Explaining Psychological Safety in Innovation Teams:
Organizational Culture, Team Dynamics, or Personality?
Amy C. Edmondson and Josephine P. Mogelof
January 7, 2004
An organization's ability to innovate – whether to develop new products,
implement new technologies, or formulate new strategies – is critical to success in a
changing world. In contrast to activities that support execution, activities supporting
innovation involve risk, uncertainty and even failure along the way to success. Team
members are often reluctant to offer novel contributions for fear of being wrong
(Edmondson, 1999), or for fear of slowing team progress and creating frustration (Ford &
Sullivan, forthcoming). One of the core challenges of innovation, therefore, is coping
with the increased risk of failure that the creative process entails. Past research has
identified an interpersonal climate characterized by psychological safety as conducive to
interpersonal risk taking and hence to creativity and innovation in teams (Edmondson,
2002; West, 1990), yet we know less about factors that give rise to psychological safety.
This paper extends past work on team learning and innovation by systematically
considering the antecedents of psychological safety in innovation teams. The results
increase our understanding of what factors enable people to experience a sense of
psychological safety at work and thereby also shed light on antecedents of organizational
Why study psychological safety in innovation teams?
Increasingly, organizations use teams to accomplish their innovation goals
(Wheelwright & Clark, 1995; Lewis et al., 2001). Innovation teams – project teams in
organizations put together to develop new products or implement substantial change –
face considerable uncertainty. Team members must embrace the creative process of
taking risks, experimenting, and frequently experiencing failure while working closely
together with each other, navigating differences in discipline, experience, status, and
other factors. This navigation of differences takes both interpersonal skill and an
environment of psychological safety.
Psychological safety describes taken-for-granted beliefs that others will respond
positively when one exposes one’s thoughts, such as by asking a question, seeking
feedback, reporting a mistake, or proposing a new idea (Edmondson, 1999). It is a
psychological state that has consequences for how members of innovation teams
behave. Effective action in innovation teams necessarily involves behaviors for which
outcomes are unpredictable – including asking questions, seeking honest feedback
about a new idea, and experimenting. Although these activities are integral to
innovation, engaging in them is risky for individuals who may be seen as ignorant,
disruptive or even incompetent by others as a result (Edmondson, 2003a). Most people
feel a need to manage this risk in order to minimize harm to their image. The
experience of psychological safety can allow team members to relax their guard and
engage openly in the behaviors that underlie learning and innovation.
In psychologically safe environments, people believe that if they make a
mistake or ask a naïve question others will not penalize or think less of them for it.
Psychological safety differs from other constructs that may be associated with creativity
such as efficacy (Bandura, 1982), trust (Kramer, 1999), and intrinsic motivation
(Amabile, 2001). Whereas efficacy is the belief that taking action will produce a
desirable effect; trust is the belief that others' actions will be favorable to one's interests,
and intrinsic motivation is the belief that engaging in a task will be inherently
rewarding; psychological safety is the belief that taking action will not lead to one’s
own denigration or humiliation. Thus, while other factors also promote creativity and
innovation in organizations, psychological safety is seen to have a unique mediating
function (see Edmondson, forthcoming, for more detail).
The construct of psychological safety dates back to early, classic research on
organizational change by Edgar Schein and Warren Bennis (1965), who discussed the
need for psychological safety to help people cope with significant change. Schein
(1985) argued that psychological safety helps people confront data that disconfirm their
expectations or hopes – virtually inevitable in the context of an innovation team –
without defensiveness. Psychological safety does not refer to a cozy environment in
which people are close friends, nor does it suggest an absence of pressure or problems.
The construct is distinct from group cohesiveness, which reduces willingness to
disagree and challenge others' views (Janis, 1982). Psychological safety instead
describes a climate in which the focus can be on productive discussion that enables
early prevention of problems and the accomplishment of shared goals because people
are less likely to focus on self-protection (see Edmondson (1999) for more detail on the
construct and its history). It is thus important to note that psychological safety does not
reduce conflict in teams, but rather allows it to be managed more productively than
when psychological safety is not present (Barsade et al., 2001).
Teams lacking psychological safety are less likely to engage in the behavioral
hallmarks of creativity: members are less likely to speak up to suggest novel ideas, criticize
others' ideas, challenge the status quo, ask naïve questions, or admit mistakes, for fear of
ridicule or more subtle forms of interpersonal rejection. Psychological safety in teams thus
enables creativity and innovation by facilitating the interpersonal risks inherent in the
innovation process (West, 1990).
Although effects of psychological safety on learning, performance, productive
handling of conflict, and experimentation have been discussed in the organizational
literature (Barsade et al., 2001; Edmondson, 1996; 1999; 2002, 2003b; Frese & Baer,
forthcoming; Lee et al., forthcoming), less work has investigated drivers of individuals'
perceptions of psychological safety. In particular, conditions affecting psychological
safety in innovation teams are not well understood.
Some researchers have examined psychological safety as a feature of
organizational culture (Schein, 1985), others as a team characteristic shaped by team
leader behavior (Edmondson, 1996; 2003b), and still others as an individual difference.
For instance, Renee Tynan (forthcoming) developed and validated an individual
difference measure of face threat sensitivity and showed its negative relationship with
upward communication in a hierarchy. Schlenker and Leary (1982) found that some
individuals have a dispositional tendency toward feeling anxious about social
interactions. These individuals doubt their ability to make a positive impression on
others (Leary, 2001; Leary & Kowalski, 1995; Schlenker & Leary, 1982), leading to
social withdrawal (Schlenker & Leary, 1985), inhibition and avoidance (Leary &
Atherton, 1986; Leary, Atherton, Hill, & Hur, 1986), and passiveness in one-on-one
interactions (Leary, Knight, & Johnson, 1987).
The aim of this paper is to show that an individual’s experience of psychological
safety in an innovation team is affected by factors at multiple levels of analysis, including
organizational culture, team leader behavior, team member interactions, and individual
differences in personality. First, we develop theoretical arguments for why this should be
the case. We then examine the effects of multilevel factors simultaneously, by analyzing
longitudinal data from 26 innovation teams in seven companies. Finally, we discuss what
we have learned from these analyses and suggest implications for future research and
Psychological Safety in the Workplace
In this section, we develop a theoretical model that posits factors at multiple
levels of analysis as influences on the psychological safety experienced by individuals
working on organizational innovation projects. We argue that these multilevel influences
imply that psychological safety will vary across organizations, as well as across groups or
teams within organizations.
