Questions to Ask in a Diversity Survey
These are questions to ask in a diversity survey of staff, students or members, written for the person or committee who has to draft the form. They follow the order a finished survey usually takes: belonging and respect first, then fairness in pay, promotion and assignments, bias and how reporting works, accommodations and access, the optional demographic questions, and open comments to close. Most are statements to rate on a five-point agreement scale, and a good survey uses a selection of them, so keep the ones a poor score would make you act on.
Want questions from the whole vault instead? Try the random question generator.
The questions
Each question, and why to ask it
Belonging
I feel a sense of belonging at this organization.
Why ask it
Score it on a five-point agreement scale, from strongly disagree to strongly agree. The overall figure says little on its own: the finding is the distance between groups, so plan to compare it across the demographic answers at the end of the form. Change the wording next year and that comparison is gone, so settle now on 'organization', 'school' or the name itself.
I am treated with respect by the people I work with.
Why ask it
This one is about the respondent's own treatment, so read it group by group: a strong score overall with one group sitting well below it is the result to take to leaders. For a school or a membership body, swap 'work with' for 'study with' or 'volunteer with'.
I can be myself here without worrying about how I will be judged.
Why ask it
Disagreement usually means someone is putting effort into hiding something, which might be an accent, a partner, a faith or a health condition. The score cannot say which. Place an optional comment box under it asking what people hold back, and expect short, careful answers.
This organization is committed to diversity and inclusion.
Why ask it
Most ready-made diversity surveys carry a line like this, and what it mostly measures is whether the message has been heard. Pair it with the later statement about what senior leaders do. When agreement here runs well ahead of agreement there, people know the commitment exists and have not yet seen it acted on.
People from all backgrounds have a fair chance to succeed here.
Why ask it
This asks what people observe around them, so everyone can answer it, including those who have never felt held back. Set it beside the personal items in the fairness section. When a group says yes to this and no to its own pay or promotion, give more weight to the personal answer.
My opinions are taken seriously, even when they differ from the majority view.
Why ask it
Cut this by job level before you cut it by anything else. If junior staff disagree and senior staff agree, you are looking at hierarchy. A split between groups at the same level is the diversity finding.
My manager makes sure everyone on the team is included.
Why ask it
Of everything in this group, a manager statement is the one a single person can change this week. That also makes it personal: roll any team below your minimum group size up into its department, and hand each manager their own score privately, beside the organization's, instead of putting a ranking on a slide.
Senior leaders show through their actions that inclusion matters here.
Why ask it
'Through their actions' is there so that nobody scores the speeches. If agreement is low, ask a focus group which actions would count, and bring leaders that list in place of the number.
I see people like me in senior roles here.
Why ask it
'People like me' is left for each respondent to define, on purpose. Expect low agreement from any group that is missing from the top, and read that as a fact about the org chart that no amount of messaging will move. A school can ask students the same thing about teachers and staff.
How often do you feel left out of conversations, meetings or social events here?
Why ask it
Use a frequency scale for this one: never, rarely, sometimes, often, always. It catches people who agree that they belong in general and are still missing from the lunch table. Follow it with a checkbox asking which of the three it was, since a meeting invite list and a social calendar are fixed by different people.
I hear jokes or comments here that put down a group of people.
Why ask it
Agreement is the bad result on this one, so flag it as reverse-scored before anyone averages the section. Because it asks what people hear, not whether they were the target, it tends to get franker answers than a direct question about harassment. High agreement at one location is a reason to visit, not to send a memo.
Fairness
I am paid fairly compared with others doing similar work here.
Why ask it
This measures whether pay feels fair, which a pay audit cannot tell you, just as this cannot tell you what the audit would find. Watch the neutral middle: when people have no way to see pay ranges, many choose it, and a tall middle bar is a finding about transparency. What an employer may or must disclose about pay differs by place, so ask before promising anything in the follow-up.
Promotion decisions here are made fairly.
Why ask it
Timing moves this score: asked the week after promotions are announced, it measures that round. Where one group scores low, pull the promotion records for that group at that level before drawing a conclusion. The perception may be ahead of the records or behind them, and you need both.
I know what I would need to do to be promoted.
Why ask it
This is about clarity, and it belongs directly after the fairness item so the two can be read together. Low on both usually points to criteria that live in managers' heads. Writing them down is among the cheaper fixes this survey can lead to.
