Comfortably Wrong: Hello, Is There Anybody in There?
“Hello, is there anybody in there?”
In 1979, Roger Waters wrote that line about someone present but unreachable, switched off to what was happening around them. In 2006, Adam Galinsky and colleagues published experimental evidence that power produces exactly this condition. High-power participants were less accurate at reading others' emotions, anchored more heavily on their own vantage point, and failed to adjust for information others didn't share. Power, the research found, makes people comfortably wrong.
In 1996, Amy Edmondson was studying medication error rates in hospital nursing units. She expected better-performing units to make fewer errors. She found the opposite: the better-performing units reported significantly more errors than the worse ones.
It took further investigation to understand why. The better units were surfacing errors. The worse units were burying them. The units that looked safest on paper were accumulating hidden risk.
When fear is a higher-order concern than performance, an organisation's output becomes a function of what people calculate is safe to say, rather than what they actually know. The gap between those two things is where performance is being lost - and in the cases this post examines, that gap cost between US$1.2 billion and US$20 billion to close.
What the Research Shows
The Frazier et al. meta-analysis (136 samples, more than 22,000 individuals) documents strong correlation between psychological safety and information sharing (corrected correlation: .52); learning behaviour (.62) and work engagement (.48).
Task performance, by contrast, shows a more moderate relationship - consistent but weaker than the information and learning outcomes. This is the finding the business case rarely uses, and it changes the commercial argument considerably.
The case for psychological safety is not primarily that safe teams perform better at the tasks they are already doing. It is that safe teams surface what is going wrong, learn from it faster, and share what they know across the people who need it. The performance gain is not in execution. It is in accuracy - in the gap between what the organisation is operating on and what it actually needs to know.
Power Is the Primary Variable
Detert and Burris's field study of 3,149 employees found that manager behaviour was the strongest predictor of whether employees spoke up. More specifically, the effect was largest for the organisation's best performers. Leaders who signalled that candour was risky were losing the input of the people most capable of providing it.
This is a power-related finding that means:
Psychological safety in a team is set primarily by the behaviour of the most senior person in the room
The most capable people make the most precise calculation about whether it's safe to contribute, and are most responsive to the signal
Culture programs and values statements do not change this; the specific moment-to-moment behaviour of leaders does
Galinsky and colleagues' experimental work adds a more confronting layer. Power physically reduces perspective-taking ability. High-power participants were less accurate at reading others' emotions, anchored more heavily on their own vantage point, and failed to adjust for information others didn't share.
The implication is that as power concentrates, the person holding it simultaneously sends a stronger signal that candour is risky and becomes less able to recognise that they are receiving an incomplete picture. This is a documented effect of power on cognition.
What This Looks Like When It Fails
The APAC region provides the clearest natural experiment in what happens when power concentrates without the structural mechanisms that would compensate for it.
Three cases stand out because the official investigations named the mechanism explicitly - as the primary cause, rather than one factor among many.
Toshiba (Japan, 2015): an independent committee found the profit overstatement of approximately US$1.2 billion over six years occurred in a 'corporate culture where employees cannot act contrary to the intent of superiors.' The formal governance system - audited accounts, board papers, management sign-off - was processing numbers that everyone below the executive level understood were wrong.
Nissan (Japan, 2019): the external advisory panel attributed the concealment of more than US$140 million in compensation directly to 'the concentration of all authority' in the former chairman. Departments that should have provided checks 'did not necessarily function effectively.' The structure had removed the mechanism by which contrary information could reach the people who needed it.
Korean Air (South Korea, 1970s-1990s): a safety record that included over 700 fatalities traced consistently to the same mechanism - subordinate crew members failing to assert against disoriented captains in cockpits where authority made contradiction structurally unsafe. The industry-wide adoption of Crew Resource Management was an explicit engineering project to compensate for this: making it formally obligatory to surface life-critical information regardless of seniority.
The global record extends this. Boeing's 737 MAX grounding (approximately US$20 billion in direct costs) was attributed by the House Transportation Committee to a 'culture of concealment.' Volkswagen's Dieselgate (more than €32 billion) involved engineers who knew and were pressured to stay quiet. Wells Fargo's fake account scandal (more than US$3 billion in settlements) persisted for over a decade while senior leaders knew.
In each case the information existed inside the organisation. The power structure determined whether it travelled.
Inclusion Is the Second Variable
Psychological safety solves whether information can flow. It does not determine which information flows.
