Psychological safety: A systematic review

of the literature
Author links open overlay
panelAlexander Newman  , Ross Donohue  , Nathan Eva 
a b b

Show more
Add to Mendeley
Share
Cite
[Link] rights and content

Highlights
•
Reviews the extant literature on psychological safety

•
Highlights gaps in the existing literature and issues with measurement of
psychological safety

•
Concludes with directions for future research on psychological safety

Abstract
Since the concept of psychological safety was introduced, empirical research on its
antecedents, outcomes, and moderators at different levels of analysis has
proliferated. Given a burgeoning body of empirical evidence, a systematic review of
the psychological safety literature is warranted. As well as reviewing empirical work
on psychological safety, the present article highlights gaps in the literature and
provides direction for future work. In doing so, it highlights the need to advance our
understanding of psychological safety through the integration of key theoretical
perspectives to explain how psychological safety develops and influences work
outcomes at different levels of analysis. Suggestions for future empirical research to
advance our understanding of psychological safety are also provided.

 Previous article in issue
 Next article in issue

Keywords
Psychological safety
Learning
Measurement issues
Work outcomes
1. Introduction
In the contemporary business world, organizations are increasingly requiring their
employees to contribute to the continuous improvement of organizational processes
and practices through behaviors that enable learning to occur (e.g., voicing new
ideas, collaborating with other members of the organization, and experimenting with
new ways of doing things; Edmondson, 1999, Nembhard and Edmondson, 2011).
While such activities may potentially benefit the organization, they carry certain risks
for the individual. For example, the voicing of new ideas might challenge the
established way of doing things and go against the vested interests of other members
of the organization (Detert and Burris, 2007, Edmondson et al., 2001). In addition,
experimentation with new approaches in the workplace might ultimately be
unsuccessful, viewed as a failure, and lead the individuals involved to be seen in a
negative light (Van Dyne & LePine, 1998). As a result, there is growing evidence to
indicate that such risks may lead employees not to contribute to learning processes,
and thereby inhibit both individual and organizational learning (Detert & Burris,
2007). The provision of a psychologically safe work environment (i.e., one in which
employees feel safe to voice ideas, willingly seek feedback, provide honest feedback,
collaborate, take risks and experiment, is one way to overcome such threats to
individual and organizational learning; Edmondson, 1999). For example, in recent
longitudinal work by Google's People Analytics Unit, psychological safety was
identified as the number one characteristic of successful high-performing teams
(Bergmann & Schaeppi, 2016). Psychologically safety is especially important in work
environments where employee and customer safety are paramount, such as the
healthcare or aviation industries, as it has been shown to be critical in reducing
employee errors and enhancing safety (Leroy et al., 2012, Nembhard and
Edmondson, 2011), and been shown to increase team and individual learning across
multiple organizations (e.g., Liu et al., 2014, Ortega et al., 2010). Given the
importance of psychologically-safe work environments to organizations, their
employees and their customers, the present article reviews prior scholarship on
psychological safety.
Although several definitions of psychological safety have been proposed, the majority
of studies have followed Edmondson (1999) by defining it as a shared belief amongst
individuals as to whether it is safe to engage in interpersonal risk-taking in the
workplace (Edmondson et al., 2007, Edmondson and Lei, 2014). In a psychologically
safe work environment, employees feel that their colleagues will not reject people for
being themselves or saying what they think, respect each other's competence, are
interested in each other as people, have positive intentions to one another, are able
engage in constructive conflict or confrontation, and feel that it is safe to experiment
and take risks (Edmondson, 1999). Behaviorally, psychological safety leads
employees to engage in open communication, voice their concerns, and seek greater
feedback; all of which are interpersonally risky behaviors (Pearsall & Ellis, 2011).
This, in turn, has been found to influence a range of workplace outcomes (e.g.,
learning and performance) at different levels of analysis (Edmondson & Lei, 2014).
Although psychological safety shares some overlap with trust, psychological safety is
conceptually different as it focuses on how group members perceive a group norm,
whilst trust focuses on how one person views another.
Since Kahn's (1990) and Edmondson's (1999) initial work on psychological safety at
the individual and team levels of analysis, empirical research on its antecedents,
outcomes, and moderators has proliferated (Baer and Frese, 2003, Kark and
Carmeli, 2009). By the end of 2015 there were > 83 published articles on
psychological safety (78 of which are empirical), including a meta-analysis of the
relationships among psychological safety and team performance/learning (Sanner &
Bunderson, 2013), a meta-analysis of the antecedents and outcomes of psychological
safety (Frazier, Fainshmidt, Klinger, Pezeshkan, & Vracheva, 2016), and a limited
review of prior work (Edmondson & Lei, 2014). Although the meta-analyses provide
us with some knowledge of the antecedents, outcomes, and boundary conditions of
the relationship between psychological safety and its antecedents/outcomes, and the
prior review provides insight into extant scholarship on psychological safety, the
present article builds on this previous work to make a significant contribution in a
number of ways.
First, by examining a larger number of published articles (83 studies compared to 25
studies for the Sanner and Bunderson meta-analysis, 78 for the Frazier et al. meta-
analysis,1 and fewer than 30 studies for Edmondson and Lei's [2014] review), we
undertake a far more systematic review of prior work, delivering a thorough analysis
of the state of the literature. As well as examining prior research on all the
antecedents and outcomes of psychological safety, many of which were not the focus
of prior empirical work, we offer a greater theoretical analysis of the past literature
than both the Frazier et al. and Sanner and Bunderson meta-analyses. In addition,
we examine the moderators of relations among psychological safety and its
outcomes, and work in which psychological safety is treated as a moderator. This
allows us to better elucidate the complex nomological network in which
psychological safety is embedded. Although Edmondson and Lei's (2014) review
looked at the antecedents and outcomes of psychological safety, their work did not
cover more recent articles that have made substantive theoretical and empirical
advancements (e.g. Gu et al., 2013, Halbesleben and Rathert, 2008, Liu et al.,
2014, Post, 2012, Roussin and Webber, 2012, Roussin et al., 2016, Singh et al., 2013),
including recent work that has begun to look at the negative aspects of psychological
safety (Pearsall & Ellis, 2011). Edmondson and Lei (2014), like Frazier et al. (2016),
also did little to highlight the theoretical perspectives that have been adopted by
researchers to explain how psychological safety develops and influences work
outcomes.
We make a second important contribution by highlighting opportunities for
theoretical advancement of the field through the integration of different theoretical
perspectives to explain the relationship between psychological safety and its
antecedents/outcomes. Currently, theoretical perspectives such as social learning,
social exchange and social identity theories have predominantly been used by
researchers to explain the processes by which psychological safety develops and
influences outcomes. Our review calls upon researchers to utilize alternate theories
such as the Conservation of Resources Theory (Hobfoll, 1989) to explain how the
psychological safety engendered through access to resources in the work
environment motivates employees to invest their resources at work to help others,
and stimulate learning, growth and development. In addition, we highlight how the
incorporation of person-situation theoretical perspectives such as Trait Activation
Theory (Tett & Guterman, 2000) might help to explain how the psychological safety
climate strengthens the behavioral manifestation of certain personality traits.
Finally, we also provide advice for advancing empirical research. In addition to
providing recommendations concerning the measurement of psychological safety, we
also call on researchers to: (1) adopt alternative methodologies to study psychological
safety, (2) conduct additional research to investigate the influence of culture on the
development and deployment of psychological safety, (3) investigate the potential
negative effects of psychological safety and (4) conduct more multi- and cross-level
work to understand the relative influence of individual-, team-, and organizational-
level antecedents on psychological safety.