The multilevel nature of influences on psychological safety
Understanding the experience of psychological safety at work requires theoretical
and empirical justification for the level of analysis at which the construct resides. We
thus start by exploring whether psychological safety at work is an experience that
characterizes the individual (shaped by a priori differences in personality), the face-to-
face working group (shaped by interpersonal experiences and shared mental models that
accumulate over time), or the broader organization (shaped by corporate culture)?
At first, this seems an unnecessary diversion. Psychological safety is an
intrapsychic state; surely it lives in the hearts and minds of individual employees. Yet,
theory and past research suggest that psychological safety differs significantly across
work groups – even for groups with highly diverse members, or in strong organizational
cultures (e.g., Edmondson, 1999; 1996). If members of a team have shared beliefs about
the psychological safety of their work environment, and other teams have different shared
beliefs, then the construct can be said to describe groups, not individuals (Edmondson,
At the same time, research finding stable individual differences in sensitivity to
face threat or social interaction anxiety suggests that some people have thicker skin than
others and thus may feel inherently psychologically safer than others, finding it easier to
speak up in any environment relative to their thinner-skinned counterparts. Social
interaction anxiety can lead to disengagement, and lowered input into group-based
activities, with negative implications for group performance, and perhaps limit the kind
of risk taking experience that supports the development of psychological safety in a team.
For instance, groups composed of half or more socially anxious individuals performed
worse on brainstorming activities than those groups comprised of individuals who were
not socially anxious (Camacho & Paulus, 1995), illustrating how an individual difference
can translate into group attributes and outcomes.
Not only do individuals differ in ways that should affect their experienced
psychological safety, organizations also may vary in interpersonal climate. Theories of
corporate culture imply that organizations can have signature levels of psychological
safety (e.g. Schein, 1985). The popular management press is replete with stories of
organizations in which speaking up is believed to be nearly impossible by employees
(e.g., Ryan & Oestrich, 1988), as well as those in which employee voice and participation
are celebrated (e.g., Collins & Porras, 1994).
The above theoretical argument, which implies that psychological safety could
vary on all three levels of analysis simultaneously, has not previously been tested.
Research has not measured variance in psychological safety across groups and
organizations in the same empirical study nor included the contribution of individual
differences. In addition to differences across levels, we might anticipate differences
across time. For example, the level of psychological safety an innovation team member
feels might vary over the course of a project. The study reported in this paper thus
includes variables measured longitudinally at all three levels of analysis, as described
below, and finds preliminary support for a multilevel model, in which contributions to
psychological safety exist at individual, group and organizational levels of analysis at
Organizational level influences
Corporate cultures have been shown to vary dramatically across companies, and
such attributes as deference to authority, outspokenness and participation are central
markers of an organization's culture (Collins & Porras, 1994; Deal & Kennedy, 1982;
Peters & Waterman, 1982; de Pree, 1987). An extensive literature on organizational
culture examines how norms, values and beliefs arise in organizations to reduce the
anxiety people feel confronting ambiguity and uncertainty (Schein, 1985). Anxiety can
arise when individuals cope with the uncertainty of innovation, for example, and an
organization's culture may exacerbate or mitigate this psychological state.
In particular, words and actions of high-level managers in an organization –
especially those that indicate their supportiveness, openness, and tolerance for error –
should affect others’ beliefs about acceptability of open discussion of threatening issues
(Detert, 2003). Apocryphal stories prevalent in many organizations capture the ways in
which senior management can powerfully influence views of psychological safety in the
organization as a whole. One such story involves Tom Watson, Jr. at IBM and a field
executive responsible for a ten million dollar mistake. Called into the chairman's office,
the executive was understandably anxious. As retold by Paul Carroll (1993, p. 51),
"Watson asked, 'Do you know why I've asked you here?' The man replied, 'I assume I'm
here so you can fire me.' Watson looked surprised. 'Fire you?' he asked 'Of course not. I
just spent $10 million educating you.' He then reassured the executive and suggested he
keep taking chances."
True or merely myth, such stories have lasting organizational effects. The sent
message is that error is inevitable and the point is to learn, to share the learning, and to try
again. As espoused by Watson, "You really aren't committed to innovation unless you're
willing to fail… The fastest way to succeed is to double your failure rate." (Farson &
Keyes, 2002, p. 64).
In addition to senior management actions, organizational structures can diminish
or increase barriers between individuals and groups. When cross-functional relationships
among peers are encouraged and enabled by organizational norms and structures, this
should increase the general climate of psychological safety. With more experience
interacting with others who have different views and expertise and who face different
pressures, organization members will become more familiar and comfortable with diverse
perspectives. This broader view of the organization's activities is likely to make it easier
for people to understand others' work and to feel comfortable communicating their own
ideas and goals, such as when seeking help or input. In this way, stronger interpersonal
ties can increase the sharing of information (Hansen, 1999) and promote a sense of
psychological safety. In contrast, when organizational "silos" present divisive barriers to
communication, peer relationships across departments or functions are likely to be
characterized by weaker ties, lower willingness to share information and less
psychological safety. In sum, organizational characteristics are likely to affect
individuals' perceptions of psychological safety, leading to differences across firms.
Proposition 1a: Psychological safety will show significant differences across
More specifically, when members of innovation teams perceive top management
in their organization as supportive of innovation and believe that collaboration among
peers is supported by the organization, they are likely to experience their work
environment as having greater psychological safety.
Proposition 1b: An organizational climate for innovation will increase the
psychological safety experienced by members of innovation teams.
Group level influences
In addition to organizational effects, within a given company we expect
individuals working closely together on a project to develop shared perceptions of how
safe their environment is for speaking up about difficult issues. Cues in the environment
about speaking up and shared perceptions of proximal or local authority figures, such as
supervisors and team leaders, will contribute to shared beliefs about psychological safety.
It is also possible that the documented tendency to fail to discuss relevant information in
groups – not due to interpersonal fear but to air time limits and failure to recognize the
salience of privately held information (e.g., Stasser, 1999; Stasser & Titus, 1985) – can
affect psychological safety indirectly. When relevant information fails to surface, group
members may infer that it is not safe to discuss certain things.
In sum, psychological safety describes beliefs about interpersonal interaction, and
these are likely shaped by the history of interactions in a team. To illustrate, teammates
of a nurse who reported being “made to feel like a two year old” when reporting a drug
error independently reported similar feelings of discomfort about speaking up, for
example commenting that “nurses are blamed for mistakes” and “[if you make a mistake
here,] doctors bite your head off.” These nurses, either from personal or vicarious
experience, came to the common conclusion that, on their team, reporting mistakes was
interpersonally penalized (Edmondson, 1996). And, when relationships within a work
group are characterized by trust and respect individuals are likely to believe they will be
given the benefit of the doubt, contributing to a sense of psychological safety
Our focus on team member interactions builds on previous research on innovation
teams where different members often have equal status, deep expertise in different
disciplines, and must work hard to overcome these differences (Dougherty, 1992). Given
the communication challenges in innovation teams and the importance of psychological
safety for enabling open learning oriented conversation (Edmondson, 1999), the team
level of analysis is especially likely to show variance in an empirical study.
