Key takeaways
- The option list does the measuring, not the question stem. Wording, order and completeness move the result more than the sentence above them.
- Option order is worth about 7 percentage points. That is the primacy bias Pew measured across 11,114 respondents, rising to 11 and 12 points among young and Hispanic respondents.
- Single-select and multi-select answer different jobs. A forced choice recovers endorsement a check-all list quietly loses.
- Four or five options is the working ceiling. Past that, people stop reading the list and start scanning it, and scanning is where order bias lives.
Most advice about multiple choice questions is about the question. The measurable failures are almost all in the answers. A Pew Research Center analysis published on 27 August 2026 ran 13 questions with randomised response options past 11,114 US adults and found respondents leaned toward whatever sat at the top of the list by about 7 percentage points. No wording changed. Only the order did.
What counts as a multiple choice question?
A multiple choice question gives the respondent a fixed set of answer options and asks them to pick from it. That covers three formats people treat as one:
- Single-select. One answer, usually radio buttons. Best when the options are mutually exclusive and you need a clean split.
- Multi-select. Any number of answers, usually checkboxes. Also called check-all-that-apply.
- Rating and Likert items. Single-select with an ordered list, but the order carries meaning, so they follow different rules.
What it is not: a yes/no or dichotomous question, a two-option special case with its own bias profile, or an open-ended question, which hands over a blank box. Multiple choice sits between them: cheaper to analyse than free text, more nuanced than a binary, and more fragile than either, because you have to guess the answers in advance.
40 multiple choice survey question examples by job
Every example is written to produce a decision, not a chart. Options are separated by slashes.
Customer satisfaction and churn
Ask these right after a completed interaction, while detail is still recoverable.
- How would you rate your experience today? Very good / Good / Neither good nor poor / Poor / Very poor
- Will you still be using us in six months? Very likely / Likely / Not sure / Unlikely / Very unlikely
- What is the main reason you considered leaving? Price / Missing feature / Support quality / A competitor / I did not consider it
- How easy was it to get your issue resolved? Very easy / Easy / Neither / Difficult / Very difficult
- Which part do you use most? Scheduling / Forms / Payments / Reporting / Something else
- How did you first hear about us? Search / Colleague / Social media / Event / Elsewhere
Product and feature research
These feed a roadmap, so pitch the options at the level you make build decisions.
- Which would you use next month? (Select all) Reminders / Payment collection / Group polls / Routing rules / None
- What stops you using it more often? Too slow / Hard to find / Does not fit my process / I did not know it existed / Nothing
- How often do you use the product? Daily / A few times a week / Weekly / Monthly / Less than monthly
- Which single improvement would matter most? Speed / Reliability / More integrations / Better mobile experience / Lower price
- What would you use if this disappeared tomorrow? A competitor / A spreadsheet / Email and calendar / Nothing / Not sure
- Which describes your team size? Just me / 2 to 9 / 10 to 49 / 50 to 249 / 250 or more
Onboarding and first-run
Field these in the first week, before the memory of setup fades.
- How long did setup take? Under 5 minutes / 5 to 15 / 15 to 60 / Over an hour / Not finished
- Where did you get stuck first? Account setup / Calendar connection / First form / Inviting the team / Nowhere
- What did you want to do first? Take bookings / Send a survey / Collect payments / Route incoming leads / Something else
- How clear were the setup instructions? Very clear / Clear / Neither / Unclear / Very unclear
- What nearly stopped you signing up? Price / Data protection / A missing integration / Too many steps / Nothing
Sales discovery and lead qualification
Put these in the form ahead of the call so the call starts further along.
- What is your timeline for a decision? This month / This quarter / This year / No fixed timeline / Just researching
- Who else is involved in the decision? Only me / One colleague / A committee / Procurement / Not decided
- What are you using today? A competitor product / Spreadsheets / Email and calendar only / A custom internal tool / Nothing
- What triggered the search now? Team growth / A retired tool / A compliance requirement / New budget / A specific problem
- Which describes your role? Founder / Team lead / Individual contributor / Procurement or IT / Agency
- Which requirement is non-negotiable? EU data hosting / An integration / Single sign-on / A price ceiling / Nothing yet
Attach these to the booking page and the answers arrive with the meeting. Our guide to lead qualification covers scoring them, and the discovery call script template covers the live conversation.
