not a measured product test.
2027 K310 SEC examination format
For the 2027 Singapore-Cambridge Secondary Education Certificate, G3 Mathematics is K310, with 4052 as the former O-Level reference code. Official SEAB information gives Paper 1 as 2 hours 15 minutes, 90 marks and 50%, and Paper 2 as 2 hours 15 minutes, 90 marks and 50%. Paper 2 ends with a question focusing on a real-world scenario. The syllabus includes measures of central tendency, quartiles, interquartile range, standard deviation, appropriate statistical representations, misleading graphs and contextual interpretation.
The official scheme expects essential working. Non-exact numerical answers generally follow the stated three-significant-figure rule unless the question specifies another accuracy. We use exact values where possible and round clearly when needed. The questions in this article are original practice; no table is taken from an official examination paper. A legitimate mathematical method should be visible even when the final answer is simple.
The fictional six-specimen dataset
Three labelled treatments were observed for 14 days under a fictional, matched inspection protocol. The reported percentage of visible surface with rust-like deposits was: P (paint system), 2, 3, 3, 4, 5, 7; Z (zinc system), 1, 1, 2, 2, 2, 4; and U (unprotected comparison), 12, 13, 14, 14, 15, 16. There are six observations in each group. These values measure an explicitly defined visible area, not iron mass loss, structural safety or a validated prediction of useful life.
Every group has the same number of observations and inspection duration. That removes two potential sources of unfair comparison, but does not guarantee the specimens were identical in every meaningful respect. We have not been given exact surface geometry, coating thickness or long-term wear. The role of mathematics is to process the stated data reliably and to notice where a requested claim reaches beyond it.
First sort, then read the question
The numbers are already ordered here, but exam data will not always be. To find a median, quartile or range, arrange observations from smallest to largest before deciding which positions matter. With six values, the median is the mean of the third and fourth ordered values. For P those are 3 and 4, giving 3.5%. For Z both central values are 2, giving 2%. For U they are 14 and 14, giving 14%.
A median describes the centre of the ordered record; it does not reveal every observation. It is resistant to some extreme values, but two groups with equal medians can have very different distributions. In a data-interpretation question, name the measure and state what quantity the percentage refers to. “The median is 2” is weaker than “The median observed affected area for Z was 2% after 14 days.”
Calculate the three means
For P, the total is 2 + 3 + 3 + 4 + 5 + 7 = 24, so the arithmetic mean is 24 ÷ 6 = 4%. For Z, the sum is 12, giving 2%. For U, the sum is 84, giving 14%. These exact results are convenient for learning, but a real question may require careful calculator use and rounding.
A mean pools all the observations in one group. The value is not necessarily an observed specimen. It also does not tell us how variable the treatment was. A group with data 4, 4, 4 and a group with 0, 4, 8 both have a mean of 4, yet their consistency differs greatly. That is why the next step is not to announce a winner, but to compare spread.
Percentage points and percentage change differ
The difference between P’s mean 4% and Z’s mean 2% is 2 percentage points. Relative to P’s mean, Z’s mean is 50% lower because (4 − 2) ÷ 4 × 100% = 50%. These sentences are not interchangeable. “Two percent lower” can be read ambiguously; “two percentage points lower” states the absolute difference between percentages. When comparing published claims, insist on a named baseline.
The wording “Z halves rusting” would still be too strong. These means concern the fraction of visible surface affected after fourteen days. They do not measure total corrosion, internal damage or protection over years. A precise numerical statement respects its measurement definition: “Z’s mean visible affected area was half P’s in this fictional trial.”
Range identifies a simple spread
For P the range is highest minus lowest: 7 − 2 = 5 percentage points. For Z it is 4 − 1 = 3 percentage points. For U it is 16 − 12 = 4 percentage points. Z has the smallest range among the three, while P has the largest. Notice how a range uses only the highest and lowest observations; it says nothing about how the middle four values are arranged.
If the examiner asks which treatment is most consistent, one can compare range when that is the available spread measure, but the conclusion should not be stronger than the evidence. A single unusually high result can enlarge a range markedly. To understand where the typical observations lie, quartiles and standard deviation provide additional perspectives. No measure is automatically the one right description for every question.
