Build Team Trust Before You Need It

Executive summary

Team trust is the willingness to accept vulnerability to colleagues because their competence, intentions, and conduct are sufficiently dependable for the situation. It is not warmth, agreement, obedience, or the absence of control. Trust becomes operational when people can predict how work will be coordinated, surface bad news without avoidable retaliation, depend on realistic commitments, challenge reasoning, and repair breaches.

Trust matters most before pressure arrives. Under deadline, uncertainty, or conflict, people rarely invent a new relationship system; they rely on patterns already learned. If everyday meetings reward silence, the crisis meeting will not produce candor because a leader asks for it. If missed commitments routinely disappear, a new dashboard will not create reliability. If apologies protect status rather than repair impact, unresolved friction becomes coordination cost.

Research offers several complementary lenses. Mayer, Davis, and Schoorman distinguish trust from trustworthiness and describe ability, benevolence, and integrity as important antecedents.^[s1] Edmondson’s work defines team psychological safety as a shared belief that interpersonal risk taking is safe and connects it with learning behavior.^[s2] A meta-analysis across 112 independent studies found a positive association between intrateam trust and performance, with effects contingent on factors such as interdependence and authority differentiation.^[s3] None of this means trust guarantees performance. Teams also need competence, resources, task design, standards, and decision clarity.

This lesson turns trust into five loops: clarity, reliability, candor, repair, and learning. It shows how to diagnose the system, install everyday practices, measure without surveillance, and distinguish healthy trust from comfort, coercive harmony, or unbounded vulnerability.

Learning objectives

By the end of this lesson, you will be able to:

• distinguish trust, trustworthiness, psychological safety, cohesion, and accountability;
• diagnose where ambiguity, broken commitments, silence, or unrepaired harm erode coordination;
• install five observable trust loops without demanding personal disclosure;
• lead a breach repair that includes impact, responsibility, remedy, and changed safeguards;
• combine psychological safety with high standards and clear decision rights;
• measure team trust through behavior, outcomes, and protected feedback rather than invasive monitoring.

Foundations: trust is a calibrated willingness to be vulnerable

Trust exists because action involves uncertainty and dependence. If a teammate can neither help nor harm the work, trust is largely irrelevant. If every action is completely observed and enforced, the relationship relies mainly on control. In real teams, people depend on others to share information, exercise judgment, represent their interests, and keep commitments when no procedure covers every case.

Mayer, Davis, and Schoorman’s model separates a trustor’s propensity, perceptions of another party’s ability, benevolence, and integrity, the willingness to accept vulnerability, and subsequent risk taking.^[s1] The separation is practical. A colleague may be honest and caring but lack the skill for a high-risk task. Another may be technically capable but willing to conceal mistakes. Trust should be specific to the domain and stake, not a permanent label attached to the person.

Trust is not the same as psychological safety

Psychological safety describes a shared team belief that speaking up, asking for help, admitting uncertainty, and making other interpersonal risks will not trigger humiliation or punishment.^[s2] It overlaps with trust but is not identical. Trust often concerns whether another person or system will act dependably when you are vulnerable. Psychological safety concerns the interpersonal climate for taking voice and learning risks. A team can feel safe to question ideas yet distrust a colleague’s follow-through. It can rely on a skilled operator yet feel unsafe challenging the operator’s decisions.

Edmondson and Lei’s review traces psychological safety across individual, team, and organizational research and links it with voice, teamwork, and learning, while also emphasizing context and level of analysis.^[s4] Treating it as a personality trait or a promise that nobody will feel discomfort misses the construct. A safe team can still conduct a demanding review, reject a proposal, or apply consequences. The standard is whether people can participate honestly and fairly, not whether every exchange feels pleasant.