Event and webinar feedback
Send within 24 hours. Recall of session detail drops fast.
- Which session was most useful? (List sessions by name, plus None)
- How was the event length? Much too short / Slightly short / About right / Slightly long / Much too long
- How likely are you to attend the next one? Very likely / Likely / Not sure / Unlikely / Very unlikely
- What would you change first? More practical content / Better speakers / Shorter sessions / More networking / Nothing
- How did you attend? In person / Live online / Recording afterwards / I registered but did not attend
Employee and internal surveys
Anonymity changes the answers, so state the rule above the first question.
- How manageable is your current workload? Very manageable / Manageable / Neither / Unmanageable / Very unmanageable
- How often do you have the information your job needs? Always / Usually / Sometimes / Rarely / Never
- Which would improve your week most? Fewer meetings / Clearer priorities / Better tools / More autonomy / More headcount
- How comfortable is raising a problem with your manager? Very comfortable / Comfortable / Neither / Uncomfortable / Very uncomfortable
- How well does team handover work? Very well / Well / Neither / Poorly / Very poorly
- Which have you hit in the last month? (Select all) Unclear ownership / Duplicate work / Blocked on someone / Rework after a reversed decision / None
Post-appointment client feedback
One screen, sent automatically when the appointment closes.
- Did the appointment start on time? Yes / Within 5 minutes / More than 5 minutes late / It was rescheduled / I do not remember
- How prepared was the person you met? Very well / Well / Neither / Poorly / Very poorly
- Was the outcome what you expected? Better / As expected / Worse / Different but fine / No clear outcome
- How would you prefer to book next time? Self-service link / Email / Phone / In person / I would not
- What is your next step? Ready to proceed / Internal discussion / Revised quote / More time / None
- Which channel suits you for follow-up? Email / Phone / WhatsApp / Video call / No follow-up needed
Why your option list is the measurement instrument
Three failure modes account for most of the damage, and all three are measured.
Options you leave out. A closed list can only collect answers you thought of in advance. The classic demonstration is the "most important problem" experiment: an open version produced far fewer crime answers than a closed version that listed crime and violence, because respondents had to generate the category themselves. Krosnick and Presser's chapter on question design collects the evidence. The fix is to pilot the question open-ended on a small sample, then close the list around the answers you get.
Options you word loosely. The label is the stimulus. Pew's guidance on writing survey questions documents a 2002 case where describing an option as the war on terrorism rather than as foreign policy moved the share selecting it from 34% to 52%.
Options in a fixed order. This is the primacy problem above, and the cheapest to fix: randomise any list whose options have no natural sequence. Ordered scales stay in order, but show the sequence reversed to half the sample so the bias distributes rather than accumulates.
Should you use single-select or multi-select?
Default to single-select. Multi-select looks generous and measures worse.
Pew's 2019 experiment comparing forced-choice and select-all formats found that asking each item as its own forced choice yields more accurate answers than one check-all list, and Pew now generally avoids select-all questions. The mechanism is ordinary satisficing: people work down a checklist until they feel they have done enough, then stop.
There is an analysis cost too. Every checkbox becomes its own variable, so an eight-option check-all lands in your export as eight columns, and any cross-tab has to be rebuilt by hand.
A working rule:
- Use single-select when the options are mutually exclusive, or you need one clean number per respondent for segmentation.
- Use multi-select when the behaviour genuinely is plural, such as which channels someone uses, and you can accept undercounting the lower half of the list.
- Split into separate forced choices when the undercount would change a decision.
How do you rewrite a bad multiple choice question?