Proposition 2a: Different teams within the same organization will have different
levels of psychological safety.
The quality of interpersonal interactions within a team is particularly salient in
establishing psychological safety (Edmondson, 1999), and thus should be associated with
the variance anticipated by proposition 2a. Over the course of a project, day to day team
member interactions should influence the extent to which members of a team feel others
are accepting of and respectful of disagreements, criticism, or new ideas.
Proposition 2b: Positive team member interactions promote psychological safety.
Past research also suggests that interactions with team leaders, who have greater
positional power and status than other members, powerfully influence psychological
safety (Edmondson, 1996). For example, team leader coaching by surgeons was
important for creating psychological safety in a recent study of operating room teams
Proposition 2c: Positive team leader interactions with the team promote
We also suggest that initial team member interactions will have an important and
lasting effect on psychological safety in an innovation team. In contrast, once beliefs
about the team's psychological safety are created, later interactions may have a limited
impact on interpersonal climate, due in part to the tendency of individuals to seek and pay
attention to data that confirm initial impressions (Ross & Nisbett, 1991).
Proposition 2d: The influence of positive project team member interactions on
psychological safety will dissipate over time.
In addition to interpersonal dynamics in a team, the supportiveness of team
structures (Hackman, 1987) may affect psychological safety experienced by innovation
team members. Supportive structures include a well-defined team task, clearly articulated
team goals, and sufficient information and resources to get the job done. Structures that
enable a team to get its work done decrease the degree to which members find themselves
facing anxiety and ambiguity, increase the chances that team members will have positive
views about the chances of success and will work together effectively. Together these
factors should decrease chances of negative events that are interpreted as violations of
camaraderie and teamwork, creating a general atmosphere of success that may promote
The availability of resources and organizational support for an innovation team
also should increase the team's ability to get work done, and also increase members' sense
of psychological safety. When resources are plentiful, there should be less anxiety and
less concern about competing with others for scarce funds, opportunities, or access.
Further, as projects approach deadlines, a lack of resources may become increasingly
salient such that the effect of resources on psychological safety is stronger over time.
Proposition 2e: Having sufficient resources with which to carry out a project will
promote psychological safety, and this effect will increase over time.
Next, we suggest that goal clarity is likely to be more important for psychological
safety later in an innovation team's tenure, than it would be earlier. Immediately
following a team's creation members would not be expect to have full clarity about their
shared aims. However, as time passes and project deadlines approach, clarity about the
team's goal should become more important in sharing a sense of psychological safety. A
team member's confidence that he or she knows not only the team's overall aims but also
his or her tasks for helping to accomplish them should make it easier to speak up with
questions, challenges and concerns, thereby reinforcing a climate of psychological safety.
Project teams at times may reconsider team goals or processes (Gersick, 1988).
We argue that as final deadlines approach in innovation teams, the increased time
pressure experienced by team members will lead such questioning and reconsideration to
be more threatening than it would be early in a project. We thus suggest that a positive
relationship between clear goals and psychological safety should increase over time for
innovation teams. That is, a clear goal is associated with psychological safety in general,
but is more important as deadlines approach.
Proposition 2f: Team goal clarity will promote psychological safety in an
innovation team, and the strength of the relationship will increase over time.
Individual level influences
As noted above, individual differences also may affect team members’
experiences of psychological safety. In addition to differences in social anxiety and
threat sensitivity, individuals differ in the extent to which they are focused on learning
versus performing (Dweck & Leggett, 1988; Elliott & Dweck, 1988). Some individuals
vigorously pursue opportunities to learn and grow whereas others instead seek validation
(Dykman, 1998; Pichanick, 2003). When individuals are more concerned with how
others evaluate them than with opportunities for learning they are likely to feel less
psychologically safe than those who are not as concerned, and they may refrain from
asking questions, experimenting with ideas, or seeking help. Individuals with “thin skin”
may be less tolerant of criticism and may go out of their way to avoid it, simultaneously
compromising learning opportunities that could lead to important innovations.
Because they are the frequently studied and recognized across fields, we use the
framework of the “Big Five” personality factors to develop specific predictions about
psychological safety at work. Further, using measures such as social interaction anxiety
or face threat sensitivity to predict psychological safety in work interactions might run
the risk of tautology. In contrast, three of the five well known personality factors –
neuroticism, extraversion, and openness – have theoretical relationships with
psychological safety and interpersonal risk taking without being nearly identical
constructs (Costa & McCrae, 1992).
First, individuals with high levels of neuroticism tend to experience persistent
negative affect including feelings of anxiety, inferiority, and shame. Pervasive concerns
such as these could lead individuals to be suspicious of others' motives and to construe
their work environments as hostile and unwelcoming. Second, extraverts tend to be
outgoing and assertive in their interactions with others, traits that may result in speaking
up regardless of the interpersonal climate. Third, individuals characterized by openness
tend to exhibit a natural curiosity about the world. Being open to new ideas and different
ways of doing things may increase the likelihood that individuals would feel safe taking
risks and exposing their vulnerabilities in a work environment.
We thus expect that individual differences in neuroticism, extraversion, and
openness influence individuals' experience of psychological safety at work. Further, we
suggest that these personality differences should be more important in determining
psychological safety earlier in the course of a team’s life than later; as team members
gain experience working together, team influences will become stronger, potentially
overwhelming initial individual differences. After a project has had a chance to get
going, personality influences thus may be less salient and less influential than
organizational and group factors that play out on a daily basis.
Proposition 3a: Neuroticism will be negatively associated with psychological
safety experienced by members of innovation teams, especially early in the team's tenure.
Proposition 3b: Extraversion will be positively associated with psychological
safety experienced by members of innovation teams, especially early in the team's tenure.
Proposition 3c: Openness will be positively associated with psychological safety
experienced by members of innovation teams, especially early in the team's tenure.
We posit multiple simultaneous influences on psychological safety in innovation
teams. Specifically, those teams with supportive organizational climates, adequate
resources, supportive team interactions, and high levels of goal clarity will have higher
levels of psychological safety than teams that do not. Although we also predict that
psychological safety will be negatively associated with neuroticism and positively
associated with extraversion and openness, we anticipate organizational and team
influences to get stronger over time, as a project unfolds, and individual differences to
become less predictive of psychological safety, over time. Figure 1 depicts the predicted
influences on psychological safety over time. In the next section we describe a field
study in which support for these relationships can be examined.
