Visit Website →
← Back to blog

Psychological Safety Survey: Questions, Scoring and Next Steps

November 10, 2026
Psychological Safety Survey: Questions, Scoring and Next Steps

A psychological safety survey is a short, anonymous set of agreement statements that shows whether people on a team feel able to speak up, ask for help and admit mistakes without being punished or embarrassed. The strongest versions have 7 to 30 items, a 5-point scale, results reported at team level, and a plan for acting on the findings before anyone is asked a single question. This guide covers what to ask, how to score it, how to protect candor and what to do once the numbers arrive.

Leaderlyapp
Turn Survey Findings Into Daily Habits
Leaderly supports leadership development with personalized microlessons and practical exercises that help employees build impactful habits.
Explore Leaderly

TL;DR

  • Psychological safety is the belief that you can take interpersonal risks, such as raising a concern or owning a mistake, without fear of embarrassment or retribution. A survey measures that climate, not individual wellbeing.
  • A workable survey is 7 questions on a 5-point agreement scale, run anonymously, and reported as a team score plus themes and actions.
  • Scores are conversation starters. Pair them with open comments and real work evidence, and treat published score ranges as discussion guides, not diagnoses.
  • The survey only works if people see something change afterward. Decide who owns the follow-up before you send it.

Table of Contents

What a psychological safety survey measures

The American Psychological Association defines workplace psychological safety as a climate in which workers are comfortable expressing themselves and believe they can take appropriate interpersonal risks, including sharing concerns and mistakes without fear of embarrassment or retribution.

That definition points to what the survey is and is not:

  • It measures climate on a team. APA's 2024 Work in America Survey used seven team-focused statements adapted from Amy Edmondson's scale and asked respondents to think about the small group they work with most closely day to day. The unit of analysis is the team, not the company and not the person.
  • It is not a wellbeing survey. Wellbeing questions ask how people feel: stressed, rested, satisfied. Psychological safety questions ask what people believe will happen if they speak: will a mistake be held against me, can I ask for help, will someone undermine my effort. A team can have decent morale and still be quiet in meetings. If you want the wellbeing side, our piece on workplace wellbeing and transformational leadership covers it.
  • It is not an engagement survey. Engagement asks about commitment and satisfaction with the job. Safety asks about the risk of speaking. They correlate, but they answer different questions. Our engagement survey design workbook explains how to keep the two from blurring into one long, unfocused form.

Psychological safety survey questions that work

APA's question set covers a consistent group of interpersonal risks. According to APA, its items address speaking up and raising concerns or problems, asking for help, and whether mistakes will be held against people. One item checks whether it is difficult to ask others for help, and another checks whether team members deliberately undermine each other's efforts.

The statements below follow those same themes. They are our own wording for a team check-in, not a copy of a published instrument. Respondents rate each from Strongly Disagree to Strongly Agree.

  1. If I make a mistake on this team, it is not held against me.
  2. People on this team can bring up problems and tough issues.
  3. It is easy to ask other team members for help.
  4. No one on this team would deliberately undermine my efforts.
  5. I can disagree with a decision in a meeting without worrying how it will affect me.
  6. When someone admits an error here, the team focuses on fixing it, not on blame.
  7. My skills and contributions are valued and used on this team.

Two of these (items 2 and 3) are positively worded and two (items 1 and 4) are risk-oriented. If you reverse-word anything, make sure the scoring flips it. If your team is mixed, make clear that "this team" means the people each respondent works with every week, not the whole department.

Add one or two open questions

Numbers tell you where to look. Words tell you what you are looking at. Add a short free-text prompt, such as "What is one thing that would make it easier to speak up on this team?" Keep it optional and tell people it will be summarized as themes, never quoted with identifying detail.

If you want a broader bank of items for other leadership topics, see our list of 45 leadership survey questions.

Scale and scoring

A simple 7-item score

The Canadian Federation of Independent Business recommends an anonymous check-in with a 5-point scale from Strongly Disagree to Strongly Agree. Its scoring is simple: add all seven ratings, which gives a range of 7 to 35. Higher means a safer climate.

A worked example: a team of six returns the following totals: 29, 31, 24, 27, 33 and 22. The team average is 27.7 out of 35, or about 4.0 per item on the 5-point scale. The spread matters as much as the average. A gap between 33 and 22 suggests people on the same team experience it very differently, which is worth exploring in the discussion.

