Psychological safety — Amy Edmondson's 25 years of research, the #1 variable from Google's Project Aristotle, 4-step team diagnosis for "safe questions, mistakes, dissent"

Psychological safety — Amy Edmondson's 25 years of research, the #1 variable from Google's Project Aristotle, 4-step team diagnosis for "safe questions, mistakes, dissent"

Amy Edmondson (Harvard Business School) defined "Psychological Safety" in her 1999 doctoral dissertation. Definition: "a shared belief among team members that interpersonal risks (mistakes, questions, dissent) are safe to take". Google's Project Aristotle (2012–2014) analyzed 180 teams → the #1 variable determining performance was "psychological safety" — far larger effect than individual skill or seniority. 4 stages (Clark 2020): ① Inclusion — I am accepted; ② Learner — questions / mistakes are safe; ③ Contributor — opinions / ideas are safe; ④ Challenger — dissent / criticism is safe. Korea's hierarchical culture (kkondae, face, nunchi) often makes even stages 1–2 difficult. Clinical effect: more psychological safety → less team depression / burnout, more innovation / performance. Measurement: Edmondson's 7-item scale. 5 leader behaviors.

TL;DR

Edmondson 1999 psychological safety. Google Aristotle: #1 team-performance variable. 4 stages (inclusion, learner, contributor, challenger). Korea's hierarchical culture stalls even at stages 1–2. Effect: less depression / burnout, more innovation. Edmondson 7-item measure. 5 leader behaviors: curiosity, self-disclose mistakes, welcome questions, reward dissent, mistake-learning rituals.

1. Edmondson's 1999 discovery

Amy Edmondson (then a Harvard doctoral student, now Novartis Professor) studied hospital medical teams and found a surprising result: the teams with "more mistakes reported" were not the teams with "more mistakes" — they were the teams "safe to report mistakes". This is the start of the "psychological safety" concept.

2. Google Project Aristotle (2012–2014)

Google analyzed 180 teams over 4 years to ask "why are some teams high-performing and others not?" Five key variables (in order of importance):

  1. Psychological safety (overwhelmingly #1)
  2. Dependability
  3. Structure and clarity
  4. Meaning of work
  5. Impact of work

Shocking finding: individual skill / IQ / seniority / diversity were not in 1–5. The same person produces high performance on a safe team and low performance on an unsafe team.

3. 4 stages of psychological safety (Timothy Clark 2020)

Stage 1: Inclusion Safety

  • "I belong on this team"
  • No exclusion by appearance / background / gender / age
  • "I don't need to read the room"

Stage 2: Learner Safety

  • "Safe to ask questions / make mistakes"
  • Can admit "I don't know"
  • Safe to ask for help

Stage 3: Contributor Safety

  • "Safe to put forward my ideas"
  • Active opinion-sharing
  • Can critically evaluate others' opinions

Stage 4: Challenger Safety

  • "Safe to disagree with existing methods / boss's opinion"
  • Can propose innovation / change
  • The hardest stage; the highest value

4. Edmondson 7-item measure

Rate your team 1–7:

  1. If you make a mistake on this team, it is often held against you (reverse)
  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 (reverse)
  4. It is safe to take a risk on this team
  5. It is difficult to ask other members of this team for help (reverse)
  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

Reverse-score the reverse items. Average 5+ = safe team; 3.5–5 = average; under 3.5 = risk.

5. The challenge in Korean workplaces

  • Hierarchy (rank, seniority) blocks Challenger safety
  • "Kkondae" culture: juniors' opinions / dissent treated as "impudence"
  • "Nunchi": reading the room before expressing your opinion
  • "Face": admitting mistakes is costly
  • "Emotional labor" of drinking parties / personnel evaluations
  • 80% of Korean office workers report "I can't speak my opinion freely" (JobKorea 2022)

6. Clinical and management effects

MetricHigh safetyLow safety
Team depression10%30%
Burnout15%40%
Turnover5%25%
Innovation / new productsHighLow
Medical-error reportingHigh (good signal)Low (concealment)
Customer satisfactionHighLow

7. 5 leader behaviors

  1. Start with curiosity: "What do you think?" / "What did you see?"
  2. Disclose your own mistakes first: "I made this mistake" — model
  3. Welcome questions: "good question" / no "silly questions"
  4. Reward dissent: openly thank those who voiced opposing views
  5. Mistake-learning rituals: regular weekly meetings on "what we learned / mistakes this week"

8. 5-step Korean application

Leader (manager) view

  1. Self-assess: Edmondson 7-item anonymous survey of your team
  2. Start by disclosing your own mistakes (in meetings)
  3. Meeting rule: "speak from the youngest first"
  4. Explicitly thank juniors who voiced dissent
  5. Convert error reporting from "penalty" to "bonus" in evaluations

Member (junior / colleague) view

  1. Measure your team's safety
  2. Provide Stage 1 (inclusion) to friends / colleagues at minimum
  3. If Challenger safety is absent, try small opinions first — refusal is information
  4. Secure 1–2 safe colleagues
  5. For severely toxic teams, transfer or change job (#252, #256)

9. Application to family / relationships

Psychological safety applies not just to teams but to family / partner / friends:

  • Is it safe to share opinions at the family table?
  • Can you tell your spouse "bad news"?
  • Do children tell their parents about mistakes / worries?
  • Is it safe to disagree with your friends' views?

Low psychological safety in these areas risks depression and relational rupture.

10. Korean resources

  • "The Fearless Organization" (Edmondson, Korean edition)
  • "The 4 Stages of Psychological Safety" (Timothy Clark)
  • "Psychological safety" workshops at some Korean large companies (Naver, Kakao, SK, etc.)
  • Management / leadership-consulting firms
  • Can be integrated into family / couples therapy
Ad

Frequently asked questions

Isn't psychological safety just "no criticism / everything is fine"?

Opposite. Edmondson is explicit: safety = an environment where "honest criticism, dissent, and admitting mistakes are safe". Not "fake kindness". The hardest of the 4 stages — Challenger Safety — is the core. No criticism = no safety = "hiding culture".

In Korean hierarchical workplaces, how can a junior create challenger safety?

Hard but small attempts are possible. 1) Express opinions in 1:1 meetings (not big meetings), 2) make arguments with data / evidence (not emotion), 3) use "question" form ("what if we tried..."), 4) form "safety alliance" with 1–2 colleagues, 5) if ineffective, the company itself is toxic — consider changing jobs (#252, #256). You can't change every Korean hierarchical workplace by your own effort.

After Google Aristotle, Korean companies are following — does it work?

Workshops / training alone don't work. "Safety" is created by the leader's "behavior". A few Korean companies (Naver, Kakao, some startups) succeed — leaders actually disclose their own mistakes and welcome dissent. Most stay "the same after training" — no leader behavior change. Real change happens when HR evaluations convert "error reporting / dissent" into bonuses.

Related reads

Mental health

Fifty Years of the Bystander Effect: Reassessing Darley·Latané (1968) with Philpot (2020)

9 min read
Mental health

The Science of Hoarding Disorder: Frost, Steketee, and the DSM-5 Standalone Diagnosis

9 min read
Mental health

Why Worry Won't Stop: Borkovec's Cognitive Avoidance Theory and the Science of GAD

9 min read
Mental health

The Stranger in the Mirror: Clark-Wells Cognitive Model of Social Anxiety and CT-SAD

9 min read