A sample is the group a source actually collected data from, and it only speaks for the wider group if it resembles it. This skill appears whenever a survey, poll or study is used as evidence for a claim about “young people”, “residents” or “students”.
It follows reading counts and rates in the module on data interpretation and limitations. The surveys here are invented.
How do you test a sample?
Ask four questions in order:
- Population: who is the claim about? (All students in a school? All adults in a city?)
- Selection: how did people end up in the sample? (Random draw, volunteers, whoever was nearby?)
- Missing voices: who could not or would not be included?
- Size: is the group big enough that a few unusual answers would not swing the result?
A sample that fails question 2 or 3 is unrepresentative, whatever its size.
Worked example (fictional)
A made-up student council reports: “We asked 60 students at the canteen during recess whether the canteen should open longer. 45 said yes, so 75% of our school wants longer hours.”
Population: all students in the school.
Selection: students who were already in the canteen at recess. People who bring packed lunch, stay in class or visit the library were less likely to be asked.
Missing voices: students who rarely use the canteen, and anyone who finishes eating quickly and leaves.
Size: 60 is small, though size is the lesser problem here.
Fair interpretation: 45 of 60 canteen users asked, which is 75%, said yes. The result may well hold for regular canteen users. It cannot be stated as the view of the whole school.
A better sentence: “Of 60 students surveyed in the canteen at recess, 75% wanted longer opening hours; students who do not use the canteen were not asked.”
What is the common mistake?
Mistaken answer: “The survey proves that 75% of students want the canteen open longer.”
Two errors sit in this sentence. It widens the claim from “students in the canteen” to “students”, and it uses “proves” for a small, convenient sample. The correction is to match the wording of the conclusion to the group that was really asked.
Check yourself
All surveys are fictional.
1. A poll about school transport is handed out only at the bus stop at 7 a.m. Who is probably missing?
Show answer
Students who walk, cycle, are driven by family or live nearby, plus anyone who arrives at a different time. The poll mostly reflects bus users who travel early.
2. A survey of 500 members of a football club asks whether the school should build a new pitch. Is 500 enough to speak for the school?
Show answer
No. The number is large but the selection is skewed: football club members are more likely to want a pitch. Size does not repair a skewed selection.
3. Rewrite this claim fairly: “Young people think social media is harmful” (based on 40 volunteers who answered an online poll).
Show answer
For example: “Of 40 volunteers who answered an online poll, most said social media can be harmful; volunteers may hold stronger views than other young people, so this cannot be taken as the view of all young people.”
Where does this lead next?
Once you know who the numbers describe, the next question is what they can show about cause: distinguish association from a justified causal claim. The practice set mixes all four skills.
A teacher can also read your own survey wording with you and help you phrase limitations clearly, which fits online one-to-one Global Perspectives tuition.