What Questions Should Be on a Diversity & Inclusion Survey?
A diversity and inclusion survey should measure two distinct constructs: belongingness (do people feel they fit and are accepted?) and uniqueness (are their distinct perspectives valued?). The validated Work Group Inclusion Scale (WGIS, Chung et al. 2020, α = .94) covers both components and links directly to engagement, job satisfaction, and reduced turnover intention (STAT-016). Because DEI results are segmented by demographics, the Rule of 5 — suppressing any demographic group below n ≥ 5 respondents — is the single most important methodology rule on this page.
Copy-Ready Questions, Grouped by Theme
Every group uses the scale that fits it. Copy one question, a whole theme, or the full set straight into your survey tool.
Belonging (WGIS — Validated)
5-pt LikertStrongly disagree → Strongly agreeBelongingness subscale of the Work Group Inclusion Scale (Chung et al. 2020). Measures whether employees feel they fit and are accepted in their immediate work group. Attribution required; seek author permission before commercial deployment.
- 1.
I belong in my work group.
Work Group Inclusion Scale, Chung et al. (2020)Core belongingness anchor item from the WGIS. The most direct measure of felt inclusion in the proximate team — action is most feasible at the team level.
- 2.
I am connected to the people in my work group.
Work Group Inclusion Scale, Chung et al. (2020)Captures relational bond quality, distinct from mere role acceptance. Low scores here alongside a higher belonging score indicate surface-level, not deep, inclusion.
- 3.
I believe that my work group is where I am meant to be.
Work Group Inclusion Scale, Chung et al. (2020)Taps a subjective sense of rightfulness and fit — a signal that the employee sees themselves as a natural, legitimate member rather than a tolerated outsider.
- 4.
People in my work group listen to me even when my views are different from theirs.
Work Group Inclusion Scale, Chung et al. (2020)The WGIS belongingness item most sensitive to dissenting-voice dynamics. A low score on this item while overall belonging scores are moderate signals performative — not genuine — inclusion.
- 5.
I feel like a valued member of my team.
Work Group Inclusion Scale, Chung et al. (2020)Bridges belongingness and uniqueness constructs; a useful complementary item that flags whether felt acceptance translates into perceived contribution value.
- 6.
My team treats everyone fairly, regardless of their background.
Work Group Inclusion Scale, Chung et al. (2020)Equity-of-treatment item. Scores tend to diverge sharply by demographic segment — the gap itself is diagnostic: if majority-group members score 4.2 and minority-group members score 3.1, that 1.1-point gap is the data that drives action.
Uniqueness & Being Valued (WGIS — Validated)
5-pt LikertStrongly disagree → Strongly agreeUniqueness subscale of the Work Group Inclusion Scale (Chung et al. 2020). Measures whether employees' distinct perspectives and differences are recognized and valued — the complement to belongingness in the full inclusion construct.
- 1.
My work group values the distinct perspective I bring.
Work Group Inclusion Scale, Chung et al. (2020)Uniqueness anchor item. Inclusion theory distinguishes groups where people are accepted only when they conform (high belonging, low uniqueness — assimilation) from genuinely inclusive groups (both high). This item isolates that second dimension.
- 2.
I am able to express my authentic self at work without fear of negative consequences.
Work Group Inclusion Scale, Chung et al. (2020)Authenticity expression is the behavioral indicator of the uniqueness construct. A gap between this item and the belongingness score reveals assimilation pressure — people feel they belong only by hiding parts of themselves.
- 3.
My differences from others in my team are treated as a strength, not a problem.
Work Group Inclusion Scale, Chung et al. (2020)Flips the uniqueness framing from value-add to risk-free differentiation. Scores below 3.5 here are a leading indicator of code-switching burden and identity concealment.
- 4.
When I share an idea or point of view that is different from the majority, it is taken seriously.
Work Group Inclusion Scale, Chung et al. (2020)Operationalizes whether uniqueness is merely tolerated or genuinely leveraged. Triangulate this against the listen-when-views-differ item in Group 1 — a matched low score confirms systemic voice suppression, not a single bad meeting.
Compiled Inclusion & Psychological Safety
5-pt LikertStrongly disagree → Strongly agreeCompiled best-practice items (QSET-020). These are NOT from a named validated instrument — do not present them as such. They measure the organizational-level inclusion climate and psychological safety that surrounds the team-level WGIS items.
- 1.
I can be myself at work.
QSET-020 compiledThe most frequently benchmarked inclusion item in practitioner surveys. Keep it here — rooted in the individual experience — and use it to contextualize the WGIS uniqueness scores above.
- 2.
My contributions are valued regardless of my background.
