The gap between what team members know and what they say in team settings is among the most expensive information gaps in organizational life. Closing it requires specific leader behaviors, not culture slogans.
Edmondson (1999) found that teams in the same organization, with comparable talent levels and organizational resources, showed dramatically different rates of speaking up about errors, concerns, and ideas based almost entirely on the psychological safety of their team climate rather than on member characteristics. The gap between what team members observe and what they share in team settings is one of the most expensive information asymmetries in organizational life: the quality problems that go unreported, the strategic risks that go unvoiced, and the process improvements that go unshared collectively represent a substantial fraction of the organizational performance that teams leave unrealized. This article reviews the specific conditions and leader behaviors that most reliably produce teams that speak up.
Detert and Burris (2007) distinguished two categories of voice behavior: promotive voice, which advocates for changes and improvements, and prohibitive voice, which raises concerns about existing practices, errors, and risks. Both categories require the employee to incur a social cost: the potential for being seen as a troublemaker, as not a team player, or as undermining the manager's authority. Most organizational research on voice finds that both categories are substantially underproduced relative to the organizational information they would provide, with prohibitive voice, specifically the reporting of errors, safety concerns, and ethical issues, more severely underproduced because its social cost is typically higher.
The cost-benefit calculation that organizational members perform before speaking up, rarely consciously but always implicitly, weighs the perceived value of the information against the perceived social risk of sharing it, weighted by the individual's assessment of whether sharing it will produce any organizational response. When the perceived social risk is high, when the individual believes that raising the concern will damage their relationship with their manager or their standing in the team, voice is suppressed regardless of the information's value. When the perceived likelihood of organizational response is low, when the individual believes that the concern will be heard but not acted upon, voice is also suppressed because the social risk is incurred without the organizational benefit that would justify it.
Psychological safety, as Edmondson (1999) defined it, addresses the social risk component of this calculation by creating the team-level belief that interpersonal risk-taking, including speaking up with concerns, questions, and ideas, is safe in this specific team context. It does not eliminate the individual cost of speaking up; it reduces the social cost component sufficiently that the organizational benefit of sharing information can outweigh the residual cost. Teams with high psychological safety do not have team members who feel no discomfort speaking up; they have team members for whom the social risk of speaking up is low enough to make the decision to speak up more frequently and across a wider range of topics.
The organizational costs of suppressed voice are systematically underestimated because they are most severe in the situations that are already most organizationally difficult. The safety concern that goes unreported in a high-pressure production environment, the strategic risk that goes unvoiced in a high-stakes executive conversation, and the ethical concern that goes unshared in a climate where raising it would damage the individual's organizational standing are precisely the voice behaviors that most determine organizational outcomes and that are most suppressed by the conditions that make them most valuable.
| Low Perceived Leader Openness | High Perceived Leader Openness | |
| High Team Psychological Safety | Selective voice: members speak up in general but withhold challenge from this leader | Full voice: members raise concerns, ideas, and challenge freely across all team interactions |
| Low Team Psychological Safety | Near-absent voice: maximum suppression; individual and team-level factors both suppress | General safety but cautious with leader: selective use of structural voice channels |
Nembhard and Edmondson (2006) found that leader inclusiveness was the single strongest leader-behavior predictor of team psychological safety and voice behavior, specifically the degree to which leaders actively and explicitly invited input from all team members and responded to that input with genuine acknowledgment and engagement rather than dismissal. The behavioral specificity of leader inclusiveness is important: it is not a general posture of openness or a statement of values about valuing all voices, but a specific pattern of conversational behavior in team meetings and one-on-one interactions that creates evidence for the team-level belief that speaking up is safe and valued in this specific context.
Detert and Burris (2007) found that perceived leader openness, specifically the degree to which team members believed their manager would respond constructively rather than defensively to challenging information, was the strongest individual-level predictor of voice behavior above and beyond team-level psychological safety. Their finding suggests that voice is determined by the interaction between team-level safety and individual-level assessment of the specific leader's likely response: a team member may believe the team climate is safe for voice in general while still suppressing voice to their specific manager if they have learned from past experience that this specific manager responds poorly to challenge or unwelcome information.
The leader behaviors most consistently suppressing voice are those creating evidence that challenging information carries social cost. Visible frustration, impatience, or dismissal in response to challenging questions or concerns; consistently attributing credit for team insights to oneself rather than to the team members who generated them; retaliating, however subtly, against team members who raised concerns that proved inconvenient; and creating the informal organizational signal that some perspectives are welcome and others are not, each teach the team through behavioral observation that speaking up in this context carries risk. These lessons are learned quickly and generalized broadly: a single visible punishing response to voice suppresses voice behavior across the entire team for extended periods.