As well as advancing theoretical and empirical knowledge of psychological safety, our

review of the literature has important practical implications for organizations.
Understanding the benefits that psychological safety brings to organizations, the
situations in which psychological safety is most influential, and the factors that may
lead to psychological safety development, will assist leaders in designing work
environments that maximize beneficial outcomes for their organizations.
Psychological safety is becoming increasingly important to organizational success in
today's business environment, given the requirements for employees to share
information and exchange ideas with other team and organizational members in the
attainment of shared goals (Edmondson & Lei, 2014). A growing proportion of work
in organizations is highly specialized and complex in nature, and therefore requires
greater collaboration among individuals than in the past. In the following sections we
set out how we conducted our literature search before reviewing how psychological
safety has been defined and measured, and examining its antecedents and outcomes.

2. Literature review process

2.1. Inclusion and exclusion criteria
We set three inclusion/exclusion criteria prior to undertaking our literature search.
First, articles had to be published between 1990 (the year in which Kahn did his
seminal work on psychological safety) and the end of 2015. Second, we only included
studies that mentioned the words psychological safety in their title, abstract, or
keywords or included psychological safety in their empirical analysis. Third, we only
included studies that focused on psychological safety in work environments; studies
with student or non-working adult samples were excluded. Finally, we only included
articles that focused on the antecedents and outcomes of psychological safety, or
those that examined how psychological safety interacted with other variables to
predict outcomes.

2.2. Literature search
In line with best practice (Short, 2009), we searched a number of academic
databases for articles (including: Web of Science, Google Scholar, EBSCO Host,
Emerald, PsycInfo, Science Direct and ProQuest) using the keywords psychological
safety. In addition, we examined the reference lists of all retrieved articles for
additional studies on psychological safety and we examined all studies that had cited
the retrieved articles to locate additional literature. This led to the identification of
83 articles for inclusion in our review, 78 of which were empirical studies.

3. A review of the psychological safety literature

3.1. Defining and measuring psychological safety
The construct of psychological safety is grounded in the seminal work conducted
by Schein and Bennis (1965) on organizational change. They described it as the
extent to which individuals feel secure and confident in their ability to manage
change. Since then other researchers have explored the meaning of psychological
safety in work settings. A quarter of a century after Schein and Bennis's
(1965) seminal work, Kahn's (1990, p.708) work led to renewed focus on
psychological safety. In conceptualizing psychological safety as an individual's
perceptions as to whether he or she is comfortable to show and employ his(her)self
without fear of negative consequences to self-image, status, or career, he argued that
people are more likely to feel psychologically safe when they have trusting and
supportive interpersonal relationships with work colleagues (Kahn, 1990). More
recently, Edmondson (1999) argued that psychological safety is better treated as a
team-level climate and defined it as the “shared belief held by members of a team
that the team is safe for interpersonal risk taking” (p. 350). This definition is
divergent from Kahn's definition that it is a perception emanating from the
individual. Based on qualitative work she developed and validated a 7-item scale to
measure team psychological safety. This measure includes items that capture shared
perceptions amongst team members as to whether they believe that others will not
reject members for being themselves, team members care about each other as
individuals, team members have positive intentions to one another, and team
members respect the competence of others.
Our review identified 78 empirical studies, conducted over the past 25 years, which
have focused on examining the antecedents and outcomes of psychological safety.
Among these, 740 studies utilized a quantitative survey methodology and the
remainder employed a qualitative interview methodology. Studies adopting a
qualitative interview methodology have helped to generate detailed explanations as
to why and how psychological safety develops and influences work outcomes. They
have typically been used as a basis from which to guide quantitative work. In
contrast, studies adopting a quantitative survey methodology have furthered our
understanding of the strength of links between psychological safety and its
antecedents/outcomes.

3.1.1. Individual-level measures
Our review identified 29 studies that have measured employees' individually-held
perceptions of psychological safety within dyadic-relationships, teams, or
organizations. First, examining individuals' perceptions of psychological safety in
dyadic relationships, studies by Tynan (2005) and Detert and Burris
(2007) adapted Edmondson's (1999) team-level measure to capture self and other
psychological safety, and manager-focused psychological safety, respectively. In
contrast, Roussin and Webber (2012) used a measure of manager-focused
psychological safety developed by Edmondson and Woolley (2003) and Hetzner,
Gartmeier, Heid, and Gruber (2011) constructed their own measures of supervisor-
and colleague-focused psychological safety, based on items from measures used in
other studies. Second, a number of studies examining individually-held perceptions
of psychological safety within organizations have used adapted versions
of Edmondson's (1999) measure, replacing the referent “team” with “organization”
(Carmeli et al., 2009, Carmeli and Gittell, 2009, Carmeli et al., 2010, Carmeli and
Zisu, 2009, Chen et al., 2014, Kark and Carmeli, 2009, Madjar and Ortiz-Walters,
2009). However, Singh et al. (2013) adopted a measure from Chrobot-Mason and
Aramovich (2004), De Clercq and Rius (2007) used a measure from Brown and
Leigh (1996), Liang, Farh, and Farh (2012) developed their own measure using items
from other studies, including May, Gilson, and Harter (2004), and Xu and Yang
(2010) constructed their own measure.

3.1.2. Team-level measures
Our review identified 42 studies that have measured psychological safety at the team
level. Most of these studies used Edmondson's (1999) 7-item Psychological Safety
Scale or abbreviated versions of this scale. However, other researchers have
developed their own measures or adapted measures that included items
from Edmondson (1999) and items from other scales (Tucker, 2007). In
contrast, Hirak, Pang, Carmeli, and Schaubroeck (2012) used a measure of unit-level
psychological strength climate taken from Anderson and West (1994). In these
studies, team members' perceptions of the psychological safety within the team were
aggregated to the team level, based on high intra-class correlation coefficients (ICCs)
between team members.

