Back
Closed Ended Questions: Examples and When to Use Them
Matthieu Saussaye
A closed ended question gives the respondent a fixed set of answers to choose from. Yes or no, one option from a list, a point on a scale, an order of preference. The respondent selects; they do not write. That constraint is the whole point: every answer lands in a category you defined in advance, so the results can be counted, compared and charted without anyone reading them first.
What you need to know before writing one:
The contrast is with open ended questions. Open questions collect text or speech in the respondent's own words. Closed questions collect a selection from your words.
There are five working sub-types. Dichotomous, multiple choice, rating scale, ranking and matrix. Each answers a different shape of question.
They dominate quantitative work for a reason. Fixed categories mean you can compute percentages, run significance tests, track a number over time and cut it by segment.
The cost is real. You only ever learn what you thought to ask. A missing option is invisible in the results: it looks like nobody mentioned it, when in fact nobody could.
Answer sets need to be exhaustive and mutually exclusive. Every respondent must find an option that fits, and no respondent should fit two.
Most good questionnaires are mostly closed. Closed questions carry the measurement; a small number of open questions carry the explanation.
The five types of close ended questions
"Closed question" is a family, not a format. Picking the wrong member of the family is one of the most common ways a questionnaire produces data nobody can use.
Type | How it works | What you get | Best for | Main risk |
|---|---|---|---|---|
Dichotomous | Two options, usually yes or no | A percentage split | Screening, eligibility, did it happen or not | Forcing a binary onto something that is a spectrum |
Multiple choice | A list of options, single or multi select | Counts and percentages per option | Categories, brands, behaviors, reasons | An incomplete option list that hides the real answer |
Rating scale | A point on an ordered scale | A distribution and a comparable average | Satisfaction, agreement, likelihood, importance | Everyone clustering in the top two boxes |
Ranking | Items placed in order | A priority order, no distances | Forcing trade-offs between options | Long lists where the middle is guesswork |
Matrix | Several items sharing one scale in a grid | Comparable ratings across many items | Batteries of related attributes | Straight-lining and mobile layouts that break |
Each of these has its own design rules. The deep versions live in the guides on multiple choice questions, single select questions, Likert scales, rank order questions and matrix questions. This article is about what they share.
Closed ended versus open ended
The difference is not length or difficulty. It is who supplies the vocabulary.
In a closed question, you supply it. You decide the categories, and the respondent's job is to locate themselves within them. In an open question, the respondent supplies it, and your job is to make sense of what comes back.
That single distinction drives everything downstream:
Analysis. Closed answers are already coded. You can tabulate them the moment fieldwork closes. Open answers have to be read and categorized before they become countable.
Comparability. Two people who both pick "Very satisfied" have given the same answer by definition. Two people who both write a sentence have not.
Discovery. Open questions can surprise you. Closed questions cannot. A closed question can only confirm, size or rank the possibilities you already had in mind.
Effort. Selecting an option takes a second. Writing a considered paragraph takes a minute, which is why most people do not.
The two formats are not competitors. They answer different questions, and a well-built questionnaire uses closed questions to measure and open questions to explain. That split is the practical face of the wider qualitative versus quantitative distinction.
Why closed ended questions in research dominate quantitative work
Nearly every number in a research report comes from a closed question. There are four reasons for that.
They produce a denominator
"38% chose option B" only means something because every respondent saw the same options and had to pick one. The fixed answer set creates a base you can divide by. Open text has no denominator: if 40 people mention price unprompted, you know 40 people mentioned it, not that 40 people care about it.
They scale without extra cost
The analysis cost of a closed question is the same whether you field 200 responses or 20,000. Coding open text scales with sample size, either in analyst hours or in machine cost.
They support statistical testing
Significance tests, cross-tabs, correlation, segmentation and regression all need variables with defined values. Closed questions produce those directly. If you want to test whether satisfaction differs between two customer segments, both segments need to have answered the same scale.
They can be tracked
A tracker only works if the question is identical wave after wave. Closed questions freeze the wording and the answer set, so a movement in the number reflects a movement in the market rather than a change in how you asked. This is why brand trackers and satisfaction programs are built almost entirely from closed questions.