The assignments that build a career are shared fairly on my team.
Why ask it
Nobody signs off on who gets the big client, the visible project or the advanced class to teach, which is why this kind of unfairness goes unrecorded. Give an example in parentheses so people know what you mean by a career-building assignment. A low score is a prompt for each manager to list who got the last five.
I am asked to take notes, organize events or show new hires around more often than my colleagues are.
Why ask it
Reverse-scored: agreement is the problem. Compare by gender and by race or ethnicity where the groups are big enough to report. If a gap shows, the next question for HR is whether that work is counted anywhere in reviews.
My work is judged by the same standards as my colleagues' work.
Why ask it
Ask it close to review season, while people remember what was said to them. If a group disagrees, have someone compare the written reviews across groups for length and for comments about personality, with names removed. That reading tells you whether the fix is in the form, the training or one manager.
I have the same access to mentoring and career advice as others at my level.
Why ask it
Mentoring and sponsorship are different things, and one statement cannot measure both: a mentor gives advice, a sponsor puts a name forward when its owner is not in the room. If sponsorship is what you suspect is uneven, add 'Someone senior here would recommend me for an opportunity' as a second item. In a school survey for students, 'a teacher or adviser who knows me well' does the same job.
We hire people from a wide range of backgrounds.
Why ask it
Staff see who arrives, not who applied or who was turned down, so this records an impression. Add a 'do not know' option, because people with no part in hiring otherwise park themselves in the middle. Read the score next to your actual hiring figures, and expect long-serving staff and recent joiners to answer differently.
Bias and reporting
In the past 12 months, have you personally experienced discrimination or harassment here?
Why ask it
Offer yes, no, not sure and prefer not to say. The time frame keeps old incidents out of this year's figure, and 'not sure' is an honest answer that deserves a box. Print where to get help directly above the question, because an anonymous form means nobody will be following up with the person who checks yes.
In the past 12 months, have you seen someone else here treated unfairly because of who they are?
Why ask it
Witnesses give you a second reading that does not rely on the people affected being willing to say so. Report it separately from the question before. The two counts cannot be added together, since six witnesses may be describing one afternoon.
What was the unfair treatment you experienced or saw related to? Select all that apply.
Why ask it
Show it only to people who answered yes to either of the two questions before it. Build the option list from the characteristics your own policy names, such as age, disability, gender, race, religion and sexual orientation, and ask your HR or legal contact which ones the law covers where you operate. Include 'something else' with a text box: it is where class, accent and job level turn up.
Was the person responsible a manager, a colleague, a customer or member of the public, or someone else?
Why ask it
Roles only, never names, and say so in the help text under the question. The pattern sends the response in different directions: a customer problem calls for backing up front-line staff, a manager problem for a route that goes around the line manager. In a school, the list becomes staff, students, parents and visitors.
Did you report what happened?
Why ask it
Count the people who answer no and set that beside the number of formal complaints received over the same period. The difference is a rough size for what the complaint log never sees. Keep the options to yes, no and prefer not to say.
If you did not report it, what stopped you? Select all that apply.
Why ask it
Each option should lead to a different fix. 'I did not know how' is a communications job, 'I did not think anything would be done' is about track record, 'I feared it would count against me' is about protection, and 'the person was my manager' is about who the route runs through. Leave room for 'I dealt with it myself', which is a reasonable answer.
If you did report it, how satisfied were you with how it was handled?
Why ask it
Very few people will reach this question, so it rarely makes a chart and should not be forced into one. Send the answers and any comments to whoever owns the complaints process, in a form that cannot identify the person. One detailed 'very dissatisfied' deserves more of that owner's time than the section average.
I know how to report discrimination or harassment if I ever needed to.
Why ask it
A knowledge item, and the quickest number on the form to move: a low score answers to orientation, posters and a link that works. Ask it of everyone, not only those with something to report. If people in their first year score lowest, the gap is in onboarding.
I could raise a concern about how I was treated without it being held against me.
Why ask it
Read it against the item before. People who know the route and do not trust it are in a worse position than people who simply have not been told. A break by job level shows where the fear sits, and it is seldom spread evenly.
When someone here raises a concern about bias, I trust it will be taken seriously.
Why ask it
People answer this from the one case they heard about in the hallway. If trust is low, find out what the organization is allowed to say about outcomes without breaking confidentiality; that limit differs by employer and by place, so put the question to HR or legal. Even a plain count of concerns received and concerns acted on gives people something better than rumor to answer from.
