1. Why a stem-and-leaf diagram is more than an unusual graph
Suppose fifteen students record how long they waited for a school workshop. A histogram can show how many waited between ten and twenty minutes, but it cannot tell you whether a particular waiting time was twelve or nineteen minutes. A stem-and-leaf diagram has another advantage: it organises the data while retaining every original numerical observation. That makes it possible to reconstruct the exact ordered list, find the median and range, and check a claim against real individual values rather than guessed class midpoints.
For the 2027 Singapore-Cambridge Secondary Education Certificate, the official G3 directory updated on 26 October 2026 lists Mathematics as K310 (replacing reference code 4052 for 2026 and earlier). The K310 syllabus includes interpretation and construction of statistical diagrams, including stem-and-leaf diagrams; it also covers descriptive measures such as the mean, median and range and their appropriate interpretation. Both K310 written papers carry 90 marks and last 2 hours 15 minutes each, with 50% of the total assessment apiece. Students should always consult the current official syllabus for detailed boundaries rather than assuming every more advanced data technique is required.
This guide moves from place value to constructing a stem-and-leaf display, then to exact numerical conclusions and finally to comparing displays that preserve different amounts of information. The worked data are original and fictional, not actual SEC questions, school waiting-time reports or published survey findings. A useful answer should be comprehensible to a student encountering the diagram for the first time yet precise enough to withstand a challenging examination question.
2. Start with the unit and the raw observations
Here is a small fictional dataset of waiting times in minutes, already sorted: 12, 14, 17, 19, 20, 21, 21, 23, 25, 27, 29, 31, 33, 34, 38. There are fifteen observations, not fourteen, because the value twenty-one appears twice and must be counted twice. The smallest is twelve and the largest thirty-eight. Every subsequent calculation depends on preserving each observation and its frequency.
Before drawing a diagram, ask what the numbers represent and how accurately they were recorded. If the source says the values are whole minutes, do not silently pretend they are measurements to the nearest second. A stem-and-leaf display is only as accurate as its source. It cannot independently correct inconsistent timing rules or missing observations. If you have times for different events, do not combine them unless the comparison is meaningful and the task permits it.
A second check is whether the data contain zero, negative numbers or decimals. These can be represented with carefully explained keys, but the stem choice may change. Here we begin with two-digit non-negative whole numbers so the place-value structure is transparent. The method is not memorising a shape. It is deciding which digits represent the stem and which represent the leaf.
3. Build the diagram without losing an observation
Choose the tens digit as the stem and the units digit as the leaf. The ordered textual display is: stem 1 has leaves 2 4 7 9; stem 2 has leaves 0 1 1 3 5 7 9; stem 3 has leaves 1 3 4 8. In a traditional diagram the stems form a left column and the leaves appear to the right. The critical key is 2 | 3 represents 23 minutes. Without the key, another reader could confuse 2 | 3 with 2.3, 203 or some other scale.
Count the leaves: four in the teens, seven in the twenties and four in the thirties. The sum is 4 + 7 + 4 = 15, matching the original list. This count is a quick integrity check. Each leaf represents one observation, including duplicates. Leaves in each row should normally be ordered from smallest to largest so the complete arrangement can be read from top to bottom.
An unordered row such as 7, 1, 5, 1, 0, 3, 9 still contains seven values, but it frustrates reading and invites mistakes in median questions. Sorting the leaves is not cosmetic. It produces the ordered list directly, which is why the diagram can help find an exact median. A full answer should state the key and use the correct units whenever the situation involves physical quantities.
4. Find the median exactly from the ordered leaves
When there are fifteen observations, the median occupies the eighth position because (15 + 1) / 2 = 8. The first stem contains four values, so the eighth observation is the fourth leaf in the next row: 23 minutes. Count carefully: positions one to four are 12, 14, 17, 19; five to eleven are 20, 21, 21, 23, 25, 27, 29. The eighth value is 23. It is an observed value, not a midpoint estimate.
With an even number of observations, take the average of the two central values after ordering. For six values 12, 14, 17, 19, 20 and 21, the middle pair is 17 and 19, giving median 18. A median need not be one of the original values. The rule comes from the count and order of the observations, not from the numerical size of the stem labels.
Do not choose the middle row and call its first or middle leaf the median without counting all leaves. Unequal row frequencies matter. A display with one value in the teens and thirty values in the twenties has a very different middle position from a display with thirty values in the teens. The stem is a grouping device, not a vote giving each row equal weight.
