Team effectiveness
·
8 min read
Key findings
Edmondson's seven-item measure records a team's shared belief and says nothing directly about behaviour in meetings.
Hospital units that rated better on Edmondson's survey had higher detected error rates, because their members talked openly about errors.
A 2025 review of 28 software studies found most relied on single-source questionnaires, which can inflate associations.
High safety with high standards produces learning, and high safety with low standards produces comfort.
A low score is checked by observing two meetings before it becomes a conversation with the leader.
In Edmondson's hospital study, units that rated better on her team survey had higher detected error rates, because their members talked openly about errors.
Psychological safety at work is a team's shared belief that speaking up, asking for help or admitting a mistake will not be punished. A seven-item survey measures it team by team, and a low score points first to how that team's meetings are run.
Edmondson introduced the construct at team level in the 1990s, from a study of about 50 teams in a manufacturing company.
The practitioner's task is to know what the resulting measure supports and what it leaves out.
What A Team Score Measures
Edmondson defines psychological safety as a belief that asking for help or admitting a failure will not bring formal or informal punishment.
The belief is held at group level, because people working together tend to have similar views of the climate.
Her survey has seven items, answered on a seven-point scale, though a five-point scale also works.
Three items are worded negatively and are reverse scored before anything is averaged.
Mistakes are held against you (reverse scored)
Problems and tough issues can be raised
People are rejected for being different (reverse scored)
It is safe to take a risk
It is difficult to ask for help (reverse scored)
No one would deliberately undermine your efforts
Your skills and talents are valued and used
Edmondson reports that the measure holds up across variations in the number and wording of items.
It has also been translated into numerous languages, including German, Spanish, Japanese and Chinese.
Responses are aggregated by team, because the construct is a group-level belief.
What The Score Leaves Out
The score records what people believe, and it says nothing directly about what happens in the meeting.
In Edmondson's hospital study, units that rated better on her team survey had higher detected error rates, because their members talked openly about errors.
A low incident count in a unit is therefore consistent with high safety or with silence, so it cannot confirm or refute a survey score.
A team average also hides who feels unsafe.
In the research Edmondson cites, lower-status team members generally feel less safe than higher-status members.
The distribution of answers belongs beside the mean, with the responses of junior members read separately where the team is large enough to protect anonymity.
The research base has its own limits.
A 2025 systematic review of psychological safety in software teams examined 28 studies.
Fifteen of the 28 studies in the review used Edmondson's scale, and most studies of links between safety and other variables relied on a single source.
The review notes that completing one questionnaire from the same respondents can inflate the observed relationships through common method bias.
It also records an absence of longitudinal studies among those reviewed.
The review is a preprint dated August 2025 and covers software engineering only.
It speaks to how researchers measure the construct, and it does not speak to HR teams in general.
Edmondson's own early evidence included independent ratings of the outcome.
In her study of about 50 teams, performance was rated by team members and also by managers and internal customers.
Teams with higher safety scored higher on both kinds of rating.
Edmondson also states that safety does not supply everything high performance needs, because leaders must still set high standards and a worthy purpose.
Three Misreadings
Edmondson lists common misconceptions about the construct, and three of them bear on how an executive reads a score.
Niceness. Edmondson describes psychological safety as candor and productive disagreement, and it does not require agreement or unequivocal praise
Lower standards. Safety and standards are two separate axes, and only high safety with high standards gives the learning and high performance zone
Trust. Trust runs between two individuals and concerns a future promise, whereas safety is a group view of the immediate consequences of speaking up
Edmondson reports companies where people are polite in meetings, disagree later in the hallway and do not carry out what was agreed.
She also cites research finding no correlation between the experience of safety and introversion or extroversion.
That places the cause in the climate of the team.
High safety with low standards is the comfort zone, and Edmondson notes that it produces little learning or innovation.
Low safety with high standards is the anxiety zone, which she names as the quadrant that worries her most.
