Home
Blog
DEI Survey Questions
How to Write Inclusive Survey Questions

How to Write Inclusive Survey Questions: A UK Best-Practice Guide

Q: Should survey demographic questions be optional?

Yes. Every demographic question should carry a genuine 'prefer not to say' and none should be mandatory. Forcing an answer to a question about disability, ethnicity or sexual orientation pushes people to abandon the survey or answer dishonestly, which harms both trust and data quality. Making the questions optional — and explaining why you are asking — is what keeps disclosure rates high enough for the data to be usable.

13 July 2026 by Mark Holt

How to write inclusive survey questions: a UK best-practice guide

To write inclusive survey questions, explain why you are asking and reassure people the survey is anonymous; use plain, neutral language that makes no assumptions; ask one thing at a time; make every demographic question optional with a genuine "prefer not to say"; offer comprehensive response options with a self-describe write-in where identity is involved; and base your categories on the UK's GSS and ONS harmonised standards rather than a bespoke list.

That is the checklist. The harder truth, and the part most guides skip, is that inclusive wording is a moving target — so the rest of this guide covers both the principles and how to keep a question set current once you have written it.

Divrsity Disclaimer
This article describes general good practice for writing inclusive survey questions in a UK context, and references how the Divrsity platform approaches each point. It summarises official guidance neutrally for practitioners; it is not legal advice. UK standards on sex, gender identity and ethnicity are under active review, so treat the specifics here as current at the time of writing and check the latest GSS guidance before you launch.

1. Explain why you are asking — and make anonymity real

People answer sensitive questions honestly only when they understand why the data is being collected and trust that it cannot be traced back to them. Say both, briefly, before the demographic section begins: what the information is for, how it will be used, and that responses are anonymous. Official UK guidance recommends offering this "question justification" close to the question itself so nobody has to guess at your motives.

The reassurance has to be true, not just stated. If people suspect an answer about disability or sexual orientation could identify them, disclosure collapses and your data becomes unreliable — which is why we treat anonymity as an engineering property rather than a promise: each response is tied only to a random identifier, with no cookies, no IP logging and no link to an email address. Inclusion and anonymity are the same problem viewed from two angles.

2. Use plain, neutral language that makes no assumptions

Inclusive wording starts with clarity. The UK Government Analysis Function's survey-design guidance is refreshingly concrete: use simple vocabulary, keep sentences to a maximum of about 25 words, avoid abbreviations, colloquialisms and even contractions (write "do not" rather than "don't" for non-native speakers), and aim for a reading age of around nine years old. A question a tired employee can read once and answer is a question far more people will actually complete.

Neutral also means assumption-free. Do not imply there is a "normal" answer, do not gender the wording ("he or she"), and do not bundle a value judgement into the phrasing. The goal is that every respondent, whoever they are, can read the question and feel it was written with them in mind — not adapted for them as an afterthought.

3. Ask one thing at a time

The most common technical flaw in a home-grown survey is the double-barrelled question — one that asks about two things at once, so a single answer cannot honestly cover both. "Do you feel respected and fairly paid?" is impossible to answer if you feel one but not the other. The UK guidance is blunt about this: if a question needs guidance notes to make sense, that is a sign it should be split into simpler components. One concept per question keeps the data clean and the respondent unconfused.

4. Make every demographic question optional

No demographic question should ever be mandatory, and every one should carry a genuine "prefer not to say". This is not just courtesy — it is what protects your response rate. Force an answer about religion, ethnicity or sexual orientation and a meaningful share of people will abandon the survey or answer untruthfully, and both outcomes corrupt the very data you are trying to gather. Optionality, counter-intuitively, is what gives you more usable data, not less.

5. Offer comprehensive, respectful options — with a self-describe write-in

Where a question touches identity, the response options must reflect the full range of people who might answer, not a shortlist the author happened to think of. For gender identity in particular, best practice is to let people self-describe rather than have an identity assigned to them: any question that offers no write-in risks erasing respondents whose identity was not anticipated. Include a non-binary option and an open "prefer to self-describe" field.

There is a balance to strike, though. Inclusivity does not mean an overwhelming wall of tick-boxes — official methodologists class a massive option list as "highly fatiguing", which itself suppresses responses. The art is a set of options that is genuinely comprehensive and scannable, with a write-in to catch the rest. Getting that balance right, question by question, is a large part of what a maintained DEI question set exists to do for you.

6. Base categories on UK harmonised standards, not a bespoke list

Wherever a standard exists, use it. The Government Statistical Service (GSS) and the Office for National Statistics maintain more than 40 harmonised standards — agreed question wordings and response options for sex, gender identity, ethnicity, religion, disability and more — specifically so that data can be compared across organisations. Base your questions about the protected characteristics on these, using the ONS ethnicity categories, and your results become comparable to national and sector benchmarks instead of a bespoke list only you can read.

A bespoke list is not just less useful — it is usually less inclusive, because the harmonised standards have been cognitively tested with real respondents in a way a question dashed off in an afternoon has not. Standing on that research is the single fastest way to make a question set both inclusive and credible.

7. Handle sex and gender identity with particular care

This is the area where UK guidance has changed most, and where getting it wrong causes the most offence. The direction of travel is to treat sex and gender identity as two distinct concepts rather than a single conflated question. The independent Sullivan Review, published in March 2025, recommended that government and other data owners collect data on both biological sex and on transgender and gender-diverse identities — rather than replacing one with the other — so the influence of each can be understood. The GSS is now developing updated harmonised standards for both.

