Psychological Safety and Leadership: Amy Edmondson, Voice, Learning, Trust, and Teams
Updated: 2 days ago
Author: Ukrainian Psychological Hub · Published: September 25, 2026 · Editorial Policy
Psychological safety in leadership is the team-level belief that people can take interpersonal risks—ask questions, admit mistakes, disagree, request help, raise concerns, and offer unfinished ideas—without expecting humiliation, rejection, or punishment for the act of speaking. Leadership matters because leaders often control attention, evaluation, access, rewards, information, and the social meaning of mistakes. Yet psychological safety is not a leadership style, not a theory of leadership effectiveness, and not a synonym for trust, engagement, inclusion, job satisfaction, or a generally pleasant workplace.
Amy Edmondson’s foundational team research defined psychological safety as a shared belief that a team is safe for interpersonal risk taking and linked it to learning behavior in work teams. In her 1999 multimethod study of 51 manufacturing teams, psychologically safer teams reported more learning behavior, and learning behavior was related to performance. The study also placed team leader coaching and contextual support among proposed antecedents. The evidence was influential because it made interpersonal risk measurable at the team level, but it was a field study rather than a randomized experiment, so its associations should not be read as proof that one leadership behavior mechanically causes better performance (Edmondson, 1999).
Two later evidence syntheses define the modern research landscape. A meta-analysis of 136 independent samples representing more than 22,000 individuals and nearly 5,000 groups mapped antecedents and outcomes of psychological safety and showed that the construct adds information beyond several neighboring positive-workplace variables (Frazier et al., 2017). A systematic review likewise found a rapidly expanding literature but emphasized persistent problems of measurement, levels of analysis, and causal inference (Newman et al., 2017). The practical message is strong enough to act on and precise enough to constrain overclaiming: leadership can shape the interpersonal risk environment, but psychological safety is an emergent property of a social system rather than a personality feature of a leader.
This article focuses on the evidence behind that claim: what psychological safety means, how leadership enters the mechanism, why voice and learning matter, how trust and inclusion differ from psychological safety, what validated measures actually assess, which findings are correlational or longitudinal, where causal evidence exists, and what leaders can reasonably do without turning a research construct into corporate folklore.
What Is Psychological Safety?
Psychological safety is a perception about the interpersonal consequences of action. The core question is not whether work is easy, agreeable, emotionally comfortable, or free from accountability. It is whether a person expects that a socially risky contribution—such as saying “I may be wrong,” “I do not understand,” “I think this plan will fail,” or “I made a mistake”—can be made without disproportionate interpersonal cost. Edmondson and Lei’s major historical review describes psychological safety as central to understanding voice, teamwork, and organizational learning under uncertainty (Edmondson & Lei, 2014).
The intellectual history predates the 1999 team construct. William Kahn’s 1990 qualitative work on personal engagement and disengagement identified psychological safety as one of the conditions influencing whether people could bring more of themselves into work roles. Kahn’s use of the idea concerned people feeling able to show and employ themselves without fear of negative consequences to self-image, status, or career (Kahn, 1990). Edmondson subsequently operationalized psychological safety at the team level, shifting the construct toward a shared climate for interpersonal risk taking.
That team-level emphasis is important. An employee can be personally confident yet work in a team where questions are punished. Another can be anxious by temperament yet accurately perceive that colleagues respond constructively to uncertainty and error. Psychological safety therefore should not be treated as resilience, self-esteem, courage, or emotional stability. Those individual differences may affect how people experience and act within a climate, but they are not the climate itself.
Psychological safety also does not mean that every contribution is good, every decision is consensual, every mistake is consequence-free, or every disagreement must be resolved gently. A psychologically safe team can have demanding standards, strong debate, formal authority, performance evaluation, and serious accountability. The defining issue is whether interpersonal risk is attached to useful participation itself. A team can reject an idea while preserving the contributor’s ability to keep contributing.
Why Leadership Matters to Psychological Safety
Leadership enters psychological safety because leaders are unusually visible sources of social information. Team members watch what leaders reward, ignore, ridicule, investigate, forgive, and punish. They also watch what happens to other people who question assumptions, report errors, ask for help, or challenge higher-status members. These reactions teach the local rules of interpersonal risk.