Insert Figure 1 about here
We obtained longitudinal data on 26 innovation teams from a study of team
events and motivation (Amabile et al., 2003). Survey measures of psychological safety
were obtained in the beginning, middle and end of team projects or significant project
phases, allowing us to compare earlier and later measures of the same construct to see
how it might change over time in innovation teams. These data also provide better
indications of causal relationships than could be obtained from survey data collected at a
single point in time.
Additional individual, team, and organizational data were obtained electronically on a
daily basis from members of the teams throughout the course of a major project. The teams
were drawn from seven organizations within three industries (three organizations were in
technology, two in chemicals/pharmaceuticals, and two in consumer products). In total, 238
individuals participated in the study. Data collection occurred between May 1996 and April
Each team was followed from the first day of a project (or major project phase) to the
last. Project periods covered by the study ranged from six weeks to ten months. Most
participating teams’ projects involved the development of new products or new processes.
Teams that participated consisted of 3-20 members, all or most dedicated at least 50% to a
particular time-bound project with specific outcome objectives. Response rates were 75%
for the daily questionnaire, and 87%, 75% and 66% for the paper surveys at beginning,
middle and end, respectively.
The first author had input regarding the wording of survey items and contributed
items to assess psychological safety. For each team, the dependent variables were
assessed at the beginning, mid-point, and end of the project, and the independent
variables were assessed on a daily basis throughout the course of the study with the
exception of the personality measures; these data were collected once, at the beginning of
A short email questionnaire was sent to team members to assess their perceptions
of organizational, team, and work factors on a daily basis. We anticipated that
individuals would tend to be consistent over time in their responses. To test this, we used
one-way analyses of variance to check whether individuals' daily responses were more
similar across time (within individual) than between individuals, and found that this was
the case: using individuals as the independent variable and daily questionnaire variables
as dependent variables, we found that significant differences existed between individuals.
We then averaged the daily measures to produce aggregate measures of individuals'
responses; specifically, we chose to create an average of each individual's daily response
data for both the first and second half of each project. Thus, we have measures of
psychological safety at three time points and two temporal measures of each independent
variable (with the exception of the personality variables), one reflecting the first half and
one the second half of each project.
Measures of organizational characteristics included organizational climate for
innovation, as described below. Group and work influences included resources, interpersonal
interactions, and goal clarity. Individual influences included neuroticism, extraversion, and
Psychological Safety (at time 1, time 2, and time 3). New scales for the dependent
variables were constructed from items from the Keys, a survey assessing organizational
climate for creativity (Amabile et al., 1996). The resulting scale consisted of five items:
“There is free and open communication within my work group;” “People in my work group
are open to new ideas,” “My supervisor is open to new ideas,” “There is a feeling of trust
among the people I work with most closely;” and “Within my work group, we challenge each
other’s ideas in a constructive way.” Items were rated on a four-point scale from “never or
almost never” to “always or almost always.” The scales had an internal consistency
reliability of .79, .78, and .84 for the measures at project beginning (psychological safety 1),
mid-point (psychological safety 2), and end (psychological safety 3), respectively.
Organizational Climate for Innovation (1, 2). The daily measure of
organizational climate included two items assessing the extent to which participants
perceived high-level management encouragement of team creativity and collaborative
idea-flow across the organization concerning their projects. These items were intended to
assess the organizational climate for innovation, and were highly correlated with each
other, such that the resulting scale had internal consistency reliabilities of .74 for the first
half of team projects and .84 for the second half.
Resources (1, 2). A single item assessed whether individuals perceived sufficient
resources to be available for their work. The use of single item measures has been shown
to be particularly effective in assessing constructs that are sufficiently narrow or
unambiguous (Sackett & Larson, 1990; Wanous et al., 1997).
Team Interactions (1, 2). Four items assessed interactions among team members
and their leaders on a daily basis. We had expected to distinguish between interactions
among peers and those between team members and leaders, using two variables. In these
data, however, team members rated both in similar ways in their daily responses, in part
because the leaders we also considered members of the teams – and, as the data
suggested, were less likely to be seen as supervisory authorities. The daily items asked
individuals to rate the extent to which they perceived “supportive interactions within the
team,” “encouragement and support from the project supervisor,” and “positive
interactions between the team and the supervisor,” and all were highly correlated. The
internal consistency reliability for the team interactions measure was .92 for the first half
of team projects and .91 for the second half.
Goal Clarity (1, 2). A single item assessed the extent to which individuals felt
they had clarity regarding the goals of the project. This item was assessed on a daily
Individual Differences. Participants’ neuroticism, extraversion, and openness
were measured in the beginning of the study using the NEO Personality Inventory (Costa
& McCrae, 1992). The response rate for this measure was 90%.
First, we conducted analyses of variance to test whether individuals' daily
responses should be aggregated across time to provide meaningful individual level
measures of certain team properties. Second, we conducted analyses of variance to test
whether significant variance in psychological safety was found at the team or
organizational level or both. We then used multivariate general linear model (GLM)
analyses of psychological safety at the project midpoint and completion (times 2 and 3),
with psychological safety 2 or 3 as the dependent variable. For each time period the
predictors were the corresponding temporally antecedent variables that assessed
organizational climate, resources, team interactions, goal clarity, and individual
differences in personality.
Variance in psychological safety
Across all 26 teams, the means of psychological safety at the beginning, mid-
point, and end of projects were, 3.06, 3.07, and 3.01, respectively, on a five-point scale.
Most teams experienced little change in psychological safety over the course of their
projects: the mean difference between psychological safety 1 and psychological safety 2
was -.04 (sd = .20); the mean difference between psychological safety 2 and
psychological safety 3 was -.07 (sd = .16).
At the beginning of each project, psychological safety varied significantly at both
the team and organizational levels. Because about half of the teams had experience
working together at the start of the study, psychological safety may have been party
established in some but not all teams measured. At the mid-point, teams within the same
organization had become more similar, such that psychological safety still differed
significantly across organizations but not across teams within organizations. At the end
of each project, there were again differences in psychological safety across both teams
and companies (Tables 1, 2, and 3 show these differences in psychological safety over
Insert Tables 1, 2 and 3 about here
These data provide consistent support for proposition 1a, that psychological safety
differs across organizations, and inconsistent support for proposition 2a, that
psychological safety varies across teams within organizations. Although psychological
safety did vary across teams at the beginning and end of projects, at project midpoints,
teams within the same organization were not significantly different. We explore possible
interpretations of these findings in the discussion section.
