Sec 1 G2 Maths Bukit Timah | Fair Surveys and Data

Did you know? A survey can count every answer correctly and still support the wrong conclusion. If ten students in one class answer a question, the result tells us about those ten responses. It does not automatically describe every pupil in Singapore—or even every pupil in the school.

For Bukit Timah families seeking a Secondary 1 G2 Mathematics tutor, the useful method is ask a clear question, define the group, count valid answers, choose a suitable graph and explain exactly what the data do and do not establish. That brings mathematical reasoning into statistics and data handling.

The MOE G2 Mathematics syllabus includes data collection, classification, tabulation and evaluating misleading statistical diagrams. The unchanged eduKateSG Secondary 1 tutorial reference describes weekly 90-minute three-student sessions near Sixth Avenue MRT. Discussion of sampling bias extends these skills; it is not presented as a separate mandatory Sec 1 examination chapter.

Fair data in one minute

Know which population the survey aims to describe.
State who actually answered and how they were selected.
Use clear, neutral wording and categories suitable for the question.
For one response per person, frequencies should sum to valid replies.
For multiple selections, mentions may exceed the number of respondents.
A percentage must name its denominator.
Bar graphs compare separate categories; line graphs show ordered changes; pie charts show shares of a whole.
Read axis scales before judging visual heights.
State what the observed data support; avoid unsupported wider or causal claims.

A ten-pupil example with three transport choices

Imagine ten fictional students each choosing one main transport mode. The raw responses produce bus five, walk three and rail two. The valid-response total is ten. The category shares are fifty, thirty and twenty percent. If a pie chart were used, the respective sector angles would be 180°, 108° and 72°, which add to 360°.

The maths is straightforward, but the wording of the conclusion matters. ‘Five of the ten surveyed students chose bus’ is supported. ‘Half of all Bukit Timah students use the bus’ is not determined by this tiny hypothetical sample. That difference is how data interpretation becomes mathematical judgement.

Thirty-five principles of data collection and credible claims

1. A clear statistical question comes before the survey

Decide what information is needed and who it should describe before collecting responses. Otherwise the resulting table may count something different from what readers expect.

Worked example: To understand the main transport mode among ten fictional students, ask each person for one main mode under an explicit definition.

Question to ask: Write the exact question and its intended group above the data table.

2. Population and sample are different groups

The population is the complete group a question concerns. A sample is the smaller group from which responses are actually collected. Findings about a sample do not automatically describe everyone else.

Worked example: If ten students are surveyed in a class of thirty, the ten are the sample and thirty is the immediate class population.

Question to ask: Do not label a result from one classroom as the preference of every secondary student in Singapore.

3. A sample’s size alone does not guarantee representativeness

A large sample from a narrow or systematically selected group can still misrepresent a wider population. The way respondents are chosen matters as well as how many answer.

Worked example: Surveying only a cycling club might overstate cycling preference among the whole school.

Question to ask: Ask how the sample was selected, not merely how many replies appeared.

4. Convenience selection can introduce bias

Choosing only the easiest people to reach may produce responses unlike the target group’s overall pattern. That does not mean their answers are false; it limits what can be claimed.

Worked example: A survey of students waiting at one bus stop may overrepresent bus users if the goal is transport preference across the whole school.

Question to ask: State where and when the responses were collected.

5. Voluntary response can skew an estimate

When people choose to respond, those with strong opinions may be more likely to participate. This is one possible source of selection bias, not a certainty about every voluntary survey.

Worked example: An online poll about a popular school activity may attract mostly highly enthusiastic participants.

Question to ask: Report that the results describe respondents, and avoid claiming the wider population shares exactly the same view.

6. A leading question can influence the response

Wording that suggests a preferred answer can steer participants rather than neutrally record their choices. The mathematical data would still be numbers, but the measurement process is weaker.

Worked example: Compare ‘Don’t you agree the new timetable is better?’ with ‘Which timetable do you prefer, A, B or neither?’

Question to ask: Ask the learner to rewrite a leading question in neutral language.

7. Double-barrelled questions confuse interpretation

A question asking about two different things at once can make a yes/no reply ambiguous. A respondent may agree with one part and disagree with another.

Worked example: ‘Do you enjoy Maths and Science?’ does not reveal a pupil’s separate opinions of Maths and Science.

Question to ask: Split the question into two clear items.

8. Categories should be mutually exclusive when one choice is intended

If every respondent must choose exactly one category, the categories should not overlap in the defined context. Then their frequencies can legitimately sum to the number of valid respondents.

Worked example: For main transport mode, a student may select only one category under a clear ‘used for most distance’ rule.

Question to ask: If multiple categories are allowed, say so and interpret totals differently.

9. Exhaustive categories cover all valid possibilities

A classification should offer a place for responses that do not fit the listed labels, perhaps ‘other’ with a specified definition. Otherwise valid observations may be silently excluded.

Worked example: A three-choice survey with bus, rail and walk may need ‘other’ for bicycle or car.

