Demographic Questions in Surveys: Wording and Examples

Two questionnaires want the same information. One asks “How old are you?”, the other asks “What year were you born?”. The answers differ in how complete they are, how precise they are and how well you can analyse them later.

Demographic questions are the items about the person that appear in almost every questionnaire: age, sex, ethnic group, education, employment status, income and location. You collect them to compare answers between groups and to judge your sample, and you should word them the way the UK harmonised standards do rather than inventing your own. By the end you will have six questions you can copy, and a clear idea of which ones to leave out.

📌 The key points

• Demographic questions describe the person, not the topic.
• The UK harmonised standards supply the tested question wording.
• Ethnicity, religion and health are special category data.
• Demographics at the start do not reduce response rates (Teclaw 2012).
• Groups below five people make anonymity impossible to promise.

Here is what such a question looks like in a finished questionnaire, as a closed item with fixed categories.

What are demographic questions?

Demographic questions are items that place a respondent socially and demographically: age, sex, ethnic group, education, occupation, employment status, income, marital status and where they live. They describe the person holding the opinion rather than the opinion itself.

That is exactly where their value comes from. Without demographic items you have one average across everyone who took part and nothing else. With them you can ask whether students answer differently from people in work, whether age groups diverge, and whether your sample is composed anything like the population you want to say something about.

Survey methodologists therefore call these background characteristics, and they are the starting point for judging how representative a sample is. For a club survey that sounds excessive. For a dissertation it is the difference between an analysis and a column of numbers.

These six items cover most surveys:

• Age, collected as a single year or as a band.
• Sex, using the harmonised UK question.
• Ethnic group, using the country-specific harmonised list.
• Highest qualification, separate from current study.
• Employment status, including students and carers.
• Income, in bands rather than exact amounts.

Demographic or socio-demographic: where the line sits

Demographic characteristics describe population structure in the narrow sense: age, sex, marital status and location. Socio-demographic characteristics add social position on top: education, occupation, employment status and income. Every demographic item is therefore also socio-demographic, but not the other way round.

In practice the two terms are used almost interchangeably, and no binding standard draws the line. When a template says “demographic questions” it usually includes education and occupation anyway. For your own work the distinction matters far less than the question of which items you will actually analyse.

A third term turns up in market research reports and means something different again: psychographic characteristics, meaning values, interests and lifestyle. Those are not descriptions of the person in the sense used here, they are attitudes, and they belong in the substantive part of your questionnaire.

Socio-demographic describes who someone is. Psychographic describes how someone thinks.

The six demographic questions and how to word them

The reference point for wording in the UK is the set of harmonised standards maintained by the Government Statistical Service and published through the Government Analysis Function policy store. They exist so that answers from different surveys can be compared, which is precisely why borrowing them beats inventing categories.

Comparability is the practical argument. If you collect qualifications in bands you made up, you cannot compare your sample against any published figure afterwards, and for a dissertation that comparison is often the only evidence you have about sample quality. The wording below is our working version for short online questionnaires.

Asking about age: single years or bands?

Single year of age is the preferred form, and the harmonised standard says so directly: use single years where data quality allows and fall back on groupings only when it does not. The reason is arithmetic. From single years you can build any band afterwards, while a band never gives you the years back.

Data protection pulls the other way in small surveys. A year of birth that appears only once identifies someone in a team of twenty immediately. A useful rule of thumb: once a group contains fewer than five people, a promise of anonymity no longer holds. In that situation, ask in broad bands from the outset.

Asking about ethnic group in the UK

Ethnic group is the item where a home-made list does the most damage. The Government Statistical Service maintains a harmonised ethnic group question, and the format differs slightly between England, Northern Ireland, Scotland and Wales, so the country-specific version is the one to use where you can.

Two properties of this item are easy to miss. Ethnic group is a self-identification measure, which means the respondent answers it for themselves and nobody answers it on their behalf. And under the UK GDPR it is special category data, so it needs an explicit consent tick box of its own rather than a place in a general list.

Asking about sex and gender identity

Sex and gender identity are two different measures and belong in two different questions. The Government Analysis Function publishes separate harmonisation guidance for sex, and the sex question used in the three UK censuses is harmonised across the UK even where the accompanying guidance differed.

What follows for a short questionnaire is simple. Ask only what your analysis needs, and if that is sex, use the harmonised item and leave gender identity out. Where you do ask about gender identity, offer a prefer-not-to-say option, and remember that in a survey of 40 people a single distinctive answer is effectively identifying.

Asking about income: bands, not exact amounts

The income question cannot be analysed without three decisions: gross or net, individual or household, weekly, monthly or annual. Leave one out and respondents answer different questions, which makes the average meaningless. A workable wording is: “What is your household's total income after tax, per month?”