Predicting psychological safety in teams at project midpoint and end
The relations among all the dependent and independent variables are shown in
Table 4. Psychological safety at all time points is moderately related to almost all of the
Insert Table 4 about here
We conducted multivariate general linear model analyses predicting
psychological safety at the project midpoint and end (times 2 and 3), using the temporally
antecedent organizational climate, resources, team interactions, goal clarity, and
individual differences in personality as predictors. Analyses were conducted with
individuals nested in teams and teams nested in companies.
Results showed that neither extraversion (proposition 3b) nor antecedent
perceptions of organizational climate (proposition 1b), nor resources (proposition 2e)
were significant predictors of psychological safety; this finding was consistent at both
project midpoint and end, and we therefore excluded these predictors from the final
models. Significant predictors of midpoint psychological safety (psychological safety 2)
were company, team interactions, neuroticism, and openness. Significant predictors of
psychological safety at project end included both company and team (that is, simply
being members of a given team or company accounted for significant variance in
psychological safety) and temporally antecedent perceptions of interpersonal interactions
and goal clarity (both in the predicted positive direction). And, as also predicted,
neuroticism was negatively associated with psychological safety 3. Tables 5 and 6 show
Insert Tables 5 and 6 about here
Although goal clarity was not a significant predictor of psychological safety at
time 2, proposition 2f was partially supported in that goal clarity was a significant
predictor at time 3, suggesting goal clarity became increasingly more important over the
course of the project. Next, the propositions regarding team interactions (2b, 2c, and 2d)
were partially supported, as follows. Because the items assessing team leader and
member interactions were highly correlated, discriminant validity between the two
measures did not exist and so we combined them to create a measure of the positivity of
team interactions. We then found that positive team interactions were significant
predictors of psychological safety at times 2 and 3.
Contrary to the expectation that the role of individual differences would diminish
over time, neuroticism (proposition 3a) was a significant predictor of psychological
safety at times 2 and 3. As predicted, openness (proposition 3c) was a significant
predictor of psychological safety at time 2 but not at time 3. Figure 2 depicts these
Insert Figure 2 about here
The findings in this study support the overarching premise that psychological
safety in innovation teams is influenced by factors at multiple levels of analysis.
Previous research has emphasized the importance of psychological safety – to allow the
freedom to imagine alternative possibilities and to take risks – but has not systematically
investigated antecedents of this psychological state. This paper does so, with an
unusually comprehensive data set drawn not only from multiple teams but also from
multiple companies and industries, while including longitudinal data that enable
examination of relationships between the same constructs at different points in time.
A number of studies have found group-level differences in psychological safety
(e.g., Edmondson, 1996; 1999; 2003). An unstated implication of this past work was that
personality was not important because social processes were the major influence on this
psychological state, effectively swamping individual differences. This paper challenges
this view, by showing that individual differences exert an independent influence on
perceptions of psychological safety in innovation teams.
At the same time, most of the individual differences were uncorrelated with group
and organizational factors (as shown in Table 4), implying that, despite personality
differences that predispose individual team members to feel more or less psychologically
safe in the teams in which they work, other influences powerfully shape the average level
of psychological safety felt by members of a team. Given that influential variables, such
as goal clarity and team interactions, are more conducive to change via managerial
intervention than personality, the finding that differences across teams are not dominantly
shaped by the personality composition of those teams is reassuring for the aims of
This study was consistent with previous work in finding group level differences in
psychological safety, and it provides new empirical evidence consistent with the
proposition that organizations differ in psychological safety at the same time. Previous
work suggested that organizational cultures would have a main effect on psychological
safety without explicitly testing this proposition with data from multiple teams and
organizations at the same time. Although the group level differences were only
significant at the beginning and end of the team projects, not at the midpoints, further
research is needed to know whether this is a meaningful and replicable result. It is
possible that when teams are first formed members are particularly attuned to the
interpersonal climate of the team, and then as teams get underway, project management
routines imposed by the organization lead to greater similarity across teams in how
members experience the work place. Subsequently, divergence may reemerge as projects
begin to experience discrepant events and focus on proximal impediments to project
execution particularly from the midpoint onward.
Finally, although previous work has emphasized the role of team leaders in
creating psychological safety (Edmondson, 1996; 2003), this study was unable to find
support for this relationship. In contrast to predictions, effects of leader behavior could
not be disentangled from effects of team member interactions. Thus, team interpersonal
interactions emerged as a single influence; members' perceptions of team member and
team leader interactions were so highly correlated that they formed a single construct,
which we referred to as positivity of team member interactions.
A central contribution of this study is its inclusion of data that span the innovation
team lifetime. This allowed us to detect changes in relationships between variables at
different points in time. Some of these were predicted and make logical sense, some of
them are more difficult to understand, and clearly some may be noise, or artifacts of data
collection. It is also possible that the perceptions of respondents, surveyed multiple
times, shifted in ways that don’t accurately capture the actual underlying processes in
these teams. We explore these issues further below in discussing the study's limitations.
As noted previously, psychological safety remained relatively stable in teams
across time. Despite this consistency, factors predicting psychological safety at the
middle and end of a team project differed. These findings run somewhat counter to what
both research and intuition suggest about how teams should approach their work. First, it
is a commonly held notion that team members should agree upon team goals from the
outset. Further, teams that focus on process and procedures to the detriment of goals may
perform less well than those that emphasize goals (Woolley, 2003). However, recent
empirical evidence that teams that actively reflect on both goals and processes (team
reflexivity) outperform those that don’t (Schippers, 2003), suggests that flexibility and
mindful attention to goals may be more important than early agreement.
We argued above that goal clarity would foster a sense of psychological safety,
because a clear and agreed upon goal would remove a potential source of anxiety in a
team. We found that this was the case, but only at later project stages, suggesting that it
may be less threatening (and hence easier to discuss) a lack of goal clarity at the outset of
a team's work than later in its tenure. In particular, as deadlines approach, a lack of goal
clarity may be particularly threatening, harming psychological safety. This possibility is
consistent with Gersick’s midpoint theory, which posits that much of the important
interdependent work accomplished by a time limited team begins with a midpoint crisis
or reevaluation. We build on these ideas by suggesting that a clear sense of team goals is
important for psychological safety later, but not earlier, in a team project.
In addition, it may often be the case that at first meetings teams pay “lip service”
to creating an atmosphere of open and supportive communication but that little is done to
ensure that such interactions occur in the course of carrying out a challenging project
(West, 1990). We found that positive team interactions were a significant predictor of
psychological safety both earlier and later in a project team. This finding suggests that
teams would do well to place a consistent focus on the nature of their interactions not just
early on but at all stages of their work.