Question to ask: Track excluded or missing responses rather than forcing them into an inaccurate group.

10. Raw data records individual observations

A raw list contains the original categories or values before they are grouped. A frequency table compresses it by counting how many observations belong to each category.

Worked example: Bus, walk, bus, rail, bus, walk becomes bus three, walk two, rail one.

Question to ask: Count all six observations in the raw list before turning them into bars.

11. Tally marks are a counting tool

Tallies record each response as it arrives and group counts for easier checking. The frequency is the count of marks, not the number of tally groups.

Worked example: Seven responses can be one completed group of five tallies plus two further marks.

Question to ask: Check the frequency sum against the number of valid single-choice responses.

12. Single-response totals can be reconciled

When each valid respondent gives exactly one answer, the sum of all category frequencies equals the number of valid replies. A mismatch indicates a count or classification issue.

Worked example: If three categories have frequencies 5,3,2, their sum is ten valid responses.

Question to ask: Do not silently equate the valid count with everyone originally invited.

13. Multiple-response counts can exceed participants

When respondents may select several categories, each person can contribute more than one mention. Category frequencies may therefore total more than the number of people.

Worked example: Ten students may make sixteen sport selections if each can pick several sports.

Question to ask: Write ‘mentions’ or ‘selections’ instead of ‘students’ when appropriate.

14. Nonresponse changes the denominator

A survey sent to a certain number of people may receive fewer valid replies. Percentages calculated from replies describe respondents, not automatically the full invited group.

Worked example: If 30 students are invited and 24 give usable single choices, a category with twelve replies is 50% of respondents, but 40% of those invited.

Question to ask: Label the denominator clearly in the sentence reporting the result.

15. A percentage needs a stated reference whole

The same category frequency can represent different percentages under different totals. Dividing by the wrong denominator creates a perfectly calculated but incorrectly interpreted percentage.

Worked example: Ten votes out of twenty are 50%; ten votes out of fifty are 20%.

Question to ask: Write part/whole×100 before using a calculator.

16. Comparing groups of different sizes needs care

Raw counts alone can make a larger sample appear to prefer something more strongly even when its proportion is smaller.

Worked example: Ten out of fifty is 20%, while five out of ten is 50%. The first raw count is larger, but its share is lower.

Question to ask: State whether the question asks for absolute frequency or relative percentage.

17. A bar chart should use labelled categories and regular scale

The axes must describe what each bar measures, and equally spaced ticks should represent equal numeric intervals. Decorative design cannot substitute for valid scale.

Worked example: With frequencies 2,4,6, ticks 0,2,4,6 are evenly spaced; ticks 0,2,5,6 placed equally would not form a uniform linear axis.

Question to ask: Read every axis label and tick before judging the heights.

18. Truncated axes can exaggerate visual contrasts

A bar chart starting near the smallest observation makes small numerical differences look much larger. The printed values may still be accurate, but the impression can mislead.

Worked example: Bars ending at 98 and 100 differ by only two, even if the axis starts at 97 and the second bar appears three times as tall.

Question to ask: Report the numerical difference and note whether the axis begins at zero.

19. Pictogram keys are part of the data

One icon can represent more than one response. A valid interpretation must multiply icon count by the value stated in the key.

Worked example: Five icons with a key of four students each represent twenty students, not five.

Question to ask: A half icon should be interpreted only if its value is clear under the key.

20. Pie charts require a meaningful whole

A pie chart represents shares of a total, typically one mutually exclusive category per observation for a simple survey. Its sector angles should add to 360 degrees.

Worked example: Five out of ten responses occupy 5/10×360=180 degrees.

Question to ask: Multiple selections from the same respondent do not automatically form a clean 100%-of-people pie.

21. Line graphs suit ordered change

A line graph is useful when observations have an order, often time. Joining unrelated category labels with a line may suggest continuity that the data do not have.

Worked example: Daily attendance 20,24,21 may be plotted over Monday, Tuesday and Wednesday to show a change.

Question to ask: Ask what the horizontal axis represents and whether the order has meaning.

22. Different graphs highlight different questions

Bar charts compare separate category frequencies, pie charts show shares of a whole, and line graphs communicate trends across ordered values. The most helpful representation depends on the purpose.

Worked example: For bus, walk and rail preferences, a bar chart makes differences clearer than a line joining category names.

Question to ask: Explain why the selected graph fits the data rather than picking it for appearance.

23. A graph may show association without proving a cause

Two quantities may rise together without one causing the other. The diagram reports observations, not necessarily the mechanism behind them.

Worked example: A class’s hypothetical reading minutes and quiz scores may both increase in one week; the graph alone cannot prove the reading caused the change.

Question to ask: Use wording such as ‘increased with’ rather than ‘was caused by’ without further evidence.

24. A local sample does not automatically describe a whole estate

A class survey conducted near Bukit Timah describes the people who actually responded under the survey method. It cannot establish the travel habits of all Bukit Timah residents.

Worked example: If six of ten surveyed pupils use a bus, the measured sample share is 60%. It is not an official figure for the neighbourhood.