3.1.3. Organizational-level measures
Only two prior studies have measured psychological safety at the organizational level
(Baer and Frese, 2003, Carmeli, 2007). Both studies measured employee perceptions
of psychological safety within their organization, based on items adapted
from Edmondson's (1999) team-level measure, replacing the referent team with
organization. Scores on the measure were then aggregated to the organizational level
based on high ICCs between organizational members. Although these studies suggest
that individual perceptions of psychological safety can be aggregated to the
organizational-level, it is questionable as to whether this is likely to be the case in all
organizations, as there should be relatively high levels of agreement between all
members of the organization for an organizational climate of psychological safety to
exist. As employee perceptions of psychological safety are typically influenced by
leadership and team characteristics (e.g., Carmeli et al., 2010, Chen and Tjosvold,
2012), high levels of agreement between organizational members is unlikely,
especially in larger firms where employees are unlikely to have shared experiences of
leadership and team norms. Although a climate of psychological safety might exist in
smaller organizations, whose members regularly work with each other or in
organizations with strong corporate cultures, this should be confirmed through
further empirical research. In general, however, we believe that psychological safety
is likely to be more potent and meaningful at the team level, rather than the
organizational level.

3.1.4. Measurement critique and recommendations

The majority of studies examining psychological safety have been conducted at the
team level and almost invariably have utilized Edmondson's 7-item scale (or short-
form versions) to operationalize the construct. More common among studies
examining individual perceptions of psychological safety is the use of measures of
similar constructs as proxies for psychological safety, or scales developed by the
authors themselves. For example, some researchers have used psychological climate
scales that capture the extent to which the climate is motivating and involving as a
proxy for psychological safety (e.g., De Clercq & Rius, 2007). The use of proxy
measures, however, is problematic as the constructs they tap may diverge
from Edmondson's (1999) precise constitutive definition of psychological safety. A
lack of correspondence between the conceptual definition of the construct and the
operational procedure to measure it, often results in compromised construct validity
(Schwab, 1980). Moreover, all of the studies that we reviewed, that used “in-house”
measures developed by researchers themselves, applied them to test substantive
questions without first establishing that their measures were psychometrically
sound. Such measures may lack validity and therefore have the potential to result in
spurious findings (DeVellis, 2003).
In contrast, Edmondson's (1999) 7-item measure was developed based on rigorous
scale construction protocols and has been subjected to extensive validation tests,
which have invariably shown that the measure has strong content, criterion, and
construct validity. In addition, Edmondson's (1999) measure of psychological safety
has consistently been found to be reliable across diverse samples as evidenced by the
fact that all of the studies that we reviewed, that used her 7-item measure, reported
very good internal consistency reliability estimates.
In our view, the continued use of multiple measures of psychological safety is likely
to hinder advancement in the field because the results of different versions of the
construct measure are not readily comparable (Hawcroft & Milfont, 2010). As a
consequence, we believe that our understanding of psychological safety would be
greatly improved if there was consensus regarding the most valid and reliable
measures. In light of the strong psychometric properties of Edmondson's (1999) 7-
item scale, which suggest that it is isomorphic (similar in meaning) at different
levels, and the fact that nearly all of the extant studies examining psychological safety
at the team or organizational levels applied both Edmondson's (1999) definition of
the construct and her measure (changing the referent to “organization” in the latter),
we recommend it as the measure of choice for use in future studies in preference to
proxy or “in-house” measures of psychological safety.
Finally, while there have been studies conducted at different levels of analysis, it is
also the case that there has been a paucity of cross-level and multilevel psychological
safety research. The scarcity of such studies not only limits our understanding of the
nomological net of psychological safety, that is inherently mixed-level involving
micro and macro antecedents and outcomes (which we discuss below in Section 5),
but also limits our capacity to understand whether psychological safety varies across
organizations, as well as across teams within organizations. The study
by Edmondson and Mogelof (2005) was the only study identified in our review that
attempted to address this question. They examined team- and organizational-level
variance in psychological safety via data from 26 innovation teams, nested in 7
organizations (i.e., data from multiple teams and organizations), collected over 3
waves. Edmondson and Mogelof (2005) found that there were team-level differences
in relation to psychological safety and that organizations differed in psychological
safety at the same time. These findings align with the mixed-level view of
psychological safety, however, we believe that further multilevel research is required
to determine whether the findings from this single study are replicable.

3.2. Antecedents of psychological safety and psychological safety as a

mediator
Our review identified 44 empirical studies that have examined the antecedents of
psychological safety at different levels of analysis. This work has predominantly
treated psychological safety as a mediator (38 of these studies) to explain how
supportive organizational practices, supportive leadership behaviors, relationship
networks, team characteristics and individual/team differences influence workplace
outcomes at the individual, team and organizational-levels through the development
of psychological safety. In examining these issues, researchers have begun to utilize
diverse theoretical perspectives including social learning theory, social exchange
theory, social identity theory, and status characteristics theory.

3.2.1. Supportive leadership behaviors

Growing research at both the individual and team level has examined the effects of
supportive leadership behaviors on work outcomes through psychological safety. For
example, at the individual level empirical work has established that leader
inclusiveness (Bienefeld and Grote, 2014, Carmeli et al., 2010), support (May et al.,
2004), trustworthiness (Madjar & Ortiz-Walters, 2009), openness (Detert & Burris,
2007) and behavioral integrity (Palanski & Vogelgesang, 2011) strongly influence
employee perceptions of psychological safety, which in turn, drive employee
outcomes including voice behaviors, involvement in creative work, job performance
and engagement. Similarly, at the team level employees' collective perceptions of
support and coaching forwarded by the team leader (Edmondson, 1999, Roberto,
2002), leader inclusiveness (Hirak et al., 2012, Nembhard and Edmondson, 2006),
trust in the leader (Li and Tan, 2012, Schaubroeck et al., 2011), and the behavioral
integrity of the leader (Leroy et al., 2012) have been found to foster team-level
outcomes such as team learning behavior, team performance, engagement in quality
improvement work, and reduction in errors amongst team members through the
development of psychological safety. Other work has found that positive leadership
styles such as transformational leadership (Nemanich & Vera, 2009), ethical
leadership (Walumbwa & Schaubroeck, 2009), change-oriented leadership (Ortega,
Van den Bossche, Sanchez-Manzanares, Rico, & Gil, 2014) and shared leadership
(Liu et al., 2014) are positively and strongly related to such outcomes as employee
voice behavior, team learning, and individual learning through the mediating
mechanism of psychological safety.
Finally, research has established that leaders who value participation, people, and
production use dyadic discovery methods rather than group-based discovery
methods (Roussin, 2008, Wong et al., 2010), and an improvement orientation
management style (Halbesleben & Rathert, 2008), are able to foster high levels of
psychological safety. In explaining why a significant relationship may exist between
supportive leadership behaviors and psychological safety, prior research has typically
relied on key tenets from social learning theory (Bandura, 1977). Based on this
theory, researchers have argued that by listening, forwarding support, and providing
clear and consistent directions to subordinates, the leader is able to model to
subordinates that it is safe to take risks and engage in honest communication
(e.g., Hirak et al., 2012, Liu et al., 2014, Nemanich and Vera, 2009, Walumbwa and
Schaubroeck, 2009). However, other researchers have argued that social exchange
processes may underlie the relationship between supportive leadership and
psychological safety, arguing that when followers are supported by the leader, they
will reciprocate with supportive behaviors themselves, creating a psychologically safe
environment for the rest of their team (e.g. Schaubroeck et al., 2011). While we
acknowledge that the social exchange process may increase psychological safety, we
believe it is likely that the effects will be stronger and more enduring, when
psychological safety is built through learning and emulating these behaviors from the
leader, rather than displaying them at a point in time in exchange for certain
leadership behaviors.
3.2.2. Supportive organizational practices
At the individual level, there is growing evidence that supportive organizational
practices are positively related to employee work outcomes such as organizational
commitment and job performance as they heighten perceptions of psychological
safety. For example, research has found that employee perceptions of organizational
support (Carmeli & Zisu, 2009), access to mentoring (Chen et al., 2014), and
diversity practices (Singh et al., 2013) foster work outcomes through the mediating
mechanism of psychological safety. Drawing on a sample of 191 medical professionals
in an Israeli medical clinic, supportive organizational practices were found to foster
psychological safety through social learning processes, similar to that of supportive
leadership behaviors (Carmeli & Zisu, 2009). However, Singh et al. (2013) relied on
social identity theory to argue that the implementation of workplace diversity
practices fosters psychological safety via employee identification with the
organization in the context of US mid-sized production firms (N = 165). Both studies
utilized a survey design, with Singh et al. (2013) using the 3-item Chrobot-Mason
and Aramovich (2004) measure of psychological safety, rather than the well
validated Edmondson (1999) 7-item measure.