Finally, these findings have consequences for research on teams that often relies
on survey data collected at single points in time. This study shows that when survey data
are collected influences – for some relationships powerfully – the results obtained. We
showed that the strength of influences on a given construct such as psychological safety
can fluctuate over time. Clearly these results need to be replicated to increase our
confidence in them; nonetheless, these preliminary findings constitute an important
contribution to our understanding of the temporal dynamics of the antecedents of
Although, as expected, most organizational and team influences measured in this
study emerged as significant predictors of psychological safety over time, several study
hypotheses were not supported by these data. Notably, different teams within the
different companies did not differ in psychological safety at project midpoints, and
neither organizational climate nor team resources, nor extraversion were significant
predictors of psychological safety.
Extraversion may not have been related to psychological safety because it may be
the case that extraverts’ perceptions of whether an environment is psychologically safe do
not differ significantly from introverts while their behavior, which we did not measure,
does differ. If we had measured speaking up instead of psychological safety (as in
Detert, 2003; Edmondson, 2003b), we perhaps would have detected a relationship
between extraversion and this behavioral manifestation of psychological safety.1
We expected that a lack of resources would be associated with lower
psychological safety, that is, with team members being more concerned about risk-taking
behavior, but this was not the case. This may be due to the need for a moderator to better
understand this relationship. Closer examination of the data suggests that some teams
lacked both resources and psychological safety while others' lack of resources seemed to
trigger creative solutions and a sense of openness and psychological safety. It is possible
that in some teams a lack of resources creates a productive sense of urgency that frees
members up to experiment and try new ways of doing things. Yet, we lack understanding
of what makes such a response more likely. Further research should explore this
relationship further to investigate the conditions under which scarce resources impedes
versus facilitates psychological safety.
Finally, we found that an organizational climate for innovation did not
significantly predict individuals' perception of psychological safety. Perhaps in some
1 Note that this is not the same disclaimer as saying survey measures are not necessarily good indices of
actual behavior. Rather, we argue that the intrapsychic state of psychological safety is not the same
construct (despite being related theoretically) as the behavior of speaking up, and that speaking up behavior
is likely to be more strongly related to extraversion than is psychological safety.
organizations, “support” for innovation may be accompanied with an intense demand for
quick or lucrative results—the result of which could be a mixed message for teams with
regard to how safe it is to ask questions, take time to experiment, and focus on learning
(e.g., see Lee et al., forthcoming). Again, further research is needed to evaluate what
organizational factors support psychological safety and innovation.
Although considerable effort was involved in the collection of extensive,
longitudinal data, concerns about common method remain. General positive or negative
affect could create stronger correlations across variables than is warranted by
relationships between the underlying constructs. On the one hand, our strategy for
aggregation of data led to more robust measures than data collected from individuals at a
single point of time would produce. On the other hand, the data are survey data that are
subject to usual concerns about imprecision and affective halos. For example, the daily
electronic surveys failed to show discriminant validity with respect to phenomena such as
peer interactions versus leader-member interactions. This may be because leaders in all
26 teams were so integrated into the teams that they were not perceived as being distinct
from other team members. In general, our data suffer from being respondents’
perceptions that are not independently validated with behavioral measures.
Similarly, aggregation from daily scores to individual-level measures of a given
construct for the period covering the first or second half of a project's duration produced
robust measures but also ones in which much data are necessarily lost. The particular
aggregation strategy that we developed was chosen after extensive contemplation of
possibilities. It is possible that alternative approaches to aggregation or to analysis of
individual daily survey data without aggregation would be preferable, but this approach
was chosen for parsimony and clarity.
We lack data on specific organizational variables that might explain differences in
psychological safety. That is, although organizational climate did not predict
psychological safety, psychological safety did differ significantly across organizations;
other factors are needed to explain these differences. Similarly, the significant
differences found across teams were not fully explained by goal clarity and team
In sum, although our abundant data produced limited marginal value for
answering certain questions, they do provide support for the premise that multiple
influences – specific, meaningful organizational, team, and personality influences – affect
psychological safety in innovation teams.
Implications for research
One implication of the findings reported in this paper is that the design of survey
studies of innovation teams must pay particular attention to timing, specifically to when
in a team's project life data are to be collected. First, we can better understand the
antecedents and consequences of a construct when we have measured these relationships
at different points in time. Second, more specifically, these findings suggest that causes
of psychological safety in innovation teams vary somewhat from the beginning to the end
of a project.
To build on the observed associations between variables reported here, further
research is needed to better understand how to create psychological safety and how to
enhance a team's ability to innovate. In particular, to complement the extensive
quantitative data collected in this study, qualitative case studies of a very small number of
innovation teams would help shed light on the leadership and interpersonal processes
through which psychological safety is created.
We combined individuals’ daily responses into measures of first and second half
of the project to create robust measures of team interaction and other variables that were
temporally antecedent to the dependent variables collected at project midpoints and end
points. Although the multiple data points aggregated to produce each team measure may
increase our confidence in the measures, they also may represent limited marginal returns
to a large investment in terms of research expenses and participant time. We thus suggest
that such quantitative longitudinal data collected periodically, such as once a week or
month, rather than daily might provide the strengths of longitudinal data with less effort
Implications for practice
We argued and found empirical support for the idea that an individual's
experience of psychological safety at work is affected both by personality and by group
and organizational attributes. This implies that many managerial roads lead to the desired
climate of psychological safety for innovation. Organizations can do much to create
environments that foster innovation, and finding ways to promote psychological safety
throughout an organization presents an important avenue for intervention-oriented
research. With respect to individual differences that may affect psychological safety, an
awareness that these influences can differentially affect team members could provide
additional information that helps leaders and teams work better together.
The pattern of results does, however, reinforce the dominance of group level
influences (e.g., Edmondson, 1999; 2002). Although organizational and individual
differences may play an important role, day-to-day interactions among peers and
supervisors have the potential to make a critical and substantial difference in
psychological safety even after accounting for these other influences. Creating and
sustaining a psychologically safe environment is largely an outcome of team members'
own behaviors – through norms they set regarding risk-taking and learning behaviors.
An effort to be explicit about goals, and what is and is not yet clear, at the outset and
especially in later stages of a project can enable open and productive discussion of
concerns, questions, and aspirations in an innovation team. Knowing that negative team
interactions and a lack of goal clarity can reduce psychological safety as team projects
unfold over time may encourage team leaders and members themselves to discuss these
important issues early and often.