Leadership, however, should be specified as behavior, relationship, role, or process rather than inferred from the outcome. Psychological safety is not evidence that a leader is globally “good,” just as low psychological safety does not by itself identify one leadership style or one personality profile. A September 2026 conceptual commentary in Industrial and Organizational Psychology makes this point explicitly: psychological safety and trust are important potential outcomes of leadership, but folding them into the definition of effective leadership risks circular reasoning and obscures temporal order and causality (Miller, 2026).
This distinction fits the broader architecture of leadership psychology. Leadership research separates leader traits, leader behavior, leader–follower relationships, formal authority, leader emergence, and leadership effectiveness. Psychological safety belongs principally to the relational and team-process side of that map. It can mediate or moderate relationships between leadership and outcomes, but it is not a substitute measure for leadership itself.
Formal hierarchy makes the leadership connection especially consequential. Leaders and supervisors frequently have authority over schedules, assignments, access to information, performance ratings, promotion opportunities, and disciplinary processes. Even when they do not exercise these powers coercively, subordinates know that the asymmetry exists. That is why the psychology of speaking up is inseparable from power dynamics at work: the perceived cost of voice depends partly on who can impose consequences and whose contribution carries status.
Amy Edmondson’s 1999 Study: What It Established and What It Did Not
Edmondson’s 1999 Administrative Science Quarterly article remains the foundational empirical study of team psychological safety. It examined 51 work teams in a manufacturing company using multiple methods and tested a model connecting team psychological safety, team efficacy, learning behavior, and performance. Team psychological safety was associated with learning behavior; team efficacy was not associated with learning behavior once psychological safety was controlled; and learning behavior statistically mediated the relationship between psychological safety and performance (Edmondson, 1999).
The conceptual contribution was larger than any single coefficient. The study made a counterintuitive problem visible: teams that detect and discuss more mistakes can look worse if error reporting is mistaken for error production. A team where people conceal uncertainty may appear orderly while learning slowly. A team where people expose problems may generate more observable bad news while improving more rapidly. Psychological safety helps explain why information about errors, uncertainty, and disagreement either enters the collective learning process or remains hidden.
The study did not establish a universal formula in which psychological safety automatically causes superior performance. The sample was finite, the organizational context was specific, and several variables were measured through perceptions. Mediation in observational data does not by itself prove causal sequence. Later research has broadened the construct across industries and methods, yet the same discipline remains necessary: correlation, longitudinal prediction, mediation, intervention, and randomized assignment answer different questions.
What the Meta-Analytic Evidence Shows
Frazier and colleagues’ 2017 meta-analysis is the central quantitative synthesis. It aggregated 136 independent samples with more than 22,000 people and nearly 5,000 groups, examining antecedents, outcomes, moderators, and incremental validity. The results supported a broad network in which psychological safety was meaningfully related to leadership and work-design antecedents and to outcomes such as task performance and organizational citizenship behavior. The study also tested whether psychological safety contributed beyond related constructs rather than simply duplicating them (Frazier et al., 2017).
For leadership interpretation, the meta-analysis supports three conclusions. First, leaders matter, but they are one part of a wider system that includes peers, work design, and organizational context. Second, psychological safety has empirical distinctiveness even though it correlates with other positive workplace experiences. Third, average associations do not imply identical effects across teams, occupations, cultures, or outcomes. A meta-analytic relationship is a population-level summary, not a guarantee for any particular manager or intervention.
Edmondson and Bransby’s 2023 Annual Review places leadership among four dominant themes in the mature literature, alongside getting things done, learning behaviors, and work experience. Their synthesis portrays psychological safety as particularly relevant in complex and interdependent work where uncertainty must be surfaced rather than hidden (Edmondson & Bransby, 2023). This helps explain why the construct appears repeatedly in research on healthcare, innovation, interdisciplinary teams, and other environments where information is distributed across people.
Psychological Safety, Voice, and Speaking Up
Voice is a behavior; psychological safety is a belief or climate that can make voice more or less likely. This distinction matters because an employee may believe it is safe to speak yet have nothing useful to add, or may speak despite perceiving substantial risk because the issue is morally urgent. Conversely, a silent employee may be strategically withholding information, disengaged, uncertain, or simply unmotivated. Silence cannot be reverse-scored into a valid psychological safety measure.
Detert and Burris studied 3,149 employees and 223 managers in a restaurant chain and compared transformational leadership with managerial openness as predictors of improvement-oriented voice. Managerial openness was more consistently related to voice, and psychological safety statistically mediated the openness–voice relationship (Detert & Burris, 2007). The study is valuable because it moves from a broad leadership label toward a behaviorally relevant mechanism: whether the manager appears genuinely open to input.
