BTT Mathematics / Primary Mathematics Learning Hub / Statistical Thinking

Primary Mathematics: Statistical Questions, Data Collection and Variability | Worked Learning Guide

Statistics begins when the answer is not one fixed number waiting to be uncovered, but a pattern hidden inside data that vary.

Primary Mathematics already asks learners to read tables, graphs, averages and percentages. A deeper statistical layer begins one step earlier: What question are we actually trying to answer, and what data would be needed?

“How old is this one student?” has one answer. “How old are the students in this group?” anticipates variation. “How many minutes do pupils in this class spend reading each day?” requires data from several pupils and will almost certainly produce different values. That expectation of variability changes how the question is investigated and how the answer should be reported.

This guide is an upper-Primary enrichment and transition bridge. Formal statistical-question terminology is not presented as a compulsory requirement for every Singapore Primary level. The purpose is to strengthen data literacy before later statistical study.

Statistical questions · Variability · Collect data · Data types · Bias and quality · 24 questions · Worked solutions

1. Statistical questions expect data that vary

Question A: How many days are in September?

Question B: How many minutes do pupils in this class take to travel to school?

A has one fixed answer. B requires data from several pupils and those travel times are expected to vary. B is a statistical question.

Worked example: one person versus a group

“What is Ali’s height?” is answered by one measurement.

“What are the heights of pupils in Ali’s class?” requires a collection of measurements and the values will differ.

The second question creates a distribution rather than one isolated value.

2. Statistical questions need a population or group

A question such as “What is the typical mass of school bags carried by Primary pupils in this class?” identifies:

• a group: Primary pupils in this class,
• a variable: school-bag mass,
• a reason to expect different values.

Those three features make data collection possible.

Weak versus stronger wording

Weak: “Are school bags heavy?”

Stronger: “What is the typical mass, in kilograms, of school bags carried by pupils in this class on Monday morning?”

The stronger question defines a measurable variable, unit, group and time condition.

3. Variability is not an error to be removed

If five pupils report reading times of 15, 20, 20, 25 and 40 minutes, the differences are part of the data.

Statistics asks:

• What values are common?
• What is a typical value?
• How spread out are the values?
• Are there unusual values?

Variation is the phenomenon to understand, not noise that should automatically be averaged away.

Worked example: same average, different variation

Data A: 10, 10, 10, 10.

Data B: 4, 8, 12, 16.

Both have mean 10. But A has no variation while B is spread widely. The average alone does not describe both sets completely.

4. Repeated measurements can vary too

Even one person may generate variable data if measured repeatedly.

Question: “How many minutes does one pupil take to complete a daily reading routine over ten school days?”

The same pupil may record different times because daily conditions differ.

A statistical question can concern variation across people, across objects, across time or across repeated trials.

5. Define the variable precisely

“Exercise” could mean many things.

Better variable definitions include:

• minutes of physical activity per day,
• number of laps completed,
• distance walked in kilometres,
• yes/no participation in a sports club.

Different variable definitions produce different data and answer different questions.

6. Decide whether to collect from everyone or from a sample

A census collects data from the whole target group. A sample collects data from part of the group.

If the class has 30 pupils and the question concerns only that class, measuring all 30 may be practical.

If the question concerns every pupil in a large school, collecting from everyone may be difficult, so a sample may be used.

A sample is useful only when it represents the target group well enough for the intended conclusion.

7. Match the data-collection method to the variable

Variable	Possible collection method
height	measure with a suitable length tool
travel time	record elapsed time
favourite fruit	survey response
number of books read	count or record
temperature	thermometer reading

Good collection methods match the mathematical quantity.

8. Fix the unit before collecting numerical data

If some pupils record height in centimetres and others in metres without clear conversion, the dataset becomes harder to interpret and easier to mishandle.

Choose one unit in advance where possible.

Worked example

Heights 1.42 m, 138 cm and 1.51 m should be converted to a common unit before analysis: 142 cm, 138 cm, 151 cm.

9. Numerical and categorical data answer different questions

Numerical data record quantities such as age, height, time or count.

Categorical data record groups or labels such as transport mode, favourite subject or yes/no response.

Averages are meaningful for numerical quantities; they are not meaningful for arbitrary category names.