Four Stages As A Diagnostic
Timothy Clark describes four stages, drawn from his consulting fieldwork.
Inclusion safety, where the person is accepted into the team
Learner safety, where asking questions and making mistakes is allowed
Contributor safety, where using skills in the role is expected
Challenger safety, where questioning the status quo is safe
He describes a natural progression, based on how much respect and permission a team grants.
A request for challenge in a team where people hesitate to ask a basic question has skipped two stages.
The stages come from fieldwork, and Clark's firm sells an assessment built on them.
They work best as a vocabulary for a diagnostic conversation, and Edmondson's seven items remain the measure with published research behind it.
Fifteen of the 28 studies in the review used Edmondson's scale, and most studies of links between safety and other variables relied on a single source.
Reporting By Team
Edmondson reports that in every organisation she has studied, safety differs substantially across groups.
She finds that safety is very much shaped by local leaders.
An enterprise average therefore reports the middle of a spread.
Four reporting rules follow.
Report by team wherever enough people respond to protect anonymity, and set the minimum count before the survey runs
Show the distribution of answers beside the mean, since a team can disagree with itself
Compare a team with its own previous score, because the organisation average is not a standard
Keep the items and the scale identical between rounds, so the scores remain comparable
The index often arrives inside an engagement report, and what to do after an engagement survey covers choosing the findings a business head can act on.
Observing Without A Survey
A practitioner can read a team in two of its regular meetings.
Who speaks, and in what order
Whether anyone disagrees with the most senior person
What happens in the ten seconds after bad news
Whether people ask questions or only answer them
Whether errors are described in the first person
Edmondson cites a pharmaceutical executive's term, a culture of telling, for a company where asking gets short shrift.
A meeting in which no one asks the leader a question is a meeting where that culture may be at work.
Errors described in the first person show who is willing to own one.
Disagreement in a leadership meeting covers how to handle the case where the observation concerns the most senior person.
What A Leader Does
Edmondson organises the leader's work into three tasks, each with observable moves.
Setting the stage. Frame the work as uncertain and interdependent, and name what is at stake and for whom
Inviting participation. Show situational humility, ask genuine questions, and build structures that make input routine
Responding productively. Thank the speaker, discuss next steps without blame, and sanction clear violations
Staff at Children's Hospital and Clinics in the Twin Cities stayed silent when Julie Morath first talked about errors.
She then asked whether everything had been as safe as they wanted it to be that week, and people began to bring up incidents.
Edmondson gives three rules for a good question.
The leader does not know the answer
The question does not limit answers to yes or no
The wording helps others share their thinking in a focused way
The obstacle is naive realism, the bias that gives an experienced person the feeling of already knowing what is going on.
Edmondson states that an angry or disdainful response to the first person who speaks up will quickly end the safety that stage setting built.
The third task also includes sanctioning clear violations, because people need to know what counts as blameworthy.
When The Score Is Low
A low score is a finding about a team and its leader that still needs checking.
The sequence below runs about two quarters and involves an HR business partner for finance, the finance director and the director's team of eight.
Confirm the score. The HR business partner reads the distribution of answers and sits in two of the team's regular meetings.
Brief the leader privately. The HR business partner describes specific events from those meetings to the finance director before any group discussion.
Agree three meeting changes. Examples are who speaks first, how bad news is received, and a question the finance director asks in each meeting.
Change what is rewarded. A status report that lists no risks is a trained behaviour, so the format changes to ask for them.
Re-measure. The same seven items are run two quarters later and compared with the team's own first score.
Escalate last. Only if the first five steps fail does the question become one of role fit.
The first five steps are cheap and reversible, and a decision about role fit is neither.
Advising a leader you disagree with covers the private briefing in step two.
Where safety scores well and the team still underperforms, team effectiveness diagnosis covers the other conditions to test.
The Executive Conversation
Three statements carry the argument without overclaiming.