Quartiles and the middle half
For a six-value set, find the median of the lower three and the median of the upper three using the stated school convention. P’s lower half is 2, 3, 3, so Q1 is 3; its upper half is 4, 5, 7, so Q3 is 5. The interquartile range is 5 − 3 = 2 percentage points. For Z, Q1 = 1 and Q3 = 2, so IQR = 1. For U, Q1 = 13 and Q3 = 15, so IQR = 2.
Quartiles indicate where the central half of observations lies. They support comparisons of concentration without giving every detail. A box-and-whisker representation could display the median, quartiles and extreme values. The learner must be careful not to confuse the position Q3 with the numerical value of Q3 or to divide all data by four mechanically.
Standard deviation with the official formula
For these six observations, use the usual syllabus formula for the standard deviation of a data set: square root of [mean of x squared minus square of mean x]. P has mean 4 and the sum of squared deviations from 4 equals 16. Its standard deviation is sqrt(16 ÷ 6), approximately 1.63 percentage points. Z has mean 2 and squared-deviation sum 6, so standard deviation equals sqrt(6 ÷ 6) = 1. U has mean 14 and squared-deviation sum 10, so its standard deviation is sqrt(10 ÷ 6), about 1.29.
These calculations use the full supplied data set and the K310 school-level formula; they are not estimates of a larger population’s unknown uncertainty. Display enough intermediate working to show the intended method, then use a suitable number of significant figures. A calculator’s answer without a labelled interpretation is incomplete in a multi-part contextual task.
The lowest mean is not a safety certificate
Z’s mean visible affected area is lower than P’s and U’s in this record. The next sentence must identify the exact scope: six specimens per group observed for fourteen days under the supplied conditions. We do not know whether a small area of visible damage conceals serious structural weakness or whether a coating will last longer outdoors. There is no conversion factor from the observed percentage to years of safe service.
This difference between a mathematical comparison and a real decision is central to good examination performance. You can rank the observed sample means confidently while refusing a claim that requires absent evidence. It is not indecisive to be exact about limits. In fact, the more complicated the story becomes, the more important it is to separate the calculation, the interpretation and the decision-maker’s additional constraints.
Why matching times and areas matters
Suppose another dataset reports P after seven days and Z after fourteen days. Direct means no longer compare the same exposure period. One might calculate an average change per day if the question defines the measurement and assumptions, but it would be unjustified to assert that corrosion grows linearly with time without evidence. Rust-like area can grow unevenly, and a percentage cannot exceed 100%.
Equal-looking percentages can also hide different total affected areas when specimen sizes differ. If one coupon has surface area 25 square centimetres and another 50, an affected percentage of 4% represents 1 square centimetre on the first and 2 on the second. Relative and absolute measurements answer different questions. An exam scenario can test whether the student notices this before applying a familiar formula.
The calculation-to-claim ladder
Start with the data definition: visible affected area as a percentage. Then show one valid calculation, including its denominator and unit. Next compare treatments using the same measure and a consistent baseline. After that identify any visible variation. End with a carefully limited conclusion. “The mean for Z was 2%, compared with 4% for P, and Z had the smaller IQR of 1 percentage point; therefore Z produced lower and less dispersed recorded values under the stated trial conditions” is a strong mathematics answer.
A weaker conclusion jumps immediately from a mean to “always best”. You may be asked which option a committee should select. In that case, read the decision criteria: a strict cost ceiling, a minimum performance threshold or an additional requirement for engineering approval can change the feasible set. The smallest sample mean is not necessarily the only relevant criterion.
How to spot a misleading graph
A chart with a vertical axis starting at 10 rather than 0 can make a small numerical difference seem dramatic. A graph showing percentages without units invites confusion between percentage points and relative changes. A chart using different scales for two panels may imply unequal effects even when the raw numbers are comparable. Check the scale, intervals, labels and whether the displayed values are observations or summaries.
For this record, a box-and-whisker plot is useful for median and spread while a table preserves exact values. A bar chart of means alone hides variation, so a responsible reader would inspect the underlying data as well. If an exam question asks why a display might mislead, name the design feature and its effect on interpretation. Do not just write “the graph is confusing”.