The cost: you only learn what you thought to ask
Every closed question is a hypothesis about what the answers could be. When the hypothesis is incomplete, the data does not tell you. It just quietly returns a clean-looking result.
Suppose you ask why a customer canceled and list five reasons: price, missing features, poor support, switched to a competitor, no longer needed it. The real reason for a slice of your churn might be that the onboarding never finished and they never got the product working. Nobody can say that. They pick the closest option, usually "no longer needed it", and your report concludes that a chunk of churn was demand-driven and unfixable.
Three specific failure modes follow from this:
Forced choice into a wrong bucket. Respondents rarely abandon a survey over a missing option. They pick something. The error enters your data silently.
Anchoring. Showing a list changes what people report. An option nobody would have volunteered can collect substantial support once it appears on screen, simply because it sounds plausible and it is there.
No intensity, no reason. "Price" as a selected option does not distinguish between "it is 10% more than I hoped" and "it is four times my budget". Both are one tick in the same column.
The standard mitigations are worth building in by default: pilot the question with a small open version first and use the answers to build the list, include an "Other, please specify" field and actually read it, and follow important closed questions with a short probe asking why.
Rules for writing the answer set
Most closed question failures are answer set failures, not question wording failures.
Exhaustive. Every respondent must find a home. Add "Other", "None of these" or "Not applicable" when the list genuinely may not cover everyone.
Mutually exclusive. Options must not overlap. Age bands of 18 to 25 and 25 to 34 put every 25-year-old in two places at once.
Balanced. A satisfaction scale needs as many negative points as positive ones. Three positive options and one negative one manufactures a positive result.
One idea per option. "Fast and affordable" is unanswerable for anyone who found it fast and expensive.
Randomize order where order is arbitrary. Items near the top of a long list get selected more often. Randomizing spreads that bias instead of concentrating it on your first option. Keep scales and any natural sequence fixed.
Keep the list scannable. Long option lists on a phone push respondents toward whatever is visible without scrolling.
Example of close ended questionnaire items
Fourteen closed questions with their answer sets, grouped by type. Adapt the wording; keep the structure.
Dichotomous
Have you purchased from us in the last 12 months? Yes / No
Are you currently employed full-time? Yes / No
Did you contact customer support about this order? Yes / No / I do not remember
Multiple choice, single select
How did you first hear about us? Search engine / Social media / Recommendation from someone I know / Advertising / Press or article / Other, please specify
Which of these best describes your role? Individual contributor / Team lead or manager / Director / Executive / Business owner / Not currently working
How often do you use the product? Daily / Several times a week / Weekly / Monthly / Less than monthly / Never
Multiple choice, multi select
Which of these features have you used in the past month? Select all that apply. Reporting / Exports / Integrations / Mobile app / Notifications / None of these
Which of these did you consider before choosing us? Select all that apply. [Competitor list] / Building it in-house / Doing nothing / I do not recall
Rating scales
Overall, how satisfied are you with the product? Very dissatisfied / Dissatisfied / Neither satisfied nor dissatisfied / Satisfied / Very satisfied
How likely are you to recommend us to a colleague? 0 to 10, where 0 is not at all likely and 10 is extremely likely
The setup process was straightforward. Strongly disagree / Disagree / Neither agree nor disagree / Agree / Strongly agree
How would you rate the value for money? Very poor / Poor / Acceptable / Good / Very good
Ranking and matrix
Rank these four factors in the order you would want us to improve them first. Price / Speed / Support quality / Feature range
Rate each of the following on the same 5-point satisfaction scale. Rows: ease of use, reliability, support responsiveness, documentation, pricing clarity
How many closed questions is too many
Closed questions are cheap for you and cheap for the respondent, which is exactly why questionnaires bloat. Every question that gets added because it "might be interesting" adds length, and length drives abandonment and careless answering. It is one of the mechanics behind falling survey response rates.
A working test for each closed question: name the decision that changes depending on the answer. If you cannot, cut it. "We would like to know" is not a decision.
