Psychological safety: what it is and what Google found
Psychological safety is the shared sense that in your team you can give your opinion, ask questions or make mistakes without paying a social price for it. Properly understood, it goes hand in hand with high standards and allows problems to be raised in time.
It is a belief shared by the whole team
In 1999, Harvard professor Amy Edmondson defined it as the belief, shared by the members of a team, that the group is safe for interpersonal risk-taking.
An interpersonal risk is anything that can make you look bad in front of others:
- Saying “I don’t understand” in a meeting.
- Admitting you made a mistake before anyone notices.
- Disagreeing with whoever makes the decisions.
- Suggesting an idea that is not fully thought through yet.
The key word is shared. It does not describe a brave person, but the climate of a group. That is why two teams in the same company can have very different levels.
In hospitals, the best teams recorded more errors
The story begins with a result that looked like a mistake. In the mid-1990s, Edmondson studied several nursing units in two hospitals. She wanted to check whether the best-led teams with the best relationships made fewer medication errors.
The data said the opposite: the best teams reported more errors, not fewer.
When she analyzed how each unit worked, Edmondson proposed another reading: the best teams probably did not make more mistakes, but talked more about their mistakes. In the units where a slip was met with blame, people stayed quiet.
She herself pointed out that her data could not fully separate errors made from errors reported.
She published this in 1996 and, three years later, turned that insight into the concept of psychological safety, with a study of 51 teams at a manufacturing company. There, it was associated with more learning behaviors: asking for feedback, discussing mistakes and seeking information from outside the team.
Google ranked it ahead of four other dynamics
In 2012, Google launched Project Aristotle to find out what made its teams effective. According to the guide the company published on its re:Work website, it studied 180 teams (115 in engineering and 65 in sales) and combined several measures of effectiveness: ratings from executives, from the team leaders themselves and from members, plus sales against target.
The result was that who was on the team mattered less than how the team worked together. Google identified five dynamics, in this order of importance:
- Psychological safety: being able to take risks without feeling insecure or embarrassed.
- Dependability: delivering what was agreed, on time.
- Structure and clarity: clear roles, plans and goals.
- Meaning: the work matters to the people doing it.
- Impact: believing the work makes a difference.
By contrast, factors such as working in the same office, seniority, team size or members’ individual performance were not significantly related to effectiveness. Journalist Charles Duhigg told the story in 2016 in The New York Times Magazine, and from then on the concept moved beyond academic journals.
The five facets we use at Personaramic
To make it more concrete, at Personaramic we divide it into five facets. They are a practical way of looking at it, not an official classification by Edmondson.
- Voice: saying what you think, including to whoever is in charge. Example: in a store, someone tells the manager that the new schedule leaves the till uncovered at midday.
- Mistakes: how slips are received. Example: in a hospital, a nurse flags that a dose was recorded wrongly and the conversation focuses on the procedure, not on her.
- Help: asking for help or admitting you do not know. Example: at a school, a new teacher asks colleagues how the grading platform works without feeling their reputation is on the line.
- Differences: different views are valued and nobody is left out. Example: in a project meeting, the quietest person finishes their argument without being interrupted.
- Risk: suggesting new ideas without fear of looking bad. Example: in a volunteer association, someone suggests changing a long-standing event and the group discusses it seriously.
It coexists with high standards and with disagreement
Psychological safety is often misunderstood. According to Edmondson herself:
- It is not being soft. She crosses safety and standards in a matrix. With low standards, safety leads to a comfort zone. With high standards and little safety, to an anxiety zone. Only the two together make up the learning zone.
- It is not always agreeing. A safe team argues more, because nobody is afraid to disagree.
- It is not the same as trust. Trust is about whether you give another person the benefit of the doubt. Psychological safety is about whether you believe the group will give it to you when you take a risk.
- It is not a personality trait. A very extraverted person can stay quiet in one team and speak freely in another.
A meta-analysis of 136 samples links it to performance
The support is broad, with nuances.
The meta-analysis by Frazier and colleagues (2017) pooled 136 independent samples, with more than 22,000 people and nearly 5,000 groups. It found a relationship between psychological safety and performance, learning and engagement. Among the associated factors were peer support, role clarity and a good relationship with whoever leads.
In neonatal intensive care teams, Nembhard and Edmondson (2006) found that psychological safety was higher when the leader explicitly invited people to speak up and valued their contributions.
The limits also matter:
- Many of the studies are cross-sectional: they show association, not that one thing causes the other.
- Project Aristotle is an internal study by a single company and was not published in a scientific journal.
- It is measured through each member’s perception. A single response does not describe the whole team.