The road to innovation is fraught with obstacles. Innovation teams manage
numerous risks, uncertainties, and failures along the way to creating new and exciting
results. Psychological safety is a key factor in helping such teams and their members to
take the interpersonal risks, conduct the experiments, and learn from the mistakes integral
to this creative process. This paper takes an important step in building understanding of
how this psychological state is created in innovation teams. We argue and show that
psychological safety is malleable, dynamic, and subject to multiple influences. Our
results thus suggest that innovation teams seeking to encourage risk taking among
members can achieve the desirable outcome of a state of psychological safety in more
than one way.
The aim of this paper was to directly address the question of what influences
psychological safety in innovation teams. Our review of related literature, together with
our new theoretical arguments about factors that might contribute to psychological safety
in project teams precluded arriving at a simple answer—a single cause or explanation of
this psychological state. Instead, we offer a multilevel, multi-variable model that
suggests psychological safety in the workplace is complexly and multiply determined. At
the same time, drawing inferences from the patterns of variance, we can conclude that the
power to change the team's climate for innovation lies primarily in the hands of those
working together in the team. Ultimately, the day-to-day interactions of team members
are the critical determinants of perceptions of how safe it is to take the risks of
Amabile, T. M. (2001). Beyond talent: John Irving and the passionate craft of creativity.
American Psychologist, 56(4), 333-336.
Amabile, T. M., Conti, R., Coon, H., Lazenby, J., & Herron, M. (1996). Assessing the
work environment for creativity. Academy of Management Journal, 39, 1154-
Amabile, T. M., Schatzel, E. A., Moneta, G. B., & Kramer, S. J. (in press). Leader
behaviors and the work environment for creativity: Perceived leader support.
Bandura, A. (1982). Self-efficacy mechanism in human agency. American Psychologist,
Barsade, S. G., Gibson, D. E., et al. (2001). To Be Angry or Not to be Angry in Groups:
Examining the Question. Academy of Management, Washington, D.C
Camacho, L. M., & Paulus, P. B. (1995). The role of social anxiousness in group
brainstorming. Journal of Personality and Social Psychology, 68 (6), 1071-1080.
Carroll, P. (1993). Big Blues. New York: Crown.
Collins, J. C., & Porras, J. I. (1994). Built to last. NY: Harper Collins Publishers
Costa, P. T., & McCrae, R. R. (1992). Revised NEO Personality Inventory (NEO PI-R)
and NEO Five-Factor Inventory (NEO-FFI) professional manual. Odessa, FL:
Psychological Assessment Resources.
De Pree, M. (1987). Leadership is an Art. East Lansing, MI: Michigan State University
Deal, T. E., & Kennedy, A. A. (1982). Corporate cultures: The rites and rituals of
corporate life. Reading, MA: Addison-Wesley.
Detert, J. R. (2003). To speak or not to speak: The multi-level influences on voice and
silence in organizations. Unpublished doctoral dissertation, Harvard University.
Dougherty, D. (1992) Interpretive Barriers to Successful Product Innovation in Large
Firms. Organization Science, 3(2), 179-202.
Dweck, C. S., & Leggett, E. L. (1988). A social-cognitive approach to motivation and
personality. Psychological Review, 95, 256-273.
Dykman, B. (1998). Integrating cognitive and motivational factors in depression: Initial
tests of a goal-orientation approach. Journal of Personality & Social Psychology,
Edmondson, A. C. (1996). Learning from mistakes is easier said than done: Group and
organizational influences on the detection and correction of human error. Journal
of Applied Behavioral Sciences, 32(1), 5-32.
Edmondson, A. (1999). Psychological safety and learning behavior in work teams.
Administrative Science Quarterly, 44(4), 350-383.
Edmondson, A. C. (2002). The local and variegated nature of learning in organizations.
Organization Science, 13(2),128-146.
Edmondson, A. (forthcoming). Psychological safety, trust and learning: A group-level
lens. For publication by Roderick Kramer and Karen Cook, (Eds) in Trust:
Emerging Perspectives, Enduring Questions, New York: Russell Sage.
Edmondson, A. C. (2003a). Managing the risk of learning: Psychological safety in work
teams. In West, M. (Ed.), International Handbook of Organizational Teamwork
and Cooperative Working. London: Blackwell, pp. 255-275.
Edmondson, A.C. (2003b). Speaking up in the operating room: How team leaders
promote learning in interdisciplinary action teams. Journal of Management
Studies. 40:6, 1419-1452.
Elliott, E. S., & Dweck, C. S. (1988). Goals: An approach to motivation and
achievement. Journal of Personality & Social Psychology, 54, 5-12.
Farson, R. R Keyes, (2002). The failure-tolerant leader. Harvard Business Review; 80(8),
Ford, C., & Sullivan, D. M. (forthcoming). A time for everything: How the timing of
novel contributions influences project team outcomes. Journal of Organizational
Frese, M., & M. Baer. (forthcoming). Innovation is not enough: Climates for initiative,
psychological safety, process innovations, and firm performance. Journal of
Gersick, C. J. G. (1988). Time and transition in work teams: toward a new model of
group development. Academy of Management Journal,31, 9-41.
Hackman, J. R. (1987). The design of work teams. In J.W. Lorsch (Ed.), Handbook of
organizational behavior (pp. 315-342). Englewood Cliffs, NJ: Prentice Hall.
Hansen, M. T. (1999) "The search-transfer problem: The role of weak ties in sharing
knowledge across organization subunits." Administrative Science Quarterly
Janis, I.L. (1982). Victims of groupthink (2nd ed.). Boston, MA: Houghton Mifflin.
Kramer, R. M. (1999). Trust and distrust in organizations: Emerging perspectives,
enduring questions. Annual Review of Psychology, 50, 569-598
Leary, M. R. (2001). Social anxiety as an early warning system: A refinement and
extension of the self-presentation theory of social anxiety. In S. G. Hofmann & P.
M. DiBartolo (Eds.), From social anxiety to social phobia: Multiple perspectives.
Needham Heights, MA: Allyn & Bacon.
Leary, M. R., & Atherton, S. C. (1986). Self-efficacy, social anxiety, and inhibition in
interpersonal encounters. Journal of Social and Clinical Psychology, 4(3), 256-267.
Leary, M. R., Atherton, S. C., Hill, S., & Hur, C. (1986). Attributional mediators of
social inhibition and avoidance. Journal of Personality, 54(4), 704-716.
Leary, M. R., Knight, P. D., & Johnson, K. A. (1987). Social anxiety and dyadic
conversation: A verbal response analysis. Journal of Social and Clinical
Psychology, 5(1), 34-50.
Leary, M. R., & Kowalski, R. M. (1995). Social anxiety. New York: Guilford Press.