3.2.3. Relationship networks
Growing research at the individual, team, and organizational levels has looked at
social support and the social capital (resources) inherent in relationship networks as
key determinants of psychological safety. At the individual level, research has
established that rewarding co-worker relationships and the extent to which members
of the organization interact with one another on an interpersonal basis, influence
individual learning and engagement through the mediating mechanism of
psychological safety (Carmeli and Gittell, 2009, Carmeli et al., 2009, May et al.,
2004). Similarly, at the team level, researchers have found that relationship
networks, and the social support and resources inherent in such networks, promote
psychological safety and contribute to team learning, performance, and innovation.
For example, researchers have found that the key drivers of psychological safety and
its outcomes are: the level of prior interaction between; and familiarity among, team
members (Roberto, 2002); the quality of social relationships between team members
as measured by trust; network ties and collective thinking (Gu et al., 2013, Huang
and Jiang, 2012, Schulte et al., 2012); high quality relationships between team
members and external parties (Brueller & Carmeli, 2011); and membership of the
inner-circle (Burris, Rodgers, Mannix, Hendron, & Oldroyd, 2009). Finally, at the
organizational level Carmeli (2007) found that the strength of social networks
between members of the organization was positively related to their ability to learn
from failure through the development of psychological safety. Although most studies
in this area did not explicitly address why strong relationship networks, and the
social resources inherent in such networks, foster psychological safety, a small
number of researchers have proposed that social relationships may engender
psychological safety through social learning processes (e.g. Carmeli, 2007, Gu et al.,
2013).

3.2.4. Team characteristics
At both the individual and team level, research has begun to examine how
individually-held and collective perceptions of team characteristics influence
perceptions of psychological safety. For example, at the individual level researchers
have found that employee perceptions of similarity in systems understanding among
team members (Bendoly, 2014), and a continuous quality improvement climate in
teams (Rathert, Ishqaidef, & May, 2009), influence project performance and
organizational commitment respectively through higher levels of psychological safety
within the team. At the team level, researchers have found that characteristics such
as shared team rewards (Chen & Tjosvold, 2012), formal team structures (Bresman
and Zellmer-Bruhn, 2013, Bunderson and Boumgarden, 2010), and team
engagement in boundary work (buffering, spanning, and reinforcement) (Faraj &
Yan, 2009) are positively associated with higher levels of team psychological safety.
However, Chandrasekaran and Mishra (2012) found that only team autonomy
influenced psychological safety when the project goals and processes of the team
were aligned with their broader organizational goals and when there were low
degrees of relative exploration (i.e., the team focused on refining existing products
and processes rather than seeking to develop new products and processes). Contrary
to what they expected, Lau and Murnighan (2005) found that the presence of strong
faultlines within teams (i.e., the existence of sub-groups with non-overlapping
demographic characteristics) led to greater psychological safety amongst team
members. They argued that this may have resulted from generalization of the
positive social effects within strong faultline groups to the entire team.
Finally, O'Neill (2009) found that when team members were collectively responsible
for bad investment decisions, psychological safety gave them the courage to admit
failure, whereas when they were individually responsible, psychological safety
escalated their commitment to bad investment decisions. They argued that this is due
to the fact that, when there is collective responsibility, team members are less
motivated to salvage a project that may fail. Although growing work has examined
the influence of team characteristics on psychological safety there has been limited
incorporation of theory to explain the effects of such characteristics.

3.2.5. Individual and team differences

Researchers have also found that individual and team differences such as adherence
to co-worker norms and self-consciousness (May et al., 2004), status differences
(Bienefeld and Grote, 2014, Nembhard and Edmondson, 2006) and team members'
sequential cognitive style (i.e., thinking in a logical sequential routine) (Post, 2012)
are associated with psychological safety. For example, utilizing perspectives from
status characteristics theory, prior research has found that an individual's perceived
status within the team (Bienefeld & Grote, 2014), and the professionally-derived
status of the team (Nembhard & Edmondson, 2006), leads to outcomes such as the
willingness of individuals to speak up and team engagement through enhancing
psychological safety. This work suggests that the higher the status of the employee or
the team, the more safe individuals will feel to speak up and share ideas.

3.3. Outcomes of psychological safety

Our review identified 62 empirical studies that focused on the outcomes of
psychological safety at different levels of analysis. Given that the early conceptual
development of psychological safety was grounded in the organizational learning
literature, early empirical work on psychological safety typically focused mainly on
learning and performance outcomes (e.g., Edmondson, 1999). Research has also
examined the relationship among psychological safety and outcomes such as
innovation, creativity, employee attitudes, communication, knowledge-sharing, and
voice behaviors. Research examining the relationship among psychological safety
and its outcomes has begun to integrate perspectives from social learning
theory, social information processing theory, social identification theory, and social
exchange theory (Carmeli, 2007, Chen et al., 2014, De Clercq and Rius, 2007, Liu et
al., 2014, Schaubroeck et al., 2011, Singh et al., 2013).