QSET-020 compiledEquity-of-contribution item. Useful for cross-group comparison: if contribution value varies significantly by race, gender, or tenure segment, that signals a specific process problem (credit attribution, visibility) rather than a climate problem.
- 3.
I feel comfortable speaking up about concerns or ideas without fear of retaliation.
QSET-020 compiledPsychological safety in the inclusion context — comfort speaking up is specifically about identity-related concerns, not just work-process feedback. Low scores here predict concealed problems that surface only in exit interviews.
- 4.
Our organization is genuinely committed to diversity and inclusion — not just in words.
QSET-020 compiledThe 'not just in words' qualifier is intentional: it forces respondents past performative agreement and captures the credibility of DEI commitment. Low scores on this item despite stated DEI programs signal an implementation gap.
- 5.
I have witnessed or experienced exclusionary behavior at work in the past six months.
Yes / NoQSET-020 compiledPrevalence screen for exclusionary incidents. Use yes-no rather than agreement scale to get a clean incidence count; follow with an open-end for context. Flag: incidents are underreported in confidential surveys — treat 'yes' responses as a floor, not a ceiling.
- 6.
What is one thing our organization could do to make everyone feel more included? (Optional)
Open textQSET-020 compiledOpen-end captures the specific, actionable changes that quantitative scores cannot surface. Mark optional — mandatory open-ends on sensitive identity topics suppress response rates and reduce data quality.
Organizational DEI Commitment
5-pt LikertStrongly disagree → Strongly agreeCompiled items measuring whether DEI commitment is visible, resourced, and embedded in day-to-day decisions. Distinct from team-level inclusion — these questions address what the organization does, not how the team behaves.
- 1.
Leaders in our organization model inclusive behaviors visibly.
QSET-020 compiledLeadership visibility is the strongest predictor of whether DEI programs translate into culture. Employees calibrate the credibility of any stated commitment against what they observe leaders doing.
- 2.
Our hiring and promotion processes are fair and free from bias.
QSET-020 compiledProcess-fairness item. Often the highest-disparity item by demographic group — a gap between majority and minority perceptions of hiring/promotion fairness is diagnostic of where systemic barriers sit.
- 3.
I am aware of the resources available to support diversity and inclusion at our organization.
QSET-020 compiledAwareness item that surfaces a comms problem masquerading as a DEI problem. If most employees can't name a DEI resource, the gap isn't belonging — it's communication.
- 4.
I feel represented in leadership — I can see people like me at senior levels.
QSET-020 compiledRepresentation perception. Analyze by demographic segment: low scores within a specific group point to specific pipeline gaps, not general culture issues. Suppress any segment below n ≥ 5 before reporting.
Segmentation & Data Quality
Multiple choiceSelect oneShort voluntary demographic and data-quality items. These are NOT scored — they are the segmentation keys that make every other group's data actionable. Apply the Rule of 5: suppress any group cell below n ≥ 5 before reporting results.
- 1.
Which of the following best describes your gender identity? (Optional, select all that apply — or prefer not to say)
QSET-020 compiledVoluntary, inclusive framing reduces non-response bias. 'Prefer not to say' is a legitimate response, not missing data — track the opt-out rate; a high rate signals trust issues worth addressing before the next survey cycle.
- 2.
How long have you been with the organization? (Under 1 year / 1–3 years / 3–5 years / 5+ years)
QSET-020 compiledTenure is a critical DEI segmentation variable: newer employees often report lower belonging, and the slope of belonging gain over time reveals whether onboarding and early culture exposure are working.
- 3.
How confident are you that your responses to this survey are confidential? (Not at all / Somewhat / Very / Completely)
QSET-020 compiledTrust-in-process item. Low confidence scores in a specific demographic group predict underreporting from that group — a bias that will distort every other result. Treat this as a data-quality diagnostic, not just a sentiment question.
- 4.
Is there anything else you would like us to know about your experience of inclusion at work? (Optional)
Open textQSET-020 compiledCatch-all open-end at the survey close. Place this last to avoid anchoring earlier quantitative responses. Thematic coding of these responses often surfaces the highest-specificity action items.
When Should You Use This Survey?
Match the survey type and cadence to your situation.
You have never run a DEI survey and need a validated baseline
Use
Avoid
Launching with an ad-hoc belonging question list assembled from blog postsA validated instrument gives you a psychometric baseline you can track year over year and defend to skeptical stakeholders. Ad-hoc lists produce scores you cannot interpret or compare.
You suspect belonging scores differ sharply by demographic group
Use
Avoid
Reporting every possible cross-tab regardless of cell sizeOver-segmenting small groups re-identifies individuals even in a confidential survey. Suppressing below n ≥ 5 is a legal and ethical floor, not a bureaucratic formality.
