--- title: "40 quantitative research question examples, by format" description: "Forty closed questions grouped by format, what each one measures, its blind spot, and four scale decisions that change the answers." url: "https://raconte.ai/en/blog/quantitative-research-questions-examples" --- [Raconte](/) ⟩ [Blog](/en/blog) # 40 quantitative research question examples, and what each format measures Forty closed questions grouped by format, what each one measures, its blind spot, and four scale decisions that change the answers. August 25, 2026 Godefroy Key takeaways * A quantitative question fixes the shape of the answer in advance: yes or no, an option from a list, a rating on a scale, a number, or statements to put in order. The test is whether two people giving the same answer produce the same value in your spreadsheet. * Two different things get called a quantitative research question: the question your study sets out to answer, and the item you put in front of a respondent. This bank holds the items. * Eight closed formats exist, and each measures something different with its own blind spot. The format you pick decides what you will not see. * Changing only the range of an answer scale moved the share of people reporting more than 2½ hours of daily television from 16.2% to 37.5%. A scale measures the respondent as much as the behaviour. * People answer a closed question because it costs them almost nothing. Ticking a box takes no writing, so the answer explains nothing. An open-ended question draws 18% nonresponse against 1% to 2% for a closed one. The response rate hits 79%, and nobody skipped a question. Overall satisfaction is 3.4 out of 5, down from 3.6, and the team has been arguing about it for a week. The survey worked. Closed questions are the ones people answer, which is all anyone asks of them. Nobody had to write a sentence, so the team learns that satisfaction fell and nothing about why. Every closed question shares that limit. They differ in what they measure: a single choice, an agreement scale and a ranking measure three different things, and each one misses something else. This page is a bank of quantitative questions you can copy, grouped by those formats rather than by industry, each with what it measures and its blind spot. The bank of [open-ended questions](/en/blog/open-ended-questions-qualitative-research) covers the questions that ask why. ## Two different things get called a quantitative question A **research question** is what your study sets out to answer. “Does the length of onboarding predict whether an account is still active at 90 days?” Nobody is ever asked that. You write it at the top of the plan, and it fixes what you go and measure. A **survey item** is the question a respondent reads and answers. “How many days passed between signing up and your first export?” That is what you put in the form, and the answers to it are what let you settle the research question. Most question banks on this subject shuffle the two together, which is how you end up with “What is the relationship between stress and physical health in retirees?” sitting in a list of things to ask retirees. The forty questions here are survey items. If you are writing the research question instead, the useful split is descriptive (how much of it is there), comparative (do these two groups differ) and relational (do these two variables move together), and each of those points at a different set of items. ## What makes a question quantitative A quantitative question fixes the shape of the answer in advance: yes or no, an option to pick from a list, a rating on a scale, a number to type, or a set of statements to put in order. Nobody writes a sentence, so every answer drops into a spreadsheet column without anyone interpreting it. The test takes a second. If two people give the same answer, do they produce the same value in your spreadsheet? A five-point scale passes. “How was your week?” does not. Quantitative question Qualitative question What the person does Ticks, rates, counts or ranks Describes something they lived through What you get back A value in a column A story you read against other stories What it answers How many, how often, how much, which one Why, and how it came about Analysis Counting and cross-tabulation, from the first response Reading, coding, comparing Population it suits Hundreds to thousands Ten to fifty Where it fails You only learn what you thought to ask Slow, and hard to compare across people Both are instruments, and the useful question is which one you need this week. A number shows you that something moved. It does not show you which change would move it back. ## The eight formats, five questions each Take the two or three formats your study needs rather than the whole bank. ### Single choice, when the categories are already known 1. Which of these best describes your role? (a list of roles) 2. Which plan are you on today? (a list of plans) 3. Where did you first hear about us? (a list of channels) 4. Which device did you use for your last session? (phone, tablet, computer) 5. Which of these describes your situation right now? (a list of mutually exclusive statuses) Single choice splits your population into groups you can count and cross with every other answer, which is more useful than reporting it on its own. The categories also have to exclude each other, or the same person fits two of them. A missing option is harder to spot. Someone whose real answer is not in the list picks the nearest one, and their answer looks exactly like a true one. You wrote the list, so the answers can only confirm categories you already had in mind. ### Multiple choice, when several answers are true at once 6. Which of these features have you used in the last month? (select all that apply) 7. Which tools do you use alongside this one? (select all that apply) 8. Which of these got in your way during setup? (select all that apply) 9. Who else was involved in the decision to buy? (select all that apply) 10. Which of these do you check before you decide? (select all that apply) Multiple choice gives you coverage and combinations: you learn which tools your users run alongside yours, and you can cross each combination with churn. Four ticked boxes do not show which one mattered, and people tick more of them as the list gets longer, so two surveys with different list lengths are not comparable. ### Yes or no, when the fact really is binary 11. Have you finished setting up your account? 