Outliers must be described, not erased
P’s highest observation is 7%, larger than its other five values. That alone does not establish that the observation is an error; the question may require a more specific criterion to call a point an outlier. Removing it changes P’s mean from 4% to 17 ÷ 5 = 3.4%. Such a change might look helpful to P but would be dishonest without a valid documented reason for exclusion.
A correct examination response may compare the mean before and after an unusual reading, explain the sensitivity and ask what further check is needed. If one specimen was damaged during measurement, the problem should say so or invite a reasoned evaluation. The learner should not manufacture facts to achieve a preferred conclusion. Keep the raw record and the proposed correction distinct.
Missing data and selective reporting
Imagine Z originally had six trials but the report prints only five. Its stated mean could be misleading because the sixth value is unknown. Replacing the missing observation with 0 would be a serious error: unknown does not mean zero. If five known Z observations total 8, the complete mean would be (8 + x) ÷ 6 where x is the missing measurement. A question can ask what range of x would keep Z’s mean below a given benchmark.
This is algebra inside statistics, a common real-world integration. To remain below 3%, (8 + x) ÷ 6 must be less than 3, so x is less than 10. If the percentage x is constrained between 0 and 100, that gives a useful but conditional result. Keep the inequality direction correct and explain what x represents.
Comparing consistency is a defined job
A student might say U is “more reliable” because its values cluster around 14. The word reliable is ambiguous: tightly grouped measurements can consistently be high. In the supplied sample, Z has the smallest standard deviation, 1 percentage point, and the smallest IQR, 1 percentage point. That supports greater numerical consistency under those descriptive measures. It does not certify the product, its manufacturing reliability or long-term safety.
When a problem asks for the more consistent set, make the comparison with the measure requested. When a problem asks which set has the lower typical value, choose mean or median as appropriate. The skill is to select the statistic that answers the reader’s question, not to recite every statistic you can calculate.
A conditional budget extension
Add a separate fictional condition: P costs $360 per group, Z costs $430, and U costs $120. Suppose the available budget is no more than $400. Z has the smallest observed mean affected area, yet it fails the given cost condition. The available options by price are P and U. If the decision requires only the lowest recorded mean among affordable options, P is preferred to U in this mathematical model. That does not make U unsafe or P certified.
This example differs from Volume 0774’s recurring inspection calculations. Here a statistical performance measure is combined with a single constraint. When two criteria pull in different directions, first identify feasibility, then compare qualifying candidates. Never rank an infeasible option as the recommended solution without acknowledging the failed condition.
Probability is not the same as frequency in six trials
The observation that two of six P specimens have a reported value of at least 5% gives a sample proportion of 2/6 = 1/3. It does not prove that every future P specimen has exactly a one-third probability of reaching that level. Such a probability claim would need a justified statistical model and more information about sampling. In a K310 context, a question may ask for an empirical fraction; state precisely that it comes from this dataset.
This distinction avoids one of the most common overclaims in a public report. A six-observation summary is an account of those six. Treating it as an exact forecast converts descriptive statistics into an unsupported prediction. In examination writing, the phrase “among the six recorded specimens” protects that boundary without adding unnecessary apology.
Real-world Paper 2 answer structure
The final K310 Paper 2 question has an explicit real-world focus. An effective response begins by defining variables and identifying required quantities, calculates cleanly, compares against the requested criterion and ends with a sentence that returns to the context. If the task includes a fixed budget, a percentage threshold and a range of acceptable variation, check all the requirements separately.
Write the working at the point where it helps the reader follow the decision. For example, show sum and denominator for the mean, then state the mean with its unit. Explain why the lowest number matters or does not matter according to the rule. A bare answer such as “Z because 2” cannot show whether the learner understands the data or has simply guessed the smallest printed value.
What changes from PSLE mathematics
At primary level, a student may already be able to divide totals to find a mean or calculate basic percentages. G3 SEC Mathematics builds on those foundations with deeper use of distributions, spread, statistical representations and mixed constraints. The learning jump is not merely more difficult arithmetic. It is deciding what the calculation means and how far the evidence reaches.
