“We’re a lovely team. Nobody argues.”
When a manager says that, it sounds like a good thing. It’s often a warning. Calm can feel like safety when it’s actually silence, and the problems three people saw coming turn up on your desk after they’ve become expensive.
Only 3 in 10 US workers strongly agree that their opinions count at work. Gallup estimates that getting that to 6 in 10 could mean 27% lower turnover, 40% fewer safety incidents and 12% higher productivity. Those are US figures and Gallup’s own projection, so read them as a direction, not a promise.
US workers strongly agree their opinions count at work. Gallup estimates that getting to 6 in 10 could mean 27% lower turnover, 40% fewer safety incidents and 12% higher productivity.
Before we get to the research, we need to go to a jury room.
The room where nobody wants to be the one
12 Angry Men, 1957
Twelve jurors in a hot room have to reach a verdict. The first vote is eleven to one for guilty. The one, Juror 8, doesn’t make a speech. He asks questions. He asks the others to go back over the evidence. At one point he asks for a secret ballot, so that anyone who’s been going along with the room can change their mind without saying so in front of everyone. Over the film he moves them one at a time.
It’s fiction, and real juries are messier. But look at what he does: questions instead of speeches, a private way to change your mind, and no rush. Those are close cousins of what the research backs.
Notice what the film is really about. The loudest jurors look the strongest and turn out to be the weakest. And a vote that reads eleven to one is only a snapshot of what people are prepared to say out loud in that room. It tells you nothing about what they think.
That’s what you’re up against in a team meeting. It’s rarely a lack of opinions. It’s a quiet calculation each person makes in about a second: is it worth it?
What psychological safety is, and what it isn’t
Psychological safety is a shared belief in a team that it’s safe to speak up: to ask a question, admit a mistake, disagree or raise a risk without being embarrassed or punished for it. It isn’t niceness, low standards or everyone agreeing with each other.
In fact the research suggests safety and standards belong together. Teams where people feel able to speak up show more learning behaviour, like asking for feedback and talking about errors, and they perform better. Most of that evidence is correlational, so it shows that safety goes with learning and performance rather than proving it causes them. There’s also a caution. In one research team’s findings, very high safety with no accountability alongside it may hurt routine performance, and that result is contested.
So the aim isn’t comfort. It’s a team that’s safe enough to say things, and stretched enough to have things worth saying.
Why your team isn’t telling you
You probably feel approachable. It doesn’t matter much. People speak up when they judge it’s safe, and that it’ll make a difference, and stay quiet otherwise.
And as the manager you weigh more than you feel. Bosses tend to underestimate how hard their requests are to refuse (Bohns, 2016), so your “any thoughts?” lands as a test and your “quick question” lands as an instruction. On top of that, telling the boss something unwelcome carries an image cost for the person doing it (Morrison, 2014; 2023). Silence is the sensible move unless you’ve made speaking up cheap.
Here’s what that looks like in the real world.
Nokia
Researchers who studied Nokia (Vuori and Huy, 2016) found fear at the top and among middle managers, with bad news held back and optimism inflated as information moved up the company. Middle managers withheld bad news from senior leaders they were afraid of.
It’s one qualitative study of one company, so it’s a story that fits the research rather than proof. But it shows what withheld bad news looks like when it’s multiplied across a hierarchy.
Five things Juror 8 would do in your meetings
1. Ask specifically, not generally. “Any thoughts?” rewards whoever is quickest and most confident. Ask one person about one thing: “What am I missing on this? I’d like to hear first from whoever’s newest to it, and pass if you’d rather.” (That wording is Ducora practice. The order of speaking hasn’t been tested directly.) It’s the Juror 8 move: a question, not a speech.
2. Give people a private way to change their minds. The secret ballot in the film does one job: it separates what you think from what you’re willing to say in front of the room. In meetings, that means everyone writes their view down before anyone speaks. There’s more on this below.
3. Respond well, every single time. Thank them. Ask for an example so you get it right. Show what you changed within two weeks. If the first honest thing someone says gets defensiveness, it’ll also be the last.
We’ve been through this. Is anyone actually going to give me something new?
Thank you for flagging that. Give me an example so I get it right, and I’ll come back to you on Thursday with what I’m changing.
4. Go first on the awkward stuff. In a field experiment, leaders who only asked for feedback didn’t make their teams feel safer a year later. Leaders who shared criticism they’d received did (Coutifaris and Grant, 2022).
Someone told me last month that I interrupt in the Monday meeting. They were right. I’m working on it, and I’d like you to tell me when I do it.
5. Keep the standard. Safety on its own isn’t enough. Say what good looks like and review work together, so speaking up leads somewhere. And listen properly while you do it: in experiments, being listened to well raised people’s sense of safety (Castro et al., 2018). Summarise what you heard, ask one open question, and hold your reaction.
Safe and Stretched
Two moves up (invite, then respond well) and two across (name the bar, review together). A 2x2 that raises psychological safety and standards at the same time. Ten minutes a week to review.
A meeting habit that costs nothing
A lot of the damage in a meeting is probably done in the first ten minutes, when the most senior person says what they think. That’s a reasonable extension of how anchoring works rather than a tested finding, so treat “speak last” as a sensible precaution. What we do know is that groups tend to talk about what everyone already knows. When key facts were spread across different members of a group, groups were about eight times less likely to find the best answer (Lu, Yuan and McLeod, 2012).
That’s a hidden-information problem, and the film plays it out: facts that were in the room all along, waiting for someone to go back over them. There’s a second problem. When people knew each other’s preferences before they discussed, groups found the best option less often (Mojzisch and Schulz-Hardt, 2010). Which is the case for the secret ballot, and for writing before talking.
