Writing Clear Ballot and Poll Questions: Rules and Examples

A clear ballot or poll question asks about exactly one issue, uses neutral, everyday language and offers answer options that cover all reasonable positions without overlapping. That sounds simple, but small wording choices can shift results noticeably. This guide sets out the rules, shows typical mistakes with invented examples and ends with a checklist. The examples are made up for illustration and do not refer to any real organisation or vote.

The six rules

1. One issue per question. If two things are bundled, people who agree with one and disagree with the other cannot answer honestly.
2. Neutral wording. No adjectives that judge (“urgently needed”, “costly”), no appeal to authority (“as the board recommends”) in the question itself.
3. Plain language. Short sentences, familiar words, no jargon, no double negatives.
4. Complete and exclusive options. Every respondent should find exactly one fitting answer.
5. Balanced scales. As many positive as negative steps, with a clear middle if you use one.
6. Clear consequences. In a decision, voters should know what a “yes” and a “no” lead to.

Before and after: invented examples

Problem	Before	After
Two issues in one	“Should the club renovate the clubhouse and increase the membership fee?”	Two separate questions, or a clear package: “Do you approve the renovation plan in the attached motion, financed by the fee increase described there?”
Leading wording	“Do you support the long overdue modernisation of our website?”	“Should the association commission a new website in 2027?”
Double negative	“Are you against not extending the opening hours?”	“Should the opening hours be extended?”
Hidden assumption	“How much do you enjoy the new training times?”	“How satisfied are you with the new training times?” with a balanced scale
Overlapping options	“Age: 18–30, 30–50, 50+”	“Age: 18–29, 30–49, 50 or older”
Unbalanced scale	“Excellent, very good, good, poor”	“Very good, good, neither good nor poor, poor, very poor”

Yes/no or several options?

Many decisions are framed as yes/no questions. That works when there is a single concrete proposal. If there are several alternatives, a yes/no format forces you to pick one in advance, and voters who prefer another option are left out.

Alternatives for multi-option decisions:

• Single choice: each voter picks one option. Simple, but with many options the winner may have a small share.
• Approval voting: voters tick every option they find acceptable. Our explainer on approval and score voting shows how this works.
• Ranking: voters order the options by preference.

Choose the format before writing the question, as the wording follows from it.

What public referendums show

Referendum questions in public votes are a good illustration of how much wording matters. In the United Kingdom, for example, the Electoral Commission assesses the intelligibility of proposed referendum questions. In Switzerland, the Federal Constitution requires that a popular initiative respect the unity of subject matter, so that one vote does not combine unrelated issues. Our explainer What is a referendum? covers the different types of public votes. For an association or a team, the same principles apply on a smaller scale.

Answer options for opinion polls

• Offer “don’t know” or “no opinion” where it is a genuine answer.
• Randomise the order of options where the tool allows it, to avoid a bias towards the first option.
• Keep free-text fields optional and short, and remember they can reveal identity, as described in our guide to anonymous polls.

Answer options for binding votes

• Follow the wording of the motion exactly; do not paraphrase it in the voting tool.
• Typical options are “yes”, “no” and “abstain”. Check your rules on whether abstentions count.
• For elections, list candidates in a neutral order, such as alphabetical or drawn by lot.

Checklist before sending

• Does each question cover exactly one issue?
• Would someone with the opposite view find the wording fair?
• Can every respondent find exactly one fitting answer?
• Is the scale balanced?
• Is it clear what happens with each result?
• Have two or three uninvolved people tested the questions?
• Does the question display well on a phone in your chosen tool? See the survey tools comparison for options.