The Survey Question Types Complete Guide: Everything You Need To Know

Survey question types – the format in which you ask a survey questions, such as multiple choice, rating scales, ranking, open-ended or demographic.

Every kind influences how folks answer, how simple the survey seems to fill out, and how valuable the insights turn out down the line.

When you know when and why to use each format, you can craft surveys that seem transparent to answer and provide accurate data to analyze.

Choosing the right survey question types is essential for collecting accurate and actionable feedback—but it doesn’t have to be complicated. With FORMEPIC, you can create surveys in seconds using AI-generated question suggestions, customizable templates, and intuitive logic. Start building your perfect survey with FORMEPIC today and get insights that truly matter. Try FORMEPIC for free

Key Takeaways

• Know that survey questions are of only two main types: closed-ended for rapid, organization-friendly information and open-ended for nuanced, in-depth feedback. You’ll generally need both for a well-rounded survey. Selecting a good combination enables you to capture both the “what” and the “why” of answers.
• Use closed-ended questions — multiple choice, yes or no, rating, Likert — when you need fast, comparable, and easy-to-analyze data. Make answer choices obvious, balanced, and non-leading. Don’t be biased or simplistic.
• Use open-ended questions when you want the depth, emotion, and context that numbers can’t capture, like to discover why, motivations, and ideas. Keep them few, one thing at a time, and clear to minimize vague responses and analysis paralysis.
• Align question types with your goal exploration, measurement, or segmentation, instead of selecting them arbitrarily. For instance, use rating and interval or ratio questions to track change over time and demographic or behavioral questions to segment various respondent groups.
• Design questions strategically by sequencing from general to specific, grouping related topics together, and controlling cognitive load with simple language and clean layouts. This keeps respondents on point, less fatigued, and your data sharper.
• Defend data quality by reducing bias and response error through neutral wording, anonymous responses where possible, and careful testing with a small pilot group. Utilize beta test feedback to iron out any unclear, confusing, or leading questions before your full launch.

Comparing Qualitative Open-Ended and Quantitative Closed-Ended questions

Survey questions fall into two broad families: closed-ended and open-ended. Closed-ended questions typically generate quantitative data that you can tally and contrast. Open-ended questions provide qualitative data that reveals the nuance, language, and context behind those numbers.

Family	Typical data type	What respondents do	Best for
Closed-ended	Quantitative	Pick from fixed options (e.g., 1–5, Yes/No)	Measuring patterns, benchmarking, fast decisions
Open-ended	Qualitative	Write answers in their own words	Exploring “why”, uncovering themes and new ideas

Both are legitimate. The correct balance varies based on your objectives, audience, and intended analysis.

Closed-Ended Questions

Closed-ended questions ask people to choose from predetermined options: a set of answers you define in advance. These are the backbone of most professional surveys because they scale well. You want thousands of responses and you want clear metrics.

Common types are multiple choice (single or multiple answer), dichotomous (yes/no, true/false), rating scales (1 to 5 satisfaction, 0 to 10 likelihood), ranking questions, and matrix or grid scales combining several ratings in one table.

The primary benefit is expediency. Respondents answer immediately, which reduces drop-off. You get organized info that software can summarize, chart, and filter with little human effort. This makes closed-ended questions perfect for monitoring KPIs, A/B testing, or comparing across markets and over time.

They further eliminate vagueness since we’re all choosing from the identical menu of response possibilities. However, they are limited. Reply’s seldom demonstrate depth, passion or nuanced trade-offs. If they are leading, unbalanced, or incomplete, you have bias built into the data from the get-go.

Even subtle rating scales reduce complex experiences to a number, making it less robust than open text in the eyes of some researchers. Badly designed closed options can shove them into options that do not really fit.

Open-Ended Questions

Open-ended questions offer the respondent a blank text box and encourage them to respond in their own words. They seek to capture thought, feeling, and context that do not easily fit into a checkbox, which is why they sit at the heart of qualitative research and exploratory surveys.

Well used, they uncover language your audience really uses, surprising pain points, and options you never would have thought to include as choices. The benefits are depth and subtlety. They can explain ‘why’ behind a rating, describe exceptions, or suggest new solutions.

A number of researchers view them as necessary to grasp complex behavior, help test new ideas, or design improved closed-ended items for subsequent surveys. However, the trade-offs are legit. Open text takes longer to answer, which can generate survey fatigue if you use too many, especially on mobile.