Trust is not cohesion, friendship, or agreement

Cohesion can help people remain committed, but close groups can also suppress dissent. Friendship can enrich work, but requiring intimacy disadvantages people with different boundaries, identities, schedules, or communication styles. Agreement may reflect genuine alignment, weak information, status pressure, or conflict avoidance. Observable trust is better tested through questions such as:

• Can people ask for clarification without being treated as incompetent?
• Are commitments realistic, visible, and renegotiated early?
• Can a junior specialist contradict a senior leader on evidence?
• Does the team distinguish an error report from a character judgment?
• After a breach, do remedy and safeguards change?
• Do decisions, handoffs, and review dates remain legible?

Trust and control are complements when designed well^[s9]

The false choice between trust and process creates two bad systems. One relies on goodwill and memory, then personalizes predictable failures. The other responds with surveillance and approval layers that signal suspicion and slow adaptation. Good controls make expectations, authority, risk, and evidence visible. They reduce the amount of blind faith required.

A deployment checklist, peer review, or dual authorization for a sensitive payment can protect trustworthy people and stakeholders. The control becomes corrosive when it is secret, excessive, selectively applied, or used to avoid managerial judgment. Explain why the control exists, what risk it addresses, who can see the data, how exceptions work, and when the control will be reviewed.

Team trust is conditional and dynamic

The 2016 meta-analysis by de Jong, Dirks, and Gillespie found that intrateam trust was positively related to performance on average, and that the relationship varied with task interdependence, authority differentiation, and skill differentiation.^[s3] This is intuitive: when work is tightly interdependent, one person’s concealment or delay propagates. When tasks are mostly independent, trust may still matter for climate and coordination but have a different performance role.

Trust also changes through evidence. Small reliable actions accumulate. One breach can matter greatly when it reveals a different intention or integrity than previously assumed. Repair depends on what was violated. A competence failure may require training, narrower authority, or verification. An integrity violation may require truthful acknowledgment, restitution, and independent safeguards. A betrayal affecting safety or rights may make restoration inappropriate.

The five-loop team trust operating system

The loops reinforce one another. Clarity without candor can produce compliant execution of the wrong plan. Candor without repair exhausts people who repeatedly raise the same issue. Reliability without learning can preserve an obsolete process. The system works when each loop produces evidence for the next.

Loop 1: clarity reduces avoidable suspicion

Ambiguity encourages people to explain behavior through motive. When priorities, authority, and quality are unclear, a missed handoff can look careless, political, or disrespectful even when two reasonable interpretations existed.

For consequential work, clarify:

• the outcome and why it matters;
• the decision owner and consultation mode;
• roles, dependencies, and handoff conditions;
• the definition of done and evidence required;
• constraints, risks, and non-negotiables;
• when to escalate, renegotiate, or stop;
• the next review date.

End a decision meeting with a “read-back.” Each owner states what they believe was decided, what they own, by when, and what could block it. The purpose is not to test memory. It is to expose divergent interpretations while correction is cheap. Link high-stakes choices to the practical decision-making system so evidence, authority, guardrails, and review rules remain explicit.

Clarity is not certainty. Leaders should label what is decided, provisional, unknown, and outside the team’s control. Pretending certainty may produce short-term calm but damages credibility when reality changes.

Loop 2: reliability turns promises into evidence

Reliability is not perfect on-time delivery. It is the practice of making realistic commitments, protecting them, surfacing threats early, and closing the loop. A team that punishes every renegotiation teaches people to hide risk until the deadline. A team that accepts every excuse teaches commitments to mean little.

Use a commitment protocol:

1. The requester explains the outcome, reason, deadline, and flexibility.
2. The owner confirms scope, dependencies, and capacity before agreeing.
3. The commitment is visible in the team’s normal system.
4. The owner signals risk at the earliest credible point, with impact and options.
5. The affected parties explicitly accept a new plan or escalate.
6. Completion is acknowledged; unfinished edges are not hidden.

Measure promise quality, not heroic rescue. If a person regularly works nights to preserve unrealistic promises, the system is borrowing reliability from health and family time. Workload design is part of trust. The energy-aligned week can help leaders stop treating chronic overload as a character test.

