A manager asks in a meeting, "What do you think about this plan?" and the room goes quiet. After the meeting, in the corridor, everyone explains why the plan won't work. More often than not, the name for that silence is a lack of psychological safety.

Psychological safety is the shared belief among team members that asking questions, admitting mistakes, asking for help and voicing a different opinion will not be punished. The concept was made measurable by Amy Edmondson of Harvard Business School; the 7-item scale she published in 1999 is still the most widely used tool. In this article we explain what psychological safety is, the Edmondson scale, scoring, reading results at team level, organization-specific sample questions and the steps you can take when results are low.

What is psychological safety?

Edmondson defines psychological safety as a shared belief held by members of a team that the team is safe for interpersonal risk taking. "Interpersonal risks" are the behaviors people avoid at work so as not to look bad:

• Saying you don't know or asking a question (the risk of looking ignorant)
• Admitting you made a mistake (the risk of looking incompetent)
• Asking for help (the risk of looking weak)
• Questioning a decision or a manager (the risk of looking negative or disruptive)
• Proposing a new or half-formed idea (the risk of looking foolish)

In teams with high psychological safety these risks feel small. In teams where it is low, people choose to stay silent. Problems are noticed late, mistakes are hidden and good ideas are never voiced.

What psychological safety is not

The concept is often misunderstood. Psychological safety is:

• Not about everyone being nice to each other. On the contrary, in psychologically safe teams people can give each other more open and more direct feedback.
• Not about low standards or a lack of accountability. Goals and expectations can stay high; what changes is that mistakes and problems can be discussed.
• Not about everyone always being right. Ideas are debated and can be rejected; what matters is that the person who voiced the idea is not excluded.
• Not an individual personality trait. The same person may speak freely in one team and stay silent in another. Psychological safety is the climate created by the team, and especially by the manager.

Edmondson explains this relationship with a two-by-two table:

This table also shows why you shouldn't look at psychological safety on its own when measuring it. A high psychological safety score is meaningful when the team also has clear goals and high expectations.

Why does psychological safety matter so much?

The study that brought psychological safety to a wide audience is Google's Project Aristotle. When Google investigated why some of its teams were more effective than others, it found that how a team works together mattered more than who was on it. Among the five dynamics that stood out, the most important was psychological safety; the others were dependability, structure and clarity, meaning of work and impact of work.

Edmondson's early studies point in a similar direction: teams with high psychological safety show more learning behavior. They seek feedback, discuss mistakes and experiment, and this is reflected in performance. Research in hospitals found that well-functioning teams reported more errors, because in these teams the problem was not making a mistake but hiding one.

From an organizational point of view, psychological safety becomes especially critical in these areas:

• Error- and quality-sensitive work: Fields such as healthcare, manufacturing, aviation, energy and finance, where catching a mistake early is vital.
• Innovation and change: Environments where new ideas and half-formed suggestions need to be voiced. We also cover this topic for change programs in our change readiness and adoption survey article.
• Quiet quitting and burnout: Employees who feel they have no voice first go silent, then withdraw. We explained how to catch these signals with a survey in our article on quiet cracking.
• Inclusion: Whether minority groups, new hires or remote employees can voice their views.

Edmondson's psychological safety scale

Edmondson used a 7-item scale in her 1999 article "Psychological Safety and Learning Behavior in Work Teams", published in Administrative Science Quarterly. It is the most commonly used tool for measuring psychological safety in both academia and organizations.

The items of the scale are listed below:

If you are using the scale in a non-English-speaking organization or in an academic study, choose a validated adaptation in that language and state which one you used. Translating the items yourself and reporting them as "the Edmondson scale" in a thesis or paper makes the validity of the measure questionable. Common translation pitfalls are covered in our article on language and culture pitfalls in global employee surveys.

How is it scored?