Responses can be vague or tangential, and analysis is slower and more qualitative, even if you use AI-assisted coding or text analytics. Different readers can interpret the same comment differently, so you need clear coding rules and coder alignment.

To make open questions work harder, keep prompts specific and tangible rather than vague and conceptual. Request one thing at a time, set expectations about length in one or two sentences, and put your most important open questions early, before respondents are fatigued. Use them sparingly, where you really need richer insight than a scale can provide.

A Deep Dive into Closed-Ended Survey Question Types

Closed-ended questions produce quantitative information by prompting respondents to select from predetermined responses. They’re quick to answer, simple to contrast, and are the heart of most significant survey research.

Type	Purpose	Typical Format / Example
Multiple-choice	Classify or segment respondents	“Which channel did you use? ☐ Email ☐ Social media ☐ Website”
Dichotomous (Yes/No)	Force a binary decision	“Have you purchased from us in the last 6 months? Yes / No”
Checklist / Multi-select	Capture multiple applicable attributes	“Which tools do you use? ☐ Zoom ☐ Teams ☐ Meet ☐ Slack”
Rating scale	Measure intensity on a scale	“Rate satisfaction: 1–5” or “0–10” or even “0–100”
Likert scale	Measure attitude agreement / frequency	“I trust this brand: Strongly disagree … Strongly agree”

Benefits are rapid response in seconds per item, higher completion rates, simpler coding, and obvious metrics for dashboards and stats. You can slice results by audience segment, time, or channel without wrangling text-mined answers.

Limitations are real. Closed options can oversimplify complex experiences, hide “why” behind the numbers, and introduce bias if options are incomplete or poorly worded. They may shove respondents into ill-fitting answers.

Designing good closed-ended questions uses specific language, eschews leading or loaded phrasing, makes answer options mutually exclusive and collectively exhaustive wherever possible, gives ‘Other, N/A’ when necessary, and maintains rating formats consistent throughout the survey to minimize confusion and noisy data.

1. Categorical Questions

Categorical questions cover nominal (no inherent order: country, industry, device type) and ordinal (ordered categories: “Beginner, Intermediate, Advanced”) structures.

They work well for segmentation. You can group responses by age band, region, or product tier and then compare behaviors, satisfaction, or intent. This makes them incredibly powerful in marketing, education, and product research.

Here’s the catch: It’s direct. For example, requesting income in large bands obscures holes within each band, and fitting a job role into a narrow list can misclassify individuals. Too many categories can overwhelm respondents.

Keep types in line with your goals, use labels your audience really uses, don’t have overlapping ranges like “1 to 10” and “10 to 20,” and add “Other (please specify)” when you know the list can’t be comprehensive.

2. Ordinal Questions

Ordinal questions order answers by a ranking, and the ‘distance’ between levels is not necessarily equal. Typical examples include satisfaction scales (“Very dissatisfied” to “Very satisfied”), preference rankings, and Likert-type agreement questions.

They allow you to survey breadth and depth of sentiment in a concise manner. They summarize easily with medians, percent positive, and trend graphs.

One individual’s “4 out of 5” may seem like a 3 to someone else, and scales can skew results if you inundate choices with positives or negatives. Handle the numbers sensitively and don’t make over-precise conclusions.

3. Interval Questions

Interval questions employ scales in which intervals are assumed to be equal. For example, a 0 to 10 likelihood-to-recommend scale or a 0 to 100 performance ratings scale enables more sophisticated statistics, such as means, standard deviations, and correlations, to compare groups and monitor nuanced changes.

They can baffle people if scale labels are ambiguous or inconsistent, or if you insert a midpoint that doesn’t fit the behavior. 0” must always be clearly defined. Overly granular scales, such as 0 to 100, can bog respondents down.

Clearly define your anchors at both ends, determine if a neutral middle point is appropriate, maintain the same scale range across related questions, and pilot your scales.

4. Ratio Questions

Ratio questions add a real zero so ‘0’ means none. For example, the number of purchases in the last month, hours per week, or revenue in a certain period.

They provide exact measurement, enable the entire suite of statistical analysis, and render meaningful phrases such as “twice as many” or “half as often.” They’re great for modeling behavior and segment value.