Lee, F., Edmondson, A., Thomke, S., & Worline, M. (forthcoming). The mixed effects of
inconsistency on experimentation in organizations. Conditional acceptance,
Lewis, M. W., Welsh, M. A., Dehler, G. E., & Green, S. G. (2002). Product development
tensions: Erxploring contrasting styles of product management. Academy of
Management Journal 45(3), 546-564.
Peters, T. J. & Waterman, R. H. (1982). In search of excellence. New York: Warner
Pichanick, J. S. (2003). The regulation of well-being: Growth and action orientations.
Unpublished doctoral dissertation, Harvard University.
Lee R. & R. E. Nisbett (1991). The person and the situation: Perspectives of social
psychology. Philadelphia: Temple University Press
Ryan, K. D., & Oestrich, D. K. (1998). Driving fear out of the workplace. (2nd ed.). San
Schein, E. H. (1985). Organizational culture and leadership. San Francisco: Jossey-
Schippers, M.C. (2003). Reflexive learning in teams. Unpublished doctoral dissertation,
Vrije Universiteit, Amsterdam.
Schlenker, B. R., & Leary, M. R. (1982). Social anxiety and self-presentation: A
conceptual model. Psychological Bulletin, 92(3), 641-669
Stasser, G. & Titus, W. (1985). Pooling of unshared information in group decision
making: Biased information sampling during discussion. Journal of Personality
and Social Psychology, 48, 1467-1478.
Stasser, G. (1999). The uncertain role of unshared information in collective choice. In L.
Thompson, J. Levine, & D. Messick (Eds.), Shared Cognition in Organizations
(pp. 49-69). Mahwah, NJ: Lawrence Erlbaum Associates.
Tynan, R. (forthcoming). The effect of threat sensitivity and face giving on dyadic
psychological safety and upward communication. Journal of Applied Social
Wanous, J.P., A. E. Reichers, M.J. Hudy, (1997). Overall job satisfaction: How good are
single-item measures? Journal of Applied Psychology, 82(2): 247-252.
West, M. A. (1990). The social psychology of innovation in groups. In M. A. West and J.
L. Farr (Eds.), Innovation and Creativity at work: Psychological and
organizational strategies (pp. 309-333). Chichester: John Wiley & Sons.
Wheelwright, S. C. and K. B. Clark (1995). Leading Product Development: The senior
manager's guide to creating and shaping the enterprise. New York: Free Press.
Woolley, A. W. (2003). The antecedents and consequences of procedural orientation in
work teams. Unpublished Doctoral Dissertation, Harvard University.
Table 1: Company and Team Variance in Psychological Safety at Time 1
Note. R2 = .37 (Adjusted R2 = .28)
Table 2: Company and Team Variance in Psychological Safety at Time 2
Square F p<
Corrected Model 12.47 24 0.52 2.38
Intercept 1368.95 1 1368.95
Company 7.54 6 1.26 5.75
Team(Company) 3.84 18 0.21 0.98 0.50
Error 33.43 153 0.22
Total 1728.08 178
Corrected Total 45.90 177
Note. R2 = .27 (Adjusted R2 = .16)
Source Sum of
Squares Df Mean
Square F p<
Corrected Model 22.13 25 0.89 4.28 0.00
Intercept 1586.32 1 1586.32 7664.5
Company 9.25 6 1.54 7.45 0.00
Team(Company) 12.02 19 0.63 3.06 0.00
Error 38.29 185 0.21
Total 2037.30 211
Corrected Total 60.42 210
Table 3: Company and Team Variance in Psychological Safety at Time 3
Square F p<
Corrected Model 19.36 25 0.77 3.24
Intercept 1122.90 1 1122.90
Company 9.27 6 1.54 6.45
Team(Company) 8.40 19 0.44 1.85 0.05
Error 32.07 134 0.24
Total 1496.24 160
Corrected Total 51.44 159
Note. R2 = .38 (Adjusted R2 = .26)
Table 4: Correlations among Psychological Safety and Independent Variables
1 2 3 4 5 6 7 8 9 10 11 12 13
1. PS 1
2. PS 2 .68**
3. PS 3 .63** .78**
4. Org Clim 1 .36** .35** .32**
5. Org Clim 2 .36** .35** .33** .82**
6. Resources 1 .25** .27** .31** .26** .26**
7. Resources 2 .13† .19* .23** .13* .22** .78**
8. Team Inter 1 .50** .48** .51** .65** .55** .33** .20**
9. Team Inter 2 .44** .47** .54** .58** .59** .32** .23** .81**
10. Goal Clar 1 .31** .25** .30** .44** .44** .32** .17* .47** .42**
11. Goal Clar 2 .25** .27** .32** .41** .50** .31** .29** .34** .47** .81**
12. Neuroticism -.17* -.25** -.35** -.08 -.13† -.11† -.04 -.08 -.17* -.13* -.14*
13. Extraversion .10 .19* .17* .15* .17* -.04 -.08 .15* .19* .14* .13† -.46**
14. Openness -.01 -.05 .13 .10 .10 -.02 -.05 .04 .02 .07 .05 -.19** .29**
† p < .10
* p < .05
** p < .01
Table 5: General Linear Model of Psychological Safety at Project Midpoint
Square F p<
Corrected Model 19.65 28 0.70 3.97 0.001
Intercept 17.47 1 17.47 98.9
Company 4.20 6 0.70 3.96 0.001
Team (Company) 2.47 18 0.14 0.78 0.75
Team Interactions 2.03 1 2.03 11.5
Goal Clarity 0.14 1 0.14 0.78 0.40
Neuroticism 2.27 1 2.27
Openness 0.62 1 0.62 3.53 0.06
Error 25.96 147 0.18
Note. R2 = .43 (Adjusted R2 = .32)
Table 6: General Linear Model of Psychological Safety at Project End
Square F P<
Corrected Model 30.06 29 1.04 6.57 0.001
Intercept 8.16 1 8.16 51.6
Company 4.78 6 0.80 5.05 0.001
Team (Company) 6.69 19 0.35 2.23 0.001
Team Interactions 1.36 1 1.36 8.59 0.001
Goal Clarity 1.65 1 1.65
Neuroticism 2.75 1 2.75
Openness 0.22 1 0.22 1.41 0.25
Error 20.21 128 0.16
Note. R2 = .60 (Adjusted R2 = .51)
Figure 1. Predicted multiple level influences on psychological safety over time.
First Half Project Influences Second Half Project Influences
First Half Project Influences Second Half Project Influences
Figure 2: Multiple level predictors of psychological safety across time.
First Half Project Influences Second Half Project Influences
First Half Project Influences Second Half Project Influences