How to read results from one team: the average is not enough
When several people on a team take the test, it is tempting to look only at the average. Research on organizational climate suggests also looking at how much the answers agree. Schneider, Salvaggio and Subirats (2002) called this climate strength: the less perceptions vary within a group, the stronger its climate.
To show it, we calculated three imaginary five-person teams, each person with an overall score from 1 to 5:
| Team | Individual scores | Average | Standard deviation | Gap between extremes |
|---|---|---|---|---|
| A | 3.8 · 4.0 · 4.0 · 4.2 · 4.0 | 4.0 | 0.14 | 0.4 |
| B | 4.8 · 4.6 · 4.6 · 3.2 · 2.8 | 4.0 | 0.93 | 2.0 |
| C | 3.0 · 3.2 · 3.0 · 2.8 · 3.0 | 3.0 | 0.14 | 0.4 |
Teams A and B have the same average, but they do not share the same climate. In A, everyone perceives roughly the same thing. In B, three people feel very safe and two much less so, and the average hides exactly the people who might be holding back.
From there, we suggest a reading guide built on five patterns:
- High average, little spread (team A). A shared climate. The useful move would be to look after what already works and check in from time to time.
- High average with one or two low scores (team B). The low scores are the most informative data. Nembhard and Edmondson (2006) found that professional status was associated with greater psychological safety, so it would be worth asking whether the low scorers are the newest members or those with least say in decisions.
- Low average, little spread (team C). Everyone agrees that speaking up is hard. That is a clear, shared diagnosis, which often makes it easier to start changing.
- One facet well below the others. For example, high “Help” and low “Voice”: the team supports each other, but disagreeing with the person in charge feels costly. The lowest facet shows where to look first.
- The coordinator scores far above everyone else. This is an expected pattern given the status effect, and a cue for leaders to invite input explicitly, which in that same study was associated with greater safety.
Two cautions: with five people, a single answer moves the average a lot, and anonymity is what keeps answers honest. The numbers are a way to open the team’s conversation, not to close it.
The team psychological safety test places your current team at a low, moderate or high level, breaks it down into the five facets and lets several people combine their answers anonymously so they can be read this way. If you are more interested in the part you play in the group, the team role test looks at it from the other side.
Sources
- Edmondson, A. C. (1996). Learning from mistakes is easier said than done: Group and organizational influences on the detection and correction of human error. The Journal of Applied Behavioral Science, 32(1), 5-28.
- Edmondson, A. C. (1999). Psychological safety and learning behavior in work teams. Administrative Science Quarterly, 44(2), 350-383.
- Nembhard, I. M., & Edmondson, A. C. (2006). Making it safe: The effects of leader inclusiveness and professional status on psychological safety and improvement efforts in health care teams. Journal of Organizational Behavior, 27(7), 941-966.
- Edmondson, A. C., & Lei, Z. (2014). Psychological safety: The history, renaissance, and future of an interpersonal construct. Annual Review of Organizational Psychology and Organizational Behavior, 1, 23-43.
- Frazier, M. L., Fainshmidt, S., Klinger, R. L., Pezeshkan, A., & Vracheva, V. (2017). Psychological safety: A meta-analytic review and extension. Personnel Psychology, 70(1), 113-165.
- Pearsall, M. J., & Ellis, A. P. J. (2011). Thick as thieves: The effects of ethical orientation and psychological safety on unethical team behavior. Journal of Applied Psychology, 96(2), 401-411. https://doi.org/10.1037/a0021503
- Edmondson, A. C. (2019). The Fearless Organization: Creating Psychological Safety in the Workplace for Learning, Innovation, and Growth. Wiley.
- Duhigg, C. (2016, February 25). What Google learned from its quest to build the perfect team. The New York Times Magazine.
- Google re:Work. Guide: Understand team effectiveness.
- Schneider, B., Salvaggio, A. N., & Subirats, M. (2002). Climate strength: A new direction for climate research. Journal of Applied Psychology, 87(2), 220-229. https://doi.org/10.1037/0021-9010.87.2.220
Want to know more about yourself?
Discover your result with one of our related tests.
Frequently asked questions
How is psychological safety measured in research?
The most cited scale is Edmondson's (1999): seven statements that each member rates, such as 'If you make a mistake on this team, it is often held against you.' Answers are usually averaged per team, after checking that members agree closely enough to speak of a shared climate.
Can psychological safety have unwanted effects?
In some cases, yes. In a study of 126 student teams, Pearsall and Ellis (2011) found that teams with a more utilitarian ethical orientation were more likely to cheat on a self-assessment, and that this tendency was stronger when psychological safety was high. The authors read it as a reminder that the climate amplifies the norms a group already has, good or bad.