Question to ask: State the measured group and the limit of generalisation.

25. One missing category can shift calculated shares

Omitting a legitimate category reduces the recorded total and may inflate the displayed percentages of categories retained. A complete count matters.

Worked example: If two bicycle respondents are omitted from a ten-person dataset, a count of four bus responses becomes 4/8=50% instead of 4/10=40%.

Question to ask: Reconcile the raw list and explain any exclusions before reporting rates.

26. A wrong response is not the same as missing data

A respondent may give an inconsistent answer, decline to answer or be accidentally excluded. These need different recording decisions if the task requires accurate classification.

Worked example: If a pupil selects two categories when only one is allowed, a teacher may seek clarification instead of silently choosing one.

Question to ask: Describe the treatment of invalid responses rather than invent a category for them.

27. Units and categories should be named correctly

A bar can represent students, choices, kilometres or other units. A numerical height without a label cannot be interpreted confidently.

Worked example: A value of ten in a chart might mean ten pupils, ten minutes or ten books, depending on the axis.

Question to ask: Ask the learner to write the unit beside each table heading and graphical axis.

28. An average is not automatically needed

Simple G2 Sec 1 data handling often focuses on collection, tabulation and interpretation of graphs. Averages may be a later topic depending on the syllabus level and year.

Worked example: A categorical survey of favourite subjects needs frequencies or percentages; averaging the arbitrary category codes would not be meaningful.

Question to ask: Choose a summary measure appropriate to the kind of data, not a familiar formula.

29. Protect unnecessary personal information

A classroom mathematics exercise can collect simple non-sensitive category counts without requesting pupils’ names, addresses or private details. Follow teacher and school guidance on any real survey.

Worked example: For a fictional transport-mode tally, anonymous counts by category are enough for the mathematical analysis.

Question to ask: Do not add personal identifiers when they provide no value for the question.

30. A claim should not outrun its evidence

A careful conclusion names the sample, the number of valid replies, the observed frequency and any limitation. Big sweeping statements can sound attractive but may not follow from the available data.

Worked example: ‘Five of ten respondents chose bus’ is supported by the table. ‘Half of all Singapore pupils travel by bus’ is not established by those ten replies.

Question to ask: Require a student to underline which exact data support each conclusion.

31. Checking totals is a powerful self-correction method

Reconcile the frequency total, percentage total and pie sector total when the table represents one complete set of single-category responses.

Worked example: If categories have frequencies 5,3,2 among ten replies, the percentages 50%,30%,20% total 100% and the pie angles 180°,108°,72° total 360°.

Question to ask: Use a second representation to check the original table instead of merely repeating the same addition.

32. A fair survey can still contain random variation

Even well-chosen questions and a thoughtful sampling plan do not guarantee that a small sample perfectly matches the population. Fair design reduces avoidable distortions, not all uncertainty.

Worked example: Two separate samples of ten students may report different percentages without either table being calculated incorrectly.

Question to ask: Avoid promising perfect representativeness from a classroom exercise.

33. The size of a reported change can be described two ways

A rise from 40% to 50% is ten percentage points, while the relative increase is 25% of the original forty-percent rate. This mathematical distinction survives regardless of survey design.

Worked example: The rates 40% and 50% differ by ten points. For relative change, (50−40)/40×100=25%.

Question to ask: Name the requested type of change before calculating.

34. An unbiased-looking chart does not repair a biased sample

Even a graph with perfect labels, honest axes and correct arithmetic cannot make an unrepresentative sample representative. Collection and presentation quality are separate questions.

Worked example: A beautifully drawn chart of cycling-club preferences is still a chart of that club, not the entire school.

Question to ask: Review who was asked and how the replies were recorded before admiring the chart.

35. A tutor can teach the logic without complex statistics

Sec 1 G2 students can reason about reference totals, clear categories, bar-chart scaling and the limits of small samples without advanced inferential methods.

Worked example: Give two small frequency tables and ask which conclusions each table actually supports. Require a sentence naming the group surveyed.

Question to ask: Keep the mathematics tied to official data-handling objectives and mark sampling-bias discussion as an interpretive extension.

36. Independent interpretation is the skill that matters

A pupil might copy a frequency table correctly yet believe its biggest bar proves something about the whole country. Feedback should focus on where the inference went beyond the evidence.

Worked example: After analysing ten fictional responses, ask the student to write one supported conclusion and one claim the data cannot establish.

Question to ask: Retest with an unfamiliar survey after a short delay, without the tutor supplying the wording.

57 worked survey, graph and interpretation clinics

Cover each answer first. Some questions are numerical; others ask which interpretation is valid. For every result, label the population, sample or valid-response total before calculating a percentage or drawing a conclusion.

Data clinic 1: A fictional single-choice survey has 10 valid replies: bus 5, walk 3, rail 2. What is bus share and pie sector?

Answer: 50% and 180°. Reasoning: Divide bus frequency 5 by all 10 valid replies, then multiply by 100% or 360°.