3.3.1. Communication, knowledge sharing and voice behavior

At the individual and team level, psychological safety has been linked to
communication outcomes such as greater reporting of treatment errors and more
interpersonal communication (Leroy et al., 2012, Peltokorpi, 2004), as well as
greater knowledge sharing among team members (Mu and Gnyawali, 2003, Siemsen
et al., 2009, Xu and Yang, 2010, Zhang et al., 2010). Psychological safety within
dyadic relationships and teams has also been shown to lead to more voice behavior
among employees (Bienefeld and Grote, 2014, Detert and Burris, 2007, Liang et al.,
2012, Tynan, 2005), and a reduction in silence behaviors (Brinsfield, 2013). For
example, Tynan (2005) found that individuals high in other-psychological safety
(i.e., they believe others feel safe in their relationships), were more likely to raise
disagreement, give candid feedback, and point out errors to their supervisor.

3.3.2. Learning behavior
A growing body of research has found positive associations among employee
perceptions of psychological safety and learning behavior at different levels of
analysis. For example, researchers have established positive links between
psychological safety and learning behavior at both the individual (Liu et al., 2014)
and team levels (Bstieler and Hemmert, 2010, Ortega et al., 2010, Roberto,
2002, Stalmeijer et al., 2007, Van den Bossche et al., 2006, Wong et al., 2010).
Psychological safety has also been shown to assist individuals to learn from failure
(Carmeli, 2007, Carmeli and Gittell, 2009). In their meta-analytical work, Sanner
and Bunderson (2013) found the correlation between team psychological safety and
team learning to be 0.42 (95% CI = 0.05 to 0.85). In examining the link between
psychological safety and learning, researchers have begun to adopt perspectives from
the social learning and social information processing literature (e.g., Carmeli,
2007, Liu et al., 2014). This literature suggests that through signaling it is acceptable
to take risks, experiment and voice their ideas and concerns to team members, a
psychologically safe work environment will foster team and individual learning.

3.3.3. Performance, innovation and creativity

As well as directly and strongly influencing performance at the individual (Singh et
al., 2013) and team levels (Schaubroeck et al., 2011), psychological safety has been
found to influence performance indirectly through facilitating learning behavior at
both the individual (Li and Tan, 2012, Li and Yan, 2009) and team (Brueller and
Carmeli, 2011, Edmondson, 1999, Hirak et al., 2012, Huang and Jiang,
2012, Kostopoulos and Bozionelos, 2011, Ortega et al., 2014) level. In their meta-
analysis, Sanner and Bunderson (2013) found the indirect effect of psychological
safety on team performance, through team learning, to be 0.17 (95% CI = 0.14 to
0.20). Other work has established that alternative mechanisms such as team
turnover also explain the effect of psychological safety on team performance
(Chandrasekaran & Mishra, 2012). Finally, at the organizational-level, Baer and
Frese (2003) found that employees' collective perceptions of psychological safety
were strongly and positively related to firm performance, as measured by return on
assets and goal achievement.
In addition to performance, there is growing evidence of a link between employee
perceptions of psychological safety within the organization and their creativity
(Carmeli et al., 2010, Kark and Carmeli, 2009). Palanski and Vogelgesang (2011) also
found evidence of a positive link between employee perceptions of team
psychological safety and both creative thinking and risk-taking. At the team-level,
research has found that team psychological safety is strongly and positively related to
innovation in R&D teams (Gu et al., 2013, Post, 2012), manufacturing process
innovation performance (Lee, Swink, & Pandejpong, 2011) and knowledge creation
(Choo, Linderman, & Schroeder, 2007). Kessel, Kratzer, and Schultz (2012) found
that team psychological safety was positively related to creative team performance
and this relationship was mediated by the sharing of two types of knowledge:
information and know-how. However, Kayes (2006) found no evidence that
psychological safety led to higher levels of critical thinking within teams. Researchers
have argued that psychological safety may influence performance outcomes through
fostering social exchange between the employee and organization, and enhancing the
extent to which the employee identifies with the organization (Schaubroeck et al.,
2011, Singh et al., 2013, Singh and Winkel, 2012).

3.3.4. Employee attitudes
At the individual level, a number of studies have established a strong and positive
link between psychological safety and the work attitudes of employees such as
organizational commitment (Chen et al., 2014, De Clercq and Rius, 2007, Rathert et
al., 2009), work engagement (May et al., 2004, Nembhard and Edmondson, 2006)
and positive attitudes towards teamwork (Ulloa & Adams, 2004). In examining the
link between psychological safety and employee work attitudes, researchers have
typically relied on social exchange theory to explain how the psychological safety
engendered by supportive practices leads employees to reciprocate in the form of
desired work attitudes (Chen et al., 2014, De Clercq and Rius, 2007).

3.3.5. Other outcomes
At the individual-level, perceptions of psychological safety within the team have also
been shown to increase the likelihood that employees will engage in “work arounds”,
defined as work procedures undertaken to deal with blocks in workflow (Halbesleben
& Rathert, 2008), and lead them reflect on events at work (Hetzner et al.,
2011). Roussin and Webber (2012) established that employee perceptions of
psychological safety in dyadic relationships with their manager were positively
related to their initial perceived trustworthiness of new team members. They argued
this resulted from the fact that psychological safety would encourage team members
to engage in the social risk taking involved in extending initial trust to a new co-
worker. At the team level, psychological safety has been linked to other outcomes
such as lower quality relationships and task conflict (Wilkens & London, 2006),
more feedback giving and seeking behavior (Wilkens & London, 2006), and frontline
system improvements (Tucker, 2007). Tucker, Nembhard, and Edmondson
(2007) also found that team psychological safety was associated with the use of
internally focused learning activities that provide employees with the knowledge
required to adapt and implement new practices in their unit, and had a subsequent
impact on the implementation success of improvement projects. Finally, Edmondson
et al. (2001) found that psychological safety positively influenced the successful
implementation of new technology.

3.4. Factors moderating psychological safety antecedents and outcomes

A limited number of studies have focused on factors that may interact with
psychological safety to predict various outcomes at different levels of analysis. In
their meta-analysis, Sanner and Bunderson (2013) found that psychological safety
was more strongly related to team learning and performance in task environments
that benefit from learning. For example, they found that association between
psychological safety and team learning to be stronger in task environments where
there were higher creativity requirements (β = 0.39, p < 0.01), sense-making
requirements (β = 0.54, p < 0.01), complexity (β = 0.63, p < 0.01), and social
contact/impact (β = 0.50, p < 0.01). Moreover, Sanner and Bunderson (2013) found
that the association between psychological safety and team performance was
stronger in task environments where there were higher creativity requirements
(β = 0.69, p < 0.01), sense-making requirements (β = 0.31, p < 0.01) and complexity
(β = 0.73, p < 0.01). Our review also identified work that examined other boundary
conditions on psychological safety-outcome relationships at different levels of
analysis. At the individual-level, Roussin and Webber (2012) found that
organizational identification moderated the relationship between manager-focused
psychological safety and initial perceived trustworthiness of new team members,
such that the relationship was positive when organizational identification was low,
and non-significant when organizational identification was high. In explaining these
results, they argued that employees with higher organizational identification are less
likely to be influenced by team psychological safety when determining initial
perceptions of trustworthiness in new co-workers as they shift their cognitive focus
to their valued relationship with the organization. Liang et al. (2012) found that
psychological safety was more strongly related to promotive and prohibitive voice
behavior, for employees high in felt obligation, and that it was more weakly related to
promotive voice for employees high in organization-based self-esteem. At the team
level, Siemsen et al. (2009) found that the relationship between psychological safety
and knowledge-sharing was positively moderated by the level of confidence that
employees have in what they know. Finally, Kostopoulos and Bozionelos
(2011) established that task conflict moderated the relationship between team
psychological safety and exploitative learning (they associated with activities that
assist the team to refine and utilize existing knowledge and skills), in such a way that
the relationship was stronger when there was higher task conflict.