Common mistakes with close ended questions
Building the option list from internal vocabulary. Your feature names, your segment labels, your internal jargon. Respondents pick the option they think they understand, and you get a clean chart of a misunderstanding.
Leading options. "How much did our excellent support help you?" has already told the respondent the answer. So has an option list where the flattering choices come first and the critical ones are buried.
Double-barreled stems. "How satisfied are you with delivery speed and packaging?" cannot be answered by anyone who liked one and not the other.
Forcing an answer with no exit. If "don't know" is a true state for part of your sample and you do not offer it, those people guess. Guesses are indistinguishable from opinions in the export.
Adding "Other, please specify" and never reading it. The other box is the only early warning that your list is wrong. Reading it is not optional.
Changing wording mid-tracker. Rewording an answer option between waves breaks the trend line. If the change is necessary, run both versions in one wave to bridge them.
Reporting a mean on a scale that does not support one. Averaging ordinal labels is common, and it hides bimodal distributions where half your customers are delighted and half are furious. Show the distribution alongside any average.
Where a probe beats another option
When a closed question comes back ambiguous, the instinct is to add options next time. Usually the better move is to add a follow-up.
If 31% select "price" as their reason for leaving, no expansion of the option list explains what "price" means to those people. One conversational follow-up does: ask why, right after they answer, while the reason is still in mind. That is the piece closed questions structurally cannot deliver, and it is the reason the best questionnaires are closed in structure with a small number of well-placed open moments.
For the full map of question types and how to sequence them, start with the survey questions guide, then read the counterpart to this article on open ended questions.
Keep the structure, recover the reason
Closed questions give you numbers you can defend. What they never give you is the sentence behind the tick. SmartInterview runs surveys by voice or text and asks the follow-up automatically: a respondent selects "price", the AI asks what about the price, and the answer comes back in their own words.
Those open answers are coded into themes as they arrive, so you can count them next to your closed data instead of leaving them in a spreadsheet column nobody opens. Surveys run in the respondent's own language, and quotas keep the sample shaped the way you specified.
Build your next questionnaire with both halves working: start free or talk to us about a study.
Frequently Asked Questions
What is a closed ended question?
A closed ended question offers the respondent a fixed set of answers and asks them to select from it rather than write their own. Yes/no questions, multiple choice lists, rating scales, rankings and matrix grids are all closed. Because every answer falls into a category you defined in advance, the results can be counted and compared without anyone reading them individually.
What is an example of a close ended questionnaire?
A customer satisfaction questionnaire is a typical one: a screening yes/no on whether the person purchased recently, a multiple choice on which product they bought, a 5-point satisfaction scale, a 0 to 10 recommendation question, and a matrix rating several attributes on the same scale. Every question has predefined answers, so the whole survey can be tabulated the moment fieldwork closes.
What is the difference between closed and open ended questions?
Closed questions supply the answer options; open questions ask respondents to answer in their own words. Closed questions produce countable, comparable data and can only confirm possibilities you already listed. Open questions can reveal reasons and issues you never anticipated, but they require coding before they become countable. Most questionnaires use closed questions to measure and a few open questions to explain.
Why are close ended questions used in research?
Because they produce variables. Fixed answer sets create a common base across all respondents, which is what percentages, cross-tabulations, significance tests and segmentation all require. They also keep wording constant across waves, which is the only way a tracking study can attribute a change in the number to a change in the market rather than a change in the question.
What are the disadvantages of closed ended questions?
You only learn what you thought to ask. If the real answer is missing from your list, respondents pick the nearest option and the error becomes invisible in the results. Closed questions also strip out intensity and reason: a selected option tells you what, never why or how much. Showing a list can additionally prompt answers people would never have volunteered.
Should every closed question include an "Other" option?
Include one whenever your list may not be exhaustive, which is most of the time for behaviors, brands and reasons. Leave it out when the categories are genuinely complete, such as a yes/no about an event you can verify. When you do include it, read the responses: a rising "Other" share is the clearest signal that the option list needs rebuilding before the next wave.