Respondents can struggle with requests for precise figures they don’t monitor (“How many minutes per day…?”) or when zero is fuzzy. Badly selected units or ranges can damage precision.

Level	True Zero	Example Question	Precision	Typical Analysis
Nominal	No	“Which region do you live in?”	Low	Counts, percentages
Ordinal	No	“Rank these features from 1–5”	Low–medium	Medians, non-parametric tests
Interval	No	“Rate satisfaction from 0–10”	Medium–high	Means, correlations, regressions
Ratio	Yes	“How many purchases did you make last month?”	Highest	Full parametric statistics, ratios

The Power of Open-Ended Survey Questions

Open-ended questions are important in survey research because they allow respondents to break out of predefined boxes and describe what they really think, do, or feel in their own words. They reveal nuance and context and edge cases that closed questions typically overlook.

Open-ended questions frequently expose concepts you didn’t even know to inquire about. This ability to gather rich, qualitative data is what makes them so powerful in understanding respondents’ true sentiments.

Uncovering “Why”

If you’re interested in the why behind behaviors, you need open-ended questions. They:

• Reveal motivations (“Why did you cancel your subscription?”)
• Uncover concealed obstacles. What almost prevented you from making your purchase?
• Capture exceptions and edge cases
• Give your language to repurpose for your messaging, training, or product copy.
• Deliver stories and examples that make the numbers meaningful.

Closed-ended questions still count here. They give you structure, speed, and statistical power. You can track trends, run comparisons, and visualize data quickly.

A closed item such as “How satisfied are you with delivery time?” on a 1 to 5 scale is easy to summarize and benchmark. An open follow-up (“What influenced your rating?”) explains the pattern behind the numbers.

Demographic questions layer on top. If you segment open-ended responses by role, region, or experience level, you find that the “why” is seldom consistent. A public school teacher and a corporate trainer might issue the same rating but for very different reasons.

Segmentation allows you to view that difference clearly, providing deeper insights into the motivations behind ratings.

Likert scale questions rest in between. They provide you a standardized scale of opinion (“strongly disagree” to “strongly agree”), which is ideal for dashboards and measuring change over time.

However, that common measure misses the personal nature of an open answer, which fosters expression, inventiveness, and distinctiveness of discussion.

Capturing Emotion

Open-ended, scaled (e.g., 1–10 satisfaction), multiple-choice, and ranking questions can all access emotion, but they access it differently. Open text captures raw voice by asking, “How did this experience feel?

Scales demonstrate intensity. Multiple-choice and ranking questions assist people in rapidly sorting what is most important.

When you capture emotion well, you get more insight into what really drives behavior. You hold respondents more engaged, and you have richer data for storytelling and decision-making.

A brief quote about frustration or delight can resonate internally far more persuasively than a bar chart alone.

There are trade-offs. Emotional questions can be biased if they are worded in a loaded way. They can also be misread across cultures, and free-text emotional answers are difficult to analyze at scale.

They require manual coding or AI-assisted classification, which can add complexity to the analysis process.

To craft emotional questions, use plain, neutral language. Don’t hint at the “correct” emotion, tie each question closely to your topic, and word prompts to elicit more than a one-word response.

For instance, “Describe a recent moment when…” rather than “How was it?” encourages a more detailed response.

Avoiding Bias

Bias in survey questions frequently results from leading phrasing. For example, “How incredible was…”. Double-barrelled items create confusion. An example is, “How satisfied are you with support and pricing?

Assumptions that not everyone shares also contribute to bias. These issues warp open and closed answers, and they can conceal genuine pain points.

It’s crucial that your language be neutral. Ask ‘How would you rate your experience with our support team?’ instead of ‘How helpful was our excellent support team?’

Continue with an open-ended ‘What is the primary reason for your rating?’ that doesn’t guide the response. This approach helps to gather more authentic feedback.

Always pilot your survey with a small, diverse group. Seek questions that perplex folks, steer them toward a particular answer, or produce many tangential open responses.

Those are cues that phrasing or format should be adjusted.

Revise relentlessly. Get rid of judgmental jargon. Break difficult questions down. Tighten open-ended prompts so they solicit detailed and relevant responses rather than general remarks.

Beyond the Basics: Strategic Question Design

Strategic question design is about aligning question types to well-defined objectives and being pragmatic about respondent attention span. All questions should serve a specific, well-defined research purpose or they likely do not belong.