Loop 3: candor makes risk visible

Candor is the responsible expression of relevant information, disagreement, uncertainty, and impact. It is not bluntness without care. The leader’s response to the first uncomfortable message is the most credible policy.^[s8]

Build candor into the sequence:

• collect independent judgments before open discussion;
• ask the most senior person to state a view later;
• invite disconfirming evidence and a plausible alternative;
• rotate a red-team role rather than assigning permanent negativity to one person;
• ask “what are we not saying because of who is in the room?”;
• thank the messenger before evaluating the conclusion;
• record dissent and the decision response.

Psychological safety research supports the importance of an interpersonal climate for learning behavior, but a leader should not claim causality from one survey score.^[s2][s4] Observe whether people raise errors, ask questions, and seek help—and whether doing so produces a fair response.

Candor has boundaries. Confidential personal information, legal privilege, security details, and another person’s story should not be disclosed merely in the name of openness. Teach employees how to escalate sensitive concerns through protected channels.

Loop 4: repair turns breach into changed behavior

Trust cannot be built by pretending breaches will disappear. A repair process begins by protecting affected people and stopping ongoing harm. Then it establishes facts, acknowledges impact, assigns responsibility, makes restitution where possible, and changes safeguards.

Kramer and Lewicki’s review distinguishes approaches to reducing organizational trust deficits and emphasizes that repair is a complex process, not a single apology.^[s5] The repair must match the breach. Use six moves:

1. Stabilize: stop the harmful action and secure safety, data, money, or work.
2. Establish facts: separate verified events, interpretations, and unknowns.
3. Acknowledge impact: name effects without debating whether they were intended.
4. Own responsibility: state what the person or system controlled.
5. Repair and remedy: correct the work, compensate harm, restore access, or change the decision.
6. Change safeguards: alter authority, review, training, workload, or controls and set a review date.

An apology that says “I am sorry you felt that way” shifts responsibility to the affected person. An apology that promises “it will never happen again” without a mechanism asks for faith. A credible repair says what happened, why it mattered, what is being done now, and how recurrence will be detected.

Restoration is not always the goal. Serious harassment, retaliation, fraud, violence, or repeated integrity breaches may require formal investigation, separation, external reporting, or legal remedies. Do not pressure harmed people to reconcile for the appearance of team unity.

Loop 5: learning makes vulnerability worthwhile

Speaking up is costly when the system never changes. Learning closes the loop by converting errors, surprises, and successful adaptations into revised practice.

Run short after-action reviews around four questions:

1. What did we expect, and why?
2. What actually happened?
3. What mechanisms and conditions explain the gap?
4. What will we change, protect, test, or stop?

The After Action Review process provides a deeper facilitation method. Keep the review separate from immediate performance discipline where possible. People need to know whether the purpose is learning, accountability, or formal investigation. Blending them without notice makes candor unsafe.

Salas, Sims, and Burke’s teamwork framework identifies team leadership, mutual performance monitoring, backup behavior, adaptability, and team orientation, supported by coordinating mechanisms including shared mental models, closed-loop communication, and mutual trust.^[s6] This reinforces an important point: trust is not a substitute for teamwork skills. It allows those skills to operate under interdependence.

Worked example: recovering from a failed enterprise launch

This is a hypothetical composite case. NilaPay is a 75-person financial-technology company. A new enterprise reporting feature misses its promised launch by six weeks. Sales has already committed the date to two customers. Engineering says product changed requirements. Product says engineering hid complexity. Quality assurance raised an integration risk three weeks earlier, but the concern was discussed in a private chat and never entered the launch record.

Diagnose the trust system

The chief product officer initially calls a “trust reset” and asks everyone to be more transparent. The team’s first useful move is to stop using trust as an accusation. It reconstructs the operating evidence:

• the launch owner was never explicitly named;
• “feature complete” meant code merged to engineering and production-ready to sales;
• product changes were accepted in chat without schedule impact review;
• engineers reported local status, but no one owned cross-system dependency risk;
• quality assurance expected the project manager to escalate;
• the last three status meetings opened with pressure to preserve the date.