Interpretation check: Verify all three frequencies sum to 10; use the correct whole. Try a different sample size or response pattern to see whether the conclusion remains justified. If the answer depends on a hidden assumption, name that assumption rather than silently inserting it into the story.

Data clinic 2: A fictional single-choice survey has 20 valid replies: bus 8, walk 7, rail 5. What is bus share and pie sector?

Answer: 40% and 144°. Reasoning: Divide bus frequency 8 by all 20 valid replies, then multiply by 100% or 360°.

Interpretation check: Verify all three frequencies sum to 20; use the correct whole. Try a different sample size or response pattern to see whether the conclusion remains justified. If the answer depends on a hidden assumption, name that assumption rather than silently inserting it into the story.

Data clinic 3: A fictional single-choice survey has 25 valid replies: bus 10, walk 9, rail 6. What is bus share and pie sector?

Answer: 40% and 144°. Reasoning: Divide bus frequency 10 by all 25 valid replies, then multiply by 100% or 360°.

Interpretation check: Verify all three frequencies sum to 25; use the correct whole. Try a different sample size or response pattern to see whether the conclusion remains justified. If the answer depends on a hidden assumption, name that assumption rather than silently inserting it into the story.

Data clinic 4: A fictional single-choice survey has 40 valid replies: bus 16, walk 14, rail 10. What is bus share and pie sector?

Answer: 40% and 144°. Reasoning: Divide bus frequency 16 by all 40 valid replies, then multiply by 100% or 360°.

Interpretation check: Verify all three frequencies sum to 40; use the correct whole. Try a different sample size or response pattern to see whether the conclusion remains justified. If the answer depends on a hidden assumption, name that assumption rather than silently inserting it into the story.

Data clinic 5: A fictional single-choice survey has 50 valid replies: bus 20, walk 18, rail 12. What is bus share and pie sector?

Answer: 40% and 144°. Reasoning: Divide bus frequency 20 by all 50 valid replies, then multiply by 100% or 360°.

Interpretation check: Verify all three frequencies sum to 50; use the correct whole. Try a different sample size or response pattern to see whether the conclusion remains justified. If the answer depends on a hidden assumption, name that assumption rather than silently inserting it into the story.

Data clinic 6: A fictional single-choice survey has 30 valid replies: bus 12, walk 9, rail 9. What is bus share and pie sector?

Answer: 40% and 144°. Reasoning: Divide bus frequency 12 by all 30 valid replies, then multiply by 100% or 360°.

Interpretation check: Verify all three frequencies sum to 30; use the correct whole. Try a different sample size or response pattern to see whether the conclusion remains justified. If the answer depends on a hidden assumption, name that assumption rather than silently inserting it into the story.

Data clinic 7: A fictional single-choice survey has 60 valid replies: bus 24, walk 21, rail 15. What is bus share and pie sector?

Answer: 40% and 144°. Reasoning: Divide bus frequency 24 by all 60 valid replies, then multiply by 100% or 360°.

Interpretation check: Verify all three frequencies sum to 60; use the correct whole. Try a different sample size or response pattern to see whether the conclusion remains justified. If the answer depends on a hidden assumption, name that assumption rather than silently inserting it into the story.

Data clinic 8: A fictional single-choice survey has 80 valid replies: bus 32, walk 28, rail 20. What is bus share and pie sector?

Answer: 40% and 144°. Reasoning: Divide bus frequency 32 by all 80 valid replies, then multiply by 100% or 360°.

Interpretation check: Verify all three frequencies sum to 80; use the correct whole. Try a different sample size or response pattern to see whether the conclusion remains justified. If the answer depends on a hidden assumption, name that assumption rather than silently inserting it into the story.

Data clinic 9: A fictional single-choice survey has 100 valid replies: bus 45, walk 35, rail 20. What is bus share and pie sector?

Answer: 45% and 162°. Reasoning: Divide bus frequency 45 by all 100 valid replies, then multiply by 100% or 360°.

Interpretation check: Verify all three frequencies sum to 100; use the correct whole. Try a different sample size or response pattern to see whether the conclusion remains justified. If the answer depends on a hidden assumption, name that assumption rather than silently inserting it into the story.

Data clinic 10: A fictional single-choice survey has 15 valid replies: bus 6, walk 5, rail 4. What is bus share and pie sector?

Answer: 40% and 144°. Reasoning: Divide bus frequency 6 by all 15 valid replies, then multiply by 100% or 360°.

Interpretation check: Verify all three frequencies sum to 15; use the correct whole. Try a different sample size or response pattern to see whether the conclusion remains justified. If the answer depends on a hidden assumption, name that assumption rather than silently inserting it into the story.

Data clinic 11: A fictional single-choice survey has 24 valid replies: bus 12, walk 8, rail 4. What is bus share and pie sector?

Answer: 50% and 180°. Reasoning: Divide bus frequency 12 by all 24 valid replies, then multiply by 100% or 360°.

Interpretation check: Verify all three frequencies sum to 24; use the correct whole. Try a different sample size or response pattern to see whether the conclusion remains justified. If the answer depends on a hidden assumption, name that assumption rather than silently inserting it into the story.

