12 rules for survey questions that get honest answers
Avoid double-barrelled, leading and vague questions. Twelve practical rules with before-and-after examples, plus tips for Filipino, Taglish and multilingual surveys.
By the Kita team · Updated · 3 min read
On this page
A survey is only as good as its questions. Most bad data isn’t caused by bad respondents — it’s caused by questions people can’t answer honestly. These twelve rules fix the most common problems.
Before and after
| Instead of… | Ask… | Why |
|---|---|---|
| How satisfied are you with our price and service? | Two questions: one on price, one on service | One idea per question |
| Don’t you agree our new menu is better? | How does the new menu compare with the old one? | Not leading |
| Do you exercise regularly? | In the past 7 days, on how many days did you exercise for 30 minutes or more? | Specific and measurable |
| Age: 18–25, 25–35, 35+ | 18–24, 25–34, 35–44, 45 or older | No overlaps |
The 12 rules
- One idea per question. “Price and service” is two questions. Double-barrelled questions produce answers you can’t interpret.
- Don’t lead. Avoid “Don’t you agree…”, “How much do you love…” and loaded words like “convenient” or “expensive” in the question itself.
- Balance your scales. Give as many negative options as positive ones, with a true midpoint if neutral is a real opinion.
- Offer a way out. Add “Not sure” or “Not applicable” when some people genuinely can’t answer.
- Make options exhaustive and non-overlapping. Every respondent should fit exactly one box. Add “Other” when in doubt.
- Be specific about time. “In the past 7 days” beats “usually” or “regularly”.
- Use everyday words. Skip jargon and acronyms; your respondents aren’t your thesis panel.
- Ask easy questions first, and sensitive ones (income, health) later — after trust is built.
- Keep it short. Every question should earn its place in your analysis plan. Aim for under 10 minutes.
- Mix in one attention check. For example: “To show you’re reading, select Somewhat agree.” One is enough; more annoys honest people.
- Limit open-ended questions to the one or two that matter most. Long text boxes are skipped or answered in a word.
- Pilot test. Five to ten people, timed, then ask them which questions were confusing. Fix those.
Surveys in Filipino, Taglish or other languages
- Translate meaning, not words. Have a native speaker review, then back-translate a few key questions to check they still mean the same thing.
- Avoid idioms and slang that differ between regions.
- Keep scale labels consistent across languages — “Lubos na sumasang-ayon” should always sit at the same end of the scale.
- Test on a phone. Most respondents in Southeast Asia answer on mobile; long matrix grids are painful on small screens.
Build it on Ask Kita
The Ask Kita builder supports single and multiple choice, rating scales, NPS, ranking, matrix grids and short or long text. It adds an attention check automatically, shows a live phone preview as you build, and warns you about questions that ask for personal data or contact details. Collecting thesis data? Read how to find respondents for your thesis.