A later meta-analysis of employee voice shows why “speaking up” also needs precision. Promotive voice proposes ideas for improving the status quo; prohibitive voice raises concerns about harmful or problematic practices. Their antecedent patterns are not identical, and psychological safety shows different relationships across forms of voice. Leader–member exchange, leader openness, ethical leadership, engagement, and workplace climate all appear in the larger nomological network (Chamberlin et al., 2017). A leader who welcomes creative suggestions may still react defensively to warnings, and those are psychologically different environments.
Psychological Safety, Learning, and Failure
Learning behavior includes asking for feedback, discussing errors, experimenting, seeking help, and reflecting on how work can improve. These actions are informationally valuable and socially risky. A question can expose ignorance. A request for help can expose dependence. An error report can expose responsibility. A challenge can threaten status. Psychological safety is therefore especially relevant where learning requires making uncertainty public.
Hirak and colleagues tested a longitudinal model in a hospital. Leader inclusiveness was positively related to members’ psychological safety perceptions at an earlier time point; unit psychological safety was associated with later learning from failures; and learning from failures was associated with subsequent unit performance (Hirak et al., 2012). The time-separated design strengthens temporal interpretation relative to a single cross-sectional survey, while still stopping short of a clean randomized causal test of the whole chain.
This mechanism also clarifies why psychological safety should not be marketed as a mood-management technique. Its organizational value is most plausible when the team actually needs information that is difficult to surface. A group doing uncertain, interdependent work has more to gain from rapid error detection and dissent than a group doing simple independent tasks with little informational interdependence. Context changes the value of the mechanism.
Hierarchy, Status, and Leader Inclusiveness
Interpersonal risk is unevenly distributed. A high-status specialist, a new employee, a temporary worker, and a junior professional do not face the same social cost when contradicting a powerful person. Psychological safety research therefore intersects with status and hierarchy rather than existing outside them.
Nembhard and Edmondson studied 23 neonatal intensive care units and introduced leader inclusiveness as words and deeds that invite and appreciate others’ contributions. Their results linked leader inclusiveness to psychological safety and examined how professional status shaped improvement work in cross-disciplinary healthcare teams (Nembhard & Edmondson, 2006). The study does not imply that one “inclusive leadership style” universally solves hierarchy. It identifies a behaviorally specific way leaders can alter the perceived interpersonal cost of contribution in status-differentiated teams.
That distinction matters because inclusion and psychological safety overlap without being identical. Inclusion usually concerns belonging, value, participation, access, and the experience of being recognized as a legitimate group member. Psychological safety concerns the expected consequences of interpersonal risk. A person may feel included socially yet still believe that questioning a senior expert is dangerous. Another may feel safe to disagree on technical matters while still experiencing exclusion from informal networks or advancement opportunities.
Which Leadership Behaviors Have the Strongest Mechanistic Case?
Inviting input and showing genuine openness
Leaders lower interpersonal risk when they explicitly invite information they may not already have and when their subsequent response shows that the invitation was genuine. Asking “What are we missing?” is more diagnostic than asking “Everyone agrees, right?” because it makes dissent task-relevant. Yet invitations work only when people observe what happens after someone accepts them. A leader who requests candor and then humiliates the first dissenter teaches the team that the invitation itself is unsafe.
Responding productively to bad news
The leader’s response to errors, uncertainty, disagreement, and failed experiments is a repeated social signal. Productive responses distinguish blame from diagnosis: they investigate what happened, identify responsibility where appropriate, and preserve the informational value of reporting. Psychological safety does not require the elimination of consequences for negligence or misconduct. It requires avoiding indiscriminate interpersonal punishment for surfacing information the team needs.
Making leader fallibility visible
Leaders can also alter risk perceptions by demonstrating that status does not require a performance of omniscience. Coutifaris and Grant studied leader feedback-seeking and feedback-sharing across three studies. In a longitudinal field experiment, randomly assigning leaders to share criticism they had received produced higher team psychological safety one year later, whereas assigning leaders to seek feedback did not produce the same enduring effect (Coutifaris & Grant, 2021). This is unusually valuable evidence because random assignment strengthens causal inference about a specific leader behavior.