The leader behaviors most consistently producing voice create evidence that speaking up is both safe and useful. Responding to every voice attempt with genuine acknowledgment and substantive engagement, even when the content of the voice is incorrect or the concern is unfounded; explicitly attributing team decisions and organizational insights to the team members who contributed them; following up on concerns raised in previous conversations to demonstrate that the voice produced an organizational response; and modeling voice behavior oneself by raising concerns and uncertainties in team settings, collectively create the behavioral track record that teams use to assess the safety and utility of speaking up in this specific context.
| Leader behavior category | Effect on voice | Time scale |
|---|---|---|
| Asks before telling in team discussions | Builds: signals input valued before assessment | Weeks |
| Responds with genuine engagement to challenge | Builds: updates prior about response safety | Repeated interactions |
| Attributes credit explicitly to contributors | Builds: voice produces recognition not appropriation | Consistent pattern |
| Visible frustration with challenging questions | Destroys: one event generalizes across team | Single event |
| Dismisses or minimizes concerns raised | Destroys: utility of voice demonstrated to be low | 2-3 events |
| Follows up on previously raised concerns | Builds: demonstrates voice produces organizational response | Consistent follow-up |
Organizational structures that support voice provide formal channels through which organizational members can raise concerns, ideas, and observations without the full social cost that direct voice to one's manager entails. Anonymous reporting systems, ethics hotlines, open-door policies, and skip-level conversations each reduce the social risk of specific voice behaviors by providing alternative channels with lower individual cost. Research on the effectiveness of these structural voice mechanisms finds that they produce incremental voice behavior above the levels that team-level psychological safety produces, but that they are not effective substitutes for psychological safety: structural channels primarily capture the voice of individuals whose discomfort with direct voice is high, not the everyday voice behavior that psychological safety enables and that determines team performance most directly.
Regular structured opportunities for voice, including brief post-meeting retrospectives, start-stop-continue reviews, and regular team health check-ins, create the organizational norm that voice is expected as part of normal team operation rather than as an exceptional behavior that carries exceptional social cost. These structural mechanisms work most effectively when they are consistently used rather than sporadically deployed, when the leader responds to the voice they produce with genuine engagement rather than with defensive minimization, and when the organizational outcomes of previous voice exercises are visible to team members so that they can assess whether speaking up in this context produces organizational response.
The measurement of voice behavior as an organizational condition requires approaches sensitive to what is not said as well as what is, which is methodologically challenging but organizationally important. Teams with low rates of error reporting, concern raising, and improvement suggestion relative to the actual rate of errors, concerns, and opportunities for improvement they encounter are underproducing voice in ways that have real organizational consequences. Survey instruments that assess the psychological safety conditions associated with voice behavior provide a leading indicator of voice production; behavioral observation of specific team interactions and the rate at which concerns and questions are raised provides a direct measure of voice behavior that the safety conditions predict.
Organizations that systematically develop leader behaviors producing team voice, and that create the structural conditions supporting it, build a genuine organizational information advantage over those that leave voice to emerge spontaneously from individual courage and interpersonal comfort. The teams that speak up most consistently are not those with the most individually courageous members but those with the most consistently inclusive leaders, the most clearly established norms of constructive challenge, and the most reliable organizational evidence that speaking up produces genuine engagement and organizational response.
The connection between team voice behavior and team performance is mediated by the quality of team decisions, the speed of error identification and correction, and the rate of innovative improvement. Teams with higher voice rates make better decisions because they incorporate more diverse information and more critical evaluation of proposed approaches before committing to them. They identify and correct errors faster because errors are reported rather than managed individually. They improve their processes more rapidly because members who notice opportunities for improvement share them rather than keeping them to themselves.
Edmondson (2012) documented through multiple organizational case studies that the quality of team performance in complex, high-uncertainty environments is determined more by the team's ability to learn from experience through voice and reflection than by the initial capability of team members. Teams with high voice rates and strong learning behavior show steeper performance improvement curves than those with lower voice rates and weaker learning behavior, even when the low-voice teams start at higher initial performance levels. The compounding advantage of voice over time is substantial: high-voice teams at year three substantially outperform low-voice teams at year three even when the low-voice teams had comparable or superior performance at year one.
The implication for organizational performance management is that measuring team voice behavior alongside team performance outcomes provides a leading indicator of team performance trajectory rather than only a lagging indicator of current performance quality. Teams currently performing adequately while showing low voice rates are teams whose performance is likely to plateau or decline as the errors they are not reporting accumulate, the risks they are not voicing grow, and the improvements they are not sharing go unrealized. Organizations that monitor voice behavior as a predictor of future performance are in a better position to identify and address declining performance trajectories before they become visible in outcome metrics.
The organizational investment case for developing leader voice-producing behaviors is therefore both a current-performance case, speaking up now produces better decisions and faster error correction now, and a trajectory case: teams with high voice rates improve faster than those with low voice rates, compounding the performance advantage over time. Both arguments justify substantial organizational investment in the leader behaviors that produce team voice, and the specificity of those behaviors, the identifiable and trainable practices of leader inclusiveness, constructive response to challenge, and consistent acknowledgment of team member contributions, makes that investment actionable in ways that generic culture initiatives toward speaking up are not.