3.5. Psychological safety as a moderator

Growing work has examined the moderating effects of psychological safety on
different relationships at the individual, team and organizational levels of analysis.
At an individual level, Tangirala, Kamdar, Venkataramani, and Parke (2013) found
that when individuals perceived their environment to be psychological safe, this
weakened the negative relationship between achievement orientation and role
conceptualization. At the team level, Martins, Schilpzand, Kirkman, Ivanaj, and
Ivanaj (2013) found that when psychological safety was low, expertise diversity had a
stronger negative relationship with team performance. In contrast, when
psychological safety was high, team expertise diversity had a stronger positive
relationship with team performance. Gibson and Gibbs (2006) found that when
psychological safety was high, the negative effects of geographical dispersion,
electronic dependence, dynamic structure, and national diversity on team innovation
were mitigated. Pearsall and Ellis (2011) revealed that psychological safety
moderated the relationship between utilitarianism and unethical outcomes, such that
members of teams high in psychological safety were more likely to engage in
unethical behaviors compared to those from teams that were low in psychological
safety. Bradley, Postlethwaite, Klotz, Hamdani, and Brown (2012) found that task
conflict only influenced team performance when team psychological safety was
high. Leroy et al. (2012) established that the relationship between team priority of
safety and reporting of treatment errors was stronger for teams higher in
psychological safety. Finally, Kirkman, Cordery, Mathieu, Rosen, and Kukenberger
(2013) found that psychological safety moderated the curvilinear relationship
between nationality diversity and performance, such that the relationship was more
positive at the higher end, and less positive at the lower end, for teams high in
psychological safety.
At the organizational level, Baer and Frese (2003) established that at high levels of
psychological safety, the relationship between process innovativeness (i.e., the use of
advanced manufacturing techniques) and profitability (i.e., return on assets), was
positive, whereas at low levels of organizational psychological safety, the relationship
was negative.

3.6. Summary
In Fig. 1 we synthesize key observations from our review to provide an overview of
the nomological network of variables to which psychological safety is related.

1. Download : Download high-res image (534KB)

2. Download : Download full-size image
Fig. 1. Network of key variables related to psychological safety based on literature
review.
Our review of the literature led to a number of key observations. As can be seen
in Fig. 1, although psychological safety is a multilevel construct, our review suggests
that it is likely to be more potent and meaningful at the team level, as opposed to the
organizational level, unless the organization is small. This may be the reason why few
studies have been conducted at the organizational level. In addition, as can be seen
in Fig. 1, most of the antecedents of psychological safety, at the individual and team
levels, can be grouped under the rubric of supportive environments (i.e., supportive
leadership, supportive relationships with colleagues, and supportive organizational
practices). In the majority of studies we reviewed, psychological safety was the
mechanism through which the effects of these supportive environments were
transmitted to desirable outcomes, such as increased knowledge sharing,
engagement, creativity, innovation, and ultimately performance. Our review also
established that psychological safety is a valuable resource, especially important in
hazardous work contexts where speaking up and providing feedback is imperative in
order to reduce errors and improve safety. As shown in Fig. 1, across all levels,
psychological safety appears to be predictive of learning and performance outcomes.
As evident in Fig. 1, task-related factors, such as the presence of task conflict or
strong task environments that motivate learning, amplify the positive effects of
psychological safety on performance and exploitative learning (respectively). At the
individual level, employee personal beliefs and attitudes (e.g., organizational
identification and felt obligation) appear to strengthen the relationships among
psychological safety and positive outcomes, such as communication and admitting
errors. As can be seen in Fig. 1 our review also indicates that the extant research has
focused inordinately on testing the moderators of psychological safety-outcome
relationships, while examination of the boundary conditions of antecedent-
psychological safety relationships has been neglected.
As displayed in Fig. 1, psychological safety appears to moderate other antecedent-
outcome relationships. For example, high levels of psychological safety were found to
intensify the positive relationship between team expertise diversity and team
performance and mitigate the negative effects of geographical dispersion, electronic
dependence, dynamic structure, and national diversity on team innovation.
Although our review revealed that growing empirical work has been conducted on
the antecedents and outcomes of psychological safety, there has been limited
integration of theory to explain the processes by which psychological safety develops
and influences work outcomes. Our review highlighted two relevant theoretical
approaches that have begun to attract attention in the psychological safety literature,
both of which offer a unique perspective of the positive work outcomes derived from
psychological safety. The first stream invokes social learning theory to argue that
supportive practices and relationships at work foster psychological safety and
influence outcomes such as learning, performance, innovation, and creativity. These
occur through facilitation of information sharing, employee voice, feedback seeking,
and experimentation (e.g., Carmeli, 2007, Liu et al., 2014).
The second stream utilizes the social exchange theory to argue that supportive
practices and relationships at work may enhance psychological safety and lead
employees to reciprocate in the form of positive work outcomes (e.g., Chen et al.,
2014, De Clercq and Rius, 2007). While some have argued that psychological safety
and subsequent positive outcomes are learned behaviors by employees, others have
argued that they are conducted in exchange for positive behaviors shown towards
them. Although these theories may provide explanations for links among
psychological safety and its antecedents/outcomes, they do not provide a detailed
and holistic understanding of the underlying processes through which psychological
safety develops and influences workplace outcomes. Nor do they assist us in
understanding the boundary conditions of the relationship between psychological
safety and its antecedents/outcomes. Our review also highlighted a general reliance
on quantitative research methodologies in the area of psychological safety and
limited exploration of alternative methodologies. Finally, our review identified a lack
of longitudinal research on the dynamic processes by which psychological safety
develops and influences work outcomes. Among extant studies, only Schulte et al.
(2012) employed a panel design where both antecedents and outcomes were
measured at all points in time.

4. Agenda for future research

4.1. Opportunities for theoretical advancement
As highlighted in our review of prior research, the majority of studies of
psychological safety have focused on its antecedents (e.g., supportive leadership
behavior) and/or its outcomes. This research offers little theoretical understanding
of the processes by which psychological safety influences positive and negative work
outcomes or the boundary conditions that affect these relationships. In light of these
limitations, we highlight two potential theoretical frameworks that researchers may
draw on to examine these issues.