Open-ended questions capture rich, spontaneous insight in respondents’ own words. They work well for exploring motivations, expectations, or the “why” behind behaviors. For example, “What is the main reason you chose this product?

The cost is greater mental labor, more time to complete, and answers that sometimes wander or echo the most vociferous or eloquent. To use them well, keep the prompt short, be very specific about what you want, and limit how many you include in a single survey.

Closed-ended questions (yes/no, single choice, limited lists) are quick to administer and simple to analyze. They assist you in measuring behaviors and opinions, like, “Have you bought from us in the last 3 months?

They lack nuance and can pigeonhole answerers. Use simple neutral wording and options that are mutually exclusive and collectively exhaustive, often with an ‘Other’ option when you can’t cover every realistic case.

Multiple choice questions are the workhorse of most surveys and provide explicit choices for swift judgments. For instance, ‘Which of the following channels did you use to reach support?’ with 4–5 thoughtfully selected options and ‘Other (please specify).’

Too many options overwhelm people, particularly in telephone surveys, and badly designed lists bring in confusion or bias. Avoid double-barreled options like “Fast and friendly support” and prevent overlap between choices. For instance, use non-overlapping ranges such as 0–4, 5–9, and 10–14 hours.

Rating scales (e.g., 1–5 or 0–10) help measure the intensity of opinions, satisfaction, or likelihood. For example, “On a scale from 0 to 10, how likely are you to recommend us to a colleague?

They enable sophisticated analysis across time or across groups. They can be hard to interpret if endpoints and labels are fuzzy. A ‘7’ can mean different things to different people.

Be sure to define anchors clearly (1 equals Very dissatisfied, 5 equals Very satisfied), keep scale lengths consistent throughout the survey, and avoid absolutes such as “always” or “never” unless you really mean them.

Good questions, of course, respect cognitive limits and minimize bias. Use short, common words rather than jargon, never use leading or loaded phrasing, such as “How helpful was our outstanding support staff?

Additionally, don’t use double-barreled questions, such as “How satisfied are you with our pricing and service?” Order effects matter. Earlier questions frame how people interpret later ones, so small wording or sequence changes can shift responses in meaningful ways.

For that reason, always pretest with a small, diverse sample. Watch where people stumble or misunderstand, then iterate until every question is unambiguous, essential, and easy to answer.

Matching Question Types to Your Goal

Selecting question types is a design choice, not a decoration. The type you utilize can reinforce your information or inconspicuously fracture your findings, so it must adhere to your investigation aim and the type of data you need.

Question type	Best for	Advantages	Limitations
Open-ended	Exploration, ideas, language mining	Rich, qualitative insight; reveals “why” and new themes	Hard to analyze at scale; higher respondent effort
Closed-ended (incl. MCQ)	Quantitative counts, quick feedback	Fast to answer; easy to compare and chart	Can constrain thinking; may miss important options
Multiple choice	Segmenting, decisions, preferences	Simple to analyze; mutually exclusive options possible	Risk of oversimplifying; options must be plausible and complete
Rating scale	Emotions, satisfaction, intensity	Quantifies attitudes; engaging for many respondents	Central tendency bias; scale labels can be interpreted differently
Demographic / firmographic	Segmentation, profiling	Essential background and grouping variables	Privacy concerns; can feel intrusive if not explained

For Exploration

For exploratory work, you rely on open-ended, closed-ended, multiple choice, rating scale, and demographic questions, generally in a combination. Open-ended items bring to light language, drivers, and surprise issues in the customer path. For example, “Tell us about the most difficult part of booking your flight.

Multiple-choice and simple closed questions then help you scale those themes, such as “Which of these steps was most confusing?” with mutually exclusive options. Each type has obvious advantages. Open-ended questions provide rich stories and context. Closed ended and multiple choice produce nice quantitative data that is easy to graph.

Rating scales measure feelings such as trust or frustration on a continuum. Demographic questions inject the ‘who’ behind the ‘what’ allowing you to notice if certain age groups or industries have different pain points. They have trade-offs. Open-ended responses are slower for people to answer and time-consuming to code.

Multiple choice can oversimplify if the options are too narrow or not believable. Rating scales can confound if anchors are ambiguous and they are susceptible to central tendency when respondents avoid extreme points. Demographic items can be privacy sensitive, particularly in relation to income or sensitive identity groups.