1. Choose the response scale. Edmondson's original study used a 7-point scale. Organizational surveys, however, usually use a 5-point Likert scale (Strongly disagree – Strongly agree) to match the rest of the survey. Whichever you choose, don't change it between measurements.
2. Recode the reverse items. Items 1, 3 and 5 are reverse-worded. On a 5-point scale, recode them so that 1 ↔ 5 and 2 ↔ 4. If you skip this step, the score becomes meaningless.
3. Take the average. Each employee's score is the average of the 7 items. The team score is the average of the scores of the employees on the team.
4. Check reliability. Calculate the internal consistency of the scale in your own data. Reverse items sometimes perform worse than expected; you can see this through the Cronbach's alpha value and item-total correlations.

In organizational reporting you can also use a favorable score alongside the average: the share of 4 and 5 answers after the reverse items have been recoded. We explain this method in detail in our favorable score article.

A note on reverse items

Reverse items are useful for spotting respondents who give the same answer without reading the items. In organizational surveys, however, they can cause two problems:

• Employees may misread the negative wording and answer in the wrong direction.
• In reports, lines like "If you make a mistake on this team, it is held against you – 30% favorable" are confusing for managers.

For this reason organizational surveys often word all items positively ("When I make a mistake on this team, it is not held against me"). If you don't intend to compare with academic research, this is an acceptable choice; if you do, use the original form as it is.

Psychological safety is a team property

The biggest mistake when measuring psychological safety is reading the result only at company level. A result like "psychological safety across the company is 72%" tells you almost nothing, because the same company has teams at 90% and teams at 40%. Edmondson's research also found large differences between teams within the same company.

So:

• Report results at team level. A team should be the group an employee works with day to day and that reports to the same manager. This requires the organizational data to be prepared correctly before the survey.
• Look at the distribution across teams. What matters more than the company average is the teams in the lowest quartile and the gap between teams. A heatmap is an easy way to see this.
• Look at agreement within teams. Academic studies check how similar team members' answers are (for example with statistics such as rwg and ICC). In an organizational setting a simple rule is enough: if answers within a team are widely scattered, there may be subgroups in the team with very different experiences.
• Protect anonymity. Teams with low psychological safety are also the teams that trust anonymity the least. Show results only for groups of at least 5–10 people and merge small teams into the unit above. See our article on anonymity in employee surveys for details.

The measurement paradox

In the teams where psychological safety is lowest, survey results may be the least accurate. In a team where people are reluctant to speak up, it is natural that they are also reluctant to answer a survey honestly. Some signs of this are:

• A response rate noticeably lower than in other teams
• The same answer, such as "Agree", given to every item
• Almost no comments in open-ended questions
• High scores, but poor turnover or complaint data

For this reason, don't interpret psychological safety results on their own; read them together with the response rate, open-ended comments, turnover data and managers' observations. We also cover these response tendencies in our article on survey response bias.

Organization-specific psychological safety questions

Instead of using the Edmondson scale as it is, many organizational surveys write statements that measure the same concept in the organization's own language. The questions below are original statements written for this purpose and let you see the different behaviors that make up psychological safety separately. All can be answered on a 5-point agreement scale.

Speaking up and raising problems

1. I can comfortably raise a problem or a risk in my team.
2. When I disagree with my manager's decisions, I can say so openly.
3. When I express a different view in meetings, it is taken into account.
4. Bad news is shared without delay in my team.

Mistakes and learning

5. When I make a mistake in my team, it is not held against me.
6. In my team, mistakes are discussed to learn, not to find someone to blame.
7. I don't hesitate to say when I don't know something.
8. When I try a new approach, I am supported even if things go wrong.

Asking for help and collaboration

9. It is easy to ask someone in my team for help.
10. My teammates don't act in ways that undermine my efforts.
11. In my team, information is shared, not withheld.

Inclusion and being valued

12. People with different backgrounds and views are not excluded in my team.
13. My skills and ideas are valued in my team.
14. My ideas are listened to regardless of my seniority or title.
15. On days when I work remotely, my views are taken into account as much as when I'm in the office.

The manager's behavior

16. My manager regularly asks the team for ideas and feedback.
17. My manager openly admits their own mistakes.
18. When I raise a problem with my manager, we look for a solution together instead of looking for someone to blame.