You are running a pulse check — not a full annual DEI census
Use
Avoid
Running the full 24-question survey quarterlyFrequent long DEI surveys produce survey fatigue and declining response rates without adding diagnostic value. Reserve the full battery for your annual or biannual census.
You are responding to a specific inclusion incident or complaint
Use
Avoid
Using survey data as a substitute for a confidential investigation or HR escalationA survey measures climate; it cannot adjudicate incidents. Treating low belonging scores as the resolution of a specific complaint exposes the organization to legal and reputational risk.
Your workforce is very small (fewer than 50 employees)
Use
Avoid
Collecting gender, race, or other identity demographics in a cohort where any cell will breach n ≥ 5In a 40-person company, collecting five demographic variables guarantees re-identification in multiple cells. The data you most want — belonging by minority group — will be the data you least safely report.
What "Good" Looks Like
Scores only mean something against a benchmark. Here are the numbers worth measuring against.
Validated criterion-related links to engagement, job satisfaction, and reduced turnover intention; overall α = .94
WGIS inclusion construct validity (Chung et al. 2020)
Chung, Ehrhart, Shore, Randel, Dean & Kedharnath (2020), Group & Organization Management
+56% job performance, −50% turnover risk, −75% sick days
High belonging linked to strong business outcomes (2019 data — flag age)
BetterUp, The Value of Belonging at Work, HBR 2019(vendor-reported)
40% of people feel isolated at work
Workers feeling isolated at work (2019 data — flag age)
BetterUp / HBR, The Value of Belonging at Work, 2019(vendor-reported)
$52M+ annual savings per 10,000 employees
Estimated savings from high belonging (2019 data — flag age)
BetterUp, The Value of Belonging at Work, HBR 2019(vendor-reported)
Survey Design Best Practices
The methodology that separates a survey people answer honestly from one they ignore.
METH-006 — Anonymous vs. confidential: DEI surveys are almost always confidential, not anonymous
Anonymous means no identifying data is collected — the response is truly untraceable. Confidential means identifiers are stored securely by a third party and reported only in aggregate. DEI surveys are almost always confidential, not anonymous, because demographic segmentation (the whole point) requires linked data. A survey asking for department, tenure, and gender is NOT truly anonymous even if individual responses are never shown. Be transparent about which you run. If you describe your DEI survey as anonymous and it isn't, you erode the trust that makes honest responses possible.
METH-006; WorkTango; Perceptyx; People Element
METH-007 — Rule of 5: set the minimum group-size threshold before you launch
Suppress any demographic cell with fewer than five respondents before reporting results. This is the load-bearing rule for DEI surveys specifically, because results are segmented by race, gender, tenure, and other identity dimensions. If your engineering team has two women and you report a gender breakdown, you have re-identified those individuals — even if you don't name them. Set the n ≥ 5 threshold in your survey platform before data collection begins. Some vendors permit n ≥ 3 or 4; n ≥ 5 is the widely accepted floor. For particularly sensitive items, raise the threshold to n ≥ 10.
METH-007; 15Five (Rule of 5); WorkTango; CultureMonkey; STAT-010
METH-003 — Avoid loaded or leading wording on sensitive identity items
Each item tests one idea. On identity-adjacent topics, even small word choices carry outsized weight. 'Do you feel excluded because of your race?' conflates a yes-no attribution with a frequency measure and will produce defensive non-responses. 'I belong in my work group' (neutral, first-person) outperforms 'Our organization values diversity' (aspirational, third-person) for diagnostic power. Pretest with a diverse pilot group of 20–50 respondents and rewrite any item with >5% non-response or heavy straight-lining.
METH-003; Formbricks; SuperSurvey; QuestionPro
Use a validated scale for belonging — not just compiled items
The Work Group Inclusion Scale (WGIS, Chung et al. 2020) measures two distinct dimensions — belongingness and uniqueness — with an overall α = .94 and demonstrated criterion-related validity against engagement, job satisfaction, and turnover intention. Many competing DEI surveys use ad-hoc item lists; WGIS gives you a defensible, published psychometric foundation. For editorial use, describe and cite; for commercial deployment, seek author permission from the original researchers.
QSET-019; STAT-016; Chung et al. (2020), Group & Organization Management
Segment by demographics — but plan your suppression rules first
The whole value of a DEI survey is the segmented data: how belonging scores differ by gender, race/ethnicity, tenure, level, and function. But that value disappears — and can cause real harm — if you over-segment into cells too small to protect. Before launching, map every intended cross-tab against your workforce headcount. If your LGBTQ+ employee group has 12 people in one region, a regional × identity cross-tab will likely breach the Rule of 5. Decide which segments are reportable before the survey closes, and communicate those thresholds to participants.
