Psychological Safety as a Multi-Level Construct: Theoretical Foundations, Measurement, and Organizational Antecedents

Theoretical origins and early conceptualization

The construct of psychological safety has roots in two distinct intellectual traditions. Schein and Bennis (1965) introduced the term in the context of organizational change, arguing that individuals can tolerate the anxiety associated with learning and change only when a sufficient sense of psychological safety exists: when the social environment is perceived as non-threatening enough that the risks of exposure, experimentation, and potential failure are manageable. In this formulation, psychological safety was an individual-level construct with implications for organizational learning, distinct from self-confidence or interpersonal trust but related to both.

Kahn (1990) reintroduced the construct in a study of the psychological conditions of personal engagement at work and identified psychological safety as one of three conditions, alongside meaningfulness and availability, that determine whether employees fully invest themselves in their work roles. Kahn's conceptualization emphasized the interpersonal dimension: people experience psychological safety when they believe they can show and employ themselves without fear of negative consequences to self-image, status, or career. This framing anticipated the social-interactional emphasis that Edmondson would develop into a team-level construct a decade later.

Edmondson's (1999) contribution was to operationalize psychological safety at the team level, develop a reliable and valid measurement instrument, and demonstrate empirically that team psychological safety predicts team learning behavior and performance independent of individual-level variables including general trust and interpersonal satisfaction. The move from individual to team level required a reconceptualization: team psychological safety is an emergent property of shared perceptions that cannot be reduced to the average of individual safety perceptions, but reflects a collectively held belief about the interpersonal climate that shapes behavior across team members.

Measurement at team and organizational levels

The most widely used measure of team psychological safety is Edmondson's (1999) seven-item scale, which assesses perceptions of interpersonal risk in team contexts across dimensions including the safety of speaking up, the response to mistakes and questions, and the respect and acceptance experienced when raising concerns. The scale has demonstrated acceptable internal consistency, test-retest reliability, and construct validity across a wide range of organizational settings and has been translated and validated in multiple languages and cultural contexts.

Attempts to extend measurement to the organizational level have produced a more heterogeneous literature. Newman, Donohue, and Eva (2017) reviewed organizational-level measures and found considerable variation in conceptualization, with some instruments targeting climate for voice, others targeting error management culture, and others assessing general perceptions of the organizational environment as supportive of honest communication. The construct validity of organizational-level psychological safety measures remains an area of active methodological work, with important questions about whether the construct is additive across teams, whether it requires its own operationalization at the organizational level, and how to handle the hierarchical nesting of individuals within teams within organizations.

Leader behavior as the primary antecedent

The most consistent finding in research on psychological safety antecedents is that leader behavior is the most proximal and most powerful predictor of team-level psychological safety. Edmondson (2004) identified leader coaching behavior, defined as a combination of availability, accessibility, and the ability to create conditions for learning, as the leader behavior most strongly associated with team psychological safety. The mechanism is straightforward: team members take their cues about what the environment will tolerate from the behavior of the most powerful person in their immediate context, and adjust their own risk-taking accordingly.

Detert and Burris (2007) examined voice behavior in organizations and found that managerial openness, defined as the degree to which managers signal receptiveness to upward communication, was a stronger predictor of employee voice than either individual-level personality factors or organizational-level climate variables. The implication is that psychological safety is produced locally, through the specific behaviors of specific leaders, rather than globally, through organizational culture or policy. This helps explain why psychological safety can vary dramatically across teams within the same organization: it is a property of the team's social environment as constituted by its leader, not a global organizational attribute.

Nembhard and Edmondson (2006) extended this analysis to the role of leader inclusiveness, defined as words and deeds that invite and appreciate others' contributions, and found that it significantly predicts team psychological safety across organizational levels and professional boundaries. Their study of medical teams, which characteristically include large status differentials between physicians and other clinical staff, found that physician leaders who explicitly invited contributions from lower-status team members produced significantly higher psychological safety and quality improvement engagement across all team members, including those whose status would otherwise suppress voice.

Organizational antecedents and multi-level dynamics

While leader behavior is the most proximate antecedent of team psychological safety, organizational-level conditions set the context within which leaders operate and establish the upper bound on how much safety individual leaders can create. Baer and Frese (2003) studied error management climate at the organizational level and found that it predicts firm performance and the proactive behavior of individual employees, with effects that operate independently of team-level processes. Their findings suggest that organizational-level psychological safety conditions are not simply aggregated team-level conditions but represent a distinct level of analysis with its own antecedents and consequences.

The organizational conditions most consistently associated with psychological safety across levels include senior leadership modeling of fallibility and intellectual humility, human resource practices that reward honest communication and protect employees who raise concerns, and accountability systems that distinguish between productive risk-taking and negligent performance. Organizations where senior leaders visibly acknowledge their own uncertainty and mistakes, where whistleblower and speaking-up behaviors are protected and recognized rather than penalized, and where post-mortem analysis of failures is normative rather than exceptional tend to produce higher average team-level psychological safety across their teams, because the organizational context enables rather than constrains individual leader behavior in that direction.