Open-ended questions

19. If there is a topic you find hard to raise in your team, could you write what stops you?
20. What would need to change for you to voice your ideas more freely in your team?

For questions on manager behavior, our articles on manager evaluation survey questions and the anatomy of first-line manager questions may help; for analyzing open-ended comments, see open-ended employee survey questions.

How many questions are enough?

It is usually more efficient to measure psychological safety not with a separate survey but as one dimension of an engagement or employee experience survey. In that case 4–7 questions are often enough: choose one or two questions each from speaking up, mistakes, asking for help and inclusion. If you are running a team study focused solely on psychological safety, you can use a larger share of the questions above. For more on survey length, see our article on how many questions an employee survey should have.

How should you interpret the results?

When reading psychological safety results, it helps to follow this order:

1. Look at the distribution across teams. Which teams are noticeably low? Is the gap concentrated in a particular function, location or seniority group? Use segmentation breakdowns for this.
2. Compare items. A team may be good at asking for help but poor at raising problems. Which behavior is weak determines what the action should be.
3. Look at the link with engagement. In many organizations psychological safety emerges as one of the strongest drivers of engagement. You can test this in your own data with key driver analysis.
4. Read the comments. Open-ended comments show the real situation behind the score: "Only the manager talks in meetings", "When we make a mistake we are criticized in front of everyone."
5. Track change over time. When comparing with last year's result, check whether the difference is really meaningful. See our article on comparing engagement survey results year over year for details.

What should you do if results are low?

Psychological safety doesn't change through training or a poster; it changes through the daily behavior of the manager and the team. Edmondson suggests three basic steps for leaders:

• Set the stage. Say openly that the work is uncertain and that mistakes are inevitable. A frame such as "We don't know everything on this project; we need to hear about problems early" makes speaking up legitimate.
• Invite participation. Instead of "Any objections?", ask questions that genuinely expect an answer, such as "Where is this plan weakest?" Let the team speak first and share your own view afterwards.
• Respond productively. Thank the person who brings bad news. Approach mistakes with "What did we learn, and what will we do next?" rather than looking for someone to blame. Someone who has been reprimanded in front of everyone once will not speak up again for a long time.

In practice, the following path can be used for a team with low results:

1. Share results with the team, but don't blame. Instead of "Your results are bad", use language like "It looks like we find it hard to talk openly about some topics as a team; I want to understand this together." Details on how to share results are in our article on sharing employee survey results with employees.
2. Start the conversation without the manager. In low-safety teams, having a facilitator from HR or another unit run the first feedback session helps people speak more freely.
3. Choose one or two concrete behaviors. For example, asking "Is there anything left unsaid?" at the end of every meeting, or sharing one mistake and what was learned from it with the team each month.
4. Support the manager. Managers of teams with low psychological safety are often unaware of it. Coaching, peer manager groups and feedback help managers change their behavior.
5. Re-measure soon. After three to six months, check progress with a 4–5 question pulse survey.

For situations where results are poor overall, we also shared a 30-day roadmap in our article on what to do when employee survey results are bad.

Common mistakes

• Reading the result only through the company average: Psychological safety is a team-level property; the real information lies in the differences between teams.
• Forgetting to recode reverse items: This makes the score completely meaningless.
• Reporting your own translation as "the Edmondson scale": Use a validated adaptation in academic work.
• Always treating a high score as good: In a team with low expectations, high psychological safety may indicate comfort rather than learning.
• Punishing the manager with a low score: This damages psychological safety across the organization and may lead managers to steer their teams toward "good answers" in the next survey.
• Measuring and then staying silent: Asking about psychological safety and not sharing the results reinforces the message "There's no point in speaking up here." We cover this topic in our article on survey myths and survey culture.

In short, psychological safety shows how early and how openly a team can talk about its problems, mistakes and ideas. When it is measured properly, that is, at team level, with anonymity protected and broken down into behaviors, it shows managers very concretely what they need to change.

Frequently asked questions