Students who are shaky with ratios and percentage baselines should practise those before standard deviation. Start with straightforward groups and transparent totals. Then add interpretation, misleading display questions and final real-world decisions. Revisiting earlier misconceptions after a few days builds stronger transfer than finishing a large worksheet once and forgetting it. The goal is dependable problem solving, not an impressive number of completed pages.
A one-minute checking routine
Before submitting, ask whether the groups have the same sample size, duration and measurement meaning. Confirm the arithmetic sums. Distinguish percentages from percentage-point differences. Check that you used the requested measure of centre or spread. Ensure a conditional statement such as “under this fictional trial” appears before a broad interpretation. Finally, make sure the answer includes the appropriate unit and has not rounded too early.
Checking is especially valuable with calculator statistics: a single data-entry error can change several later answers while preserving the appearance of precision. Re-read the table and verify that every number entered belongs to the correct group. A quick plausibility check also catches impossibilities: an average must lie between the smallest and largest data values.
Thirty-two original worked mathematics questions
1. Sample size
Question. How many observations were recorded for treatment P?
Worked response. Six. Count the listed percentages rather than treating the number of treatments as the sample size. Each group contains six separate values. Check that the interpretation keeps the measured quantity, comparison group and time period visible. Arithmetic earns its place in a contextual answer when it supports the particular conclusion requested, not when it merely produces a tidy number.
2. P total
Question. Find the sum of the six P observations.
Worked response. 2 + 3 + 3 + 4 + 5 + 7 = 24. The unit is a sum of percentage-valued readings; use the sum to find their arithmetic mean. Check that the interpretation keeps the measured quantity, comparison group and time period visible. Arithmetic earns its place in a contextual answer when it supports the particular conclusion requested, not when it merely produces a tidy number.
3. P mean
Question. Calculate the mean visible affected area for P.
Worked response. 24 ÷ 6 = 4%. State that this is the average recorded area percentage after 14 days, not a rate per day. Check that the interpretation keeps the measured quantity, comparison group and time period visible. Arithmetic earns its place in a contextual answer when it supports the particular conclusion requested, not when it merely produces a tidy number.
4. Z mean
Question. Calculate Z’s mean.
Worked response. (1 + 1 + 2 + 2 + 2 + 4) ÷ 6 = 12 ÷ 6 = 2%. The denominator is six observations. Check that the interpretation keeps the measured quantity, comparison group and time period visible. Arithmetic earns its place in a contextual answer when it supports the particular conclusion requested, not when it merely produces a tidy number.
5. U mean
Question. Calculate U’s mean.
Worked response. (12 + 13 + 14 + 14 + 15 + 16) ÷ 6 = 84 ÷ 6 = 14%. Do not mistake the middle reading 14 for the only value contributing. Check that the interpretation keeps the measured quantity, comparison group and time period visible. Arithmetic earns its place in a contextual answer when it supports the particular conclusion requested, not when it merely produces a tidy number.
6. Mean difference
Question. By how many percentage points does P’s mean exceed Z’s?
Worked response. 4% − 2% = 2 percentage points. This describes the difference between two percentages. Check that the interpretation keeps the measured quantity, comparison group and time period visible. Arithmetic earns its place in a contextual answer when it supports the particular conclusion requested, not when it merely produces a tidy number.
7. Relative reduction
Question. What is Z’s relative reduction in mean compared with P?
Worked response. (4 − 2) ÷ 4 × 100% = 50%. The baseline is P’s 4%, not Z’s 2%; switching denominators gives a different answer. Check that the interpretation keeps the measured quantity, comparison group and time period visible. Arithmetic earns its place in a contextual answer when it supports the particular conclusion requested, not when it merely produces a tidy number.
8. P median
Question. What is P’s median?
Worked response. The ordered central pair is 3 and 4. Their mean is 3.5%. There are six observations, so do not select the third item alone. Check that the interpretation keeps the measured quantity, comparison group and time period visible. Arithmetic earns its place in a contextual answer when it supports the particular conclusion requested, not when it merely produces a tidy number.
9. Z median
Question. What is Z’s median?