4.1.1. Conservation of resources theory

Our review highlights the need for a more holistic understanding of the underlying
processes through which psychological safety develops and influences workplace
outcomes. Therefore, we encourage researchers to integrate perspectives from the
Conservation of Resources (COR) theory (Hobfoll, 1989). We believe that the COR
theory provides a succinct explanation as to how resources in the work environment,
that have been identified in our review of the literature (e.g., supportive
organizational practices and relationship networks), positively influence work
outcomes. Moreover, COR theory provides an explanation for the development of
psychological safety and how both individual and team resource investment and
depletion may explain the psychological safety – work outcomes relationships.
The core tenet of COR theory is that individuals seek to gain resources to protect
against resource loss (Hobfoll, 1989). Resources may include social support, rewards,
autonomy, or job security and can either be gained or taken away by the
organization, supervisor, team, or individuals. According to the 1st and 3rd
corollaries of COR theory (Hobfoll, 2011), individuals with access to greater
resources (e.g., relationship networks found within a psychologically safe work
environment) are less vulnerable to resource loss (depletion) and are more capable of
orchestrating resource gain (investment) through using their existing resources. By
obtaining these additional resources, individuals are better equipped to meet their
work demands and achieve their work goals (Halbesleben, Neveu, Paustian-
Underdahl, & Westman, 2014). For example, when provided with adequate support
in the workplace, employees will feel psychologically safe to share their ideas and
knowledge with others in the workplace in the belief that this will assist them in
obtaining further resources (e.g., positive feedback and suggestions from others in
the workplace). This, in turn, will enable them to attain their work goals and is likely
to foster an environment where individual and team learning occur and performance
improves.
Based on this theory we present a conceptual framework in Fig. 2 to highlight the
mechanisms by which psychological safety develops and influences work outcomes.
As can be seen in Fig. 2, supportive job resources engender a climate of psychological
safety, conferring protection from resource loss, which in turn is associated with
negative individual outcomes such stress and strain, as well as undesirable team
outcomes such as conflict (the health impairment pathway). A climate of
psychological safety also appears to distinguish high performing teams from their
counterparts, as members in such teams are motivated to invest resources (e.g.,
through communication and knowledge sharing), which in turn leads to positive
work outcomes such as learning, innovation, and performance at the individual- and
team-level (the motivational pathway).

1. Download : Download high-res image (247KB)

2. Download : Download full-size image
Fig. 2. An integrative theoretical framework of psychological safety.

4.1.2. Trait activation theory

We also call on researchers to incorporate prominent person-situation theoretical
perspectives, such as the trait activation theory (TAT) (Tett & Guterman, 2000), to
further our understanding of the boundary conditions on the relationships among
psychological safety and work outcomes. According to Tett and Guterman (2000, p.
398), “the behavioral expression of a trait requires arousal of the trait by trait-
relevant situational cues”. TAT suggests that the influence of personality traits may
depend on inducements offered by the context (situational cues), and therefore
provides an explanation as to how organizational climates, such as psychological
safety climate, might interact with the personality traits of the employee to predict
their work behaviors and attitudes. Based on TAT we argue that a climate of
psychological safety would amplify the positive effects of an individual's
predisposition to act proactively, embodied in personality traits such as extraversion,
proactive personality, and learning goal orientation. This would lead them to be more
likely to engage in relevant work behaviors such as voice behavior or information
sharing in environments, with high levels of psychological safety, as such an
environment provides cues and opportunities for expression of their traits. Prior
work has drawn on TAT to examine how organizational or team climates arouse
individuals to engage in more positive behaviors at work when they possess certain
personality traits (e.g., Byrne, Stoner, Thompson, & Hockwarter, 2005).
Together COR theory and TAT explain the processes by which psychological safety
influences outcomes and the ways in which personality traits and psychological
safety may interact to influence outcomes. We call upon researchers to draw on such
perspectives in future studies in order to advance our understanding of psychological
safety.

4.2. Opportunities for empirical advancement

In the following sections, we develop an agenda for advancing empirical research. As
well as (1) providing recommendations around the measurement of psychological
safety; we call for (2) the greater use of alternative methodologies to study
psychological safety; (3) additional research to investigate the influence of culture on
employee and team responses to psychological safety; (4) greater investigation of the
potential negative effects of psychological safety; and (5) more multi-level and cross-
level work to understand the relative influence of individual, team and
organizational-level antecedents on psychological safety. Research in these areas will
have important practical implications through enabling managers to design work
environments that are conducive to the development of psychological safety and
maximize the benefits of psychological safety to the organization.

4.2.1. Measurement of psychological safety

While considerable empirical work on psychological safety has relied on the use of a
cross-sectional quantitative survey methodology, limited longitudinal research has
been conducted. Longitudinal studies, focusing on newcomer socialization in teams,
involving multiple waves of data collection and the application of latent growth
modeling (Duncan & Duncan, 2004) would afford the opportunity to systematically
examine the dynamic nature of psychological safety, especially individuals' and
teams' developmental trajectories in psychological safety, and inter-individual and
team variability in its development. Research along these lines would be insightful as
we know very little about the aetiology and temporal change of psychological safety
in teams.
As well as using consensus models to measure psychological safety at the team and
organizational-levels of analysis, researchers might consider measuring the degree of
consensus between team members as to psychological safety within their dyadic
relationships and map these using social network methods as network ties (Roussin
et al., 2016). This will allow for a better understanding of how psychological safety
dynamics within sub-teams influence learning and performance outcomes within
teams, and address Roussin et al.'s (2016) concerns that the use of consensus models
has led researchers to ignore team contexts in which members do not share similar
beliefs about psychological safety.
In addition, we also call on researchers to investigate factors that
influence psychological safety strength (i.e., the degree of consensus or agreement of
individuals' perceptions of psychological safety within a team or organization;
see Schneider, Salvaggio, and Subirats [2002]).2 Although researchers have justified
conceptualizing psychological safety as a team- or organization-level construct
through measuring the degree of agreement between members, they have yet to
examine the influence of psychological safety strength on team or organizational
outcomes, or identify the factors that predict psychological safety strength. These are
surprising omissions given the topic of team climate strength has gained increasing
attention in the literature (González-Romá et al., 2009, Schneider et al., 2002). We
might expect variables such as team or organization size, geographic dispersion, and
diversity among team and organizational members to influence the degree of
psychological safety strength. Researchers might also examine whether psychological
safety strength changes across the different stages of team development, and whether
this construct moderates the relationships between team psychological safety and its
outcomes.

4.2.2. Alternative methodologies to study psychological safety

Given the skew towards a quantitative survey methodology in extant research,
researchers should also consider using alternative methodologies in future work to
gain a more holistic understanding as to how psychological safety develops and
influences work outcomes. In addition to using qualitative interviews, which have
been adopted in prior research to gain a deeper understanding as to how
psychological safety develops within teams and individuals, researchers might
consider using observational techniques to gauge levels of psychological safety within
a team or dyadic context. This may be done through looking at the verbal and non-
verbal communication people use when interacting with one another. More
specifically, face reading software might be used to determine the extent to which
people seem comfortable when talking with one another and their willingness to
engage in interpersonal risk taking.