There are individual errors, but the pattern is systemic. Clarity was weak, commitments were not renegotiated, candor had no reliable route, and learning from earlier launch misses had not changed the release process.

Stabilize and repair

The launch owner informs customers of the revised range, explains which functions are safe, offers a phased pilot, and provides a weekly update. The company does not blame an unnamed “technical delay.” Internally, the chief product officer acknowledges that repeated insistence on the original date made escalation costly. Product owns the unreviewed scope changes. Engineering owns the failure to translate complexity into schedule impact. Quality assurance is not blamed for failing to bypass an undefined escalation path.

The team changes safeguards: one launch owner, a shared definition of ready, explicit change-control thresholds, a weekly dependency review, and a red-risk rule that requires acknowledgement by product and engineering. A customer-impact threshold triggers executive review. These are not punishment rituals; they reduce ambiguity and dependence on courage.

Rebuild evidence over three cycles

At the next launch, the team uses read-backs and a visible risk log. One engineer flags a data-migration assumption four weeks early. The owner narrows the pilot rather than demanding silent recovery. The feature launches within the revised range. The team then reviews what the new system missed: sales still learned about the scope change too late.

After three launches, on-time performance improves, schedule ranges become better calibrated, and risk entries occur earlier. A pulse survey also improves, but the company does not claim that the meeting caused trust. The stronger evidence is behavioral: risks surface sooner, owners renegotiate before failure, decisions remain visible, and the review changes procedures.

One relationship remains damaged. A product manager repeatedly mocked a quality analyst in meetings. That issue is handled through the formal people process, not absorbed into a general workshop. The distinction protects both fairness and psychological safety.

Action Plan: build trust through 30, 60, and 90 days

Do not launch with a retreat or a promise of total transparency. Start with one real work system where interdependence and cost are visible.

Days 1–30: establish a protected baseline

1. Name the work system. Choose one recurring process: weekly planning, customer escalation, launch readiness, hiring, or shift handoff.
2. Map dependencies. Record who needs what from whom, by when, with what evidence, and how risk is escalated.
3. Collect protected input. Use an anonymous pulse where appropriate, one-to-one interviews, and artifact review. Ask for behavior and examples, not personality labels.
4. Select two behaviors. Choose one start and one stop behavior, such as “risk is raised with impact and options within one day” and “leaders do not rebut concerns before clarifying them.”
5. Clarify protections. Explain confidentiality limits, anti-retaliation routes, data access, and who will see raw comments.
6. Run read-backs. End consequential meetings with owners restating decisions, commitments, and blockers.

Publish a short baseline: what was heard, what will change, what will not change, who owns each action, and when the team will review. Do not publish identifiable quotes without permission.

Days 31–60: strengthen reliability and candor

1. Install the commitment protocol in the selected work system.
2. Review promise risk weekly without public shaming. Ask what changed and what support or trade-off is needed.
3. Collect independent views before two high-stakes decisions.
4. Have the leader name one error, its impact, and the process change—not a vulnerability story that asks employees for reassurance.
5. Train managers to respond to bad news: thank, clarify, assess, protect, decide, and close the loop.
6. Audit workload. Cancel or re-sequence work where chronic overcommitment makes reliability impossible.

At day 60, compare the timing of risk signals, missed handoffs, and renegotiations with the baseline. If people report greater safety but risk still appears late, the practice has not yet changed operating behavior.

Days 61–90: practice repair and institutionalize learning

1. Run an after-action review on a real completed cycle.
2. Select one breach or recurring friction suitable for repair. Do not use a public session for confidential misconduct.
3. Apply the six-move repair process and record changed safeguards.
4. Define decision rights, escalation thresholds, and protected reporting routes.
5. Train a second facilitator so the system does not depend on one leader.
6. Review measures and remove any instrument that encourages surveillance or cannot change a decision.