If I saw someone here being treated unfairly, I would feel able to speak up.
Why ask it
This one is about bystanders, and speaking up is a skill that can be taught, which makes it a fair test of whether training did anything. Run it before the training and again after. Low agreement among people who answered yes to the witnessing question means they saw something and stayed quiet.
Access
I have the adjustments or accommodations I need to do my work well.
Why ask it
Add 'I do not need any' as a separate answer. Without it, everyone who needs nothing agrees, and the handful who are going without disappear into a good average. Use whichever term your own policy uses, since 'accommodation' and 'adjustment' are each the normal word in different countries.
I know who to ask if I need an accommodation.
Why ask it
Plenty of people who have never needed one will say no, and that matters on the day something changes for them. A low score is fixed with a named contact in onboarding and on the intranet. For a school, ask families as well as students, since the request often comes from a parent.
If you have asked for an accommodation here, how easy was it to get?
Why ask it
Offer a scale from very easy to very difficult, plus 'I have not asked'. Do not ask what the accommodation was for: that is health information the survey has no use for. A comment box asking what slowed things down will usually name a step, such as a form, an approval or a wait for equipment.
The buildings, software and documents I need here are accessible to me.
Why ask it
Follow it with a checkbox for which of the three fell short and a line for the specific thing. The answers tend to be concrete: a heavy door, a video with no captions, a form a screen reader cannot get through. Pass each named item to whoever owns it, with a date.
I can observe my religious or cultural practices without it causing difficulty here.
Why ask it
It covers prayer times, dress, food at events, fasting and holidays that are not on the official calendar. Ask in a comment box which of those came up. Then check next year's calendar of all-staff events and exams against the dates people name.
Meetings and events here are scheduled and run in a way that lets me take part.
Why ask it
The people this reaches are caregivers who cannot do evenings, part-time and shift staff, remote workers, and anyone for whom an evening built around drinking is not an option. Add a short list of what got in the way: time of day, location, cost, format, food or drink. Cut the result by work schedule as well as by demographic group.
The forms and communications I get from this organization use language that includes me.
Why ask it
Think of a form with two gender boxes, a letter addressed to 'mother and father', or a system that cannot hold a name with an accent or a hyphen. Ask people to name the document. Each one named is a correction that can usually be made within the month and reported back as done.
Demographics
Which of these best describes your gender?
Why ask it
Offer man, non-binary, woman, 'prefer to self-describe' with a text box, and 'prefer not to say'. It is optional, like every question in this group. Have an employee network or student group read the options before launch: they know which terms are in use where you are and whether last year's list still reads right.
Do you identify as transgender?
Why ask it
It is its own question because gender and being trans are two different things: a list that puts 'transgender' beside 'woman' and 'man' forces a trans woman to pick one. The group will often fall below your reporting threshold, so decide in advance whether you can use the answer at all. Whether you may collect it is a question for your legal or privacy contact, country by country.
Which racial or ethnic groups do you identify with? Select all that apply.
Why ask it
Where your country's statistics office publishes ethnic categories, start from those, so that your figures can be set against the local population, and add a self-describe line. Some countries restrict or forbid collecting this, so confirm the position for every place the survey will go. Decide before launch how people who select several groups will be counted, and say so in the report.
Which age band are you in?
Why ask it
Bands, not birth years, and wider bands in a small organization so that no band holds only two people. Age gets left off diversity surveys more often than it should, and the promotion and respect items are the first ones worth cutting by it. Line the bands up with the ones your HR or enrollment records use, so the people who answered can be checked against the people who were invited.
Do you consider yourself to have a disability or long-term health condition?
Why ask it
'Consider yourself' lets people answer who have never told their employer, so this figure and the HR record will often disagree, and the gap is worth knowing. Legal definitions of disability vary by country: ask which definition your policy follows and whether to name neurodivergence and mental health conditions in the wording. Yes, no and prefer not to say are enough for comparing scores, and asking which condition adds health data this survey does not need.
Which of these best describes your sexual orientation?
Why ask it
Typical options are heterosexual or straight, gay or lesbian, bisexual, a self-describe line and prefer not to say. In some places asking this is unlawful or could put a respondent at risk, so clear it for each country before it goes out. State beside it who will see the raw data, and keep it the easiest question on the form to skip.