Worked example

Travel modes: bus, walk, car, bus, train.

We can count frequencies, but “average transport mode” is not a meaningful arithmetic quantity.

10. Discrete and continuous numerical data

Counts such as number of siblings or books are usually discrete whole numbers.

Measurements such as height, time and mass can vary continuously within the measurement precision.

This distinction affects what values are possible and how data may be displayed.

11. Recording precision belongs to the collection plan

If travel time is measured to the nearest minute, 7 min and 7 min 20 s may both be recorded as 7 min.

The recorded data are therefore less precise than the underlying time.

Do not later report conclusions to a precision finer than the data support.

12. A fair question avoids leading language

Leading: “Don’t you agree that school lunch is too expensive?”

More neutral: “How would you rate the price of school lunch: very low, low, reasonable, high or very high?”

Question wording can influence responses.

13. Convenience samples may be biased

Suppose the question is about how all pupils travel to school, but the survey is conducted only at the bicycle racks.

The sample is likely to overrepresent cyclists.

The problem is not sample size alone. It is how the sample was selected.

14. Voluntary-response samples can be biased

An online poll asking “Should homework be reduced?” may attract pupils with especially strong opinions.

The results describe the respondents accurately, but may not represent all pupils.

15. Missing data should not disappear silently

If 30 pupils are surveyed but only 24 respond, report that fact.

The missing six responses may matter if non-response is related to the question.

16. Measurement procedures should be consistent

If bag mass is measured before lunch for some pupils and after lunch for others, the procedure changes.

If the investigation asks for a fair comparison, collection conditions should be as consistent as practical.

17. Data should answer the question that was posed

Question: “How long do pupils take to travel to school?”

Collecting only distance data does not directly answer travel time.

Distance may help explain variation later, but it is not the requested variable.

18. A statistical question can have several valid summaries

Question: “How long do pupils take to travel to school?”

Possible summaries include:

• mean time,
• median time,
• range,
• most common interval,
• full distribution.

The best summary depends on what the question is trying to communicate.

19. A single extreme value can change some summaries more than others

Data: 10, 12, 13, 14, 15.

Mean=12.8, median=13.

Replace15 with40:

Mean rises sharply; median remains13.

This is one reason to inspect the full dataset rather than rely on one statistic.

20. Statistical conclusions should keep variability visible

Weak: “Pupils take 18 minutes to get to school.”

Stronger: “In this class sample, travel times ranged from 6 to 42 minutes, with a median of 18 minutes.”

The stronger statement does not turn a distribution into a false single-value rule.

21. Correlation-like patterns do not automatically show cause

If pupils who travel farther tend to take longer, the data show an association between distance and time.

They do not by themselves prove that distance is the only cause. Transport mode, traffic and route conditions may also matter.

This is an enrichment-level caution in evidence-based reasoning.

22. Data collection can create new questions

Suppose travel times vary widely.

New statistical questions might include:

• Does travel mode relate to travel time?
• Are morning times more variable than afternoon times?
• How different are median times for different transport modes?

Statistics is often a cycle rather than one isolated calculation.

23. Common statistical-question errors

• Calling every question involving numbers statistical.
• Collecting data unrelated to the stated variable.
• Mixing units.
• Using leading survey language.
• Assuming a convenient sample represents everyone.
• Ignoring missing responses.
• Reporting excessive precision.
• Collapsing a variable dataset into one number without acknowledging spread.

24. A statistical-question protocol

1. Question: Does the answer require data that vary?
2. Group: Who or what is the target population?
3. Variable: What exactly will be measured or recorded?
4. Method: How will the data be collected consistently?
5. Sample: Who will actually be observed?
6. Unit: What units or categories will be used?
7. Quality: What bias, missing data or measurement limits may matter?

25. Practice: 24 original questions

Questions 1–8: Statistical or not?

1. How many days are in February 2028?
2. How many minutes do pupils in this class spend reading each evening?
3. What is the height of one named pupil?
4. What are the heights of pupils in this class?
5. What is the favourite fruit of one named pupil?
6. What fruits are most commonly preferred by pupils in this class?
7. How many bricks are in this one wall?
8. How many books do pupils in this class own at home?