At day 90, decide what to continue, adapt, or stop. Extend the practices to another work system only after the first produces behavioral evidence. Use the mutual understanding and trust lesson for coaching relationships and the team trust foundations lesson for a shorter introduction.

Measurement: triangulate climate, behavior, and outcomes

Trust is not directly visible. Surveys, behavior, and outcomes are imperfect indicators. Use multiple measures and state their limitations.

Climate measures

Use a short, stable pulse focused on the specific team and work context. Items may address whether people can ask for help, report mistakes, challenge a view, depend on commitments, and expect fair handling. Psychological-safety measures should retain the construct’s interpersonal-risk focus rather than combining every positive workplace sentiment into one score.^[s2][s4]

Report distributions and protected subgroup patterns when sample size and privacy allow. An average can hide that junior staff, contractors, remote employees, or members of a minority group experience the team differently. Do not publish subgroup results that could identify individuals.

Behavioral measures

• time from risk discovery to recorded escalation;
• percentage of commitments explicitly accepted and closed;
• percentage renegotiated before the deadline rather than missed silently;
• questions, dissenting views, and help requests in sampled meetings;
• repair actions completed by the promised date;
• after-action reviews that produce an owned system change;
• recurrence of the same failure mechanism.

Behavior counts need interpretation. A rise in reported errors can indicate deterioration, better detection, or safer reporting. James Reason’s work on human error contrasts person-focused blame with system approaches that examine defenses and conditions.^[s7] Treat reporting as information about the system, not a target to suppress.

Outcome and guardrail measures

Track rework, avoidable delay, quality escapes, customer escalation, unwanted turnover, sick leave patterns where lawfully aggregated, safety incidents, and delivery calibration. These outcomes are affected by many variables; do not attribute improvement to trust practice without stronger design.

Guardrails include retaliation complaints, confidentiality breaches, coercive disclosure, biased discipline, excessive monitoring, and workload transfer. A team can improve delivery by creating fear. That is not a successful trust intervention.

Review cadence and decision rules

Review work-system behavior every two weeks, climate monthly or quarterly depending on team size, and serious safeguards immediately. Set preconditions:

• investigate if a reported concern is followed by adverse treatment;
• escalate if the same high-severity risk is concealed twice;
• redesign if commitments remain unreliable after workload and clarity changes;
• pause a public repair process if it risks identifying or pressuring an affected person;
• seek independent review when leaders implicated in the breach control the investigation.

Measure managers on how they handle bad news and repair commitments, not on producing high survey scores. Otherwise the instrument becomes pressure to report happiness.

Failure modes and corrective actions

1. Trust as a personality verdict

Why it happens: “untrustworthy” is easier than diagnosing a domain and behavior. Warning sign: labels replace evidence. Correction: specify the task, stake, observed behavior, impact, and safeguard needed.

2. Forced vulnerability

Why it happens: leaders imitate intimacy rituals. Warning sign: employees must disclose trauma, identity, or private feelings to demonstrate commitment. Correction: make personal sharing voluntary and build trust through work-relevant clarity, reliability, candor, repair, and learning.

3. Psychological safety without standards

Why it happens: discomfort is mistaken for harm. Warning sign: weak work cannot be challenged. Correction: state demanding standards, critique the work with evidence, and ensure the person can respond without humiliation.

4. Accountability without safety

Why it happens: pressure creates visible compliance. Warning sign: risks appear only after failure. Correction: reward early escalation, collect independent input, and review the leader’s response to dissent.

5. Apology without remedy

Why it happens: words are cheaper than changed authority or restitution. Warning sign: the same breach repeats. Correction: stabilize, establish facts, acknowledge impact, own responsibility, repair, and change safeguards.^[s5]

6. Process as surveillance

Why it happens: leaders want predictive control. Warning sign: private messages, keystrokes, or camera behavior are monitored to infer trust. Correction: collect minimum work-system data, explain purpose and access, use aggregation, and prohibit covert scoring.