Safety is measured by team, and it differs substantially across teams in the same organisation
In low-scoring teams, problems and errors are less likely to be raised, which makes the score a risk statement
The intervention changes how a named set of leaders runs meetings, with a review after two quarters
The financial evidence is limited.
Edmondson cites a study of 47 mid-size German firms.
Process innovation raised performance there only where psychological safety was present.
Performance was measured by change in return on assets and by executive ratings.
One study of 47 firms is a thin basis for a return claim, so the case rests on learning behaviour and error reporting.
The Google study known as Project Aristotle is an example executives often raise.
Edmondson reports it through a 2016 New York Times Magazine feature, which found that safety explained why some teams outperformed others.
An HR reader therefore holds a press summary and not the study, and the claim is best attributed that way.
State the limits before an executive does.
The measure records belief, most of the research relies on one questionnaire, and safety does not replace competence, structure or a workable plan.
Programmes And Their Limits
Edmondson states that the three tasks must be used repeatedly and interactively, and that leaders at all levels carry the responsibility.
The evidence cited here does not show that it changes what happens in meetings.
With a limited budget, the spend goes to the leaders of the lowest-scoring teams.
Where The Tools Sit
In Trusted Talent Advisor, the diagnostic that separates a capability gap from a conditions problem sits in component 01, The Core Practitioner Guide.
The meeting behaviours above are handled in component 02, The Stakeholder Conversation Toolkit.
The Talent Template Pack includes the conditions audit and the review canvas.
Sources
1. Edmondson, A. The Fearless Organization: Creating Psychological Safety in the Workplace for Learning, Innovation, and Growth. Wiley.
2. Edmondson, A. Psychological Safety and Learning Behavior in Work Teams. Administrative Science Quarterly, 44(2), 1999, cited in The Fearless Organization.
3. Clark, T. The 4 Stages of Psychological Safety. Berrett-Koehler.
4. Santana, B. and colleagues. Psychological Safety in Software Workplaces: A Systematic Literature Review. arXiv:2508.03369, August 2025 (preprint).
From the system
Trusted Talent Advisor sets out the full method: seven components, 51 tools, The Talent Template Pack and The Prompt Library.
See the system
Questions this answers
What is psychological safety at work?
Amy Edmondson defines it as a belief that asking for help or admitting a mistake will not bring formal or informal punishment. The belief is shared by a team, so the unit of measurement is the team and not the organisation.
How do you measure psychological safety in a team?
Edmondson's measure uses seven survey items on a seven-point scale, though five points also works. Three items are reverse scored. Responses are aggregated by team, and the distribution of answers is reported beside the mean.
What are examples of psychological safety at work?
A team member admits an error in a meeting and the leader thanks them and asks what would prevent it. A junior colleague disagrees with the director and the point is discussed on its merits. Someone asks for help early.
How do you improve a low psychological safety score?
Confirm the score by observing two meetings, brief the leader privately with specific events, and agree changes to who speaks first and how bad news is received. Re-measure with the same items after two quarters.
Is psychological safety the same as trust?
No. Trust runs between two individuals and concerns a future promise. Psychological safety is a group view of the immediate consequences of speaking up, such as asking for help, and it differs from team to team.
What are the four stages of psychological safety?
Timothy Clark describes inclusion, learner, contributor and challenger safety, a progression based on the respect and permission a team grants. The stages come from his fieldwork and work as a diagnostic vocabulary.
Keep reading
Measuring HR impact
Internal Mobility Metrics Without A Platform
Seven measures an HR adviser can build from existing records, each with a stated definition, a base and a way to read it.
7 min read
Workforce and future skills
Job Rotation As A Talent Gap Strategy
A job rotation programme closes a talent gap when the role, the person, the cover and the evidence are agreed with both managers before it starts.
8 min read
Talent reviews and succession
Seeing Talent Below The Top Team
Visibility of leadership potential drops below the top team. A cross-unit forum, run on one page per unit and a written evidence standard, is the method for seeing it.
8 min read