Psychological Safety

why your team won't tell you what you need to know

Learning Objectives

• Describe what psychological safety is and why it predicts team performance.
• Identify specific behaviors that create or destroy it.
• Change one practice in your own project to make it safer to admit uncertainty.

Before

Jess had been noticing for a while that Ren only brought problems to her after they had become crises. When Jess asked him about it, his answer was that the last time he'd admitted he didn't know how to do something, she had said, "You're the programmer, figure it out." He had spent three hours figuring it out, filed it away as "don't bring Jess your problems," and adjusted his behavior accordingly. Jess had no memory of saying it.

What Psychological Safety Is and Isn't

[Edmondson1999] studied 51 work teams in a hospital and found that psychological safety—the shared belief that it is safe to take interpersonal risks—was the strongest predictor of team learning behavior. Teams with high psychological safety admitted errors, asked questions, and flagged problems early. Teams with low psychological safety did the same amount of work but missed more problems and learned less from them. The difference was not the team members' skill or motivation; it was whether they believed that speaking up would be held against them.

[Forsgren2018] found the same pattern in software delivery: the highest-performing engineering teams were distinguished not by their tools or processes but by whether team members felt safe to admit mistakes and ask for help. Low psychological safety produced two predictable failure modes: problems discovered late and silent workarounds that nobody else knew existed.

Psychological safety is not the same as niceness or the absence of conflict. High-safety teams disagree often and directly. The difference is that the disagreement is about ideas and work rather than about people's competence or standing. A team where nobody ever says, "I don't think that's right," is not a psychologically safe team—it's a team where people have learned that disagreement is dangerous.

What Managers Do to Destroy It

The most common manager behaviors that destroy psychological safety are small, repeated, and often unintentional:

• Responding to bad news with visible frustration ("why didn't you catch this earlier?")
• Dismissing questions in public ("that's covered in the documentation")
• Solving someone's problem for them rather than helping them solve it, which signals that you don't trust their judgment
• Only seeking input from the same two people in every meeting
• Changing direction without explaining why, which teaches people that their work can be cancelled without warning

What Managers Do to Build It

• Respond to bad news with curiosity rather than frustration: "tell me more about how this happened"
• Publicly acknowledge your own mistakes and what you learned from them
• Ask for input explicitly from people who haven't spoken
• When someone brings you a problem, ask what they've already tried before offering a solution
• Close the loop: when someone flags a risk and it turns out to be real, acknowledge that they were right

After

After the conversation with Ren, Jess started keeping track of how she responded when people brought her bad news. She was surprised to find that her first instinct was often, "Why didn't you catch this earlier?" She didn't intend it to sound like blaming, but it was heard that way. She started replacing that question with, "What was the first sign something was wrong?" The answers were more useful, and the conversation stopped feeling like a debrief and more like problem-solving.

You cannot survey your team and ask them whether they feel psychologically safe: if psychological safety is low, they won't tell you honestly. The best proxies are behavioral: How quickly do problems surface? How often do people ask questions in meetings rather than afterward? How many of your sprint retrospectives produce the same items as last time? If the answers are "late," "rarely," and "all of them," psychological safety is the likely culprit.

Exercises

Diagnose Your Team (10 min)

Prompt the LLM like this:

I manage a small research software team of [N] people. I'm trying to assess whether psychological safety is an issue. Based on the following observable behaviors, tell me which are warning signs and which are neutral: [list three to five things you've observed recently in your team, e.g., "problems come to me late", "the same person dominates code review discussions", or "nobody has disagreed with a technical decision in months"].

Steps:

1. Run the prompt. Note what the LLM flagged as warning signs and whether you agree.

2. Individually, write:

   • One behavior from your team that might indicate low psychological safety.
   • One thing you have said or done in the last month that could have made it less safe to bring you problems.
   • One change you will make in how you respond to bad news.

Practice Responding (10 min)

1. In pairs: one person describes a realistic bad-news scenario, such as a test suite has been silently failing for two weeks. The other person practices responding in a way that makes it more likely the person will bring problems sooner next time.

2. Switch roles. Observers: note one phrase the manager used that would make you more likely to bring problems early, and one that would make you less likely. Compare your answers with those given by the LLM in the previous exercise.