12. Did you contact support in the last three months? 13. Do you use the mobile app? 14. Was your problem resolved on the first contact? 15. Is there a budget allocated for this in the current year? A binary question is the cleanest screener there is, and it filters the rest of the survey without costing the respondent any effort. Keep it for facts. As soon as it touches an attitude, everything between the two options disappears, and the same yes covers the person who would recommend the tool tomorrow and the one who has been hesitating for six months. Jon Krosnick, professor of communication and political science at Stanford, and Stanley Presser, sociologist at the University of Maryland, go further in their [chapter on questionnaire design](https://web.stanford.edu/dept/communication/faculty/krosnick/docs/2010/2010%20Handbook%20of%20Survey%20Research.pdf) and recommend avoiding yes/no items for opinions altogether, because a yes/no opinion item is an agree/disagree item in disguise, and people lean towards yes. ### Agreement scales, when you are after an attitude 16. “The tool does what I expected it to do.” (strongly disagree to strongly agree) 17. “I can find what I need without asking anyone.” 18. “The time this takes me each week is reasonable.” 19. “I would be comfortable recommending this to a colleague.” 20. “I understand what I am being charged for.” Agreement scales are the fastest way to place a population on an attitude and to watch that position move between two waves of the same survey. They also carry the best documented bias in survey research. Krosnick and Presser summarise the evidence: across ten studies, 52% of people agreed with an assertion while only 42% disagreed with the opposite assertion, and across seven studies an average of 22% agreed both with a statement and with its reversal. They put the pull towards agreement at about 10 percentage points. The fix is to stop asking people to agree and ask them about the behaviour directly. Rather than “I can find what I need without asking anyone”, ask “when you needed to find something last week, did you work it out yourself, ask a colleague, or contact support”. Willem Saris and Melanie Revilla at Universitat Pompeu Fabra, with Krosnick and Eric Shaeffer, a researcher at Ohio State University, ran that comparison across fourteen countries in the European Social Survey and measured [item-specific options at a measurement quality of roughly 0.74 to 0.76 against 0.18 to 0.40](https://ojs.ub.uni-konstanz.de/srm/article/view/2682) for the agree/disagree versions of the same items. ### Rating scales, when each attribute needs its own answer 21. How would you rate the speed of the tool? (1 to 5) 22. How would you rate its reliability? (1 to 5) 23. How easy was it to learn? (1 to 5) 24. How would you rate the value for what you pay? (1 to 5) 25. How satisfied were you with your last support exchange? (1 to 5) Rating each attribute separately exposes the weak one that an overall score hides, and you see sooner what to fix. The attributes you did not list cannot be rated, and the trade-offs between them never show up, so a customer who rates speed 2 and stays for the reliability looks the same as one who is about to leave. Krosnick and Presser conclude that seven points are probably optimal for a rating scale, and Revilla, Saris and Krosnick found that [moving an agreement scale from five points to seven costs 0.139 of measurement quality](http://repositori.upf.edu/bitstream/10230/28305/1/RECSM-WP-005-Choosing%20the%20number%20of%20categories%20in%20agree-disagree%20scales.pdf), so five is the safer choice on agreement specifically. ### Frequency, when you want to know how often 26. How many times did you use the tool last week? 27. When did you last export a report? (today, this week, this month, longer ago, never) 28. How many times have you contacted support in the last six months? 29. Last week, how many people on your team opened the tool? 30. How many of the last five releases did you read the notes for? Crossing frequency with satisfaction usually produces the single most informative table in a survey, because it separates the people who are unhappy from the people who barely showed up. Reading the options, the respondent works out what counts as a normal answer and places themselves against it. Asking for a number over a dated window removes that reference, because there are no options left to read. Norbert Schwarz and Hans-J. Hippler, then at ZUMA, the German survey methods centre in Mannheim, with Brigitte Deutsch and Fritz Strack, put the same television viewing question to two groups and changed nothing but the range of the scale. On a scale running up to “more than 2½ hours”, 16.2% of people landed in the top box, while on a scale running up to “more than 4½ hours”, [37.5% reported