For a workplace DEI survey, the practical implications are straightforward and neutral: only ask about sex and gender identity where you genuinely need the data; where you do, ask them as separate, clearly-worded questions rather than one ambiguous one; make both optional; and offer a self-describe option alongside "prefer not to say". Because this specific guidance is actively evolving, it is exactly the kind of question you should not write once and forget.

8. Keep the questions current — inclusive language does not stand still

This is the point every "how to write inclusive questions" article should end on but rarely does. The language of identity moves, and official UK standards move with it. In the space of a single year the Sullivan Review reshaped how sex and gender data should be collected, and the GSS ran a public consultation, from October 2025 to February 2026, on whether the ethnicity standard needs new response options. A question set that was model practice in 2023 can be quietly out of date today.

That is why inclusive wording is best treated as an ongoing maintenance task rather than a document you finish. It is also, candidly, the reason a maintained platform earns its place: Divrsity tracks this evolving guidance and updates its question templates continuously, so the set you send this year reflects this year's standards — something a static spreadsheet, however well written, can never do on its own.

A worked example: from exclusive to inclusive

Principles land better with an example. Here is a demographic question as it often appears in a home-grown survey, and the same question rewritten to the standard above:

Before: "Gender: ☐ Male ☐ Female (required)". This forces a binary, offers no self-description, makes the question mandatory, and conflates sex with gender identity.

After: two separate, optional questions. First, gender identity, with options that include a "prefer to self-describe" write-in and "prefer not to say". Then, only if you genuinely need it, a separate sex question following the current GSS harmonised wording, again optional. A short line above both explains why you are asking and confirms the survey is anonymous. Same intent, far more honest data — and nobody is erased or cornered.

Why inclusive wording is really about data quality

It is tempting to file inclusive questions under etiquette. In reality it is a measurement issue. Every forced binary, every missing option, every question with no "prefer not to say" pushes a slice of your workforce into non-response — and it is disproportionately your smaller, more marginalised groups who drop out, precisely the people a DEI survey exists to hear from. Inclusive wording is how you keep disclosure high enough that those groups remain visible in the data. It is the front end of measuring diversity accurately at all.

Frequently Asked Questions

How do you write inclusive survey questions?

Explain why you are asking and reassure people the survey is anonymous; use plain, neutral language that makes no assumptions; ask one thing at a time; make every demographic question optional with a genuine "prefer not to say"; offer comprehensive response options including a self-describe write-in where identity is involved; and base your categories on the UK's GSS and ONS harmonised standards rather than a bespoke list. Then review the wording regularly, because inclusive language changes over time.

What makes a survey question inclusive?

An inclusive question lets every respondent answer honestly and see themselves in the options, without feeling judged, cornered or erased. In practice that means neutral wording free of assumptions, response options that cover the full range of identities (with a write-in for anything unanticipated), a "prefer not to say" so no question is forced, and plain English readable at around a nine-year-old reading age. Inclusion is also about accuracy: a question people feel safe answering produces better data.

How should you ask about gender in a UK survey?

UK official guidance has moved toward treating sex and gender identity as two distinct concepts. The 2025 Sullivan Review recommended that data owners collect information on both biological sex and gender identity rather than replacing one with the other, and the Government Statistical Service is developing updated harmonised standards for both. For a workplace survey, ask sex and gender identity as separate questions only where you genuinely need both, keep them optional, and offer a self-describe option alongside "prefer not to say".

Should survey demographic questions be optional?

Yes. Every demographic question should carry a genuine "prefer not to say" and none should be mandatory. Forcing an answer to a question about disability, ethnicity or sexual orientation pushes people to abandon the survey or answer dishonestly, which harms both trust and data quality. Making the questions optional — and explaining why you are asking — is what keeps disclosure rates high enough for the data to be usable.

How do you ask about ethnicity inclusively in the UK?

Use the Office for National Statistics ethnicity categories that underpin the GSS harmonised standard, so your data is comparable with national and sector benchmarks rather than a list nobody else can interpret. Present options clearly, include a write-in for anything not listed, and add "prefer not to say". Note that the ethnicity standard is under active review — a GSS consultation on additional response options ran from October 2025 to February 2026 — so the recommended categories should be checked rather than assumed fixed.

Why do inclusive survey questions need updating?

Because the language of identity does not stand still. Terminology around gender, ethnicity, neurodiversity and disability shifts, new communities ask to be recognised, and official UK standards change — the 2025 Sullivan Review and the 2025-26 GSS ethnicity consultation are recent examples. A question set written three years ago and never revisited will quietly become dated and risk causing offence, so inclusive wording is a maintenance task, not a one-off.

Conclusion: inclusive is accurate — and accurate needs maintaining

Writing inclusive survey questions is not complicated in principle: be clear about why you are asking, use plain neutral language, ask one thing at a time, keep every demographic question optional, offer comprehensive options with a way to self-describe, and stand on the UK's harmonised standards rather than inventing your own. Do those things and more people answer, more honestly — which is the whole point.

The part that catches organisations out is that "inclusive" is not a state you reach once. The standards move, the language moves, and a question set has to move with them. That maintenance is exactly what Divrsity was built to take off your plate: a GSS-aligned, protected-characteristic question set, kept current as the guidance evolves, delivered through a genuinely anonymous survey and analysed automatically. If you would rather send battle-tested inclusive questions than write and re-write your own, that is the service in one sentence.