Did you know that a survey can collect thirty perfectly legible answers and still fail to answer its own question? A pupil who likes Science lessons but dislikes Science homework may have no honest way to answer a form asking whether they enjoy both. That problem begins before anybody adds up the responses.
A double-barrelled survey question asks about two separate things but allows only one answer. To fix it, identify the two ideas, write one neutral question about each, keep the response scales understandable, and pilot the revised wording with someone who did not write it. Do not treat responses to the original combined question as separate evidence about either topic.
The diagnosis: what does one tick actually mean?
Consider this invented class project question: “Do you enjoy Science lessons and Science homework?” The answers are Yes, No and Not sure. The pupil who enjoys practical lessons but finds homework difficult cannot express that split. Nor can a pupil who enjoys quiet homework but becomes anxious during a busy lesson. The problem is not that those pupils answered incorrectly. The form provided the wrong shape of answer.
A helpful test is to imagine one person saying “Yes to the first part, no to the second.” If that response is coherent, but the form offers only a single Yes or No, the question probably contains two targets. Words such as and can be a clue, although not every question containing and is automatically faulty. The real test is whether two independently answerable ideas have been forced into one response.
Worked example: thirty responses that cannot prove either conclusion
Illustrative numbers only, not an actual student survey: A class of 30 answers the combined Science question. Suppose 18 select Yes, 7 select No and 5 select Not sure. The totals add to 30. It would be correct to report that 18 of the 30 selected Yes to the combined wording. It would not be correct to announce that 60% enjoy Science homework, or that 60% enjoy Science lessons. The form did not measure those opinions separately.
Now redesign it as two independent questions: “Do you enjoy your Science lessons?” and “Do you enjoy your Science homework?” Each can use the same short, well-labelled response scale, with an appropriate option such as Prefer not to answer when a class project genuinely needs it. If the project seeks degrees of enjoyment, a consistent scale such as Very much, Somewhat, Not much and Not at all may be more useful than forcing a binary answer. Use a scale that the pupils can understand without guessing.
Imagine that a new, separately administered survey produces 24 pupils expressing enjoyment of lessons and 11 expressing enjoyment of homework, out of 30 respondents to each question. Those are new fictional results, not values you can derive from the earlier 18 Yes responses. They suggest different experiences worth discussing, but neither set proves why pupils feel that way. The class would still need to consider who answered, whether wording was neutral and whether the responses reflect more than the moment.
A practical repair sequence for pupils and teachers
- Define the decision: What will the class change after seeing the result? A question about homework instructions calls for a different measure from a question about enjoying lessons.
- Circle each independent idea: In “clear and enjoyable”, clarity and enjoyment are not the same property. In “safe and affordable”, safety and price are different judgements.
- Split the prompt: Give each idea its own question. Remove praise, blame and cues that pressure pupils to agree.
- Check the answer choices: Decide whether each question needs one choice, a rating scale or a written response. Do not mix incompatible scales without explanation.
- Pilot with two or three readers: Ask what each question means in their own words before wider use; revise any wording they interpret differently.
- Report the denominator: State how many actually answered each question, especially if anyone skipped one. Keep personally identifying details out of an ordinary classroom report.
Why this matters beyond questionnaires
A poorly designed question produces numbers that look persuasive because they are tidy. A clear table or colourful chart does not repair a confusing question. This is why strong information skills start with measurement design, not with graph design. Students learn to ask what was actually observed before claiming what the figures mean.
Parents can practise the same habit at home. Ask, “Did that question have one job or two?” If a child says the family wants a holiday that is cheap and exciting, invite the child to compare cost and enjoyment separately. There may be trade-offs, and the most useful choice becomes visible only when the criteria are distinct.
Common mistakes and the limits of a repaired question
Splitting a question does not automatically make a survey fair. A leading prompt such as “How helpful was our excellent lesson?” can still bias the response. A class interviewing only its friends may not represent everyone. One pupil may also misunderstand a word, or refuse a question for sensible reasons. Neutral wording, a suitable sample, privacy and honest reporting remain separate checks.
Another error is rewriting the question after looking at the answers and then presenting the old responses as though they were collected under the improved wording. Keep the original wording in the report, note its limitation, and treat the revised form as a new measurement. If pupil feedback involves personal concerns, teachers should use appropriate school procedures rather than promise anonymity that a small group cannot provide.
How to notice real improvement
Progress is visible when a child can diagnose a new double-barrelled question without being prompted and produce two neutral replacements. Try three unfamiliar prompts: one about music and sport, one about an app being fast and reliable, and one about homework being clear and interesting. The child should explain why each split matters, not merely underline the word and. Follow up a week later with an independently written survey item.
FAQs parents and pupils ask
Is a question with “and” always wrong? No. “Did you bring a pen and a ruler?” may be a practical check of whether both items are present. It becomes double-barrelled for research when the response is meant to measure separate beliefs, experiences or ratings that could differ.
Can we keep the original class results? Yes, if the report says exactly what was asked and makes no unsupported inference about its parts. The original responses may show that the question needs redesign.
Should pupils ask many detailed questions instead? Not necessarily. Every extra question adds effort. Split only independent concepts that genuinely matter to the project, and keep the survey brief.
Keep the learning connected
For the next step after improving question wording, read how to compare class surveys when response counts differ and how to collect safer anonymous feedback. The eduKate education library connects these skills to learning more broadly. Survey-question guidance from Pew Research Center also explains why asking one concept at a time matters.
One good question should give one interpretable answer. The core aim is not to collect the largest pile of ticks. It is to help children ask something precise enough that the response can genuinely inform what they do next.