Worked response. The third and fourth values are both 2, so the median is 2%. This matches Z’s mean but need not always do so. Check that the interpretation keeps the measured quantity, comparison group and time period visible. Arithmetic earns its place in a contextual answer when it supports the particular conclusion requested, not when it merely produces a tidy number.
10. P range
Question. Find P’s range.
Worked response. 7 − 2 = 5 percentage points. Range is maximum minus minimum, not sum of deviations. Check that the interpretation keeps the measured quantity, comparison group and time period visible. Arithmetic earns its place in a contextual answer when it supports the particular conclusion requested, not when it merely produces a tidy number.
11. Z range
Question. Find Z’s range.
Worked response. 4 − 1 = 3 percentage points. Z has a smaller observed range than P in this dataset. Check that the interpretation keeps the measured quantity, comparison group and time period visible. Arithmetic earns its place in a contextual answer when it supports the particular conclusion requested, not when it merely produces a tidy number.
12. U range
Question. Find U’s range.
Worked response. 16 − 12 = 4 percentage points. A larger mean does not automatically imply a larger range. Check that the interpretation keeps the measured quantity, comparison group and time period visible. Arithmetic earns its place in a contextual answer when it supports the particular conclusion requested, not when it merely produces a tidy number.
13. P Q1
Question. Find P’s lower quartile using the median-of-halves convention.
Worked response. The lower three ordered observations are 2, 3, 3. Their median is 3%, so Q1 = 3%. Check that the interpretation keeps the measured quantity, comparison group and time period visible. Arithmetic earns its place in a contextual answer when it supports the particular conclusion requested, not when it merely produces a tidy number.
14. P Q3
Question. Find P’s upper quartile.
Worked response. The upper three observations are 4, 5, 7. Their median is 5%, so Q3 = 5%. Check that the interpretation keeps the measured quantity, comparison group and time period visible. Arithmetic earns its place in a contextual answer when it supports the particular conclusion requested, not when it merely produces a tidy number.
15. P IQR
Question. Find P’s interquartile range.
Worked response. Q3 − Q1 = 5 − 3 = 2 percentage points. It describes the width of the middle half. Check that the interpretation keeps the measured quantity, comparison group and time period visible. Arithmetic earns its place in a contextual answer when it supports the particular conclusion requested, not when it merely produces a tidy number.
16. Z IQR
Question. Find Z’s interquartile range.
Worked response. Lower-half median 1%; upper-half median 2%; IQR = 1 percentage point. This is less than P’s IQR. Check that the interpretation keeps the measured quantity, comparison group and time period visible. Arithmetic earns its place in a contextual answer when it supports the particular conclusion requested, not when it merely produces a tidy number.
17. P SD
Question. Find P’s standard deviation to three significant figures.
Worked response. With mean 4, the squared-deviation sum is 16. Standard deviation = sqrt(16 ÷ 6) = 1.63 percentage points, using the full-data formula. Check that the interpretation keeps the measured quantity, comparison group and time period visible. Arithmetic earns its place in a contextual answer when it supports the particular conclusion requested, not when it merely produces a tidy number.
18. Z SD
Question. Find Z’s standard deviation.
Worked response. With mean 2, squared deviations sum to 6. Standard deviation = sqrt(6 ÷ 6) = 1 percentage point. Check that the interpretation keeps the measured quantity, comparison group and time period visible. Arithmetic earns its place in a contextual answer when it supports the particular conclusion requested, not when it merely produces a tidy number.
19. U SD
Question. Find U’s standard deviation to three significant figures.
Worked response. With mean 14, squared deviations sum to 10. Standard deviation = sqrt(10 ÷ 6) = 1.29 percentage points. Check that the interpretation keeps the measured quantity, comparison group and time period visible. Arithmetic earns its place in a contextual answer when it supports the particular conclusion requested, not when it merely produces a tidy number.
20. Consistency
Question. Which group is most consistent by standard deviation?
Worked response. Z, because 1 is less than approximately 1.29 for U and 1.63 for P. State the measure used; consistency is not the same as suitability. Check that the interpretation keeps the measured quantity, comparison group and time period visible. Arithmetic earns its place in a contextual answer when it supports the particular conclusion requested, not when it merely produces a tidy number.