The mechanism appears to be more than theatrical vulnerability. Sharing criticism can normalize the idea that feedback survives contact with authority and can create an obligation to respond rather than merely solicit. The broader principle is that leader behavior should reduce uncertainty about the consequences of candor, not simply advertise that the leader values candor.
Protecting dissent when hierarchy is salient
In high-power-distance or strongly professionalized environments, a general invitation may be insufficient because the downside of dissent is asymmetric. Leaders may need structural practices that distribute airtime, solicit lower-status expertise before senior opinions anchor discussion, separate problem reporting from immediate blame, and create multiple channels for raising concerns. These practices are not replacements for psychological safety; they change the social and procedural conditions from which psychological safety can emerge.
Following through
Voice without response can become a training program in futility. Employees learn not only whether they can speak but whether speaking changes information flow, decisions, or investigation. A leader need not accept every suggestion. A credible response can be “we considered this and chose a different path because…” The important signal is that contribution enters a legitimate process rather than disappearing into silence or triggering retaliation.
Psychological Safety Is Not the Same as Trust
Trust and psychological safety both involve vulnerability, so they are often confused. Trust is usually framed as a willingness to accept vulnerability based on positive expectations about another party. Psychological safety is more specifically a belief about the interpersonal consequences of taking risks in a social setting, often studied as a shared team property. A person can trust a colleague’s competence with confidential information while still believing that team meetings punish disagreement. Conversely, a team may have a norm that permits questions even among members who have not yet built deep interpersonal trust.
Reviews of the psychological safety literature treat trust as a neighboring but distinct construct rather than an interchangeable label (Edmondson & Lei, 2014; Newman et al., 2017). This distinction is practically important because “build trust” and “make it safer to challenge an assumption in Monday’s review meeting” are different intervention targets. One is broad and relational; the other can be specified behaviorally and contextually.
Psychological Safety Is Not Engagement, Job Satisfaction, Inclusion, or General Climate
Engagement concerns energetic psychological investment in work. Job satisfaction is an evaluation of one’s job. Inclusion concerns belonging, value, access, and participation. General workplace climate can include many dimensions, from support and fairness to innovation and safety. Psychological safety is narrower: it concerns interpersonal risk. A team can enjoy high satisfaction because pay and workload are favorable while still suppressing dissent. An intensely engaged team can work very hard while hiding mistakes. A diverse team can have formal inclusion policies while junior members remain unwilling to challenge senior experts.
The empirical reason to keep these constructs separate is not terminological neatness. If measures collapse several positive experiences into one halo, leadership effects can look stronger than they are and mechanisms become impossible to identify. Frazier and colleagues explicitly examined incremental validity beyond related variables, while Newman and colleagues identified measurement and construct-clarity problems as continuing research priorities (Frazier et al., 2017; Newman et al., 2017).
Psychological Safety Is Not a Leadership Style or Leadership Theory
Search results often package psychological safety as if it were a “psychological safety leadership style.” That phrasing is misleading. A leadership style is a recurring pattern of leader behavior, and a leadership theory specifies relationships among constructs that explain leadership processes or outcomes. Psychological safety is a social-perceptual construct. It may be influenced by transformational behavior, leader inclusiveness, managerial openness, leader–member exchange, ethical leadership, humble behavior, or other leader actions, but it should not be reclassified as a style merely because leaders influence it.
This also prevents cannibalization between research questions. The article What Makes a Good Leader? asks about leadership effectiveness across traits, behavior, relationships, and outcomes. This article asks a narrower question: how leadership relates to the interpersonal-risk climate that affects voice and learning. High psychological safety can be one desirable outcome without becoming the definition of good leadership.
Does Psychological Safety Cause Better Performance?
The evidence supports a relationship between psychological safety and several valuable processes and outcomes, but “psychological safety causes performance” is too broad. Performance is not one thing. It can mean task output, learning, innovation, error detection, adaptation, quality, safety, or supervisor ratings. Psychological safety may matter more for some outcomes and tasks than others, especially when work depends on information sharing and learning.
Meta-analytic evidence links psychological safety with task performance and citizenship behavior, while the original Edmondson study and later longitudinal work support learning pathways (Frazier et al., 2017; Edmondson, 1999; Hirak et al., 2012). These findings support prediction and plausible mechanisms. They do not justify the claim that raising a survey score by a fixed amount will reliably raise productivity by a fixed amount.