- Name the decider before the meeting. Arguments about who decides do real damage.
- Everyone writes their view first. On paper, before anyone speaks. Your private ballot.
- Ask what only you know. One round, one thing each. “One thing you know about this that the rest of us might not.”
- Ask what kills it. A five-minute written premortem: “It’s a year from now and this decision has failed badly. Write down why.”
- Speak last. If you’re the senior person, hold your view until everyone else has spoken.
Decision: which supplier. Decider: me, after hearing from you. Write your pick and your confidence out of ten before anyone speaks. I’ll go last.
DRAW
Decider named. Record alone first. Air what only you know. What kills it? A protocol for any meeting where a real decision gets made. Adds about ten to fifteen minutes.
The meeting card
The five steps above on one card, with the scripts. Stick it in your notebook and run your next decision meeting off it.
- The five steps, in order, with timings
- The exact sentences to say
- A tick-box to review how the meeting went
Where the film gets it wrong
It’s a film, so it’s tidier than life. A few honest caveats.
One brave voice isn’t a system. Juror 8 is lucky: he’s articulate, he’s got the evidence and he never gets punished. Real dissenters don’t always get that. If your plan is to wait for a hero, you’ve got a team that depends on luck.
Process matters as much as courage. When women were few in a group, unanimity rules closed a gender gap in voice that majority rule left open (Karpowitz, Mendelberg and Shaker, 2012). It’s a reminder that the way a group decides changes who gets heard, not just what it decides.
Not every voice carries the same weight. A 2024 study of 449 mock juries found that dissent from jurors of colour carried less weight (Karpowitz et al., 2024). If you only measure whether people speak, you’ll miss whether they’re heard.
How to tell if it’s working
Don’t rely on your sense of the room. You’re the person least able to tell. Ask the team, anonymously, before and after you change something, using a short pulse. The most widely used safety measure is Edmondson’s seven-item scale (cite her 1999 paper if you use it). It’s adequate as a measure, though the evidence on its stability over time is thin.
Then watch behaviour. In your next three meetings, note who spoke, who was heard, who got the credit and who got the chance. If it’s the same three people every time, you have your answer.
What the evidence says, and how sure we are
- Strong, correlationalPsychological safety goes with learning behaviour and with task performance (Frazier et al., 2017).
- Moderate, associationLeaders seen as inviting and appreciating input had teams with higher safety and a smaller status gap, in a survey of neonatal units (Nembhard and Edmondson, 2006).
- Moderate, single studiesThe moves with causal support so far are listening well (Castro et al., 2018) and sharing criticism you’ve received (Coutifaris and Grant, 2022).
- ContestedVery high safety without collective accountability may hurt routine performance (Eldor, Hodor and Cappelli, 2023).
- StrongGroups mostly discuss information everyone already shares, and are far less likely to find the best answer when the key facts are spread across members (Lu, Yuan and McLeod, 2012).
- ModerateKnowing others’ preferences before discussing lowered decision quality (Mojzisch and Schulz-Hardt, 2010).
- ModerateWhen women were few, unanimity rules gave them more voice than majority rule (Karpowitz, Mendelberg and Shaker, 2012).
- EmergingPremortems cut overconfidence and surfaced more risks in small controlled studies (Veinott et al., 2010; Gallop et al., 2016).
- EmergingAt Nokia, fear at the top and in middle management went with withheld bad news and over-optimism (Vuori and Huy, 2016). One qualitative study of one company.
- EmergingDissent from jurors of colour carried less weight in a study of mock juries (Karpowitz et al., 2024).
- UntestedA senior person speaking last, and “newest first”. Both are Ducora practice built on the evidence above.
Frequently asked questions
Is psychological safety the same as being nice?
No. It’s a shared belief that speaking up is safe. Teams can be warm and silent, or direct and safe. Safety goes with high standards, not low ones.
Can a team be too safe?
Possibly, if there’s no accountability alongside it. One research team found that very high safety without collective accountability may hurt routine performance, though that finding is contested. The practical answer is to pair safety with a clear standard and regular review.
How long does it take to build?
There’s no reliable number. You can change the in-meeting habits straight away, but trust accumulates through how you respond when someone does speak up. Expect to be judged over weeks and months, not days.
What if nobody speaks even after I’ve changed the meeting?
Then the written round is doing its job: it tells you what people think without needing anyone to be brave. Read the notes, act on one thing visibly, and say so in the next meeting. People watch what happens to the first honest answer.
Where do I start?
Change one meeting. Name the decider, have everyone write their view first, and speak last. Then ask one person privately about one thing you could do differently, and thank them for the answer.
Sources
- Gallup, on employees who strongly agree their opinions count at work (US data).
- Frazier et al. (2017), on psychological safety, learning and performance.
- Nembhard and Edmondson (2006), on leader inclusiveness and safety in hospital units.
- Castro et al. (2018), on listening and safety.
- Coutifaris and Grant (2022), on leaders sharing criticism.
- Eldor, Hodor and Cappelli (2023), on very high safety and accountability.
- Lu, Yuan and McLeod (2012), on information sharing in groups.
- Mojzisch and Schulz-Hardt (2010), on knowing others’ preferences before discussion.
- Karpowitz, Mendelberg and Shaker (2012), on decision rules and women’s voice.
- Karpowitz et al. (2024), on dissent in mock juries.
- Veinott, Klein and Wiggins (2010) and Gallop, Willy and Bischoff (2016), on premortems.
- Vuori and Huy (2016), on Nokia.
- Morrison (2014; 2023), on employee voice.
- Bohns (2016), on underestimating the power of a request.
- Edmondson (1999), on the team psychological safety scale.