5. Range and spread, not merely the most common row
The range is the greatest value minus the smallest. For the fifteen observations, 38 − 12 = 26 minutes. It is an exact range for the recorded list. It says how far apart the extremes lie; it does not say how typical either extreme is. Two classes could have the same range yet very different concentrations in the middle.
To make that distinction tangible, compare 10, 10, 10, 20, 30 with 10, 15, 20, 25, 30. Both have range 20, but the first has repeated values near one end while the second is evenly spaced. A range never describes every aspect of the distribution. Use it together with a typical-value measure and, where relevant, additional spread measures covered by the syllabus.
Read the graph before applying arithmetic. If there is an isolated leaf at the far end, ask whether it is legitimate data or a copying mistake; never remove it merely because it makes the range inconvenient. The question may require you to calculate the range including that observation. A student should distinguish correcting a documented transcription error from deciding to discard an extreme value.
6. Calculate an exact mean while the values are still available
Add every observation, counting duplicates: 12 + 14 + 17 + 19 + 20 + 21 + 21 + 23 + 25 + 27 + 29 + 31 + 33 + 34 + 38 = 364. Divide by fifteen to obtain mean 364 / 15 = 24.266… minutes, or about 24.3 minutes to three significant figures. Retain enough precision until the final answer and state any rounding. The mean differs from the median of 23 because the data are not perfectly symmetric around their centre.
The stem-and-leaf display retains the raw values, so the exact total 364 is available. That is a major contrast with a grouped table containing only class counts. A grouped table with class intervals 10 to under 20, 20 to under 30, and 30 to under 40 cannot reconstruct the exact sum. It can estimate a mean by using interval midpoints, but that estimate is not the true raw-data mean unless further information is supplied.
When an examination problem gives only the diagram, read out every stem-and-leaf pair before summing. A mistaken key can make the entire calculation wrong even when the arithmetic is flawless. A good independent check is to confirm that the mean is between the minimum and maximum values and that its magnitude fits the centre of the distribution.
7. The difference between exact and estimated statements
For our raw data, exactly nine of fifteen observations are at least 23 minutes: 23, 25, 27, 29, 31, 33, 34, 38 gives eight, not nine. This is an intentional reminder that a visual estimate should be verified by recounting. The correct count is eight, because 23 is the eighth observation and positions eight through fifteen contain eight values. The corresponding proportion is 8 / 15, about 53.3%.
A statement such as ‘at least five observations exceed thirty minutes’ can be checked directly. The values exceeding 30 are 31, 33, 34 and 38, so the claim of five is false. The exact number is four. By contrast, if you only knew there were four observations between 30 and under 40, you could count values at least thirty, but you could not say exactly how many exceeded thirty-five without seeing individual values.
Pay attention to inequalities. ‘At least 25’ includes 25; ‘more than 25’ does not. In the dataset, at least 25 gives seven observations: 25, 27, 29, 31, 33, 34, 38. More than 25 gives six. A one-word difference changes the count, and a stem-and-leaf diagram contains the raw detail needed to settle it.
8. From a stem diagram to equal-width grouped classes
Now group the fifteen values into the intervals 10 to under 20, 20 to under 30 and 30 to under 40. The frequencies are 4, 7 and 4. These equal-width classes could support a frequency histogram with adjacent bars, as explained in Vol 0198. The histogram has not changed the population of observations, but it has discarded details. The four values in the first class are not visible individually in the grouped display.
Using class midpoints 15, 25 and 35, the grouped mean estimate is (4 × 15 + 7 × 25 + 4 × 35) / 15 = (60 + 175 + 140) / 15 = 375 / 15 = 25 minutes. Compare that estimate with the true mean 364 / 15, approximately 24.27 minutes. The difference arises because class midpoints stand in for unknown individual observations. Neither result should be labelled as a measurement error; they answer questions using different levels of information.
Do not claim that a histogram is always inferior. It can communicate a large dataset clearly and quickly. The cost is lower granularity. A stem-and-leaf diagram may become unwieldy when there are hundreds of observations or many digits. Choose or interpret a display based on its purpose. The mathematics examination can reward explaining what a representation reveals and what it can no longer recover.
9. Read frequency from leaves, not from the size of stems
The stem labels 1, 2 and 3 denote tens, not frequencies. The stem 2 contains seven leaves, so its class frequency is seven. The stem 3 contains four leaves, even though the label 3 is larger than 2. Students who visually associate the largest stem label with the modal class invert the function of the diagram.