21. Truncated axis
Question. What is problematic about a bar chart starting at 12% for all groups?
Worked response. The scale excludes most values and visually magnifies differences among readings near 12. Some data may not even fit. Check the origin and display before drawing conclusions. Check that the interpretation keeps the measured quantity, comparison group and time period visible. Arithmetic earns its place in a contextual answer when it supports the particular conclusion requested, not when it merely produces a tidy number.
22. Unequal groups
Question. A fourth group has only two observations. Can its mean be compared without caveat?
Worked response. It can be calculated, but the amount of evidence is unequal. Do not imply the smaller group represents variability as well as six independent observations. Check that the interpretation keeps the measured quantity, comparison group and time period visible. Arithmetic earns its place in a contextual answer when it supports the particular conclusion requested, not when it merely produces a tidy number.
23. Unequal days
Question. P is measured after 7 days and Z after 14. What is the main confounder?
Worked response. Time. The measurements are not from the same exposure period. A simple division by days does not prove the process is linear. Check that the interpretation keeps the measured quantity, comparison group and time period visible. Arithmetic earns its place in a contextual answer when it supports the particular conclusion requested, not when it merely produces a tidy number.
24. Area conversion
Question. A 50 cm² coupon has 4% affected area. Find the area.
Worked response. 0.04 × 50 = 2 cm². Percent of area and absolute area are not interchangeable. Check that the interpretation keeps the measured quantity, comparison group and time period visible. Arithmetic earns its place in a contextual answer when it supports the particular conclusion requested, not when it merely produces a tidy number.
25. Outlier removal
Question. Calculate P’s mean without the reading 7.
Worked response. The remaining total is 17 across five readings; mean = 17/5 = 3.4%. Explain why an observation is excluded before presenting a changed mean as definitive. Check that the interpretation keeps the measured quantity, comparison group and time period visible. Arithmetic earns its place in a contextual answer when it supports the particular conclusion requested, not when it merely produces a tidy number.
26. Missing value
Question. If the known five readings total 8 and sixth is x, give the complete mean.
Worked response. (8 + x)/6. Writing 8/5 answers a different question; replacing x with zero invents evidence. Check that the interpretation keeps the measured quantity, comparison group and time period visible. Arithmetic earns its place in a contextual answer when it supports the particular conclusion requested, not when it merely produces a tidy number.
27. Missing threshold
Question. Find x if the six-reading mean must be less than 3.
Worked response. (8 + x)/6 < 3 gives x < 10. Since x is a percent, it should also satisfy the physical definition and any constraints provided. Check that the interpretation keeps the measured quantity, comparison group and time period visible. Arithmetic earns its place in a contextual answer when it supports the particular conclusion requested, not when it merely produces a tidy number.
28. Budget test
Question. Which of P $360, Z $430 and U $120 meet a $400 ceiling?
Worked response. P and U. Z does not. The best sample mean cannot override a stated hard constraint. Check that the interpretation keeps the measured quantity, comparison group and time period visible. Arithmetic earns its place in a contextual answer when it supports the particular conclusion requested, not when it merely produces a tidy number.
29. Conditional choice
Question. Among affordable options, which has lower recorded mean affected area?
Worked response. P at 4%, compared with U at 14%. This is a trial-data decision under the fictional rule, not an engineering certification. Check that the interpretation keeps the measured quantity, comparison group and time period visible. Arithmetic earns its place in a contextual answer when it supports the particular conclusion requested, not when it merely produces a tidy number.
30. Empirical frequency
Question. What fraction of P observations are at least 5%?
Worked response. Two of six, so 1/3. Report it as the observed sample fraction, not as a proven future probability. Check that the interpretation keeps the measured quantity, comparison group and time period visible. Arithmetic earns its place in a contextual answer when it supports the particular conclusion requested, not when it merely produces a tidy number.
31. Best conclusion
Question. State the defensible overall comparison.
Worked response. Among six matched observations per treatment after 14 days, Z had the lowest mean recorded visible affected area and the smallest spread under the named measures. Long-term safety and lifetime remain untested. Check that the interpretation keeps the measured quantity, comparison group and time period visible. Arithmetic earns its place in a contextual answer when it supports the particular conclusion requested, not when it merely produces a tidy number.