What is your religion or belief, if any?
Why ask it
Where the national census asks about religion, its list is a sound starting point; where it does not, build the list with a chaplain, a faith network or the groups themselves. Give 'no religion' a full place in it and not a spot at the bottom. Include the question only if an answer could change something, such as prayer space, catering or the holiday calendar.
Do you have caregiving responsibilities for a child or an adult?
Why ask it
Options: none, children, an adult, both, prefer not to say. HR systems seldom hold this, so the survey may be your only source. It is the cut that makes sense of the answers on meeting times, assignments and promotion.
Were you in the first generation of your family to attend college or university?
Why ask it
A short stand-in for social class, which most forms skip and which tends to surface under 'something else' in the bias questions. Give yes, no, do not know and prefer not to say. Where few of the jobs need a degree, a question about the work a parent or guardian did may fit better, so try both wordings on your review group.
How long have you been part of this organization?
Why ask it
The first-year group is the one to read closely, since those people can still compare what they were told at interview or enrollment with what they found. Three or four bands are plenty, and fewer in a small organization: each extra band is one more way for two answers together to describe a single person. Keep this out of any chart that is also split by department.
Which department, site or school are you part of?
Why ask it
Without it, nothing on this form can show whether a low score on respect or on a manager sits in one place. Build the list from units big enough to report on their own and fold the rest into the next level up before launch, not after the answers are in. Leave the question off if most units would fall under your minimum group size.
Which of these best describes your position?
Why ask it
For employees: individual contributor, manager of people, senior leader. A school would list teaching staff, support staff, administrators and students; a membership body would list members, volunteers and board. Show leaders the distance between their own answers and everyone else's before you show them anything else.
Closing
What is one thing we could do to make this a more inclusive place?
Why ask it
'One thing' makes people choose, which gives you answers you can sort into themes and count. Report the five most common themes with their counts, in the respondents' kind of wording and not HR's. If the same small fix appears a dozen times, do it before the full report is out.
What does this organization already do that helps you feel included?
Why ask it
The answers show what to protect when budgets or leaders change, and they are far more specific than the scores. Managers and colleagues get named in these. Pass on the praise only where the writer could not be worked out from the story; if in doubt, pass on the behavior and leave out the incident.
If you are willing, describe a time you felt excluded or unfairly treated here.
Why ask it
Put a line above the box asking people not to name anyone, and say plainly that this is not a way to make a complaint, with the route that is. Settle before launch who reads these and what happens when one describes something serious. What an organization must do with an anonymous account of harassment depends on where it is, so get that answer from HR or legal first.
Is there anything about who you are that this survey gave you no way to tell us?
Why ask it
The answers rewrite next year's option lists: a missing ethnicity, a faith not listed, a kind of family or job the form did not imagine. Read them before you finalize the report, because a group that could not identify itself is also missing from every comparison in it.
How confident are you that your answers to this survey cannot be traced back to you?
Why ask it
Place it last, on a scale from not at all confident to very confident. Where confidence is low, treat every other score from that group as the cautious version. It also tells you what to explain better next time: who hosts the survey, who sees raw responses and the smallest group you will report.
How to write a diversity survey people will answer honestly
Practical guidance for the conversation itself
Before you write it
Start from what a result could change
For every item you are considering, name the decision a bad score would lead to: a pay review, a change to how assignments are handed out, a new route for complaints. An item with no decision behind it asks people to disclose something for nothing. The test matters most for the demographic questions, where each one you add also makes small groups easier to pick out.
Anonymous or confidential: pick one and say which
An anonymous survey cannot be linked to a person by anyone, including you. A confidential one can be linked, usually so that answers can be matched to HR or student records, and it rests on a promise about who gets to see the link. Either can work, but calling a confidential survey anonymous is a quick way to lose the next round's respondents. Check the survey tool's settings too: personal invitation links, IP logging and timestamps can each undo an anonymity promise, so ask the vendor or your IT team what is stored.
Set the smallest group you will report
Before any data comes in, choose a number below which a group's results are not shown, and write it into the survey's introduction. Five and ten are both common choices; the right figure for you depends on your size and on what your privacy contact advises. Watch for subtraction as well: if you publish the total and every group but one, anyone can work out the one you hid.