In this dataset 21 appears twice, whereas every other listed value appears once. Therefore the mode is 21 minutes. The modal stem is the twenties because that row has seven observations. These are different statements: the most frequent individual value is 21, and the most populated tens group is 20 to 29. A histogram may reveal the modal class without revealing the exact mode.
An examiner may ask you to write the most common waiting time, the most common interval or the number of observations in an interval. Treat those as three distinct prompts. Always interpret the noun attached to ‘most common’ before answering.
10. Quartiles need a stated ordering convention
Quartiles divide ordered data into parts. In the fifteen-value example, the median is 23. Under the common school method of taking the median of the lower seven values and the median of the upper seven, Q1 is 19 and Q3 is 31, making the interquartile range 12 minutes. This calculation uses the two halves excluding the overall median. Students should follow their teacher’s and question’s stated convention when quartile methods differ.
Write the seven lower values explicitly: 12, 14, 17, 19, 20, 21, 21. Their middle value is 19. The seven upper values are 25, 27, 29, 31, 33, 34, 38, whose middle value is 31. The arithmetic 31 − 19 = 12 is simple; selecting the right data subset is the harder part. If a question specifies a percentile location or another method, use that method and explain it.
This guide does not replace Vol 0103 on cumulative frequency or Vol 0107 on box plots. It explains how the exact ordered observations underpin those other representations. The stem diagram’s special value is keeping individual values recoverable instead of replacing them with interval-only counts.
11. Repeated values, zero leaves and unusual stems
Zero is a legitimate leaf. The recorded value 20 appears as stem 2, leaf 0. It must not be skipped as if it were an empty placeholder. Likewise, the two observations of 21 require two leaves labelled 1. If you write just one, the row count falls and the median, mean and frequency calculations can change. Repeated values are observations, not accidental typographical clutter.
Consider measurements such as 4, 4, 5, 6, 10, 11. One reasonable key is 1 | 0 = 10, with stems 0 and 1. Another diagram using a different scale may be valid if it includes a clear key and retains all data. Do not invent leading zeros or interpret a blank stem as necessarily indicating a missing observation; some stems may be intentionally empty to show the scale continuously.
A key is especially important when decimals appear. For 2.3, 2.5 and 3.1, a diagram might use 2 | 3 = 2.3 units. If the key is instead 2 | 3 = 23 units, the same printed symbols mean a different quantity. Never assume the scale without reading the legend.
12. Missing stems and complete interval coverage
Suppose a dataset has observations 12, 19, 31 and 38, with none in the twenties. A complete stem arrangement can show stem 1 with leaves 2 and 9, stem 2 with no leaves, and stem 3 with leaves 1 and 8. The empty middle row reveals a gap rather than a missing part of the table. Whether an exam expects the empty row depends on the diagram conventions given.
When you build a diagram, determine the smallest and largest stems first. Then write a sensible sequence so a reader can detect gaps. A missing value is not itself a missing leaf if the dataset genuinely contains no observation there. Conversely, a stem skipped by mistake could hide an entire interval and make the frequencies wrong. Check totals against the original data count.
In interpretation questions, avoid claiming that an empty row proves nobody could ever have such a value. It only says no such value was recorded in this dataset. Sampling and measurement scope matter. A mathematical data display is a description of observations, not a universal law.
13. Check what a comparison really establishes
Two groups may have equal medians but different spreads. That does not make their experiences identical. For group A, 10, 15, 20, 25, 30 has median 20 and range 20. For group B, 19, 20, 20, 20, 21 has median 20 and range 2. Group B is much more tightly clustered. The same central value can coexist with strikingly different variability.
When comparing distributions, check that units, observation definitions and sampling periods match. Waiting times measured in minutes cannot be compared numerically with waiting times measured in seconds without conversion. A set of weekday observations and a set from a special holiday event may differ for reasons not recorded in the diagram. The graph itself does not explain the cause of that difference.
Use precise language: ‘In these recorded observations, group B had a smaller range’ is supported. ‘Group B runs a universally better service’ is not established. The first statement describes the sample. The second asserts quality and generalises beyond the evidence.
14. How a bad key can break an otherwise good answer
Suppose the written key says 2 | 4 means 2.4 kilograms, but a student reads it as 24 kilograms. Their median and range may be numerically ten times too large. The leaf pattern did not change; the unit and place value did. Before starting any calculation, translate two arbitrary pairs into full measurements and see whether they make sense in the scenario.