How to write exploratory questions well: Use simple language, avoid leading wording and focus on one topic per item. Make closed and multiple-choice options mutually exclusive when possible and add ‘Other (please specify)’ when you’re unsure that you know all the relevant answers.

Employ a thoughtful combination of question types to maintain interest and to balance qualitative insight with information you can summarize.

For Measurement

If you’re measuring attitude or performance over time, your workhorses are Likert scale, semantic differential, and numeric rating questions. Likert questions use agreement statements such as “Strongly disagree” to “Strongly agree.” Semantic differentials measure position between two poles, such as “Very difficult” and “Very easy.

Numeric ratings use numbers, often from 0 to 10, for rapid scoring. These formats are powerful because they provide numeric data, facilitate statistical analysis, and allow you to measure change over months or campaigns. A zero to ten recommendation score, for example, can be averaged, trended, and compared across segments.

Measurement questions are not without their flaws. They trap nuance into a handful of points on a scale, can encourage response patterns such as always selecting the middle, and can be biased if items or anchors nudge toward a ‘desirable’ answer. Understanding nuances (say, 6 versus 7) can be tricky without a referent internal rubric.

To design robust measurement items, employ explicit, specific language, maintain balanced scales with equal positive and negative options, and indicate the meaning of each point with descriptive labels. Don’t make double statements like ‘useful and easy’.

Pilot your scales with a small group to determine whether people interpret them consistently in your context.

For Segmentation

For segmentation, you depend on demographic, geographic, psychographic, and behavioral questions that allow you to group respondents in useful ways. Demographic and firmographic questions address elements such as age, position, business scale, or industry. Geographic questions capture country or region.

Psychographic questions map values or lifestyle, and behavioral questions emphasize actions like how often your customers purchase or use your product. The primary benefit is refined targeting. You can compare results between segments, target campaigns toward particular groups, and enhance precision by filtering out noise from heterogeneous populations.

For example, satisfaction of new buyers compared to veteran customers typically appears quite different. There are restrictions. A large number of segmentation questions can come across as intrusive, diminish completion rates, and divide your sample into such small groups that they are no longer statistically analysable.

Badly framed questions can bias or pigeonhole people into categories that don’t reflect their reality. Good segmentation questions are relevant, easy, and polite. Request only variables for which you will actually perform analysis, provide inclusive options as well as “Prefer not to say,” and briefly explain why you are collecting sensitive data when necessary.

When you match segmentation types to your goal—for example, tailoring a product roadmap to specific industries—you are left with data that actually informs decisions.

Choosing Your Survey Questions

Picking Your Survey Questions Identify your subject, audience and theory upfront, then select question types that can feasibly provide that insight. Each item should be pertinent, unbiased and single-topic so the data does not become ambiguous and noisy.

Ensure Clarity

Crystal questions decrease drop-offs and sloppy data. Use short, direct sentences: “How satisfied are you with our delivery time?” is better than “To what extent does our multi-channel fulfillment process meet your expectations?” No jargon, brand-in-house terms, or acronyms unless you’re confident all respondents have that knowledge.

If you have to use a technical term, provide a brief explanation in brackets. Once you’ve decided your survey questions, don’t ask ‘How satisfied are you with our price and quality?’ because respondents can feel one way about each. Break it into two separate questions.

Specify scales and context: say “In the last 30 days” or “On a scale from 1 (very dissatisfied) to 5 (very satisfied).” If you’re asking for rankings, present the entire list of choices so you don’t bias the results toward the small number you recalled to include.

At a structural level, remember the three basic types: single-select (one answer), multi-select (several answers), and open-ended (free text). Each is clear or confusing depending on wording. For instance, ‘Which channel did you first hear about us through?’ must be single-select, whereas ‘Which channels do you regularly use to follow us?’ ought to be multi-select.

Aspect	Advantages of clear questions	Limitations / trade‑offs
Respondent experience	Faster to read and answer; less frustration	Can feel blunt or oversimplified
Data quality	More consistent responses; easier comparisons	May miss subtle context or edge cases
Analysis and reporting	Cleaner datasets; fewer manual fixes	Sometimes needs follow‑up questions for nuance
Translation / localization	Easier to translate and adapt across regions	Ultra-simple wording may not capture cultural nuances

Maintain Neutrality

Various question types process neutrality in different ways. Open-ended questions solicit free-form text answers and are best when you don’t know the answer choices or want rich opinions. For instance, “Why did you pick our service?” Closed-ended questions, like single-select, multi-select, and rating scale questions, enable speedier completion and straightforward analysis.