7. Harmony mistaken for trust

Why it happens: silence feels efficient. Warning sign: meetings contain no substantive disagreement but decisions are criticized privately. Correction: use independent pre-work, invite alternatives, and record dissent with a response.

8. Heroic reliability

Why it happens: sacrifice protects dates. Warning sign: chronic overtime keeps promise metrics high. Correction: measure workload and planning quality; renegotiate scope before consuming recovery time.

9. Team intervention for structural injustice

Why it happens: workshops feel less threatening than changing power. Warning sign: trust language is used after discriminatory pay, retaliation, or unsafe conditions without remedy. Correction: address policy, authority, investigation, restitution, and protection first; use team practices only as a supplement.

10. Universal trust as the goal

Why it happens: distrust is treated as cultural failure. Warning sign: employees are told to trust despite missing evidence or active risk. Correction: calibrate trust, use verification, and accept that some relationships require boundaries or separation.

Ethics, governance, and limits

Trust work changes power. Leaders can ask for openness while controlling evaluation, pay, work allocation, and access to remedies. Therefore participation must be voluntary where possible, and confidentiality claims must be precise. Tell people who will see raw data, whether comments can be quoted, what must be escalated, how long data are kept, and how retaliation is handled.

Do not use anonymous feedback as unverified proof against an individual. It can identify patterns and trigger fair inquiry, but the person affected needs a defensible process. Likewise, do not expose a reporter in pursuit of “both sides.” In cases involving harassment, discrimination, safety, fraud, or violence, use qualified investigators and applicable legal or regulatory channels.

Culture and identity shape how candor appears. Direct public disagreement may be more costly for junior employees, contractors, caregivers, remote staff, people using a second language, or members of marginalized groups. Offer multiple routes: written pre-work, private escalation, structured turn-taking, asynchronous review, and protected reporting. Do not make one extroverted communication style the standard of trust.

Trust should remain calibrated. High-trust teams still verify critical calculations, protect credentials, separate duties, review medical or safety decisions, and document financial approvals. The relevant question is not “Do we trust them?” but “What vulnerability, evidence, and control are appropriate for this task?”

This lesson is not therapy, legal advice, or a substitute for formal employee relations. Team routines can improve coordination; they cannot diagnose trauma, resolve every interpersonal conflict, or legitimize unsafe employment conditions. When harm is severe or authority is compromised, independent support and structural remedy come before cohesion.

Checklist: prepare, run, and review a team trust intervention

Before starting:

• The work system, team boundary, and business risk are explicit.
• Trust is defined through observable behavior, not personality or closeness.
• The leader’s authority, conflicts, and role in the problem are acknowledged.
• Participation, confidentiality, data access, retention, and escalation limits are explained.
• Formal misconduct, safety, or legal issues are routed appropriately.
• Baseline measures include climate, behavior, outcomes, and guardrails.

For clarity and reliability:

• Outcomes, owners, decision rights, dependencies, definitions of done, and review dates are visible.
• Commitments are accepted rather than assigned by assumption.
• Risks are surfaced with impact and options before the deadline.
• Renegotiation produces an explicit new agreement.
• Chronic overload is treated as a system problem, not loyalty evidence.

For candor and repair:

• Independent views are collected before senior opinions dominate.
• Leaders clarify bad news before rebutting it.
• Dissent and the decision response are recorded.
• Sensitive concerns have protected routes.
• Breaches are stabilized, investigated fairly, acknowledged, remedied, and followed by changed safeguards.
• Affected people are not pressured to forgive or reconcile.

For learning and measurement:

• Reviews separate expected, actual, explanatory, and corrective information.
• At least one owner and date are attached to each learning action.
• A rise in reported errors is interpreted before being judged.
• Survey scores cannot affect individual performance ratings.
• Subgroup privacy and differential experience are protected.
• The intervention continues only if evidence changes the work system.

