Writing fair questions
Good survey work starts with fair questions. If wording is leading, vague or unbalanced, the findings can mislead even when fieldwork is carefully done.
Why fair wording matters
A questionnaire is not just a list of prompts. It is the tool that shapes how people understand the topic, recall information and choose an answer. If a question pushes people towards a position, uses loaded terms or leaves out a reasonable option, the result says more about the wording than about public opinion.
In South Africa, fair wording matters even more because research often runs across provinces, languages and very different living conditions. A phrase that sounds ordinary in Braamfontein may land differently in Polokwane, Gqeberha or a rural area where service access and daily routines are not the same. Clear neutral language helps reduce avoidable bias and makes comparison more useful.
Fair questions also support transparency. When clients, stakeholders or readers can see that a questionnaire was framed evenly, they can place more confidence in the interpretation of the findings.
Start with one clear idea per question
Each question should ask about one thing only. When two ideas are bundled together, respondents may agree with one part and disagree with the other. That creates confusion and noisy data. A common problem is the double-barrelled question, where service quality and price, safety and fairness, or trust and performance are mixed in a single line.
Keep the wording short and direct. Avoid long introductions unless they are necessary for context. If background is needed, state the minimum facts in plain language and keep the same framing for everyone.
- Ask about one topic at a time.
- Use words that ordinary respondents would use in conversation.
- Avoid jargon, acronyms and technical policy terms unless they are explained first.
- Do not assume knowledge that some respondents may not have.
- Make the time frame clear, such as now, in the past month or in the next few years.
If a topic is complex split it into separate questions rather than trying to squeeze every angle into one prompt. That usually produces cleaner answers and easier reporting.
Remove leading and loaded language
Leading language nudges respondents towards approval, disapproval or a particular emotional reaction. Loaded words can make a policy, product or public issue sound obviously good or obviously bad before the person has answered. The fix is usually simple: strip the question back to neutral wording and remove claims that have not been established for the respondent.
Words such as "failed", "unfair", "responsible", "harmful" or "necessary" can each steer the answer unless the purpose of the question is explicitly to test reactions to that exact language. The same goes for opening a question with a strong statement and then asking whether the respondent agrees. In many cases, a cleaner version starts with "Do you support or oppose" or "How favourable or unfavourable is your view of" followed by a plain description.
- Avoid praise words and criticism words in the question itself.
- Do not present one side's argument without the other side's context.
- Remove assumptions such as "given the obvious benefits" or "despite the damage caused".
- Be careful with emotionally heavy topics where tone can shift responses fast.
Question order matters too. If respondents have just answered several negative questions on the same issue, a later item can pick up that mood. Good questionnaires are reviewed not only question by question but as a full sequence.
Build balanced answer options
Fair wording is only half the job. Response options must also be balanced and complete. If the choices lean in one direction, force agreement or leave no room for uncertainty, the measure will be distorted. For closed questions, every respondent should be able to find the option that best matches their view.
A balanced scale normally offers matching positive and negative sides, with a middle point only when that midpoint is meaningful for the objective. In some studies a forced choice is useful, but it should be used deliberately, not by accident. For factual or awareness questions, include options like "don't know" where appropriate rather than pushing a guess.
- Make sure positive and negative choices are symmetrical.
- Do not overlap ranges or categories.
- Use the same scale direction throughout where possible, so respondents do not slip up.
- Include "other" only when it is likely to capture a real missing category.
- Use "don't know" or "not sure" when respondents may reasonably lack the information.
For multilingual fieldwork, answer categories need the same care as the question text. A scale that is easy to read in English may need adjustment when spoken in another official language so that the gradations still feel distinct and natural.
Check context, order and local meaning
No question sits alone. Earlier items can prime later ones, sensitive topics can affect willingness to continue and translation can shift tone if local meaning is not checked. A fair questionnaire therefore needs a practical review before fieldwork starts.
In South Africa, context checks should include geography, language and mode. A question that works online for an urban panel may be too abstract in a phone interview, and a face-to-face version may need shorter wording to sound natural when spoken. References to municipal services, grant systems, transport or household arrangements should be tested for local understanding rather than assumed.
It also helps to watch for hidden assumptions about internet access, banking, work patterns or electricity supply. During load-shedding or after a local service interruption people may interpret "normal daily life" very differently, so the wording should be precise.
- Read the questionnaire in full to spot priming and repetition.
- Check whether any item assumes a level of income, access or formal employment.
- Review translations for meaning, not just literal wording.
- Make sure interviewers can read the question aloud without stumbling.
Use a simple fairness checklist before fieldwork
Before a questionnaire goes live, run a final fairness check. This is where many small problems are caught: a missing option, an unclear time frame, a phrase that sounds fine in the boardroom but not in an interview. Even experienced researchers benefit from a slow read-through and a short pilot because little wording issues can change answers more than people expect.
A useful checklist is straightforward. Ask whether the question is neutral, specific and easy to answer. Ask whether a respondent with a different background, language or level of knowledge would understand it in the same way. Then ask whether the answer options cover the likely range of views without nudging the result.
- Is the wording neutral and free of judgement?
- Does the question ask only one thing?
- Is the time frame clear?
- Would respondents understand the key terms?
- Are the answer options balanced and complete?
- Is a "don't know" option needed?
- Could earlier questions be influencing this one?
- Has the wording been checked for all interview modes being used?
When a question passes this test it is more likely to measure opinion rather than create it. That's the standard good survey research should aim for, and its often the difference between findings that are merely interesting and findings that are genuinely useful.
What our Opinion Polling customers say
Verified Google reviews
Questions we get asked
What is a leading question in a survey?
A leading question is worded in a way that nudges respondents towards a particular answer. This can happen through loaded terms, one-sided context or answer choices that lean more heavily in one direction.
Should every survey question include a neutral option?
Not always. A neutral or middle option is useful when it reflects a real position, but some research needs a clear choice between alternatives. The important point is to decide this deliberately and keep the scale balanced.
Why do questions need testing before fieldwork?
Testing helps find unclear wording, missing response options and phrases that sound different across languages or interview modes. A short pilot can prevent avoidable bias before the full study starts.
Also worth a look
Tell us about the job
Send sizes, materials and where you are through our contact form, and we will come back with the practical options.
Open the contact form