Did you know that a school survey can count every answer correctly and still describe the wrong group? A class leader announces, “Eighty per cent of our class wants the library to open earlier.” The arithmetic looks tidy. The hidden question is whether the people who were asked actually represent the whole class.
A representative sample is a set of respondents selected in a way that gives a fair picture of the population you want to describe. To check one, define the population, inspect who could be included, look for groups missing from the replies, and limit the conclusion to what the responses support. A correct percentage is not a substitute for fair selection.
First define who the conclusion is about
“Our class”, “library users”, “students who arrive before school” and “everyone in the school” are four different populations. A survey could be useful for one and misleading for another. Before collecting answers, students should write a plain sentence: We want to understand the preferences of all 40 students in Class A about the library opening time. That sentence sets the boundary for who should have a fair opportunity to contribute.
Selection matters because experiences differ. Students who already use the library in the morning may have stronger opinions about its opening time than students who arrive later, use other study spaces or have different transport arrangements. Neither group is wrong. The mistake is allowing one group to stand silently for everybody.
Worked example: ten early visitors are not the whole class
Imagine a fictional class of 40 students. A student asks the first 10 classmates who enter the library before morning assembly. Eight want the library to open earlier; two prefer the present arrangement. The calculation 8 ÷ 10 × 100 = 80% is correct.
The accurate conclusion is: Eight of the ten early library visitors surveyed, or 80% of those respondents, preferred earlier opening. It is not evidence that 80% of all 40 classmates share that preference. Thirty classmates were not asked. It would also be misleading to say “only 20% of the class supports earlier opening” merely because 8 is 20% of 40; the other 30 opinions are unknown.
The missing responses make the uncertainty visible. If all 30 unasked classmates opposed the change, support would be 8 out of 40, or 20%. If all 30 supported it, support would be 38 out of 40, or 95%. Those are logical extremes, not predictions. They show how little the original survey tells us about the full class. The survey may still be valuable as feedback from actual early users; it just cannot answer a broader question by itself.
Compare this with the related Core Aim article on leading survey questions. That guide tests whether the words of a question influence the reply. Here the problem arises even with perfectly neutral wording: it is about who had the chance to answer.
The four checks students can perform before trusting a survey
- Population: Who is the stated conclusion supposed to cover? Name the class, club, age group or user group clearly.
- Selection: How were participants reached? A convenient doorway, friend group or online chat may leave out people with different experiences.
- Response: How many were invited, how many replied, and might nonrespondents think differently? Invitation and response are not the same thing.
- Claim: Does the final sentence say “respondents”, “invited students” or “all students”? The wording must match the actual evidence.
A random or otherwise carefully designed selection process can improve coverage, but it is not magic. If the invited students do not respond, or important groups are absent, a claim about everyone may still be weak. Privacy also matters: collect only necessary information, allow participation where appropriate, and follow school rules for surveys.
Repairing the class example without pretending the first survey was useless
A better next step is to invite the entire class through the same channel and use one neutral question. Suppose 28 of the 40 students reply, and 19 of the 28 support earlier opening. The correct respondent percentage is 19 ÷ 28 × 100 ≈ 67.9%. The survey covers more of the class than before, but 12 classmates still have unknown preferences.
A suitable report would say: “Of the 28 classmates who responded, 19 preferred earlier opening (about 68%). The views of 12 nonrespondents are unknown.” This is stronger than the first claim because the invitation covered everybody and the remaining gap is disclosed. It still cannot guarantee what all 40 would say. The teacher may decide to collect more feedback or discuss practical constraints before considering any change.
The numbers are illustrative, not a report of an actual school survey. The aim is to teach the difference between counting replies and describing a population. In a larger project, sampling design may need more formal statistical guidance; a classroom checklist is a starting point, not a full survey methodology.
A quick diagnosis for students, parents and teachers
When a student gives a confident percentage
Ask two small questions before correcting the arithmetic: “Eighty per cent of whom?” and “Who could not or did not answer?” If the learner cannot name the denominator or selection process, the difficulty is interpretation rather than calculation. It is usually better to repair that distinction before giving a longer worksheet of percentages.
When a parent sees a survey in a school project
Invite the child to explain one sentence they are prepared to defend. A good explanation might be: “This describes students who answered the form; it may not describe classmates who missed the message.” Avoid turning an elementary class assignment into an expensive or intrusive research exercise. The habit of honest reporting matters more than a dramatic-looking percentage.
When a teacher wants stronger evidence
Provide two short surveys with the same result but different selection methods: one where ten friends answer and one where students across the class had a fair chance to respond. Ask pupils to write separate conclusions. Look for whether they restrict their language to the people actually represented.
Three common mistakes—and what to say instead
- “The sample is large, so it must be representative.” Size helps with some kinds of uncertainty but does not automatically correct a biased selection method. Check coverage as well as count.
- “Everyone who replied agreed, so the whole class agrees.” Unanswered invitations are unknown opinions, not automatic agreement or disagreement.
- “The survey is flawed, so it is worthless.” A limited survey may still describe its respondents honestly and reveal a question worth investigating further.
Progress check: can the learner transfer the idea?
Give this fresh fictional case without the earlier solution: a sports club has 60 members; 15 members who stayed after training answered a poll, and 12 supported a later finish. Ask the student to calculate the respondent percentage and write the strongest defensible one-sentence conclusion. The answer is 12 ÷ 15 = 80% of those 15 respondents—not necessarily 80% of all 60 members. Then ask which members might be missing and how to invite a more balanced response.
Observe progress in the quality of the conclusion, not only the percentage. Can the student identify the intended population unaided? Do they describe who was asked? Do they distinguish nonresponse from disagreement? Can they suggest a respectful way to gather better evidence? These behaviours are more informative than memorising the word “bias”.
Frequently asked questions
Is a class census always better than a sample?
Asking everyone can remove some selection problems, but it does not ensure everyone answers honestly or responds at all. Question wording, privacy, response conditions and missing replies still matter.
Can ten respondents ever be useful?
Yes. Ten responses can reveal what those ten people experienced and suggest issues worth exploring. Whether they justify a wider claim depends on the question, selection method and purpose.
Does a representative sample prove that a policy should change?
No. A survey describes reported preferences under its conditions. Decisions may also require feasibility, accessibility, cost, wellbeing and the views of people affected but not captured well by the data.
The core aim: evidence that knows whom it describes
A student who learns sampling judgement has gained more than another formula. They can read a school survey, a headline or a project chart and ask a gentle but decisive question: Whose voices are actually in this number? That habit turns arithmetic into fairer, more useful reasoning.
For a broader route through learning and checking, explore the eduKate Education Hub and the Clementi Secondary 1 Mathematics learning benchmark, which places explanation, targeted practice and independent checking ahead of mere answer volume.