Causal evidence is growing but narrower. The Coutifaris and Grant field experiment demonstrates that a specific leader feedback-sharing intervention can affect later team psychological safety (Coutifaris & Grant, 2021). A systematic review of healthcare interventions found 14 interventions targeting psychological safety, speaking up, or voice, but results were mixed and often limited by self-report measures and the difficulty of changing entrenched speaking-up patterns through education alone (O’Donovan & McAuliffe, 2020). This is a more realistic intervention picture than a universal checklist.
A 2026 systematic review of randomized and quasi-experimental interprofessional simulation studies in healthcare reported improvements across teamwork and communication outcomes and some psychological-safety-related outcomes, while also finding mixed effects for several other experiences. The evidence is useful but domain-specific and should not be generalized uncritically to every workplace (Sarıköse et al., 2026).
How Psychological Safety Is Measured
The most widely used team measure comes from Edmondson’s seven-item psychological safety scale. It asks respondents to evaluate the interpersonal-risk climate of a specific team, including reactions to mistakes, the safety of taking risks, acceptance of difference, ability to ask for help, and whether members’ skills are recognized. Researchers commonly aggregate individual responses to represent a team-level climate only when aggregation is conceptually and statistically defensible.
A scale score is not a diagnosis. Psychological safety is not a mental disorder, clinical syndrome, or personality type. There is no clinically meaningful cutoff at which a team becomes “psychologically safe” in the medical sense. Scores depend on the measure, wording, reference group, response scale, sampling, timing, and context.
Measurement also carries unresolved psychometric questions. Plouffe and colleagues note that Edmondson’s seven-item scale dominates the literature while aspects of its original factor-analytic reporting left room for improvement; they developed and validated a broader Psychological Safety Inventory to address measurement needs (Plouffe et al., 2023). This does not invalidate the classic scale. It means researchers and organizations should know exactly which instrument they are using and what level of analysis it supports.
For organizations, repeated measurement is usually more informative than treating one pulse survey as a final verdict. Scores can be compared across teams or time only when administration is sufficiently consistent and when confidentiality is credible. If employees believe a supposedly anonymous survey can be traced back to them, the measurement process itself changes the interpersonal-risk environment it is attempting to observe.
Common Measurement and Research Problems
Common-method inflation
Many leadership studies ask the same employee, in the same survey, to rate the leader and the team climate. Shared rater perspective, mood, and general positive or negative evaluation can inflate associations. Stronger designs separate sources, use time lags, include behavioral measures, or manipulate a leader behavior experimentally.
Cross-sectional mediation
A mediation model can be statistically consistent with a theory without proving its causal sequence. If leadership ratings, psychological safety, and outcomes are all collected at one moment, alternative temporal orders remain possible. Teams that are performing well may rate both their leaders and climate more positively. Longitudinal and experimental designs reduce this ambiguity but do not erase every confound.
Aggregation problems
Team psychological safety is theoretically shared, but researchers often begin with individual responses. Treating an individual perception as if it were automatically a team climate, or averaging a team despite strong internal disagreement, can distort the construct. The level of theory, measurement, and analysis must align.
Outcome substitution
A leader should not be classified as effective merely because followers report high psychological safety. The 2026 conceptual critique by Miller is useful precisely because it separates leadership from downstream relational states. Psychological safety can be valuable while leaving unanswered whether the team makes sound decisions, meets its obligations, behaves ethically, develops expertise, adapts, or performs over time (Miller, 2026).
Can a Team Have High Psychological Safety and Poor Leadership?
Yes. Psychological safety is neither necessary nor sufficient as a complete definition of leadership effectiveness. A team can feel free to speak while suffering from weak strategy, poor technical judgment, vague priorities, resource failures, or slow execution. A leader can be receptive and humane yet lack the expertise or decision quality required for the role. Conversely, a leader can produce strong short-term output in a low-safety environment by relying on fear, concentrated expertise, or external pressure, even if learning and voice deteriorate.
The correct evaluation therefore uses multiple criteria. Ask whether people can surface information, whether decisions improve when they do, whether tasks are completed, whether ethical and safety obligations are met, whether expertise develops, whether errors recur, and whether performance is sustainable. Psychological safety supplies information about one crucial part of the system: the interpersonal cost of contributing information.
Can Psychological Safety Be Too High?