32. Wrong baseline
Question. Why is “Z is 100% lower than P” wrong for these means?
Worked response. The reduction is (4 − 2)/4 = 50%, not 100%. Dividing by the lower value answers a different relative comparison. Check that the interpretation keeps the measured quantity, comparison group and time period visible. Arithmetic earns its place in a contextual answer when it supports the particular conclusion requested, not when it merely produces a tidy number.
One complete extended problem and worked solution
A school maintenance group is shown the six-per-treatment record and told it may spend at most $400 on its initial option. The simple prices are P $360, Z $430 and U $120. The group wishes to minimise the observed mean percentage of affected area while respecting the cost ceiling. A good solution first calculates or reads the group means: 4%, 2% and 14%. Next it tests the ceiling and eliminates Z. Finally it compares the two remaining means and selects P for this mathematical model. The explanation should explicitly say that the more effective recorded sample mean for Z is not irrelevant; it simply belongs to an option that fails the stated cost constraint.
Now the committee asks whether the selected option is certified safe. No conclusion about certification follows from these data. The quantitative question was only a constrained comparison of recorded observations. A mathematically correct student should answer the two questions separately, rather than allowing the practical story to make the calculation appear to prove more than it does.
Common errors and exact repairs
Error one: announcing a “2% reduction” when the difference is two percentage points. Repair: name the measure, choose the baseline and calculate the relative change separately. Error two: treating the IQR as Q1 − Q3. Repair: higher quartile minus lower quartile. Error three: reporting a calculated standard deviation as a percentage of the mean without being asked. Repair: the spread measure here is in percentage points because each datum is a percentage-valued measurement. Error four: deleting an extreme reading because it inconveniences the preferred treatment. Repair: retain all measurements unless there is a documented reason to exclude one.
Error five: giving a real maintenance recommendation solely from the lowest sample mean. Repair: state the fictional criteria and acknowledge missing engineering evidence. Error six: using different observation periods and pretending a direct comparison is fair. Repair: compare like with like or explicitly identify the needed adjustment and its assumptions.
Four-week practice programme
Week 1 begins with ordering lists, finding means and medians, and checking percentages. Write a complete interpretation sentence after each numerical answer. Week 2 adds range, quartiles and interquartile range with graphical questions. Compare what each measure can and cannot reveal. Week 3 uses standard deviation and mixed sets of values: enter the data carefully, calculate the spread, then decide which conclusion is justified. Week 4 combines a multi-part dataset, a misleading chart, an inequality or budget condition, and a final paragraph explaining the decision.
Revisit errors at spaced intervals rather than immediately copying the correct answer. A simple record of error types—baseline, ordering, formula, unit, interpretation, or evidence—reveals whether a student needs arithmetic revision or conceptual work. When moving back to English and Science, retain the same discipline: give the evidence first and limit the claim to what has been shown.
FAQ for learners
Does a lower mean imply lower risk in all future situations? No; the mean describes the given data and may not represent a different setting. Is standard deviation in the 2027 K310 syllabus? Yes, both calculation and comparison of data sets are listed. Does a smaller standard deviation mean a smaller mean? No; these are different properties. Can an exam ask for a real-world conclusion? Yes, Paper 2 explicitly has a final applied scenario. Must all numbers be rounded immediately? No; retain accuracy during working and follow the official rounding or question-specific rule for the final response. What if different quartile conventions exist? Use the one taught and appropriate to the examination context, stating the ordered data and chosen method where ambiguity matters.
The broader lesson is that statistics support decisions only when their meaning and assumptions are made visible.
Verified syllabus sources and continued route
The SEAB 2027 K310 G3 Mathematics syllabus contains the statistics content, the two-paper scheme and the applied Paper 2 requirement. The official G3 list confirms the 2027 code K310 and previous 4052. The SEC overview explains the 2027 transition. Return to the earlier Chemistry guide if the question asks about oxygen, water, barriers or zinc mechanisms; numbers alone cannot explain the material science.