watching more than 2½ hours](https://dornsife.usc.edu/norbert-schwarz/wp-content/uploads/sites/231/2023/11/99_ap_schwarz_self-reports.pdf), because people read the middle of the scale as the normal amount. ![The same television question answered on two scale ranges, 16.2% against 37.5%](/.netlify/images?url=_astro%2Fscale-range-effect-en.BMwKuoui.jpg&w=1376&h=768&dpl=6a9479917fbe52000823da63) Vague words do the same damage in a smaller way. Tarek Al Baghal, at the Institute for Social and Economic Research in Essex, asked people to translate frequency words into numbers of times per year. [“Very often” averaged 15.5, with individual answers running from 7 to 37](https://pdfs.semanticscholar.org/936a/f67b87292a259a1d221250f04c5d05f40632.pdf). Everyone agrees that “very often” is more than “often”, but the number behind the word changes from one person to the next. ### Numeric entry, when the answer is a real quantity 31. How many people are on your team? 32. How many hours a week do you spend on this task? 33. How many months have you been using the tool? 34. What is your annual budget for tools of this kind? 35. How many suppliers did you evaluate before choosing? Numeric entry is the only format on this page that offers no ready-made answers, which is why it escapes the scale-range effect entirely and why it can be checked against your own records. You pay for that in accuracy. People estimate, they round to fives and tens, and they answer for a typical week that may never have happened. Ask for a count over a window someone can actually reconstruct, and expect the distribution to have a long tail you will need to trim. ### Ranking, when you need priorities rather than scores 36. Rank these five features from most to least useful to you. 37. Rank these four criteria by their weight in your buying decision. 38. Rank these three problems by what they cost you. 39. Rank these channels by how you prefer to be contacted. 40. If only one of these could ship this quarter, rank them. Most respondents rate five features 4 or 5 on a rating scale. A ranking forces them to separate those five, which makes it the format to reach for before a roadmap decision. A ranking does not record distance: first and second may be nearly tied or worlds apart, and you get the same ranking either way. Keep the list to five items, because ranking gets hard to hold in your head past that, and expect the analysis to be more work than any other format here. ## The eight formats side by side Format What it measures What it leaves out Reach for it when Single choice Which group someone belongs to Anyone whose answer is missing from the list The categories are known and exclusive Multiple choice Coverage and combinations Weight and order between the ticked boxes Several answers can be true at once Yes or no A binary fact you can filter on Every degree between the two options The question is factual Agreement scale Where a population sits on an attitude How much of the agreement is real You are tracking the same wording over time Rating scale The weak attribute an overall score hides The attributes you did not list, and the trade-offs You need to compare parts of one experience Frequency Intensity of use, crossed with anything else Accuracy, if the options suggest a normal answer You ask for a count over a dated window Numeric entry A quantity you can average and verify Accuracy, since people estimate and round The person can genuinely count it Ranking Priority under constraint The distance between two adjacent ranks You have to choose and want the respondents to choose too graph TD A\["What do you need from the answer?"\] --> B\["A category"\] A --> C\["A quantity"\] A --> D\["A degree"\] A --> E\["A priority"\] A --> N\["A binary fact"\] N --> O\["Yes or no"\] B --> F{"Can someone be in two at once?"} F -->|"No"| G\["Single choice"\] F -->|"Yes"| H\["Multiple choice"\] C --> I{"Can they count it exactly?"} I -->|"Yes"| J\["Numeric entry"\] I -->|"They would estimate"| K\["Frequency over a dated window"\] D --> P{"One attribute or an attitude?"} P -->|"An attribute"| L\["Rating scale, five to seven points, midpoint kept"\] P -->|"An attitude"| Q\["Agreement scale, five points"\] E --> M\["Ranking, five items at most"\] ## Four decisions that move the number before anyone answers Teams treat all four as formatting details and settle them by copying the last survey. * **Keep the midpoint.** On a five-point scale, the midpoint is “neither agree nor disagree”. Dropping it forces an even number of positions, so everyone has to lean one way. Colm O’Muircheartaigh at the University of Chicago, with Krosnick and Armen Helic, modelled what that produces: genuinely neutral people scatter at random onto the moderate points on either side, and reliability drops. * **Drop the “don’t know” filter.** Mikael Gilljam at the University of Gothenburg and Donald Granberg at the University of Missouri asked three questions about building nuclear power plants. On the one offering “don’t know”, 15% of respondents picked that option. The two without it left only 3% and 4% blank, and those answers predicted how the same people voted in a referendum months later. The 15% who took the “don’t know” had an opinion too, they simply took the way out they were offered. * **Ask items one at a time rather than in a grid.