METH-007; METH-006
Close the loop — DEI data is the one type where silence is perceived as confirmation
Sharing DEI survey results and specific actions within two to three weeks is standard practice. But for DEI data specifically, silence after the survey is interpreted as confirmation that nothing will change — and for underrepresented employees, that confirmation is especially corrosive to trust. Share aggregate results even when they are uncomfortable. If a demographic group's belonging score is low, name that in the debrief and commit to a visible action. Vague commitments ('we will look into this') damage credibility faster than honest acknowledgment of a problem.
METH-008; Perceptyx; Gallup; Culture Amp
Survey Mistakes That Wreck Your Data
Running an 'anonymous' DEI survey that is actually confidential — and calling it anonymous
If your survey collects department, tenure, function, or any demographic variable, it is not anonymous — it is confidential at best. In a team of 10 with one non-binary employee, a gender breakdown re-identifies that person even without a name. Calling a confidential survey 'anonymous' breaks trust when employees realize the truth, and that broken trust is very hard to rebuild before your next survey cycle.
Reporting demographic breakdowns for groups below n ≥ 5
If your only Black female employee in a 50-person department scores 2.1 on the belonging scale, and you publish that department's gender × race cross-tab, you have published her individual responses — regardless of whether her name appears. This is the most serious data-protection failure in DEI measurement and the most common one.
Measuring only belongingness and skipping uniqueness
Many DEI surveys ask 'do you feel you belong?' and stop there. But you can belong in a team that demands you assimilate — fitting in by hiding or suppressing your identity. The WGIS deliberately measures both components: belongingness (are you accepted?) and uniqueness (are your differences valued?). A team with high belongingness and low uniqueness scores is an assimilation culture, not an inclusion culture. You cannot diagnose that without both dimensions.
Using consultant-branded belonging stats without attributing the instrument
A large number of belonging and inclusion frameworks circulating in HR (Gallup's global engagement belonging item, McKinsey's inclusion driver lists, Coqual's belonging survey) are not based on published, peer-reviewed validated scales. Using them as if they were is a sourcing and credibility risk. When a stakeholder asks 'how valid is this scale?' you should have an answer.
Collecting DEI data without a plan for what you will do with a low score
A belonging score of 2.8 out of 5 for a specific demographic group is a finding that demands a response. If you launch a DEI survey without a pre-agreed protocol for what happens when scores fall below a threshold — who sees the data, who owns the action, what the timeline is — you will collect the data, stall on the response, and watch trust erode among the very people you were trying to support.
Surveying too frequently without visible action between cycles
DEI survey fatigue is real and asymmetric: the employees most affected by belonging gaps are the most likely to disengage from future surveys when they don't see change. Running a full DEI census quarterly, or a pulse every month with no feedback loop, signals that data collection is the goal — not improvement.
Where These Questions Come From
Validated instruments have owners. Here's what's adapted from what — and how to use each one without stepping on a license.
Work Group Inclusion Scale (WGIS)
The WGIS is a published academic scale with 10 items measuring two components: belongingness and uniqueness. Overall α = .94 (Chung et al. 2020). For editorial use and description, cite the source below. For commercial deployment of the full validated instrument verbatim, seek author permission from the original research team before administering.
Launch & Follow-Up Templates
The invite, the reminder, and the results share-back — the messages that drive response rates.
DEI Survey Launch — Email Invitation
Subject: Help us understand how included you feel at [Company] Hi [Name], We are running our [Year] belonging and inclusion survey from [Start Date] to [End Date]. It takes about 8 minutes. Why this matters: We want to understand how included every person feels at [Company] — not just in the aggregate, but across different groups. The only way to do that is to hear from you. What you need to know about confidentiality: - This is a confidential survey, not fully anonymous. - Your individual responses are held by [Survey Platform / Third Party] and are never seen by your manager or HR in individual form. - Results are only reported for groups of 5 or more people. Any group below that threshold will show 'insufficient responses' rather than a score. - We will share results — including any groups with low scores — company-wide by [Results Date]. Take the survey: [Link] If you have questions about confidentiality, contact [HR Contact Name] at [Email]. Thank you, [Executive Sponsor Name] [Title]
Send from the executive sponsor, not HR. Lead with the confidentiality specifics — DEI is the one survey type where employees most need to understand exactly what is protected before they answer honestly.
Frequently Asked Questions
Related Survey Guides
A Survey Only Helps If You Act On It
The fastest way to tank your next response rate is to collect feedback and do nothing. Actify turns survey findings into action — recognition, engagement activities, and wellbeing benefits employees actually choose.
No credit card required. 15-minute setup.