Data clinic 12: A fictional single-choice survey has 32 valid replies: bus 16, walk 10, rail 6. What is bus share and pie sector?

Answer: 50% and 180°. Reasoning: Divide bus frequency 16 by all 32 valid replies, then multiply by 100% or 360°.

Interpretation check: Verify all three frequencies sum to 32; use the correct whole. Try a different sample size or response pattern to see whether the conclusion remains justified. If the answer depends on a hidden assumption, name that assumption rather than silently inserting it into the story.

Data clinic 13: A fictional single-choice survey has 36 valid replies: bus 18, walk 12, rail 6. What is bus share and pie sector?

Answer: 50% and 180°. Reasoning: Divide bus frequency 18 by all 36 valid replies, then multiply by 100% or 360°.

Interpretation check: Verify all three frequencies sum to 36; use the correct whole. Try a different sample size or response pattern to see whether the conclusion remains justified. If the answer depends on a hidden assumption, name that assumption rather than silently inserting it into the story.

Data clinic 14: A fictional single-choice survey has 12 valid replies: bus 4, walk 5, rail 3. What is bus share and pie sector?

Answer: 33.33% and 120°. Reasoning: Divide bus frequency 4 by all 12 valid replies, then multiply by 100% or 360°.

Interpretation check: Verify all three frequencies sum to 12; use the correct whole. Try a different sample size or response pattern to see whether the conclusion remains justified. If the answer depends on a hidden assumption, name that assumption rather than silently inserting it into the story.

Data clinic 15: A fictional single-choice survey has 48 valid replies: bus 24, walk 16, rail 8. What is bus share and pie sector?

Answer: 50% and 180°. Reasoning: Divide bus frequency 24 by all 48 valid replies, then multiply by 100% or 360°.

Interpretation check: Verify all three frequencies sum to 48; use the correct whole. Try a different sample size or response pattern to see whether the conclusion remains justified. If the answer depends on a hidden assumption, name that assumption rather than silently inserting it into the story.

Data clinic 16: Survey sent to 30 pupils; 24 valid single-choice replies, of which 12 chose A. Share among respondents versus invited?

Answer: 50% of respondents; 40% of invited. Reasoning: The denominators are different: category/24 and category/30.

Interpretation check: State explicitly which group each percentage describes; the answers are not interchangeable. Try a different sample size or response pattern to see whether the conclusion remains justified. If the answer depends on a hidden assumption, name that assumption rather than silently inserting it into the story.

Data clinic 17: Survey sent to 40 pupils; 32 valid single-choice replies, of which 16 chose A. Share among respondents versus invited?

Answer: 50% of respondents; 40% of invited. Reasoning: The denominators are different: category/32 and category/40.

Data clinic 18: Survey sent to 50 pupils; 40 valid single-choice replies, of which 20 chose A. Share among respondents versus invited?

Answer: 50% of respondents; 40% of invited. Reasoning: The denominators are different: category/40 and category/50.

Data clinic 19: Survey sent to 100 pupils; 80 valid single-choice replies, of which 32 chose A. Share among respondents versus invited?

Answer: 40% of respondents; 32% of invited. Reasoning: The denominators are different: category/80 and category/100.

Data clinic 20: Survey sent to 60 pupils; 45 valid single-choice replies, of which 18 chose A. Share among respondents versus invited?

Answer: 40% of respondents; 30% of invited. Reasoning: The denominators are different: category/45 and category/60.

Data clinic 21: Survey sent to 80 pupils; 60 valid single-choice replies, of which 24 chose A. Share among respondents versus invited?

Answer: 40% of respondents; 30% of invited. Reasoning: The denominators are different: category/60 and category/80.

Data clinic 22: Survey sent to 25 pupils; 20 valid single-choice replies, of which 10 chose A. Share among respondents versus invited?

Answer: 50% of respondents; 40% of invited. Reasoning: The denominators are different: category/20 and category/25.

Data clinic 23: Survey sent to 120 pupils; 90 valid single-choice replies, of which 36 chose A. Share among respondents versus invited?

Answer: 40% of respondents; 30% of invited. Reasoning: The denominators are different: category/90 and category/120.

Data clinic 24: A graph’s labelled values rise from 98 to 100. Find numerical difference and relative percentage increase.

Answer: 2 units and 2.04%. Reasoning: Subtract 100−98=2; percentage change=2/98×100%.

Interpretation check: Do not infer the relative change from decorative bar heights; use the labelled values. Try a different sample size or response pattern to see whether the conclusion remains justified. If the answer depends on a hidden assumption, name that assumption rather than silently inserting it into the story.

Data clinic 25: A graph’s labelled values rise from 40 to 50. Find numerical difference and relative percentage increase.

Answer: 10 units and 25%. Reasoning: Subtract 50−40=10; percentage change=10/40×100%.

Data clinic 26: A graph’s labelled values rise from 12 to 18. Find numerical difference and relative percentage increase.