Reading the ranges

Reading the ranges

APA's 2024 survey combined responses to its seven statements into one score and used the median to split respondents into higher and lower groups. The result was 51% higher and 49% lower. That is a population split around a median, not a benchmark your team has to hit.

The University of South Carolina offers practical bands in its Psychological Safety Reflection and Action Framework: consensus scores of 4.0 or above suggest higher psychological safety, roughly 3.0 to 3.9 moderate, and below 3.0 lower. The framework itself says these ranges are discussion guides rather than validated diagnostic cut scores. Use them the same way: to decide which teams to talk to first, never to rank managers or tie anything to performance reviews.

Average item score (1 to 5)Total on a 7-item surveyHow to treat it
4.0 or above28 to 35Higher: ask what the team does that others could learn from
3.0 to 3.921 to 27Moderate: pick one or two items with the lowest ratings and discuss them
Below 3.07 to 20Lower: hold a facilitated conversation soon and look at the comments first

The third column is our suggested handling. The score bands come from the South Carolina framework, and the totals are the same bands converted to the 7-item, 35-point scale.

Choosing a reliable questionnaire

You have two honest options: a short, widely used check-in, or a longer validated inventory.

Short check-in (about 7 items). Quick, familiar and easy to repeat. Good for team-level conversations. Most are adapted from Edmondson's scale, as in the APA survey above.

Longer validated inventory. Plouffe and colleagues (2023) built the Psychological Safety Inventory by giving a 70-item version to 497 employees in Canada, the United States and the United Kingdom, then cutting it to a 30-item, five-factor scale with high reliability and convergent validity. A 30-item form fits a one-off organizational diagnosis, where a team would not be asked to repeat it often.

Choosing a reliable questionnaire

Adapting without breaking it

Small edits to wording, such as swapping "team" for "squad" or "unit", are usually harmless. Bigger changes carry risk. If you drop items, merge two instruments or invent your own scoring bands, you can no longer say your results match published norms. The South Carolina framework is explicit on this point: its adapted scoring guidance has not been independently validated as a combined instrument, and it recommends developmental, discussion-based use over formal evaluation.

The practical rule is to decide the purpose first. If the goal is a team conversation, a short adapted check-in is plenty. If you plan to compare across divisions or report to a board, use a validated instrument unchanged and say which one.

For educators and facilitators who want a discussion-first approach, developmental reflection and facilitated discussion are a natural companion to the survey, since the conversation after the results is where most of the learning happens.

Running the survey so people answer honestly

The most carefully built questionnaire fails if people suspect the answers will be traced. Candor is a design problem, not a motivation problem.

  1. Make it anonymous, and say how. CFIB recommends anonymity and protecting confidentiality by sharing only the overall score, themes and actions, never individual responses. Tell people exactly what leaders will see.
  2. Set a minimum group size. Do not report results for a group so small that a respondent is identifiable. Many organizations choose a threshold of five or more responses. That number is a common practice, not a rule from the sources above, so set the threshold with your own HR and legal advisers.
  3. Explain why before you ask. One paragraph: what you are measuring, who sees it, when you will share results and what will happen next.
  4. Keep it short. Seven rating items and one or two comments take under five minutes, which helps response rates, particularly on mobile.
  5. Collect no unnecessary demographics. Each extra field (tenure, role, location) narrows the pool and makes people less honest. Ask only for what you will actually use.
  6. Do not link it to evaluation. If people think a low score reflects on their manager's rating, managers will coach their teams toward high answers.
  7. Set a response window. One to two weeks is typical. Send one reminder, not five.

For teams working on trust more broadly, our guides on earning employees' trust and building a culture of trust cover what managers can do day to day.

What the results cannot tell you

A psychological safety score is a self-report. APA notes that survey results are subject to ordinary survey limits: errors from question wording and response options, coverage error, nonresponse and adjustments made after collection. There is a further, conceptual problem: the people who feel least safe are the least likely to say so, even anonymously.

