Psychological safety survey: questions, scoring and how to run it

Psychological safety is the shared belief, held by the members of a team, that the team is safe for interpersonal risk taking. That is Amy Edmondson's definition from her 1999 study of work teams, and it is worth keeping the exact wording in mind. It is not about being nice, lowering standards or avoiding disagreement. It is about whether people expect to be punished or embarrassed for speaking up, asking for help, admitting a mistake or proposing something unusual.

Because it is a belief, you cannot observe it directly. You have to ask. This guide gives you the seven-item scale Edmondson validated, explains how to score it, and covers the part that determines whether the survey helps or harms: how you run it and what you do afterwards.

The seven psychological safety survey questions

Edmondson (1999) measured psychological safety in 51 work teams at a manufacturing company with the seven statements below. Each member rated how accurately each statement described their team. Three of the items are worded negatively and are reverse-scored, which is marked with (R). The scale has been reused in hundreds of studies since, and there is no good reason to rewrite it.

1. If you make a mistake on this team, it is often held against you. (R)
2. Members of this team are able to bring up problems and tough issues.
3. People on this team sometimes reject others for being different. (R)
4. It is safe to take a risk on this team.
5. It is difficult to ask other members of this team for help. (R)
6. No one on this team would deliberately act in a way that undermines my efforts.
7. Working with members of this team, my unique skills and talents are valued and utilized.

Notice what the items do and do not ask. They ask about consequences (what happens when you make a mistake), about behaviour (can people raise tough issues), and about belief (is it safe to take a risk). They do not ask whether people like each other, whether the manager is friendly, or whether the team is high-performing. Keep it that way. Adding "we are a close team" or "my manager supports me" to a psychological safety survey mixes in cohesion and leadership and makes the score harder to interpret.

Scoring a psychological safety survey

On a five-point scale from 1 (strongly disagree) to 5 (strongly agree):

1. Reverse the three negative items (1, 3 and 5): a 1 becomes 5, a 2 becomes 4, 3 stays 3, 4 becomes 2, 5 becomes 1. After reversing, a high number means more safety on every item.
2. Average the seven items for each respondent. That is the person's psychological safety score, from 1 to 5.
3. Average the individual scores across the team. That is the team score.
4. Compute the spread. The standard deviation of individual scores, or simply the lowest and highest, tells you whether the team shares one experience or several.

Look at the item level as well as the total. A team that scores well overall but low on "if you make a mistake, it is held against you" has a specific problem with how errors are handled, and that is a much more useful finding than a 3.8 average. Low scores on item 7 (skills valued and used) often point to a staffing or role problem rather than a climate one.

Anonymity and minimum responses

A psychological safety survey is the one survey where anonymity is not optional. If people suspect their answers can be traced, the survey measures how safe they feel admitting they feel unsafe, which is a different and much smaller number. Three rules keep it honest:

• No identifying fields. No name, no role, no tenure, no "which sub-team". In a team of eight, any one of those can identify someone.
• A minimum number of responses before anything is shown, three by default, and more if the team is large enough that three is still identifying. Results stay hidden below that threshold.
• Aggregates only. Report averages and spread. Never show individual rows, even to the team lead, even when asked.

Say all of this before people answer, in writing, and then behave that way. The second survey depends entirely on how the first one was handled.

How to run it

1. Explain why. One short message: what psychological safety is (use Edmondson's definition), why the team is measuring it, who will see what, and when results will be discussed.
2. Keep it short. Seven items take under two minutes. If the survey is part of a wider team survey, put these items together and do not pad them.
3. Leave it open for a week. Long enough for everyone, short enough that it is answered in roughly the same conditions.
4. Have the lead answer too. The lead's score counts as one response like any other. Leads often rate safety higher than their teams do, and seeing that gap is part of the point.
5. Share results with the whole team within two weeks, in a meeting, not by email.

How to discuss the results

The discussion is where a survey either builds safety or damages it. A team lead who reacts to a low score with "I don't understand, I always say my door is open" has just demonstrated the problem. Some ground rules that help:

• Show the team average, the spread and the item scores. Do not speculate about who answered what.
• Let the team interpret before the lead does. Ask: "Which of these matches your experience? Which surprises you?"
• Ask for examples at the level of situations, not people: "When is it hardest to raise a concern here?" rather than "Who makes it hard?"
• Treat the lowest item as the agenda. One item, one or two concrete changes, revisited in the next survey.
• The lead goes last and keeps it short: what they heard, what they will change in their own behaviour. Not a defence.

The term is meant to suggest neither a careless sense of permissiveness, nor an unrelentingly positive affect but, rather, a sense of confidence that the team will not embarrass, reject, or punish someone for speaking up.

What to do with a low score

Low psychological safety is almost always learned from specific experiences: a mistake that was punished, a question that was mocked, a dissenting view that cost someone a project. The repair is also specific. Generic team-building rarely moves the score. These do, over two or three cycles:

• Change how mistakes are handled, visibly. The next time something goes wrong, the lead's first question is "what do we learn" and the first action is fixing the system, not finding the person. One well-handled incident does more than any workshop.
• Ask for dissent explicitly. In decisions, the lead asks "what would make this fail?" and waits. Silence is not consent; it is data.
• Admit your own mistakes first. Leads who say "I got that wrong" make it possible for others to. This is the cheapest intervention available and the one most leads skip.
• Separate help-seeking from performance judgement. If asking for help is scored at review time, item 5 will stay low no matter what else you do.
• Fix the role problem if item 7 is low. When people feel their skills go unused, the issue is often the work allocation, not the climate.
• Measure again in a quarter. Tell the team that is the plan, so the survey is seen as a cycle, not a one-off audit.

Psychological safety and the Trust & Safety KPI

In the PEAKTEAMS team survey, psychological safety sits inside the Trust & Safety KPI, one of 18 KPIs across input, process and output. The KPI combines psychological safety with interpersonal trust: whether members rely on each other to do what they say. The two are related but distinct. A team can trust each other's competence and still be unable to raise hard issues, and a team can be very open and still miss its commitments.

Google's Project Aristotle, which studied what distinguished effective teams at Google, identified five dynamics: psychological safety, dependability, structure and clarity, meaning, and impact. Psychological safety was ranked first and described as underpinning the other four. That matches what you would expect: a team that cannot talk about its problems cannot fix any of them. If you measure only one process KPI, measure this one. If you measure all 18, the team performance metrics guide explains how they fit together.