The most widely used method for measuring psychological safety is a 7-item survey developed by Harvard professor Amy Edmondson in 1999. It takes less than five minutes for team members to complete, uses a simple 1-to-5 scoring scale, and gives you a concrete number you can track over time. But surveys aren’t the only approach. Observing real behavior in meetings can reveal what people won’t always say on a questionnaire.
Edmondson’s 7-Item Survey
This is the standard tool used in organizational research and the one Google adapted for its well-known Project Aristotle study on team effectiveness. Each item is a statement that team members rate on a 5-point scale from “strongly disagree” to “strongly agree.” The seven statements are:
- If you make a mistake on this team, it is often held against you.
- Members of this team are able to bring up problems and tough issues.
- People on this team sometimes reject others for being different.
- It is safe to take a risk on this team.
- It is difficult to ask other members of this team for help.
- No one on this team would deliberately act in a way that undermines my efforts.
- Working with members of this team, my unique skills and talents are valued and utilized.
Notice that some of these statements are negatively worded. Items 1, 3, and 5 describe problems. A person who strongly agrees with “mistakes are held against you” is reporting low psychological safety, not high. These items need to be reverse-scored before you calculate results, which is explained below.
How to Score the Results
For the positively worded items (2, 4, 6, and 7), score responses as you’d expect: strongly disagree = 1, disagree = 2, neutral = 3, agree = 4, strongly agree = 5. For the three negatively worded items (1, 3, and 5), flip the scale: strongly agree = 1, agree = 2, neutral = 3, disagree = 4, strongly disagree = 5. This way, a higher number always means more psychological safety.
Once every response is on the same scale, average all seven items for each person. That gives you an individual score between 1 and 5. Then average those individual scores across the whole team to get a team-level score. A team averaging around 3 is essentially neutral. Scores below 3 suggest people don’t feel safe speaking up, admitting mistakes, or asking for help. Scores above 4 indicate strong psychological safety.
Looking at individual item averages is often more useful than the overall number. If your team scores 4.2 on most items but 2.1 on “it is safe to take a risk on this team,” that tells you exactly where the problem is. You don’t need to guess.
Running the Survey Effectively
How you administer the survey matters as much as the questions themselves. The most important factor is anonymity. People won’t honestly report that they feel unsafe if their name is attached to the response, especially if their manager will see it. Use a tool that collects responses without identifying who said what.
This creates a practical tension with small teams. If only four people are on the team and one person gives unusually low scores, it may be easy for a manager to guess who responded that way. A common threshold is to only report results for groups of five or more. For smaller teams, you can combine results with another team or use broader categories (percentage who agree vs. disagree) rather than precise averages.
Keep the survey short. Edmondson’s original tool is just seven items for a reason. Tacking on 30 additional engagement questions dilutes focus and drops completion rates. If you want to measure psychological safety specifically, let it stand alone.
Observing Behavior in Meetings
Surveys capture what people believe. Observation captures what people actually do. Researchers at BMC Medical Research Methodology developed an observational measure with 31 specific behaviors across seven categories that serve as real-time indicators of psychological safety. You don’t need to use all 31, but knowing what to look for changes how you read a room.
Signs of high psychological safety include people openly disagreeing with each other, asking questions, offering unsolicited ideas, admitting their own mistakes, asking for help, and requesting feedback. You’ll also notice supportive responses: active listening, inclusive language like “we,” enthusiastic reactions to other people’s input, and acknowledging achievements.
Signs of low psychological safety are often quieter and easier to miss. Watch for closed body language, facial expressions that suggest disengagement or anxiety, and people who only speak when directly asked. Defensive behaviors are also telling: denying fault, blaming others, showing aggression when challenged, or deflecting by focusing only on positives. Unsupportive group dynamics like frequent interrupting, side conversations in small sub-groups, and cold reactions to humor all point to an environment where people are guarding themselves.
Two behaviors are particularly diagnostic. The first is whether anyone voluntarily discloses a mistake or raises a problem they caused. The second is conversational turn-taking: in psychologically safe teams, speaking time is distributed relatively evenly rather than dominated by one or two voices. Google’s Project Aristotle research identified this equal distribution of conversational turn-taking as one of the clearest markers of effective teams.
Combining Surveys and Observation
Neither method is complete on its own. Surveys give you a number you can compare across teams and track over time. But they only reflect what people are willing to report, and they’re snapshots taken at a single moment. Observation gives you richer, more nuanced data, but it’s harder to standardize and scale.
A practical approach is to use the 7-item survey as your baseline and primary tracking tool, then supplement it with periodic observation during team meetings. You don’t need a formal checklist every time. Simply paying attention to who speaks, who stays silent, how mistakes are discussed, and whether disagreement is met with curiosity or defensiveness will tell you a great deal about whether survey scores reflect reality.
How Often to Measure
Psychological safety shifts with team composition, leadership changes, and organizational stress. Measuring once a year during an engagement survey isn’t frequent enough to catch problems early or to see whether interventions are working. Quarterly pulse surveys using the same seven items give you enough data points to spot trends without creating survey fatigue. If your team has gone through a significant change, like a new manager, a restructuring, or a high-profile failure, it’s worth measuring again four to six weeks afterward rather than waiting for the next scheduled cycle.
The most important thing is consistency. Use the same questions every time so you’re comparing the same thing. Changing the wording or adding items between rounds makes it impossible to tell whether a shift in scores reflects a real change in team dynamics or just a change in what you asked.