Did you know? Two class surveys can produce 18 “yes” answers and 30 “yes” answers yet show exactly the same share of respondents agreeing. Without knowing how many people answered each survey, the larger raw count can tell the wrong story.
To compare class-survey results fairly, put the “yes” count over the number of people who actually answered that question, calculate the percentages, and check who was invited and who did not respond. State whether you are comparing the number of supporters, the proportion among respondents, or the proportion of the whole invited group. These are different questions.
Start by naming the quantity being compared
Suppose one pupil says, “More people liked the activity this time,” while another says, “The level of support stayed the same.” They may be using different denominators. Before judging who is right, write the question in full: more people in total, or a higher fraction of respondents? The first can rise even while the second stays level.
The quickest diagnostic question is: “Thirty out of how many?” If a student cannot finish that sentence, calculating a percentage should wait until the missing count is found.
Worked example: two fictional class surveys
In a made-up student project, Class A invited 28 pupils to answer a question about a new reading activity. Twenty-four answered, and 18 of those selected “yes.” Class B invited 50 pupils, 40 answered, and 30 of those selected “yes.” The wording and context are assumed to be comparable only for this arithmetic illustration; these are not real research findings.
- Class A among respondents: 18 ÷ 24 × 100 = 75%. There were 18 yes responses.
- Class B among respondents: 30 ÷ 40 × 100 = 75%. There were 30 yes responses.
- Difference in yes counts: 30 − 18 = 12 additional recorded yes answers in Class B.
- Difference in yes percentages among respondents: 75% − 75% = 0 percentage points.
- Response rate for Class A: 24 ÷ 28 × 100 ≈ 85.7% of invited pupils answered.
- Response rate for Class B: 40 ÷ 50 × 100 = 80% of invited pupils answered.
So the number of “yes” answers is larger in B, but the percentage among people who responded is equal. Neither calculation establishes that the classes have identical underlying opinions. Four pupils in A and ten pupils in B did not respond; their views are unknown.
Notice an additional subtlety: “18 yes answers out of 28 invited” gives about 64.3%, and “30 out of 50 invited” gives 60%. Those are percentages of everyone invited who actually recorded a yes, not estimates of the opinions of those who did not answer. Label this denominator explicitly so it is not mistaken for a response-based support rate.
Why different response counts matter
If the same percentage is calculated from very different numbers of answers, the underlying evidence is not equally extensive. Sample size affects sampling variability under an appropriate sampling design, while nonresponse and selection can introduce other weaknesses. It is unsafe to invent a margin of error for a casual classroom convenience survey, because the usual probability-sampling assumptions may not hold. For background, Pew Research Center explains why sample sizes and survey design matter.
The group composition also matters. If Class B includes different ages, more pupils from one interest group, or respondents who saw a different question, the comparison is no longer like for like. This article is about the denominator and response-count check; our earlier representative-sample guide addresses who was actually surveyed.
A repeatable five-part classroom comparison
- Question: Copy the precise survey wording for both groups and check whether the meaning stayed the same.
- Counts: Record invited pupils, pupils who answered the item, yes answers, no answers and any other option separately.
- Denominator: Decide whether the task asks about respondents or everyone invited, and explain the choice.
- Calculation: Compute percentages independently for each group; show the fraction before rounding.
- Claim: State exactly what the numbers show and add a limitation about nonresponse or group differences.
A good pupil statement for the worked example is: “Class B had twelve more recorded yes answers, but both groups had 75% yes among respondents. Class B also had a lower response rate, so the results should not be treated as a firm measure of all invited pupils’ opinions.” It is more informative than either “support increased” or “the surveys proved no change.”
Mistakes that reveal the next teaching move
- Comparing 30 with 18 while forgetting that the two class sizes differ.
- Dividing one yes count by invited pupils and the other by respondents.
- Writing “a 12% increase” after finding twelve additional pupils.
- Treating two matching respondent percentages as proof that every pupil thinks alike.
- Assuming nonrespondents would vote in the same proportions as respondents.
- Declaring a difference statistically significant without a justified sampling and analysis method.
If a child switches denominators, practise two questions that share the same small data set but ask different things: “What percentage of respondents agreed?” and “What percentage of those invited recorded agreement?” The computation is simple; the reading of the question is the actual skill.
What students, parents and teachers can do next
Students can draw a three-column note headed invited, answered, yes before drawing a chart. Teachers can hand out two surveys with deliberately unequal group sizes and ask for two sentences: one about counts, one about percentages. Parents can ask, “Who is missing from this figure?” rather than supplying a calculator result.
Progress becomes visible when learners independently identify the correct denominator across several new examples and keep counts, proportions and conclusions separate. A useful new test is 9 yes out of 12 compared with 24 yes out of 32: both are 75%, despite different totals. Then change the fractions to 15 out of 20 and 20 out of 50; that is 75% versus 40%, a difference of 35 percentage points, not simply “35 pupils.”
For help reading percentages in a more complex layout, see our row-versus-column table guide. More learning and data-reading routes appear in the eduKate Education Hub.
Frequently asked questions
Is the bigger yes count ever useful?
Yes. If the decision is about how many pupils requested an activity, absolute counts matter. If the claim is that the activity is more popular relative to the group, show an appropriate proportion as well.
Does a 75% result from each group prove equal popularity?
No. It proves equality of the calculated percentages for these respondents and this question. Representativeness, nonresponse and sampling variability remain unresolved.
What if the answer options changed between surveys?
Do not compare the headline percentages as if the question had stayed fixed. Explain the wording change and consider whether a direct comparison is appropriate at all.
The habit worth learning
The core aim is a student who asks “out of whom?” before believing a number. That single question supports stronger school reports, more honest class discussions and better everyday reading of surveys.