** Alexandru Cernat, survey methodologist at the University of Manchester, and Mingnan Liu compared the two layouts on the same items and found straightlining, where someone picks the same answer down a whole block, at [1.3% item by item against 6.4% in a grid](https://journals.sagepub.com/doi/pdf/10.1177/0894439316674459) on four-point scales. Chan Zhang and Frederick Conrad at the University of Michigan measured it in the field on a Dutch probability panel of 5,523 people and found rates up to 49.6% on a single grid among respondents who were already rushing. * **A preset list sets the answers you get, even with an “other” box.** Howard Schuman, sociologist at the University of Michigan, and Presser asked what the most important thing is for children to prepare them for life. With a list that included “to think for themselves”, [61.5% picked it. With no list, only 4.6% volunteered anything of the kind](https://dornsife.usc.edu/norbert-schwarz/wp-content/uploads/sites/231/2023/11/99_ap_schwarz_self-reports.pdf). People stay inside the options you offered. ## Where the number stops People answer a quantitative question because it costs them almost nothing. The Pew Research Center measured [18% nonresponse on an open-ended question against 1% to 2% on a closed one](https://www.pewresearch.org/decoded/2021/10/14/why-do-some-open-ended-survey-questions-result-in-higher-item-nonresponse-rates-than-others/) across 92 open questions on its American Trends Panel, and Pew works with a paid panel used to writing answers. On a survey sent cold, the gap widens. That ease has a price. Ticking a box takes no writing, so the answer explains nothing. Satisfaction drops from 3.6 to 3.4, and that is as far as you get. You then move on to other questions, open ones this time: * What happened, in order, on the occasion that produced this score? * What did the person do instead, once the tool did not do what they wanted? * What would have had to change, and by when? Our post on [customer satisfaction surveys](/en/blog/customer-satisfaction-survey) covers what a satisfaction score hides. Measure with closed questions when you already understand the behaviour and want to know how much of it there is. When the number moved and nobody can say why, go and ask. ## Getting the why without booking twenty calls The answer to a surprising number is a conversation. Teams send a survey to five thousand people and then run only four interviews, because a conversation is expensive: a slot to book, an interviewer who is free, and someone taking notes. That cost is what we set out to remove with [Raconte](/en). You write what you want to ask once, as a brief, then send a link. The person opens it in their browser whenever it suits them and talks for a few minutes. The AI listens, follows up as soon as an answer stays vague, and stops once it has covered everything you asked for. You get the transcript, the audio, a summary and per-message sentiment within seconds of each call, plus aggregated reports once several people have been through the same brief. Keep the survey, and put the follow-ups in a conversation. Ask your closed questions, then send a link to the people whose answers you cannot explain, with a brief built from what happened, what they did instead, and what would have had to change. The [API and the MCP server](/en/docs/api) push the results into your CRM or into an agent without a manual export. If all you need is one countable fact from two hundred people, a form is faster for everyone, and the forty questions in this bank are enough. An AI works from the voice alone, so it misses the look on someone’s face. A negotiation or a sensitive subject still calls for a human interviewer. And a conversation gives you depth rather than a series you can plot, so it complements the survey. The first 60 interview minutes each month are free with no card, and the [pricing page](/en/pricing) covers what happens after that. ## Where to start Take the survey you are about to send and mark each question with what it produces: a category, a binary fact, a quantity, a degree, or a priority. Any question that produces none of those is an open question sitting in a closed instrument, and it will come back blank about one time in five. Then make the three fixes that cost nothing before sending. Replace every agree/disagree item with one that names the options directly. Replace every frequency word with a count over a window, “in the last 30 days” rather than “regularly”. And break any grid into separate questions. For the questions none of that fixes, because you want a reason rather than a measure, the [45 open-ended questions](/en/blog/open-ended-questions-qualitative-research) are grouped by research goal, and the [interview guide](/en/blog/interview-guide) gives the structure to run them in. ## Frequently asked questions ### What is an example of a quantitative question? “How many times did you use the tool last week?” is a quantitative question. So are “How would you rate its reliability, from 1 to 5?” and “Which plan are you on today?”. In each case the person picks from options you wrote or gives a number, and every answer drops into a spreadsheet column without anyone having to interpret it. ### What are quantitative questions? Quantitative questions are closed questions. They ask the respondent to pick an option, give a rating, give a number or order a list, and every answer is therefore a value you can average, segment and compare. They come in eight common formats: single choice, multiple choice, yes or no, agreement scale, rating scale, frequency, numeric entry and ranking. ### What are 5 examples of quantitative research questions? Pick one from five of the eight formats: “Which plan are you on today?” (single choice), “Which of these features have you used in the last month?” (multiple choice), “How many times did you contact support in the last six months?” (frequency), “How would you rate the speed of the