Answer: 6 units and 50%. Reasoning: Subtract 18−12=6; percentage change=6/12×100%.

Data clinic 27: A graph’s labelled values rise from 75 to 80. Find numerical difference and relative percentage increase.

Answer: 5 units and 6.67%. Reasoning: Subtract 80−75=5; percentage change=5/75×100%.

Data clinic 28: A graph’s labelled values rise from 50 to 52. Find numerical difference and relative percentage increase.

Answer: 2 units and 4%. Reasoning: Subtract 52−50=2; percentage change=2/50×100%.

Data clinic 29: A graph’s labelled values rise from 30 to 42. Find numerical difference and relative percentage increase.

Answer: 12 units and 40%. Reasoning: Subtract 42−30=12; percentage change=12/30×100%.

Data clinic 30: A graph’s labelled values rise from 16 to 24. Find numerical difference and relative percentage increase.

Answer: 8 units and 50%. Reasoning: Subtract 24−16=8; percentage change=8/16×100%.

Data clinic 31: A graph’s labelled values rise from 22 to 30. Find numerical difference and relative percentage increase.

Answer: 8 units and 36.36%. Reasoning: Subtract 30−22=8; percentage change=8/22×100%.

Data clinic 32: Five pictogram icons, key four students per icon: frequency?

Answer: 20 students. Reasoning: 5×4=20.

Interpretation check: Read the key before counting. Try a different sample size or response pattern to see whether the conclusion remains justified. If the answer depends on a hidden assumption, name that assumption rather than silently inserting it into the story.

Data clinic 33: Two and a half icons, key six per icon: frequency?

Answer: 15 students. Reasoning: 2.5×6=15.

Interpretation check: Only interpret fractional icons when the display permits them. Try a different sample size or response pattern to see whether the conclusion remains justified. If the answer depends on a hidden assumption, name that assumption rather than silently inserting it into the story.

Data clinic 34: A pie-sector represents one quarter of replies: angle?

Answer: 90°. Reasoning: 360°÷4=90°.

Interpretation check: Four quarters form a full circle. Try a different sample size or response pattern to see whether the conclusion remains justified. If the answer depends on a hidden assumption, name that assumption rather than silently inserting it into the story.

Data clinic 35: A bar axis has ticks 0,2,4,6. Value between four and six?

Answer: 5. Reasoning: The axis intervals represent two units.

Interpretation check: Consistent ticks and labels matter. Try a different sample size or response pattern to see whether the conclusion remains justified. If the answer depends on a hidden assumption, name that assumption rather than silently inserting it into the story.

Data clinic 36: A bar axis reads 0,2,5,6 at equal spacing: regular numerical scale?

Answer: No. Reasoning: Equal distances on a linear scale must represent equal numerical increments.

Interpretation check: The intervals here are unequal. Try a different sample size or response pattern to see whether the conclusion remains justified. If the answer depends on a hidden assumption, name that assumption rather than silently inserting it into the story.

Data clinic 37: A dataset has frequencies 5,3,2 with one reply per respondent: total?

Answer: 10. Reasoning: 5+3+2=10.

Interpretation check: Compare with valid response count. Try a different sample size or response pattern to see whether the conclusion remains justified. If the answer depends on a hidden assumption, name that assumption rather than silently inserting it into the story.

Data clinic 38: A dataset has 5 bus mentions and 4 walk mentions, multiple choices allowed. Participants exactly nine?

Answer: Not determined. Reasoning: Participants may have chosen more than one category.

Interpretation check: Mentions are not necessarily distinct people. Try a different sample size or response pattern to see whether the conclusion remains justified. If the answer depends on a hidden assumption, name that assumption rather than silently inserting it into the story.

Data clinic 39: Ten respondents, six chose bus: can we claim 60% of all Bukit Timah residents use bus?

Answer: No. Reasoning: The sample of ten does not establish a population percentage.

Interpretation check: State the measured group. Try a different sample size or response pattern to see whether the conclusion remains justified. If the answer depends on a hidden assumption, name that assumption rather than silently inserting it into the story.

Data clinic 40: Survey cyclists only; infer whole-school cycling preference?

Answer: Not justified. Reasoning: The sampling frame excludes many non-cyclists.

Interpretation check: A convenience sample can be unrepresentative. Try a different sample size or response pattern to see whether the conclusion remains justified. If the answer depends on a hidden assumption, name that assumption rather than silently inserting it into the story.

Data clinic 41: Which is more neutral: ‘Isn’t the new timetable better?’ or ‘Which timetable do you prefer?’

Answer: Second question. Reasoning: It does not suggest the expected answer.

Interpretation check: Neutral wording supports fairer collection. Try a different sample size or response pattern to see whether the conclusion remains justified. If the answer depends on a hidden assumption, name that assumption rather than silently inserting it into the story.

Data clinic 42: ‘Do you enjoy Maths and Science?’ is a good one-variable question?

Answer: No. Reasoning: It combines two potentially different preferences.