A team in which everyone feels safe to speak but shares the same background, assumptions and blind spots will be comfortable and informationally narrow. The channel is open. The range of what travels through it is limited by who is in the room and whether their different perspectives are genuinely treated as valuable.
The research on this is precise. The Categorization-Elaboration Model - the dominant integrative framework for diversity effects in teams - establishes that diverse teams outperform homogeneous ones only when information elaboration occurs: when different perspectives, knowledge and experiences are actually exchanged and integrated. Social categorisation - the in-group/out-group dynamics produced by exclusion - suppresses this elaboration.
Woolley and colleagues' study found that collective intelligence in teams - the factor that predicts performance across diverse problem types - correlates with three things, rather than average individual IQ:
Social sensitivity (ability to read others)
Equal distribution of conversational turns
Proportion of women in the group
All three are inclusion indicators. Groups get smarter when everyone's information enters the conversation, and specifically when the structure of the conversation doesn't systematically advantage the voices already at the top.
Shore and colleagues define inclusion as satisfying two distinct needs: belonging and valued uniqueness. The important distinction is between assimilation (high belonging, suppressed uniqueness - the condition where people fit in by hiding what makes them different) and genuine inclusion (where what makes someone different is the thing being valued). A psychologically safe team that operates through assimilation produces comfort.
A psychologically safe team that operates through genuine inclusion produces a wider, more accurate picture of what the organisation is dealing with. These are meaningfully different commercial outcomes.
Three Questions This Evidence Raises for Any Organisation
The positive cases are instructive. SEB's equity capital markets segment, following a psychological safety and perspective-taking intervention with its management teams, finished 2022 approximately 25% over revenue targets. Toyota's Andon cord culture - where workers are obligated, not just permitted, to stop the production line when they detect a problem - produced a quality and rework record that the NUMMI experiment showed was replicable with the same workforce under different conditions.
The common element: a structural change in who has the right and the obligation to surface a problem, regardless of where they sit in the hierarchy. This is a genuine power change.
First: where in your organisation do problems consistently surface at the point of crisis rather than earlier? That pattern tells you where the power structure is suppressing information, and it’s a measurable indicator.
Second: whose voices are genuinely shaping the decisions your most senior teams make? This is distinct from whose voices are present. The Toshiba executives had subordinates. The information those subordinates held did not reach the decisions being made. Presence is not contribution.
Third: does the regional or cultural context change what's required? The evidence says yes - but not in the direction most assume. High power distance raises the bar for what leaders must do to make speaking up feel genuinely safe. It does not make it impossible.
Psychological safety is the condition under which an organisation's capability - the knowledge, concern and judgement of the people inside it - reaches the decisions being made. Power determines whether that condition exists. Inclusion determines whether it's comprehensive. Both are commercial variables.
The Bottom Line
The organisations in this post - Boeing, Toshiba, Nissan, Korean Air, Wells Fargo, Volkswagen - paid between US$1.2 billion and US$20 billion for a failure of information flow. In each case the information existed inside the organisation. The power structure determined whether it travelled.
The question worth taking into your next senior leadership review is a simple one: what do the people closest to your biggest current risks know that hasn't yet reached the people making decisions about those risks? And what is it about your organisation's power structure that is making it safer for them to stay quiet than to say so?
SOURCES & FURTHER READING
Detert JR, Burris ER. 'Leadership Behavior and Employee Voice.' Academy of Management Journal 50(4), 2007
Edmondson AC. 'Learning from Mistakes Is Easier Said Than Done.' Journal of Applied Behavioral Science 32(1), 1996
Frazier ML et al. 'Psychological Safety: A Meta-Analytic Review and Extension.' Personnel Psychology 70(1), 2017
Galinsky AD et al. 'Power and Perspectives Not Taken.' Psychological Science 17, 2006
Google re:Work. 'Understand Team Effectiveness'
MIT Sloan Management Review. 'Skills Training Links Psychological Safety to Revenue Growth' (2023)
Nissan External Advisory Committee Report (March 2019)
Shore LM et al. 'Inclusion and Diversity in Work Groups.' Journal of Management 37(4), 2011
Toshiba Third-Party Committee Independent Investigation Report (2015)
US House Transportation Committee. 'Design, Development and Certification of the Boeing 737 MAX' (September 2020)
van Knippenberg D, De Dreu CKW, Homan AC. 'Work Group Diversity and Group Performance.' Journal of Applied Psychology 89(6), 2004
Woolley AW et al. 'Evidence for a Collective Intelligence Factor.' Science 330, 2010
Comments