The phrase “too much psychological safety” can obscure the real problem. Safety for interpersonal risk does not logically require low standards, endless discussion, lack of hierarchy, or freedom from consequences. If a team becomes complacent, avoids accountability, tolerates poor performance, or treats every opinion as equally valid, those are failures of goals, norms, decision rules, expertise, or accountability. They should be measured directly rather than relabeled as excessive psychological safety.
At the same time, high safety should not be assumed to solve coordination automatically. A team needs clarity about goals, roles, authority, evidence, and decision rights. The strongest practical model is therefore not “safety instead of accountability.” It is an environment where people can raise relevant information and where the organization can still decide, prioritize, evaluate, and act.
What Leaders Can Do: Evidence-Informed Practices
Start by making uncertainty discussable. Leaders can state what is known, what remains uncertain, and where they need information from others. This reduces the status pressure to perform certainty and creates a legitimate role for questions. In complex work, saying “I need the people closest to this process to tell me what I cannot see” is not ceremonial humility; it clarifies an information dependency.
Ask questions that create space for disconfirming evidence. “What could make this fail?”, “What is the strongest argument against our current plan?”, “Where are we relying on an assumption?”, and “Who sees a risk we have not named?” signal that dissent serves the task. The mechanism is stronger when leaders wait for answers rather than filling silence themselves.
Respond to the first difficult contribution carefully. Team norms often crystallize around visible episodes. When someone reports a mistake, contradicts a senior person, or raises a weakly formed concern, the leader’s reaction becomes evidence for everyone watching. Curiosity, clarification, and proportionate accountability preserve information flow. Ridicule, defensiveness, or retaliation suppress it.
Close the loop on voice. A contribution can be rejected without invalidating the contributor. Explain what was considered, what decision was made, and why. Where action is delayed, state what would trigger reconsideration. Where a concern reveals a real problem, acknowledge the contribution publicly without turning the person into a symbolic mascot for “speaking up.”
Make feedback to leaders observable. The strongest causal evidence in this literature comes from a specific form of leader behavior: sharing criticism leaders themselves have received. In the Coutifaris and Grant field experiment, feedback-sharing had a lasting positive effect on psychological safety, whereas merely asking for feedback did not (Coutifaris & Grant, 2021). The lesson is behavioral: leaders demonstrate whether criticism can survive contact with power.
Design for hierarchy rather than pretending hierarchy is absent. In status-differentiated teams, solicit junior or low-status expertise before high-status opinions dominate; use structured rounds when necessary; establish routes for escalating concerns; and separate factual investigation from immediate identity threat. Nembhard and Edmondson’s work on leader inclusiveness illustrates why leader behavior is especially consequential when professional status differences are salient (Nembhard & Edmondson, 2006).
Treat psychological safety as a team condition, not a leader branding exercise. Peers can punish dissent even when the formal leader is open. Workload can make questions costly. Incentive systems can reward silence. Cross-team dependencies can expose employees to retaliation outside the immediate group. Leadership matters most when it aligns local behavior with structures that make candid information usable.
Why One Workshop Is Usually Not Enough
Psychological safety is learned from repeated consequences, so a single workshop competes with the accumulated history of what actually happens in the organization. If employees attend a seminar on speaking up and then watch a colleague lose opportunities after challenging a decision, the lived contingency will dominate the slide deck.
The intervention literature supports this caution. O’Donovan and McAuliffe’s systematic review found mixed results across healthcare interventions and concluded that longitudinal, multifaceted approaches and more objective measures are needed. The review also emphasized visible leader support and interventions that operate at group and organizational levels rather than relying on education alone (O’Donovan & McAuliffe, 2020).
A stronger implementation strategy therefore combines leader behavior, response norms, meeting design, escalation channels, accountability rules, and repeated measurement. The target is not to persuade employees to report feeling safe. It is to change the contingencies from which safety judgments are formed.
What Recent Research Adds in 2026
Recent work reinforces both the importance of leadership and the need for methodological restraint. A 2026 qualitative study based on 11 employees from diverse Slovak workplaces explored how leadership behaviors shaped perceptions of psychological safety and voice. Participants described relational dynamics, trust, leader responses, and barriers to speaking up, offering context-rich evidence but not population-level causal estimates (Konečná et al., 2026).
The same year, a new conceptual commentary argued that leadership research should resist defining effective leadership through psychological safety and trust because doing so collapses antecedents and outcomes and can create circular measurement (Miller, 2026). These two contributions point in the same useful direction from different methods: leadership is central to how interpersonal risk is experienced, while psychological safety remains a distinct construct that must be measured and explained rather than used as a catch-all label for good leadership.