Likewise, a key with 12 | 3 = 123 centimetres uses a two-digit stem and one-digit leaf. A student’s assumption that stems must be single digits can produce nonsense. The choice of stem width depends on data magnitude and the precision the task requires. The printed key, not a remembered template, controls the interpretation.
If an exam question gives an incomplete diagram and asks for a missing leaf, use the relevant conditions—for example a specified median or total count—to solve the unknown. Then rebuild the whole dataset to check that the inserted leaf preserves ascending order. A correct algebraic answer that violates the ordering is not a valid stem-and-leaf completion.
15. Find a missing value from a stated mean
Consider five observations 10, 12, 15, x and 23 with a stated mean of 16. The total must be 5 × 16 = 80. The four known values total 60, so x = 20. This is an exact reconstruction from the mean. In an ordered display, check that 20 fits between 15 and 23, as it does. If the answer had been 8, the displayed order would have been inconsistent with the premise.
Now suppose the key indicates measurements to one decimal place and the missing leaf belongs to the stem 2. The possible numerical values depend on the scale, such as 2.0 to 2.9 rather than 20 to 29. Do not calculate a missing digit without first interpreting the stem. When the value cannot be reconstructed uniquely, state the range or possible leaves instead of guessing.
Missing-value questions can combine logic and arithmetic, but one canonical principle remains: all stated conditions must hold simultaneously. The result needs the correct total, the correct ordering, the allowed stem and a sensible context.
16. The trap of percentage claims from small counts
Four of fifteen observations lie in the thirties, so that share is 4 / 15, approximately 26.7%. It is inaccurate to round this to 27% and then treat 27% as an exact fraction for reconstructing a sample count. Similarly, 8 / 15 gives about 53.3%, not exactly 53%. Maintain the exact fraction while doing subsequent calculations.
If a class writes ‘Most students waited at least 23 minutes,’ eight out of fifteen does form a bare majority of this small fictional dataset. But if it writes ‘Most students at the school wait at least 23 minutes,’ the sentence generalises to a population not established by the sample. The same arithmetic cannot repair a sampling problem.
The safe habit is to name the reference group: ‘Eight of the fifteen recorded waiting times were at least 23 minutes.’ It is neither awkward nor weak. It is specific. In mathematics and science, specificity is often the difference between a correct conclusion and an unsupported inference.
17. Twenty-eight worked interpretation checks
These original checks isolate common decision points and go beyond memorising one dataset. Work out each prompt before reading its answer. A tutor can select four cases per session and ask for both calculation and a sentence explaining its limit. Where a value is not recoverable, saying so is a mathematical result, not an admission that the student gave up.
Write the observations in order whenever a question feels uncertain. For a small display that takes less than a minute and protects the rest of the solution. The cases vary the key, value repetition, exactness, order and comparison so a correct method must transfer to new data.
Worked check 1. Stem 2 has leaves 1 1 3 7; key 2|3=23. Question: How many observations and which repeated value? Answer: Four observations; 21 appears twice. Reasoning: Each leaf is one observation. A repeated leaf is not deleted during transcription. Independent check: Return to the key, sample count and unit before accepting a numerical or verbal conclusion. Use only the precision retained by the stated representation.
Worked check 2. Stems 1: 4 9 and 2: 0 3 8; key 2|3=23. Question: Find median. Answer: 20. Reasoning: There are five ordered values 14, 19, 20, 23, 28, so the third is the median. Independent check: Return to the key, sample count and unit before accepting a numerical or verbal conclusion. Use only the precision retained by the stated representation.
Worked check 3. The ordered observations are 12, 12, 14, 16. Question: Find median. Answer: 13. Reasoning: An even count uses the average of two central values, here 12 and 14. Independent check: Return to the key, sample count and unit before accepting a numerical or verbal conclusion. Use only the precision retained by the stated representation.
Worked check 4. The minimum is 8 and maximum 35. Question: Find range. Answer: 27. Reasoning: Range means maximum minus minimum, not the number of possible integers. Independent check: Return to the key, sample count and unit before accepting a numerical or verbal conclusion. Use only the precision retained by the stated representation.
Worked check 5. A row labelled 4 contains six leaves. Question: Interpret its frequency. Answer: Six. Reasoning: The stem label gives place value; the leaf count gives frequency. Independent check: Return to the key, sample count and unit before accepting a numerical or verbal conclusion. Use only the precision retained by the stated representation.
