What it measures
Psychological safety decides whether your other survey results can be trusted. On a team where people don’t feel safe to speak, engagement and culture scores are polite rather than honest.
It is a team-level property. Two teams in the same department can sit at opposite ends, which is why results are reported by team (anonymously) rather than only as one company average.
Manager safety climate
Whether the team’s manager invites challenge, responds well to bad news and treats mistakes as something to learn from.
Psychological safety
Whether people can raise problems, admit mistakes, ask for help and take a reasonable risk without being punished or embarrassed.
The 10 questions
Exactly what respondents see, grouped by theme. Each statement is answered on a five-point scale:
Manager safety climate
- My manager responds constructively when I point out a problem.
- My manager makes space for disagreement in team discussions.
- I hesitate to admit a mistake to my manager. (reverse-worded: agreeing lowers the score)
Psychological safety
- I can raise a concern with my team without worrying about how it will reflect on me.
- When something goes wrong on this team, the first question is who is to blame. (reverse-worded: agreeing lowers the score)
- If I disagreed with the rest of the team, I would say so openly.
- I can suggest an untested idea here without being made to look foolish.
- I would feel awkward asking a teammate to explain something I should already know. (reverse-worded: agreeing lowers the score)
- People on this team can be honest about what they don’t know.
- Different backgrounds and ways of thinking get a real hearing on this team.
In the live survey, questions are shown one at a time.
How it is scored and kept anonymous
- Each theme’s score is the average of its statements on a 1–5 scale, worked out per person first and then averaged, so everyone counts equally.
- Every score shows a 95% confidence range, so a small team’s result isn’t read as more precise than it is.
- Any team or group with fewer than 5 responses is hidden on every chart and export. Their answers still count in the organisation-wide score.
- Answers are stored against a single-use survey link, never a name or email. Leaders see averages and how answers are spread, never an individual’s answers.
- When you run the survey again, each organisation-wide change is tested for statistical significance and marked as a real change or as noise.
When to use it
Run it before a larger engagement or culture survey, after a reorganisation or leadership change, or on teams where you suspect problems are going unreported. Repeat every 6 to 12 months to see whether action worked.
Research basis
Amy Edmondson’s research on team psychological safety (Administrative Science Quarterly, 1999), and later research on how a manager’s behaviour shapes the climate a team works in. The statements are written for this module rather than taken from the published scale.
See what leaders get
The sample report shows a full run of this survey for Meridian Tech Services, a fictional 800-person IT services company with simulated responses, scored by the same engine your organisation would use: scores with confidence ranges, team comparisons, real-versus-noise changes since the last round, and where to act first.
Questions about this survey
How does this relate to Edmondson’s research?
The core theme measures team psychological safety as Edmondson defined it in her 1999 study. A second theme looks at the part the team’s manager plays in that climate, so results point to who can change it.
Can a manager see how their own team answered?
Only as a combined team score, and only when at least five people on that team responded. Below five, the team’s result is hidden everywhere, and individual answers are never shown to anyone.
How do we know if a change between surveys is real?
Every organisation-wide change is tested for statistical significance, so a movement is marked as real only when it is unlikely to be noise. Small teams are flagged as low confidence rather than over-read.