One practitioner, writing on a personal blog, suggests treating results as a cluster of signals and comparing them with operational evidence such as near-miss reports and how long corrective actions take to close. That is a sensible habit for any team:

  • Compare the score with what you can observe: Who speaks in meetings? Are problems reported early or late? Do post-mortems name causes or people?
  • Read the comments before the number. A 3.8 with sharp comments about one manager is different from a 3.8 with none.
  • Track the same team over time rather than comparing unlike teams. A new team, a remote team and a team mid-reorganization will not share a baseline.
  • Never use a single survey to make a personnel decision.

How often to measure and what to do with the findings

The sources we reviewed do not set a required frequency, so treat this section as our practical guidance, not a standard. A reasonable rhythm for most teams is a baseline, a pulse check after you have had time to act (a quarter or two later), and then a yearly repeat. Measure sooner after a reorganization, a merger or a leadership change. Measuring monthly is usually too often: people see no change between rounds and stop answering.

What matters more than frequency is the follow-up loop:

  1. Share results within a few weeks. Silence after a survey teaches people that honesty is pointless.
  2. Share the team-level score, themes and the actions you will take. This mirrors CFIB's advice to share only the overall score, themes and actions.
  3. Hold a short discussion. Ask what the team recognizes in the results, what surprised them and what one change would help most.
  4. Pick one or two actions. Examples: start meetings with a round where everyone speaks, run blameless reviews after incidents, or have leaders name one of their own mistakes each month.
  5. Tell people what changed. At the next pulse, open with what you did since the last survey.
  6. Support managers. A manager with a low score needs coaching and practice, not a reprimand. Habits such as inviting dissent, thanking people who raise problems and responding calmly to bad news are skills, and short regular practice builds them. Peer coaching is one low-cost way to do it.

If you are comparing options for culture measurement and improvement, our overview of workplace culture tools describes the types available and what each is good for.

Our perspective

We see the same pattern again and again: organizations treat a psychological safety survey as the finish line. They collect the scores, present a slide, and wait. Nothing changes, and the next survey gets fewer and franker answers.

We think the survey is the first rep, not the workout. Safety is built in small, repeated behaviors: asking a real question, thanking someone for bad news, saying "I got that wrong." Those are habits, and habits form through frequent, practical, personalized practice rather than an annual workshop.

That is why Leaderly AI pairs self-assessments and surveys with short microlessons and behavioral exercises, so a team that learns where it is stuck can start practicing the fix the same week. It is also why we design for every employee, not just managers: safety is made by peers as much as by bosses, and a leader who practices alone cannot change a team's climate alone.

Measure honestly, share what you learn, and give people something concrete to do on Monday morning.

Drew

Turn survey results into daily leadership habits

Once you know where a team stands, the next question is how to help managers and colleagues act on it. Leaderly AI is a mobile-first platform that delivers personalized microlessons, behavioral habit-building exercises, self-assessments, surveys and analytics dashboards, so HR teams can follow a survey with practice instead of another meeting. Organizations in business, nonprofit and higher education settings use customizable learning journeys to roll it out to everyone, not only managers. See how it works on the Leaderly AI site, or explore the pages for nonprofits and higher education.

Leaderlyapp

If you want to try it with a single team first, the pilot page is the place to start.

FAQ

How many questions should a psychological safety survey have?

Seven is enough for a team check-in. APA's 2024 survey used seven statements adapted from Edmondson's scale, and CFIB's check-in also uses seven. A validated inventory such as the 30-item Psychological Safety Inventory suits a one-time organizational diagnosis.

Should a psychological safety questionnaire be anonymous?

Yes. CFIB recommends anonymous responses and sharing only the overall score, themes and actions. Tell respondents exactly who sees what, and avoid reporting on groups so small that individuals could be identified.

What is a good psychological safety score?

It depends on the instrument. On a 5-point scale, the University of South Carolina's framework treats 4.0 and above as higher, 3.0 to 3.9 as moderate and below 3.0 as lower, but it stresses these are discussion guides, not validated cut scores. Look at your own trend over time.

What is the difference between a psychological safety survey and an engagement survey?

An engagement survey asks how committed and satisfied people are with their work. A psychological safety survey asks whether people believe they can speak up, ask for help and admit mistakes without negative consequences. Many organizations run both and read them side by side.

Can we write our own psychological safety survey questions?

Yes, for a team conversation. Keep to the core risks: speaking up, asking for help, mistakes and undermining. If you need to compare against published results or across divisions, use a validated instrument without changing it.

Sources