Find out what you may collect
Information about race, health, religion and sexual orientation is specially protected in many places, and some countries limit whether an employer or a school may ask at all. The rules differ by country and sometimes by state or sector, and nothing on this page can tell you which apply to you. Take the draft to whoever handles legal and privacy questions for your organization before it goes out, and again for each country if the survey crosses borders. If students under 18 will answer, ask what consent from parents is needed as well.
Let the people you are asking about read the draft
Ask an employee network, a student council or a few members from different backgrounds to go through the wording. They will catch the option list that leaves them out and the phrase that sounds fine to the committee and wrong to everyone else. It also means some respondents already know why the survey exists when it lands in their inbox.
Wording, scales and order
Keep one agreement scale all the way through
The statements on this page are written for five labeled points: strongly disagree, disagree, neither agree nor disagree, agree, strongly agree. Label every point, keep the direction the same on every screen, and do not switch to a longer scale halfway. Two items here, the one about jokes and the one about note-taking, run the other way, so that agreement is the bad result. Mark them in your analysis sheet before the first export.
Give 'does not apply' its own box
The neutral midpoint should mean 'I have a view and it is in the middle'. If people who have never asked for an accommodation, never sat on a hiring panel or never reported anything have nowhere else to go, they land there, and the middle stops meaning anything. Add 'not applicable' or 'do not know' wherever a respondent might honestly have no experience to answer from.
Wording the demographic questions
Every one is optional and every one offers 'prefer not to say'. Offer a self-describe line wherever identity is involved, allow more than one selection for race and ethnicity, and list options alphabetically so the form does not appear to rank them, with 'prefer to self-describe' and 'prefer not to say' at the end. Put one sentence above the group saying why you are asking and who will see the answers. Ask only for what you will compare results by, since each extra question costs some trust and makes individuals easier to identify.
Easy statements first, identity near the end
Open with belonging and respect, which anyone can answer, then move to fairness, and to bias and reporting once people have settled in. Demographics come near the end with their explanation, and open comments last, when respondents know what you did and did not ask. Use skip logic for the follow-ups in the bias group, so nobody is shown a question about an incident they never had.
Send a selection, not the whole list
Pick the items that pass your decision test, keep at least one from each group, and aim for a form that takes about ten minutes. Say the expected time on the first screen and keep to it. A shorter survey that people finish gives you better comparisons between groups than a thorough one that half of them abandon at the demographic section.
Reading and reporting the results
An average hides the people the survey is for
An organization where most people belong to one group will get a comfortable average on almost every item, because the average is mostly that group. The useful view is each item broken down by the demographic answers, with the widest gaps at the top. Report a gap in plain terms: this share of one group agreed, against this share of another.
Check perceptions against records
The survey says how pay, promotion and complaints feel. Your payroll, promotion lists and complaint log say what happened. Put them side by side for the same groups. Where both point the same way you have a case for action, and where they disagree you have learned that either the records or the communication needs work.
Handle comments as sensitive data
Open answers contain names, incidents and turns of phrase that colleagues would recognize. Limit who reads the raw text, remove names before anything is shared, and never quote a comment that only one person could have written. Sorting comments into themes and reporting the themes with counts gives leaders the substance without the fingerprints.
Publish what you found and what you will not do
Tell respondents what you found within a few weeks, and lead with the widest gap between groups, not the best average. Name two or three changes, who owns each and when you will report progress. If something people asked for is not going to happen, say so and say why. Then run the same core items again on a fixed schedule so that any change shows up in the numbers.
Where diversity surveys go wrong
Making identity questions compulsory
A required question about sexual orientation, disability or religion invites abandoned surveys and made-up answers, and depending on where you are it may not be allowed. An honest 'prefer not to say' is better data than a forced guess.
Slicing until a group is one person
Gender by department by level sounds like thorough analysis until a cell contains the one woman director in engineering. Apply the minimum group size to every combination, not only to single cuts, and merge categories or drop the cut when a cell falls under it.
Letting the survey stand in for a complaints route
People will use the comment boxes to describe things that should have been reported. An anonymous survey cannot investigate, protect anyone or reply. Say that on the form, print the real route beside the bias questions, and agree in advance what the readers of comments do when one alarms them.
Borrowing another country's categories
Template surveys found online are mostly written for one country, and their race, ethnicity and disability lists carry that country's census and law with them. Sent somewhere else, they leave people with no box to check. Start from the categories your own country's statistics office uses, where it has them, and from your own policy, then have local staff or members check the list.