Worked check 6. A key says 3|2=3.2 litres. Question: Interpret 3|7. Answer: 3.7 litres. Reasoning: The key fixes one decimal place; it is not 37 litres. Independent check: Return to the key, sample count and unit before accepting a numerical or verbal conclusion. Use only the precision retained by the stated representation.
Worked check 7. A key says 12|5=125 centimetres. Question: Interpret 13|1. Answer: 131 centimetres. Reasoning: Multi-digit stems are valid when the key says so. Independent check: Return to the key, sample count and unit before accepting a numerical or verbal conclusion. Use only the precision retained by the stated representation.
Worked check 8. A display has stems 1, 2 and 3; the stem 2 row is empty. Question: Interpret its gap. Answer: No recorded observation in the twenties. Reasoning: An empty row describes this dataset, not every possible future observation. Independent check: Return to the key, sample count and unit before accepting a numerical or verbal conclusion. Use only the precision retained by the stated representation.
Worked check 9. Stems 1: 2 5, 2: 0 4 4, 3: 1. Question: Find mode. Answer: 24. Reasoning: The individual value 24 appears twice. The modal stem and individual mode differ. Independent check: Return to the key, sample count and unit before accepting a numerical or verbal conclusion. Use only the precision retained by the stated representation.
Worked check 10. In fifteen observations, what is the median position? Question: Position? Answer: Eighth. Reasoning: Use (15+1)/2 for an odd count of observations. Independent check: Return to the key, sample count and unit before accepting a numerical or verbal conclusion. Use only the precision retained by the stated representation.
Worked check 11. Exactly nine of fifteen values exceed 20. Question: Find exact fraction. Answer: 3/5. Reasoning: Reduce 9/15 before rounding to 60%. Independent check: Return to the key, sample count and unit before accepting a numerical or verbal conclusion. Use only the precision retained by the stated representation.
Worked check 12. A grouped interval 20 to under 30 has frequency 7. Question: Can you tell how many values equal 23? Answer: No. Reasoning: Grouping preserves the count in the interval but discards individual values. Independent check: Return to the key, sample count and unit before accepting a numerical or verbal conclusion. Use only the precision retained by the stated representation.
Worked check 13. The raw sum is 364 for fifteen values. Question: Find the mean to 3 s.f. Answer: 24.3. Reasoning: Divide first; round only at the end, retaining original unit. Independent check: Return to the key, sample count and unit before accepting a numerical or verbal conclusion. Use only the precision retained by the stated representation.
Worked check 14. Four values lie in 10–under 20 with midpoint 15. Question: Their estimated sum contribution? Answer: 60. Reasoning: Multiply class midpoint by class frequency; this is an estimate. Independent check: Return to the key, sample count and unit before accepting a numerical or verbal conclusion. Use only the precision retained by the stated representation.
Worked check 15. The grouped midpoint sum is 375 over fifteen values. Question: Find grouped mean estimate. Answer: 25. Reasoning: This is not the exact raw mean because interval detail was lost. Independent check: Return to the key, sample count and unit before accepting a numerical or verbal conclusion. Use only the precision retained by the stated representation.
Worked check 16. Values are 10, 15, 20, 25, 30. Question: Find median and range. Answer: 20 and 20. Reasoning: The centre and the extreme spread describe different features. Independent check: Return to the key, sample count and unit before accepting a numerical or verbal conclusion. Use only the precision retained by the stated representation.
Worked check 17. Values are 19, 20, 20, 20, 21. Question: Find median and range. Answer: 20 and 2. Reasoning: Same median as the preceding example, much smaller range. Independent check: Return to the key, sample count and unit before accepting a numerical or verbal conclusion. Use only the precision retained by the stated representation.
Worked check 18. Five values total 80 and four sum to 60. Question: Find missing value. Answer: 20. Reasoning: Use mean or total-sum logic; verify the value fits the diagram. Independent check: Return to the key, sample count and unit before accepting a numerical or verbal conclusion. Use only the precision retained by the stated representation.
Worked check 19. The original data contain two 21s but a diagram shows one leaf. Question: What was lost? Answer: One observation. Reasoning: Omitting a repeated leaf can change total count and statistics. Independent check: Return to the key, sample count and unit before accepting a numerical or verbal conclusion. Use only the precision retained by the stated representation.
Worked check 20. The count at or above 25 is seven, and total is fifteen. Question: Find exact proportion. Answer: 7/15. Reasoning: A fraction retains exactness when percentages would be rounded. Independent check: Return to the key, sample count and unit before accepting a numerical or verbal conclusion. Use only the precision retained by the stated representation.