Questions 9–16: Collection plans

9. For the question “How long do pupils take to travel to school?”, name the variable and a unit.
10. Which is better for measuring pupil height: survey guess or measurement? Explain.
11. A school-travel survey is taken only at bicycle racks. Identify the likely bias.
12. Rewrite “Don’t you think homework is too much?” more neutrally.
13. Heights are recorded as 1.42 m,138 cm,1.51 m. Convert to one common unit.
14. Thirty pupils are surveyed;24 respond. What fact should be reported?
15. Classify “favourite subject” as numerical or categorical data.
16. Classify “number of siblings” as numerical data and state whether whole-number values are expected.

Questions 17–24: Variability and conclusions

17. Data A=10,10,10,10 and B=4,8,12,16. Compare their means and variation.
18. Data 10,12,13,14,15. Find median.
19. Replace15 by40 in Question18. What happens to median?
20. Explain why “Pupils take18 minutes to school” may be too strong when travel times vary.
21. Give one reason distance and travel time may be associated without distance being the only cause.
22. Create one statistical question about school reading habits.
23. Create one non-statistical question about the same topic.
24. Design a simple data-collection plan for your statistical question, including group, variable and unit/category.

26. Worked solutions

1. Not statistical: fixed calendar answer. 2. Statistical: collect variable reading-time data. 3. Not statistical in this form: one measurement. 4. Statistical: multiple heights vary.

5. Not statistical in this form: one response. 6. Statistical: categorical preferences vary across pupils. 7. Not statistical in this form: one count. 8. Statistical: household book counts vary.

9. Travel time, e.g. minutes. 10. Measurement is more appropriate because height is a physical quantity requiring observed data. 11. Cyclists are overrepresented. 12. Example: “How would you describe the amount of homework you receive?” with balanced response options.

13. 142 cm,138 cm,151 cm. 14. Report 24 responses from 30 pupils and note 6 non-responses. 15. Categorical. 16. Numerical discrete count, usually whole numbers.

17. Both means10; A has no variation, B varies widely. 18. Median13. 19. Median remains13. 20. A single value hides the distribution; specify a summary and spread or range.

21. Transport mode, traffic or route conditions may also influence time. 22. Example: “How many minutes do pupils in this class read for pleasure on a typical weekday?” 23. Example: “How many pages are in this one book?” 24. Answers vary; valid plans must match the stated variable and target group.

27. Investigation laboratory: from question to dataset

Question: “How many minutes do pupils in this class spend travelling to school?”

Plan:

• Target group: all pupils in the class.
• Variable: one-way travel time on a normal school morning.
• Unit: minutes.
• Collection: record one value per pupil using the same definition of start and end time.
• Quality note: unusual disruptions such as road closures may create atypical values.

After collection, the next guide in this layer asks how the resulting distribution should be described.

28. Parent and tutor guide

When teaching graphs, begin one step earlier than the graph. Ask what question the data are supposed to answer and why different values are expected. This prevents graphing from becoming a drawing exercise detached from statistical reasoning.

Use everyday class data carefully and protect privacy. Synthetic datasets can teach the same mathematics when real personal data are unnecessary.

29. Mastery receipt

• I distinguish fixed-answer and statistical questions.
• I identify the target group and variable.
• I expect variability where appropriate.
• I choose suitable units and collection methods.
• I distinguish numerical and categorical data.
• I recognise simple sample and wording bias.
• I report missing data and precision honestly.
• I keep variability visible when summarising a dataset.

Sources and scope

Illustrative Mathematics — Identifying Statistical Questions describes statistical questions as questions answered by collecting data in which variability is expected.

Illustrative Mathematics — Buttons: Statistical Questions provides a related distinction between fixed-answer and variable-data questions.

For Singapore Primary curriculum scope, use the MOE Primary Mathematics Syllabus P1–P6, updated October 2025. This page is an upper-Primary enrichment/transition guide rather than a claim that formal statistical-question terminology is compulsory at every Primary level.

Continue the Statistical Thinking collection

• Distributions, Center, Spread and Typical Values
• Comparing Data Groups, Consistency and Variability
• Data Claims, Sampling, Misleading Graphs and Evidence
• BTT Primary Mathematics Learning Hub

The Quiet Return

Before asking what a graph says, ask why the data exist. A strong statistical investigation begins with a question that expects variation and ends with a conclusion that respects it.