Interpretation check: Ask separate questions. Try a different sample size or response pattern to see whether the conclusion remains justified. If the answer depends on a hidden assumption, name that assumption rather than silently inserting it into the story.

Data clinic 43: Bar graph of four types of fruit: best basic choice?

Answer: Bar chart. Reasoning: Fruit types are discrete categories.

Interpretation check: A line graph suggests ordered continuity that is absent. Try a different sample size or response pattern to see whether the conclusion remains justified. If the answer depends on a hidden assumption, name that assumption rather than silently inserting it into the story.

Data clinic 44: Temperature measured on consecutive days: useful chart?

Answer: Line graph. Reasoning: Dates give meaningful order for change.

Interpretation check: Label time and temperature units. Try a different sample size or response pattern to see whether the conclusion remains justified. If the answer depends on a hidden assumption, name that assumption rather than silently inserting it into the story.

Data clinic 45: One clean whole divided into three categories: useful share chart?

Answer: Pie chart. Reasoning: A pie shows part-to-whole sectors.

Interpretation check: Confirm categories do not overlap. Try a different sample size or response pattern to see whether the conclusion remains justified. If the answer depends on a hidden assumption, name that assumption rather than silently inserting it into the story.

Data clinic 46: A survey of 12 people gets 10 valid answers: can all 12 be used as respondent denominator?

Answer: No. Reasoning: Two do not have valid recorded replies.

Interpretation check: Distinguish invited from respondent totals. Try a different sample size or response pattern to see whether the conclusion remains justified. If the answer depends on a hidden assumption, name that assumption rather than silently inserting it into the story.

Data clinic 47: Five of ten students use a bus. Bus share?

Answer: 50%. Reasoning: 5/10×100=50%.

Interpretation check: This is a claim about ten respondents. Try a different sample size or response pattern to see whether the conclusion remains justified. If the answer depends on a hidden assumption, name that assumption rather than silently inserting it into the story.

Data clinic 48: Ten of fifty students use a bus. Bus share?

Answer: 20%. Reasoning: 10/50×100=20%.

Interpretation check: A larger raw count can have a smaller share. Try a different sample size or response pattern to see whether the conclusion remains justified. If the answer depends on a hidden assumption, name that assumption rather than silently inserting it into the story.

Data clinic 49: A pie has sectors 180°,108°,72°. Complete?

Answer: Yes. Reasoning: They add to 360°.

Interpretation check: Their shares are 50%,30%,20%. Try a different sample size or response pattern to see whether the conclusion remains justified. If the answer depends on a hidden assumption, name that assumption rather than silently inserting it into the story.

Data clinic 50: A line graph rises from 20 to 24. Does it prove why values increased?

Answer: No. Reasoning: The graph states a change, not its cause.

Interpretation check: Avoid ungrounded causal claims. Try a different sample size or response pattern to see whether the conclusion remains justified. If the answer depends on a hidden assumption, name that assumption rather than silently inserting it into the story.

Data clinic 51: The category ‘other’ has two replies but is omitted. Would the remaining percentage denominators need review?

Answer: Yes. Reasoning: Excluding valid responses changes the total.

Interpretation check: Explain excluded data before reporting percentages. Try a different sample size or response pattern to see whether the conclusion remains justified. If the answer depends on a hidden assumption, name that assumption rather than silently inserting it into the story.

Data clinic 52: An online voluntary poll attracts enthusiasts: is it automatically representative?

Answer: No. Reasoning: Selection may overrepresent strong opinions.

Interpretation check: Report who actually responded. Try a different sample size or response pattern to see whether the conclusion remains justified. If the answer depends on a hidden assumption, name that assumption rather than silently inserting it into the story.

Data clinic 53: A tall bar shows 100 versus 98 with axis starting 97. Numeric difference?

Answer: 2. Reasoning: 100−98=2.

Interpretation check: Visual exaggeration cannot change the numbers. Try a different sample size or response pattern to see whether the conclusion remains justified. If the answer depends on a hidden assumption, name that assumption rather than silently inserting it into the story.

Data clinic 54: A one-choice table shows frequencies 6,4,5 but only 14 valid replies: consistent?

Answer: No. Reasoning: The frequencies add to fifteen, not fourteen.

Interpretation check: Check for double-counting or an error. Try a different sample size or response pattern to see whether the conclusion remains justified. If the answer depends on a hidden assumption, name that assumption rather than silently inserting it into the story.

Data clinic 55: What is stronger: ‘5 of 10 respondents chose bus’ or ‘Half of every pupil chooses bus’?

Answer: First. Reasoning: The observed count supports only the sample statement.

Interpretation check: Generalisation needs additional evidence. Try a different sample size or response pattern to see whether the conclusion remains justified. If the answer depends on a hidden assumption, name that assumption rather than silently inserting it into the story.

Data clinic 56: A rate changes from 40% to 50%. Percentage-point change?

Answer: 10 percentage points. Reasoning: 50−40=10.