Each kind has compromises. Rating scales, likert-type items (“Strongly disagree” to “Strongly agree”) and multiple-choice questions are fast, but the answer list can bias if it tilts positive or omits realistic options. Multiple-choice options can flatten nuance: respondents pick the “least wrong” answer.

To stay neutral, steer clear of wording that implies a “right” answer, such as “How much do you love our new feature?”. Use balanced language: “How satisfied or dissatisfied are you with the new feature?”. Provide complete, mutually exclusive alternatives and “Other (please specify)” whenever possible.

Keep each question focused closely on your objectives so you don’t stray into leading or off-topic territory that skews your data.

Test Everything

Once you have a draft, test every question type in context: open-ended, closed-ended, multiple-choice, Likert scale, demographic questions, and matrix/grid items. Open-ended questions provide flexibility and show language your audience really uses. They are much more difficult and slower to analyze at scale.

Closed-ended questions are simple to tabulate and graph. They run the risk of shoehorning people into categories that don’t exactly match their reality. Grids and matrix questions, where statements sit in rows and a common response scale sits in columns, can reduce survey length and appear sharp.

They work well for a small, focused set of items (“Rate the following features: speed, reliability, design”). Big matrices are exhausting, convoluted, and extra-tortuous on mobile devices; they increase abandon rates. Use them judiciously and make them brief.

In testing, watch for three things: clarity (“Did anyone ask what this meant?”), neutrality (“Did the wording push people?”), and relevance (“Does this actually move us closer to our goals?”). Run a small pilot, check completion time and item nonresponse, and tweak until every question justifies its presence in the survey.

Conclusion

Survey question types are not a totem run through as well. They are the foundation that forms the richness of insights you receive in return. Closed-ended questions provide structure, efficiency, and simpler analysis. Open-ended questions bring context, nuance, and the ‘why’ behind the numbers.

What really counts is the fit between your objective and your question construction. Defined goals, well-crafted language, and a mix of types typically beat hodge podge lists of random stuff. When in doubt, pilot your survey with a small group, polish and repeat.

With the proper blend of question types, your surveys graduate from rudimentary data gathering to significant insight. Better questions result in crisper decisions and eventually a more profound insight into your audience.

The right survey questions can make all the difference in understanding your audience. FORMEPIC helps you design, launch, and analyze surveys quickly, turning responses into actionable insights. Try FORMEPIC for free and create surveys that deliver real results for your business.

Frequently Asked Questions

What are the main types of survey questions?

There are two main families: closed-ended and open-ended questions. Closed-ended questions provide set answer options. Open-ended questions allow respondents to answer in their own words. Most effective surveys are a mix of the two.

When should I use closed-ended questions in a survey?

Use closed-ended questions when you need quantifiable, comparable data. They’re great for gauging satisfaction, ranking, and measuring trends. They make analysis quicker and more precise, particularly in conjunction with large samples.

What are examples of closed-ended question types?

Popular varieties are multiple choice, rating scales, Likert scales, ranking, and yes/no questions. Each will help you organize responses so you can measure the results. Selecting the appropriate one hinges on the level of detail and accuracy you require.

Why are open-ended questions important in surveys?

While open-ended questions expose context, motivations, and ideas you didn’t anticipate, they help illuminate the ‘why’ behind scores. They’re amazing for uncovering new insight and require more time to analyze.

How do I match question types to my survey goal?

First define your goal: measure, compare, or explore. Use closed-ended questions to gauge and benchmark. Use open-ended questions to discover the reasons, ideas, and language customers use. Match every question to a specific decision you need to make.

How many open-ended questions should a survey include?

Use them sparingly to avoid fatigue. Most brief surveys fare fine with between one and three open-ended questions. Save your most important one for near the end, behind important closed-ended questions.

What makes a survey question well-designed?

A good question is easy to understand, unbiased, and covers a single concept. Try to steer clear of leading words, double meanings, and assumptions. Test your questions with a few people first to check clarity and remove bias.