4.2.3. Cultural influences on psychological safety

As recommended by Edmondson and Lei (2014), greater work might be undertaken
to ascertain whether psychological safety has a stronger influence on outcomes for
individuals, teams, and organizations operating in different cultural settings. At
present most of the research investigating the influence of psychological safety in the
workplace has been conducted in Western cultures, such as the US which are
characterized by low levels of collectivism, power distance, and uncertainty
avoidance (Hofstede, 2001). In such cultures individuals are more likely to be direct
in voicing new ideas and engage in experimentation, without the presence of
psychological safety. As a result, the influence of psychological safety on work
outcomes such as learning, performance, and creativity might be expected to be more
pronounced for individuals and teams embedded in organizations operating in
cultures characterized by higher levels of collectivism, power distance, and
uncertainty avoidance. In such cultures speaking out or experimenting with new
ideas is less common, as engaging in such behavior leads to more social costs than in
Western cultures and causes individuals to lose face and risk being ostracized by
other group members (Friedman et al., 2006, Liu et al., 2010). Examining
psychological safety in these contexts will allow a more robust test of the predictive
validity of psychological safety as the variance between members' perceptions of
psychological safety may be higher than that of Western culture where there are
minimal social costs to speaking out.

4.2.4. Negative effects of psychological safety

There has been an inordinate focus in prior empirical work on examining the positive
outcomes of psychological safety. Recently, based on pervasive and counter-
theoretical findings in some areas of management research, Pierce and Aguinis
(2013) proposed a meta-theoretical principal they refer to as the too-much-of-a-
good-thing effect (TMGT effect). Essentially, the TMGT effect occurs when levels of
typically beneficial antecedents reach tipping points, after which their relations with
desired outcomes shift from being linear and positive to curvilinear (inverted U-
shape) and negative. Evidence of the TMGT effect has been found for assertiveness in
terms of its relationship with leadership effectiveness (Ames & Flynn, 2007) and for
conscientious with regard to its association with performance (Whetzel, McDaniel,
Yost, & Kim, 2010). Pierce and Aguinis (2013) argued that other management
constructs are likely to have non-monotonic relations with desired outcomes and
therefore we consider here the possibility that too much psychological safety within a
team could lead to detrimental outcomes.
In the extant psychological safety literature, only one prior study by Pearsall and Ellis
(2011) has highlighted its potential negative influence, demonstrating that teams
high in psychological safety are more likely to engage in unethical behavior. More
specifically, Pearsall and Ellis (2011) argue that teams high in utilitarianism may
utilize the psychological safety of their environment to choose the most beneficial
option when making decisions, even if that option may be unethical. Once the
unethical possibility is raised as a potential option, other team members may feel
willing to support the idea, given a psychological safe climate within the team. These
findings suggest that psychological safety might not always have a positive influence
on team and organizational-level outcomes.
Hence, we also call on researchers to explore possible negative consequences of “too
much” psychological safety at the individual, team, and organizational levels and
potential curvilinear effects of psychological safety on outcomes at different levels of
analysis. For example, psychological safety generally encourages change and
disruption to the status quo, which can carry a certain degree of risk for individuals
(e.g., “a damaged reputation if the initiative fails or disapproval if it is seen as
inappropriate or threatening” [Morrison & Phelps, 1999, p. 405]). In addition, recent
work has shown that high levels of trust can lead to lower levels of team performance
due to lower levels of monitoring within autonomous teams (Langfred, 2004). Thus,
we expect psychological safety to have a negative influence on team learning and
performance for teams that have high autonomy within their organization.

4.2.5. Interactive effects of individual, team and organizational-level antecedents on

psychological safety
There has been limited investigation as to how individual-, team-, and
organizational-level factors may interact to influence psychological safety, and
subsequently influence outcomes, at different levels of analysis. Cross-level designs
have only been previously used to examine how psychological safety at the team-level
predicts individual-level outcomes (Walumbwa & Schaubroeck, 2009), and how
psychological safety at the team level interacts with individual-level variables to
predict team outcomes (Liu et al., 2014).
We call on researchers to conduct additional research to improve our understanding
of how variables at different levels of analysis interact to predict psychological safety.
Better understanding of how cultural, organizational, and team-level factors combine
to predict the development of psychological safety will enable organizations to design
effective work environments and practices that assist individuals and teams to work
effectively. For example, scrutinizing cross-level effects of organizational-level
variables, future research may investigate whether the influence of positive
leadership behaviors (e.g., transformational or participative leadership; Yukl, 2002)
on team-level psychological safety (team-level variables) is stronger or weaker when
the organization has human resource (HR) practices in place that facilitate employee
voice or involvement (e.g., Holland, Pyman, Cooper, & Teicher, 2011). On one hand,
we might expect HR practices that enable “direct voice” (e.g., regular meetings with
senior management or quality circles; Holland et al., 2011) to trigger positive
leadership effects on team-level psychological safety by enabling employee
engagement or involvement. In other words, transformational and participative
leadership might only have a positive influence on psychological safety when the
organizational environment is supportive of such leadership styles. On the other
hand, it might also be argued that HR practices that facilitate employee voice or
engagement will substitute for leader influence. In other words, in the absence of
transformational or participative leadership, such HR practices may have a greater
influence on psychological safety than where there is a transformational or
participative leader.
Future research might also explore same-level direct effects of organizational-level
antecedents on organizational-level psychological safety. Indeed, our review suggests
that psychological safety and its antecedents are routinely scrutinized at the
individual or team levels; yet, similar findings might also be expected at the
organizational level. Thus, future research might focus on HR practices that
engender direct voice arrangements (Holland et al., 2011), empowerment (e.g.,
structures and procedures that encourage initiative-taking; Seibert, Silver, &
Randolph, 2004), employee discretion and group collaboration (“high involvement”
HR practices; Batt & Colvin, 2011) as antecedents of psychological safety at the
organizational level, given that such practices are likely to provide a shared structure
of cooperation that facilitates the willing contribution of information among
organizational members.

5. Conclusion
To conclude, we believe that psychological safety is important given it is a key
cognitive state that allows learning processes to occur, and contributes to improved
work outcomes at different levels of analysis (Edmondson, 1999, Edmondson et al.,
2007). From a practical perspective, this review has highlighted the myriad of
positive workplace outcomes associated with psychological safety that indicate to
managers the importance of engaging in supportive leadership behaviors, fostering
bonds between team members, and leveraging supportive organizational practices to
build psychological safety at work. Due to the importance of psychological safety for
organizations today, research that helps us to understand the antecedents of
psychological safety, and how and when psychological safety contributes to work
outcomes, is important for individuals, teams, and organizations alike. Through
offering suggestions to advance theoretical understanding of psychological safety and
opportunities for advancing empirical research, we hope this review will stimulate
future research.