Interpretation check: The relative change is 25%. Try a different sample size or response pattern to see whether the conclusion remains justified. If the answer depends on a hidden assumption, name that assumption rather than silently inserting it into the story.

Data clinic 57: A rate changes from 40% to 50%. Relative change?

Answer: 25%. Reasoning: 10/40=0.25.

Interpretation check: A different reference calculation from percentage points. Try a different sample size or response pattern to see whether the conclusion remains justified. If the answer depends on a hidden assumption, name that assumption rather than silently inserting it into the story.

Three ethical, small-scale classroom investigations

Rewrite leading questions

Compare the questions ‘Don’t you agree our new timetable is better?’ and ‘Which timetable do you prefer: A, B or no preference?’ Discuss how the first wording hints at a preferred answer while the second more clearly records the response. Do not assume either survey will be perfectly unbiased; explain why the neutral version is a better starting point.

Check frequency totals against actual replies

Use a fictitious set of twenty single-choice responses: eight bus, seven walk and five rail. The frequencies add to twenty. Calculate bus share 40%, walk 35%, rail 25%. Convert them to pie-sector angles 144°, 126° and 90°. Confirm both percentages sum to 100 and angles to 360.

Make a misleading graph and then fix it

Draw a bar chart comparing counts ninety-eight and one hundred. First use a vertical axis starting at ninety-seven; the taller bar may look dramatically larger. Then use an axis starting at zero and consistent intervals. In both drawings the numeric difference is two. Ask the learner to explain how design changes perception without changing the printed data.

A four-week data-reasoning plan

Week 1: raw lists and honest counts

Practise one clearly defined question with mutually exclusive categories. Turn a small raw list into a frequency table and reconcile the total. Avoid unnecessary personal information.

Week 2: sample, denominator and percentages

Compare invited, valid responses and total category mentions. Practise selecting the appropriate denominator and stating exactly which group a percentage describes.

Week 3: graphs and visual choices

Draw bar charts, pictograms, line graphs and pie charts from suitable tables. Read every key and axis scale, and compare a truncated display with a clear one.

Week 4: what the data can actually tell us

Discuss neutral question wording, unrepresentative samples and unsupported generalisations. Require every learner to state one valid conclusion and one conclusion the table does not establish.

How a three-student Maths tutor can diagnose an interpretation error

One pupil may double-count a multiple-choice response, another may use the invited count instead of the valid-response denominator, and a third may claim the sample result proves a neighbourhood-wide preference. The numerical skills and reasoning repairs needed are different.

A tutor can use a simple dataset, ask each learner to justify the denominator and graph choice, and then test the conclusion on a fresh example without help. This turns chart reading into independent mathematical judgement rather than a vocabulary test.

The eduKateSG Secondary 1 small-group reference describes weekly 90-minute learning in groups of three near Sixth Avenue MRT. Class suitability and availability require direct confirmation; this page does not claim survey findings about real Singapore residents.

Frequently asked questions

Are fair surveys part of Secondary 1 G2 Maths?

The MOE G2 data strand includes collecting, classifying and tabulating data and interpreting statistical diagrams. Bias and representativeness here are interpretive extensions, not a claim that formal sampling theory is a separate tested Sec 1 topic.

What is the difference between a population and a sample?

The population is the entire target group; a sample is the portion from which replies were collected.

Is a larger sample always fair?

No. Selection methods, question wording and nonresponse can affect representativeness even when many people are asked.

What is a leading survey question?

Wording that nudges a preferred answer rather than asking neutrally, such as ‘Don’t you agree this is better?’.

Why do valid response totals matter?

They determine the denominator for percentages based on respondents and help reconcile single-choice frequency counts.

What if respondents can select multiple categories?

The number of selections can exceed the number of people. Interpret those counts as mentions, not automatically as distinct respondents.

When should a pie chart be used?

When categories form clear shares of one meaningful whole and are suitable for a part-to-whole display.

Can a bar chart be misleading while its numbers are true?

Yes. A truncated or inconsistent scale can exaggerate the apparent differences.

Can survey data prove a cause?

Usually not from a chart alone. It may show an association or change without establishing why it happened.

Should pupils collect names and private details?

Avoid unnecessary personal information. For practice, simple anonymous category counts are usually sufficient, subject to teacher and school guidance.

How can parents help with survey interpretation?

Ask ‘Who answered?’, ‘What was counted?’, ‘What is the denominator?’ and ‘Which conclusion follows from these numbers?’.

How does a three-student tutorial support this?

A tutor can diagnose miscounted tallies, invalid chart choices and unsupported claims, then ask each learner to explain a fresh dataset independently.

Official syllabus and next articles

See the MOE G2 Mathematics syllabus for the core data-handling requirements and the unchanged eduKateSG tutoring reference for the teaching setting. These related guides extend statistics and critical interpretation:

The best statistical question often comes after the arithmetic: who was counted, what does this percentage refer to, and what exactly can we conclude? Those habits make every graph and table more trustworthy.

Secondary 1 G2 Mathematics Tutor Bukit Timah | Fair Surveys, Biased Samples and Data Claims