Take a real one:
How satisfied are you with our excellent customer support? Very satisfied / Satisfied / Unsatisfied
Four faults, in order of damage. The stem is leading, telling the respondent the answer before asking. The scale is unbalanced, two positive options against one negative. There is no midpoint, so anyone neutral picks a side or drops out. And it assumes contact with support happened at all.
Rewritten:
Have you contacted our support team in the last three months? Yes / No
(if yes) How would you rate that support? Very good / Good / Neither good nor poor / Poor / Very poor
The screener removes the assumption, the scale is balanced with a genuine midpoint, and the adjective is gone. The trade is one extra click for a number you can defend.
How many answer options should a question have?
Four or five for anything attitudinal. Pew keeps answer choices to roughly four, or five at the outside, because people cannot hold much more than that in mind while deciding. Factual and demographic questions are the exception: a country or job-role list needs to be complete more than short, since an incomplete list there produces a wrong answer rather than a rough one.
Two additions to decide deliberately rather than by default:
- Other, please specify. Add it whenever the list may not be exhaustive, and read the free text before the next wave rather than after the study. It is the cheapest possible pilot.
- A midpoint or a no-opinion option. A midpoint belongs on any balanced scale. A separate "don't know" is different: it gives people who do hold a view an easy exit, so add it only where genuine ignorance is likely.
Where do these questions live in your form?
A multiple choice answer is a usable branch condition, which is why the format survives its own bias profile. One selection can decide which questions come next, or which calendar the respondent lands on.
meetergo runs forms, surveys and scheduling as one flow rather than as separate products, which matters when the answer needs to do something. Its survey tool supports multiple choice and single selection with a live results dashboard, anonymous responses, Excel or PDF export, no participant login and no response cap, free forever. Its routing forms carry 25+ field types and conditional logic with EQUALS, CONTAINS, BLANK and GREATER THAN operators, so a qualifying selection can show the right follow-up questions and send the respondent to the right calendar. Invitee questions attach the same items to a booking page.
Two limits are worth knowing before you plan around it. Funnels-based smart routing is a Premium feature at €49.90/month, one tier above the €29.90 Suite plan (checked 2026-08-31), which is more than a single feedback form needs. More importantly for this article, the survey tool does not currently expose option-order randomisation, so any study where primacy would change your conclusion belongs on a dedicated research platform.
Responses are held under EU law, on European servers, under a company with no US corporate parent. That matters as soon as an options list starts capturing something personal about the person choosing. The data privacy overview sets out how.
Build a multiple choice survey on meetergo's free plan
Booking + video conferencing in one tool.
Booking + video conferencing in one tool.
Frequently asked questions
Is a multiple choice question qualitative or quantitative?
Quantitative. Each option maps to a category you can count, which is the reason to close the list at all. The qualitative material lives in the "Other, please specify" box and any open-ended follow-up underneath.
What happens to your data when you allow multiple selections?
Each option becomes a separate variable, so an eight-option check-all arrives as eight yes/no columns. Plan the analysis before you ship the question.
Should you include an "Other, please specify" option?
Yes, whenever you are not certain the list is exhaustive. It is the cheapest way to find options you missed. Drop it only where the list genuinely is closed, such as a five-point rating.
Do multiple choice questions need a "don't know" option?
Only where real ignorance is plausible. On attitude questions it gives people who do hold a view an easy way out, which costs you data. A neutral midpoint on a balanced scale is a different thing and usually belongs.
How do you avoid leading answer options?
Balance the set: equal numbers of positive and negative choices with a genuine midpoint. Keep evaluative adjectives out of the stem, and check that no option is longer than the others, because length reads as endorsement.
Are multiple choice questions good for measuring satisfaction?
Yes, provided the scale is balanced and the wording is reused across waves. Comparability matters more than perfection, so once a scale is live, resist improving it. Our overview of client feedback surveys covers timing.
How many multiple choice questions should one survey have?
Fewer than you want. On a post-appointment or post-event form, five to eight is the realistic ceiling before completion rates fall. In a research study the limit comes from your method instead, and the survey method overview covers choosing one.