Psychological Safety and Leadership: A Practical Evidence Map
Established evidence: psychological safety is a well-developed organizational construct with substantial observational, longitudinal, and meta-analytic evidence. It is associated with learning behavior, voice, performance-related outcomes, and several leadership or relationship variables. The 2017 meta-analysis and the 2014 and 2023 Annual Review articles provide the strongest broad syntheses (Frazier et al., 2017; Edmondson & Lei, 2014; Edmondson & Bransby, 2023).
Stronger causal evidence for specific behaviors: randomized field evidence indicates that leader feedback-sharing can increase later team psychological safety (Coutifaris & Grant, 2021). This supports a causal claim about that intervention in the studied setting, not about every behavior marketed as “psychological safety leadership.”
Longitudinal evidence: time-separated hospital research links leader inclusiveness, psychological safety, learning from failures, and later performance in a theoretically coherent sequence (Hirak et al., 2012). Longitudinal ordering strengthens inference but remains open to unmeasured confounding.
Preliminary or context-bounded evidence: qualitative studies, single-industry studies, and healthcare intervention reviews can identify mechanisms and feasible practices while requiring care before generalization. The 2026 Slovak qualitative study and healthcare-specific intervention reviews belong here (Konečná et al., 2026; Sarıköse et al., 2026).
Contested or popularized claims: universal stage models, fixed “best practice” recipes, claims that psychological safety automatically maximizes performance, and the idea that a high score proves a leader is effective go beyond the evidence. These claims may be useful hypotheses or practitioner frameworks, but they should not be presented as conclusions of the core research literature.
FAQ
What is psychological safety in leadership?
It is the perceived team climate for interpersonal risk taking as it relates to leadership and leader behavior. Leaders can influence that climate through how they invite input, respond to mistakes and disagreement, distribute attention, use authority, and react to criticism. Psychological safety remains a team or interpersonal construct rather than a leadership style.
How can a leader create psychological safety?
Evidence supports practices that reduce uncertainty about the consequences of candor: invite input, show openness, respond constructively to bad news, make fallibility and feedback visible, protect lower-status voice, and follow through on concerns. The best causal evidence is narrower than the full advice literature, so these practices should be implemented as testable changes rather than guaranteed formulas.
Is psychological safety the same as trust?
No. Both involve vulnerability, but trust usually concerns positive expectations about another party, whereas psychological safety concerns the interpersonal consequences of risk taking in a social setting and is often studied as a shared team climate. They correlate and can influence one another, but they should be measured separately.
What are the four stages of psychological safety?
The widely circulated four-stage model—Inclusion Safety, Learner Safety, Contributor Safety, and Challenger Safety—comes from Timothy R. Clark’s practitioner framework, first published as a book in 2020 and revised in 2026 (Clark, 2026). It is not Amy Edmondson’s original operationalization and should not be treated as if the core research literature had established a universal four-stage developmental sequence.
How do you measure psychological safety?
Research commonly uses Edmondson’s seven-item team psychological safety scale, with careful attention to the level of analysis and whether individual responses can legitimately be aggregated. Newer instruments, including the Psychological Safety Inventory, have been developed to address psychometric and conceptual questions. A workplace survey score is not a clinical diagnosis and has no universal medical cutoff.
Does psychological safety improve performance?
It is associated with performance-related outcomes and with learning and voice processes that can improve performance, especially in complex and interdependent work. The strength and pathway depend on context, outcome definition, measurement, and study design. Meta-analysis supports meaningful associations; it does not establish a universal productivity law.
Does psychological safety mean avoiding accountability?
No. Safety for speaking, questioning, admitting error, and raising concerns can coexist with demanding standards and proportionate consequences. Accountability concerns obligations and performance; psychological safety concerns the interpersonal cost of contributing information. Confusing the two makes both constructs less useful.
Can a psychologically safe team still make bad decisions?
Yes. Psychological safety improves the conditions for information to enter discussion; it does not guarantee expertise, evidence quality, strategy, or decision discipline. A team can speak freely and still reason poorly. Effective leadership requires additional capabilities and outcomes beyond psychological safety.
Is psychological safety an individual mental-health condition?
