Psychological safety at work
Psychological safety is one of the best-supported ideas in team research, and one of the most misunderstood. It is not about being nice, lowering standards, or avoiding conflict. This note covers where the concept comes from, what the evidence actually shows, how it is measured, and the misreadings that turn a rigorous idea into a slogan.
Psychological safety is the shared belief that a team is safe for interpersonal risk-taking: that members can ask questions, admit mistakes, raise concerns, and challenge ideas without fear of embarrassment or punishment. Harvard professor Amy Edmondson defined it in 1999, and Google's Project Aristotle brought it to prominence after the study, launched in 2012 and published in 2016, found it was the strongest differentiator of its highest-performing teams. The critical nuance most summaries miss: psychological safety raises performance only when it is paired with high standards. Safety without accountability produces comfort, not results. Safety with accountability produces a learning team.
Where the concept comes from
Edmondson introduced team psychological safety in a 1999 study of clinical teams, building on earlier work by Schein, Bennis, and Kahn. Her counterintuitive finding: better hospital teams reported more errors, not fewer, because they were open about mistakes and learned from them, while weaker teams hid errors. Psychological safety was the difference between a team that surfaced problems and one that buried them. The concept is fundamentally about learning behavior in conditions of uncertainty, which is why it matters most in complex, high-stakes, or fast-changing work.
The Project Aristotle finding, in context
In 2012 Google set out to find what made its teams effective, studying 180 teams across engineering and sales, more than 200 interviews, and 250-plus attributes. Team composition, shared hobbies, personalities, tenure, did not predict effectiveness. Five dynamics did: psychological safety, dependability, structure and clarity, meaning, and impact. Psychological safety was the strongest of the five and underpinned the others. The lesson leaders should take is not that talent does not matter, but that how a team works together predicts performance beyond who is on it.
How it is measured
The standard instrument is Edmondson's seven-item survey, which asks team members to rate beliefs such as whether it is safe to take a risk on the team, whether mistakes are held against people, and whether it is difficult to ask others for help. It has been used widely enough across industries, including a large body of software-team research, that its reliability is well established. Two cautions apply. Psychological safety is a property of a team, not an individual, so it should be measured and acted on at the team level. And self-report scales can be shaped by who is answering and the culture around them, so scores are a starting point for conversation, not a verdict on a manager.
The misreadings to avoid
The most damaging misreading is that psychological safety means lowering the bar. Edmondson's own framing is the opposite: safety and accountability are two independent dimensions. A team can be high on both, and that combination, high safety plus high standards, is the learning zone where performance is strongest. High standards with low safety produces anxiety and hidden problems. High safety with low standards produces a comfortable team that does not improve. The second misreading is that safety means the absence of conflict. It means the opposite: teams that feel safe surface disagreement and error earlier, which is uncomfortable in the moment and valuable over time.
What an employer can actually do
The levers are within a manager's control. Frame the work as a learning problem rather than an execution problem when the situation is genuinely uncertain. Model fallibility by admitting your own mistakes and gaps. Ask genuine questions and respond to bad news with curiosity rather than blame, because the response to the first person who speaks up sets the norm for everyone after. Hold the standard while you build the safety: name what good looks like clearly, and make it safe to fall short honestly on the way there. Middle managers, notably, often report the least psychological safety themselves, so the support has to run up the org chart, not only down.
Psychological safety is the shared belief that a team is safe for interpersonal risk, and it is one of the strongest predictors of team learning and performance. It is a team-level property, measured by Edmondson's seven-item scale, and it works only alongside high standards. Treat it as the condition that lets a demanding team learn, not as permission to demand less.
Where the science comes from
Primary sources
- Edmondson, A. C. (1999). Psychological safety and learning behavior in work teams. Administrative Science Quarterly, 44(2), 350-383. The founding study: the definition, the seven-item measure, and the finding that better teams surface more errors because they are safe enough to report them. doi.orgChecked 21 July 2026
- Kahn, W. A. (1990). Psychological conditions of personal engagement and disengagement at work. Academy of Management Journal, 33(4), 692-724. The earlier work Edmondson built on: psychological safety as a condition for people to bring themselves fully into a work role. doi.orgChecked 21 July 2026
- Frazier, M. L., Fainshmidt, S., Klinger, R. L., Pezeshkan, A., and Vracheva, V. (2017). Psychological safety: A meta-analytic review and extension. Personnel Psychology, 70(1), 113-165. The meta-analysis establishing the links to learning behavior, information sharing, engagement, and performance across 136 samples. doi.orgChecked 21 July 2026
- Newman, A., Donohue, R., and Eva, N. (2017). Psychological safety: A systematic review of the literature. Human Resource Management Review, 27(3), 521-535. The systematic review behind the team-level-property and measurement cautions in this note. doi.orgChecked 21 July 2026
- Google re:Work. Understand team effectiveness (Project Aristotle). Google’s own write-up of the 180-team study: the five dynamics, psychological safety as the strongest, and the team survey items quoted in this note. rework.withgoogle.comChecked 21 July 2026
- Edmondson, A. C. (2018). The Fearless Organization: Creating Psychological Safety in the Workplace for Learning, Innovation, and Growth. Wiley. The book-length treatment, including the safety-and-standards two-by-two this note’s learning-zone framing comes from. wiley.comChecked 21 July 2026
Psychological safety is one of the best-evidenced team constructs in organizational research, and one of the most misread. The sources above are the primary record: the founding study, the meta-analytic evidence, and Google’s own publication of the Project Aristotle findings. Where this note warns against a misreading, it is Edmondson’s own framing, not a gloss.
Common questions
Psychological safety is the shared belief that a team is safe for interpersonal risk-taking: that members can ask questions, admit mistakes, raise concerns, and challenge ideas without fear of embarrassment or punishment. Harvard professor Amy Edmondson defined it in 1999.
Google’s Project Aristotle brought psychological safety to prominence in 2012 after finding it was the strongest differentiator of its highest-performing teams, ahead of who was on the team or how it was structured.
No. Psychological safety is not about being nice, lowering standards, or avoiding conflict. It raises performance only when paired with high standards. Safety without accountability produces comfort, not results; safety with accountability produces a learning team.
It is most commonly measured with Amy Edmondson’s seven-item team psychological safety scale, a short self-report survey that asks team members how safe it feels to take interpersonal risks on their team.