Bands are the right answer format because almost no analysis needs the exact figure. Six to eight bands are enough, with the top one left open-ended. A prefer-not-to-say option belongs there too, otherwise you lose people who simply do not want to answer.

Where do demographic questions go in a questionnaire?

Demographic questions sit at the end in most questionnaires, and that is a sensible default. It is not, however, a rule supported by evidence.

The common claim is that demographics at the start put people off. Robert Teclaw, Mark C. Price and Katerine Osatuke tested this experimentally in the Journal of Business and Psychology in 2012: in an online employee survey, one group received the demographic items first and the other received them last. The result was that placing them at the beginning raised the response rate for those items, with no effect on response rates for the remaining questions or on their mean scores.

What follows is not a new blanket rule but a decision based on purpose. Three cases cover almost everything.

1. You need to screen or set quotas. Then the items go first, otherwise people outside your target group fill in the whole questionnaire for nothing.
2. Your topic is sensitive. Then an easy opening helps, and the personal items go at the end.
3. You only need a handful of items. Then position barely matters and you decide by reading flow.

⚠️ Watch out

Push every personal item to the end and you lose all of them whenever someone drops out. The characteristics you would need in order to judge who dropped out are exactly the ones missing. In long questionnaires it pays to move two or three key items to the front and leave the rest at the back.

Which demographic questions the UK GDPR restricts

Racial or ethnic origin, religion, health, political opinions, trade union membership and sexual orientation are special category data under Article 9 of the UK GDPR. Processing them is prohibited unless a specific condition applies, and in a survey the only condition that realistically applies is explicit consent, which needs a tick box of its own.

Article 9 is where many lists of demographic questions go wrong, including an earlier version of this article: religion and sexual orientation simply sat between age and marital status. Legally they are not the same kind of question, and the difference can cost you the whole data collection.

The second yardstick is data minimisation. Personal data must be adequate, relevant and limited to what is necessary, which gives you a very practical test: drop any item you cannot attach a planned analysis to. Asking for a full date of birth when you will report age bands is exactly the case the principle is aimed at.

Then there is the combination trap. Age, sex and department are harmless on their own, but together they usually point at one person in a team of twelve. If you promise anonymity, keep demographic items coarse and drop one rather than adding one. The detail is in the piece on anonymous surveys, and the lawful bases plus a privacy notice template are in the piece on survey data protection.

Where you are unsure, the position of the Information Commissioner's Office or your university's research governance rules is what counts. This article is not legal advice.

Common mistakes in demographic questions

The four mistakes below turn up again and again in questionnaires written by students, clubs and small teams. All four only become visible during analysis, by which point it is too late.

What they have in common is that none of them comes from carelessness. They come from writing the question before deciding what the analysis will look like.

A required field with no way out

Forcing an answer to a personal question produces two reactions and both hurt: some people abandon the survey, others tick anything. The second group is the more dangerous one, because you cannot tell their answers apart from genuine ones. Every personal item therefore needs a prefer-not-to-say option, even when the question is set as required.

Survey tools such as Google Forms, SurveyMonkey and empirio.ai all let you make individual questions mandatory. The opt-out does not appear by itself, though: you have to add it as a response option.

Age bands that overlap or leave gaps

“18 to 25, 25 to 35, 35 to 45” is the classic: anyone who is 25 can tick twice. Gaps are just as common, typically when nothing follows “up to 65”. Both come from typing quickly and both take two minutes to fix if you read the boundaries out loud once.

Answer lists that do not fit the audience

An employment status list without “in full-time education” is useless for a student survey. The same goes for qualification lists with no option for people still studying, and for income bands that start at a level your respondents never reach. Check every list against the people who will actually receive the link.

Home-made categories instead of harmonised ones

Invent your own qualification or ethnicity categories and you cannot compare your sample against any published statistic afterwards. That removes the simplest statement you can make about sample quality, and it is exactly what markers ask about. Copying the harmonised standards takes five minutes and avoids the problem entirely.

Conclusion

Demographic questions are the part of a questionnaire where creativity pays off least. Take the harmonised categories, collect age as a single year, give every personal item a prefer-not-to-say option, and drop anything you have not planned an analysis for.

Every personal detail you collect is one you then have to protect.

Where to go next

• Building the questionnaire right now? How to create a questionnaire, step by step
• Unsure which question types to use? Question types and answer formats explained
• Wondering what level of measurement age has? Levels of measurement, explained simply

Want to build a questionnaire without retyping the standard items every time?

empirio.ai, an online survey tool from Germany, processes responses in the EU and lets you set required fields and opt-out options question by question.

Create a survey for free