tool, from 1 to 5?” (rating scale), and “Rank these five features from most to least useful to you.” (ranking). ### What are 5 examples of quantitative data? Quantitative data includes the number of active users in a month, the minutes spent in a session, a satisfaction score out of five, the number of support tickets opened per account, and the monthly revenue per customer. All five are values you can add, average and plot over time, which is what separates quantitative data from a transcript or an open-text answer. ### What is the difference between a quantitative research question and a survey question? A research question is what your study sets out to answer, such as “does onboarding length predict retention at 90 days”. Nobody is ever asked it. A survey question is the item a respondent reads and answers, such as “how many days passed between signing up and your first export”. You write the research question first, then design the survey items whose answers will settle it. ### How many points should a rating scale have? Krosnick and Presser conclude that seven points are probably optimal for rating scales in general. On agreement scales specifically, Revilla, Saris and Krosnick measured that going from five points to seven costs measurement quality, so five is the safer choice there. Past seven points nothing is gained, and keeping the midpoint matters more than the exact count, because removing it makes genuinely neutral people pick a nearby point at random. ### Should a survey question offer a 'don't know' option? Usually not. Gilljam and Granberg found that a nuclear power question offering “don’t know” collected 15% of answers there, while the same people answered two similar questions without the option at 3% and 4% nonresponse, and those answers predicted their referendum vote months later. The opinion existed. Offer the escape hatch and the people who did not feel like deciding take it. ### What is a qualitative question example? “Walk me through the last time you ran an export” is a qualitative question. It cannot be answered with yes, no or a number, and it asks the person to reconstruct something they lived through rather than to place themselves on a scale. Our bank of open-ended questions groups its 45 items by research goal, for discovery, churn and pricing among others. ### Can you mix quantitative and qualitative questions in the same study? Yes, and the usual pattern puts the closed questions first. Closed questions show you what moved and on which segment, then a conversation with a subset of those people explains it. An open text box at the end of a form gets left blank far more often than a closed question, and nobody is there to follow up on a two-word answer. ![40 quantitative research question examples, by format](/.netlify/images?url=_astro%2Fthumbnail.DrAy0qp2.jpg&w=1376&h=768&dpl=6a9479917fbe52000823da63) Table of contents [1\. Two different things get called a quantitative question](#two-different-things-get-called-a-quantitative-question)[2\. What makes a question quantitative](#what-makes-a-question-quantitative)[3\. The eight formats, five questions each](#the-eight-formats-five-questions-each)[4\. The eight formats side by side](#the-eight-formats-side-by-side)[5\. Four decisions that move the number before anyone answers](#four-decisions-that-move-the-number-before-anyone-answers)[6\. Where the number stops](#where-the-number-stops)[7\. Getting the why without booking twenty calls](#getting-the-why-without-booking-twenty-calls)[8\. Where to start](#where-to-start)[9\. Frequently asked questions](#frequently-asked-questions) Table of contents 1\. Two different things get called a quantitative question 2\. What makes a question quantitative 3\. The eight formats, five questions each 4\. The eight formats side by side 5\. Four decisions that move the number before anyone answers 6\. Where the number stops 7\. Getting the why without booking twenty calls 8\. Where to start 9\. Frequently asked questions Raconte gets your customers and experts talking. The AI runs the interview. Describe your interview, share the link: an AI agent runs it by voice with each participant. You get back the transcript and analysis. [Learn more](/en) [Try for free](/login) Ready to launch your first interview? Describe your interview, share the link: an AI agent runs it by voice with each participant. You get back the transcript and analysis. [Try for free](/login) ## Continue reading [![45 open-ended questions for qualitative research](/.netlify/images?url=_astro%2Fthumbnail.x5EdDQ3f.jpg&w=1376&h=768&dpl=6a9479917fbe52000823da63) August 22, 2026 ## 45 open-ended questions for qualitative research A bank of 45 open-ended questions grouped by research goal, the reason each one is phrased the way it is, and the four ways a question closes back up. ](/en/blog/open-ended-questions-qualitative-research)[![Customer satisfaction survey: an example and its limits](/.netlify/images?url=_astro%2Fthumbnail.BVaPkpjh.jpg&w=1376&h=768&dpl=6a9479917fbe52000823da63) August 19, 2026 ## Customer satisfaction survey: an example and its limits A customer satisfaction survey template you can copy, what every question tells you, and the four blind spots of a score out of 10. ](/en/blog/customer-satisfaction-survey)[![Interview guide: the method and a full example](/.netlify/images?url=_astro%2Fthumbnail._6IslZAl.jpg&w=1376&h=768&dpl=6a9479917fbe52000823da63) July 29, 2026 ## Interview guide: the method and a full example The three interview families, a five-block structure, a complete guide you can copy, and the five question types that quietly ruin a study. ](/en/blog/interview-guide)