Worked check 21. The count strictly above 25 is six. Question: Why differs from at least 25? Answer: The observation 25 itself is excluded. Reasoning: A strict inequality changes membership at the boundary. Independent check: Return to the key, sample count and unit before accepting a numerical or verbal conclusion. Use only the precision retained by the stated representation.
Worked check 22. A display records participants only at one school event. Question: Can it prove all students wait similarly? Answer: No. Reasoning: The recorded sample alone does not establish the population pattern. Independent check: Return to the key, sample count and unit before accepting a numerical or verbal conclusion. Use only the precision retained by the stated representation.
Worked check 23. A stem key 2|0=20 gives value 20. Question: Should the leaf 0 be omitted? Answer: No. Reasoning: Zero is a legitimate leaf and observation. Independent check: Return to the key, sample count and unit before accepting a numerical or verbal conclusion. Use only the precision retained by the stated representation.
Worked check 24. Four leaves occur in stem 3 and seven in stem 2. Question: Which is modal stem? Answer: Stem 2. Reasoning: Compare row frequencies, not numerical sizes of stem labels. Independent check: Return to the key, sample count and unit before accepting a numerical or verbal conclusion. Use only the precision retained by the stated representation.
Worked check 25. A student treats key 2|3=2.3 as 23. Question: What is the source of error? Answer: Place-value misreading. Reasoning: All later statistics inherit the wrong scale until the key is corrected. Independent check: Return to the key, sample count and unit before accepting a numerical or verbal conclusion. Use only the precision retained by the stated representation.
Worked check 26. Two groups use minutes and seconds, respectively. Question: May raw numerical means be compared directly? Answer: No. Reasoning: Convert to the same unit and confirm comparable observation definitions. Independent check: Return to the key, sample count and unit before accepting a numerical or verbal conclusion. Use only the precision retained by the stated representation.
Worked check 27. A diagram has an extreme genuine observation. Question: Delete it to tidy the range? Answer: No. Reasoning: Valid recorded observations cannot be discarded merely for convenience. Independent check: Return to the key, sample count and unit before accepting a numerical or verbal conclusion. Use only the precision retained by the stated representation.
Worked check 28. A histogram and stem display use the same fifteen observations. Question: Which can reveal every exact observation? Answer: The stem-and-leaf display. Reasoning: The histogram aggregates values into intervals and loses within-class detail. Independent check: Return to the key, sample count and unit before accepting a numerical or verbal conclusion. Use only the precision retained by the stated representation.
18. Extended practice: audit a school waiting-time notice
Imagine a fictional school sign showing the stem diagram from the opening example and saying, ‘Every student waits about twenty-five minutes.’ The statement is not justified. The data are fifteen recorded waits whose values range from twelve to thirty-eight minutes. Their exact average is about 24.27 minutes and their median is 23 minutes. Those summaries are useful but cannot describe each student’s individual wait, especially when the spread is substantial.
Rewrite the sign as a responsible explanation: ‘For fifteen recorded workshop waits, the median was 23 minutes and observed times ranged from 12 to 38 minutes.’ If the school wants to make an operational decision, it should also ask whether the data cover typical sessions, what circumstances affected the longest waits and whether additional observations are needed. The statistics support the summary; they do not by themselves identify why waiting occurred.
Next imagine the same sign printed only with a three-bar histogram. It could state the class frequencies, but it could not quote the exact median and range of the original raw data unless another source preserved them. The question of what is knowable comes before the calculation. Identifying missing information is often the highest-value step in an examination problem.
19. A clear revision routine for G3 examination transfer
Begin with three minutes reading and reconstructing a key. Next sort one set of leaves, count observations and compute mean, median, mode and range. Then deliberately transform it into grouped intervals and ask which precise facts have disappeared. Finally, compare two datasets and write one defensible sentence using actual sample counts and units. A small number of thoroughly checked cases teaches more than a large number solved mechanically.
If a student struggles with the median, check whether they correctly counted duplicate leaves. If their mean is wrong, inspect the place-value key before checking arithmetic. If their conclusion overstates a histogram, ask whether the data still reveal the exact individual values. The teacher or tutor should identify the first mistaken step and give a new example to test the repair.
This chapter is a bridge to the Secondary 4 Mathematics grouped-data owner, the G3 histogram guide Vol 0198, Vol 0190 on spread and the PSLE Learning Guide. Use the official K310 syllabus to check the examination’s scope. The final rule is not complicated: preserve each observation, define the key and never claim information that the diagram no longer contains.