No. In organizational research it is a social and team-level construct about interpersonal risk. It should not be used as a psychiatric label, and low workplace psychological safety should not be converted into a diagnosis of an employee, manager, or team.
Related Articles
References
Chamberlin, M., Newton, D. W., & LePine, J. A. (2017). A meta-analysis of voice and its promotive and prohibitive forms: Identification of key associations, distinctions, and future research directions. Personnel Psychology, 70(1), 11–71. https://doi.org/10.1111/peps.12185
Clark, T. R. (2026). The 4 Stages of Psychological Safety, Second Edition. Berrett-Koehler Publishers. Publisher page
Coutifaris, C. G. V., & Grant, A. M. (2021). Taking your team behind the curtain: The effects of leader feedback-sharing and feedback-seeking on team psychological safety. Organization Science, 33(4), 1574–1598. https://doi.org/10.1287/orsc.2021.1498
Detert, J. R., & Burris, E. R. (2007). Leadership behavior and employee voice: Is the door really open? Academy of Management Journal, 50(4), 869–884. https://doi.org/10.5465/amj.2007.26279183
Edmondson, A. C. (1999). Psychological safety and learning behavior in work teams. Administrative Science Quarterly, 44(2), 350–383. https://doi.org/10.2307/2666999
Edmondson, A. C., & Bransby, D. P. (2023). Psychological safety comes of age: Observed themes in an established literature. Annual Review of Organizational Psychology and Organizational Behavior, 10, 55–78. https://doi.org/10.1146/annurev-orgpsych-120920-055217
Edmondson, A. C., & Lei, Z. (2014). Psychological safety: The history, renaissance, and future of an interpersonal construct. Annual Review of Organizational Psychology and Organizational Behavior, 1, 23–43. https://doi.org/10.1146/annurev-orgpsych-031413-091305
Frazier, M. L., Fainshmidt, S., Klinger, R. L., Pezeshkan, A., & Vracheva, V. (2017). Psychological safety: A meta-analytic review and extension. Personnel Psychology, 70(1), 113–165. https://doi.org/10.1111/peps.12183
Hirak, R., Peng, A. C., Carmeli, A., & Schaubroeck, J. M. (2012). Linking leader inclusiveness to work unit performance: The importance of psychological safety and learning from failures. The Leadership Quarterly, 23(1), 107–117. https://doi.org/10.1016/j.leaqua.2011.11.009
Kahn, W. A. (1990). Psychological conditions of personal engagement and disengagement at work. Academy of Management Journal, 33(4), 692–724. https://doi.org/10.2307/256287
Konečná, L., Lisá, E., & Čiriková, V. (2026). The role of leadership in shaping psychological safety: A qualitative study from Slovakia. Scientific Reports, 16, 7249. https://doi.org/10.1038/s41598-026-38706-1
Miller, C. F. (2026). But what is good leadership? A conceptual critique of defining leadership through psychological safety and trust. Industrial and Organizational Psychology, 19(3), 281–287. https://doi.org/10.1017/iop.2026.10097
Nembhard, I. M., & Edmondson, A. C. (2006). Making it safe: The effects of leader inclusiveness and professional status on psychological safety and improvement efforts in health care teams. Journal of Organizational Behavior, 27(7), 941–966. https://doi.org/10.1002/job.413
Newman, A., Donohue, R., & Eva, N. (2017). Psychological safety: A systematic review of the literature. Human Resource Management Review, 27(3), 521–535. https://doi.org/10.1016/j.hrmr.2017.01.001
O’Donovan, R., & McAuliffe, E. (2020). A systematic review exploring the content and outcomes of interventions to improve psychological safety, speaking up and voice behaviour. BMC Health Services Research, 20, 101. https://doi.org/10.1186/s12913-020-4931-2
Plouffe, R. A., Ein, N., Liu, J. J. W., St. Cyr, K., Baker, C., Nazarov, A., & Don Richardson, J. (2023). Feeling safe at work: Development and validation of the Psychological Safety Inventory. International Journal of Selection and Assessment, 31(3), 443–455. https://doi.org/10.1111/ijsa.12434
Sarıköse, S., Sengul, T., Lopez, V., & Kirkland-Kyhn, H. (2026). Fostering healthy work environments through interprofessional simulation: Teamwork, communication, psychological safety—Systematic review of randomised controlled and quasi-experimental studies. Journal of Advanced Nursing. https://doi.org/10.1111/jan.70564