Practice exercise

Choose one missed commitment from the last month. Describe it without names or motive: expectation, observed event, impact, and unknowns. Map which loop failed first—clarity, reliability, candor, repair, or learning—and which later loops failed to contain the problem. Design one behavior and one system change. For example, “owners signal delivery risk with impact and options within one working day” plus “the planning board records dependency owners and escalation dates.”

Then rehearse a leader response to bad news: “Thank you for raising it. What do we know, what remains uncertain, who is affected, what needs protection now, and when will I close the loop?” Ask a colleague to identify any phrase that minimizes impact, blames the messenger, or promises certainty. Use the revised response in the next genuine situation and review what happened.

Reflection prompts

• Where does the team currently require blind faith because the system is unclear?
• Which commitments are routinely saved by invisible overtime?
• What information is costly to say in front of the most powerful person?
• Which breach needs repair rather than another values statement?
• What did the last retrospective change in the operating system?
• Who experiences the team differently from the average survey result?
• What should remain verified even in a highly trusted team?

Key takeaways

• Trust is a calibrated willingness to accept vulnerability, not a demand for closeness or obedience.
• Psychological safety supports interpersonal risk taking; it does not remove standards or consequences.
• Clarity, reliability, candor, repair, and learning convert trust from a sentiment into an operating system.
• Small commitments and leader responses create evidence before pressure arrives.
• Repair must match the breach and change safeguards; forgiveness cannot be compelled.
• Trust and proportionate controls can reinforce one another.
• Measurement should triangulate climate, behavior, outcomes, and guardrails without surveillance.
• Structural injustice, serious misconduct, and safety threats require formal remedy, not a team-building workshop.

References

[s1] Roger C. Mayer, James H. Davis, and F. David Schoorman. “An Integrative Model of Organizational Trust.” Academy of Management Review 20(3), 1995, pp. 709–734. https://doi.org/10.2307/258792

[s2] Amy C. Edmondson. “Psychological Safety and Learning Behavior in Work Teams.” Administrative Science Quarterly 44(2), 1999, pp. 350–383. https://doi.org/10.2307/2666999

[s3] Bart A. de Jong, Kurt T. Dirks, and Nicole Gillespie. “Trust and Team Performance: A Meta-Analysis of Main Effects, Moderators, and Covariates.” Journal of Applied Psychology 101(8), 2016, pp. 1134–1150. https://doi.org/10.1037/apl0000110

[s4] Amy C. Edmondson and Zhike Lei. “Psychological Safety: The History, Renaissance, and Future of an Interpersonal Construct.” Annual Review of Organizational Psychology and Organizational Behavior 1, 2014, pp. 23–43. https://doi.org/10.1146/annurev-orgpsych-031413-091305

[s5] Roderick M. Kramer and Roy J. Lewicki. “Repairing and Enhancing Trust: Approaches to Reducing Organizational Trust Deficits.” Academy of Management Annals 4(1), 2010, pp. 245–277. https://doi.org/10.5465/19416520.2010.487403

[s6] Eduardo Salas, Dana E. Sims, and C. Shawn Burke. “Is There a ‘Big Five’ in Teamwork?” Small Group Research 36(5), 2005, pp. 555–599. https://doi.org/10.1177/1046496405277134

[s7] James Reason. “Human Error: Models and Management.” BMJ 320(7237), 2000, pp. 768–770. https://doi.org/10.1136/bmj.320.7237.768

[s8] Kurt T. Dirks and Donald L. Ferrin. “Trust in Leadership: Meta-Analytic Findings and Implications for Research and Practice.” Journal of Applied Psychology 87(4), 2002, pp. 611–628. https://doi.org/10.1037/0021-9010.87.4.611

[s9] F. David Schoorman, Roger C. Mayer, and James H. Davis. “An Integrative Model of Organizational Trust: Past, Present, and Future.” Academy of Management Review 32(2), 2007, pp. 344–354. https://doi.org/10.5465/AMR.2007.24348410