ContentServer 3
ContentServer 3
Psychological Safety
Psychological Safety
that something did not work as planned, the ability to discuss them productively has been associated with organizational effectiveness (Michael, 1976; Sitkin, 1992; Schein,
1993). On one hand, if feedback seeking, experimentation,
and discussion of errors individually promote effective performance, learning behaviorwhich includes all of these activitiesis also likely to facilitate performance, whether for individuals or teams. On the other hand, learning behavior
consumes time without assurance of results, suggesting
that there are conditions in which it may reduce efficiency
and detract from performance, such as when teams are responsible for highly routine repetitive tasks with little need
for improvement or modification. For teams facing change or
uncertainty, however, the risk of wasting time may be small
relative to the potential gain; in such settings, teams must
engage in learning behavior to understand their environment
and their customers and to coordinate members' actions effectively. Moreover, teams that perform routine production
tasks may still require learning behavior for effective selfmanagement as a team and for intermittent process improvement:
Hypothesis 1 (H1): Learning behavior in teams is positively associated with team performance.
Team Psychological Safety
Team psychological safety is defined as a shared belief that
the team is safe for interpersonal risk taking. For the most
part, this beiief tends to be tacittaken for granted and not
given direct attention either by individuals or by the team as
a whole. Although tacit beliefs about interpersonal norms are
sometimes explicitly discussed in a team, their being made
explicit does not alter the essence of team psychological
safety. The construct has roots in early research on organizational change, in which Schein and Bennis (1966) discussed
the need to create psychological safety for individuals if they
are to feel secure and capable of changing. Team psychoiogicai safety is not the same as group cohesiveness, as research has shown that cohesiveness can reduce willingness
to disagree and challenge others' views, such as in the phenomenon of groupthink (Janis, 1982), implying a lack of interpersonal risk taking. The term is meant to suggest neither a
careless sense of permissiveness, nor an unrelentingly positive affect but, rather, a sense of confidence that the team
will not embarrass, reject, or punish someone for speaking
up. This confidence stems from mutual respect and trust
among team members.
The importance of trust in groups and organizations has long
been noted by researchers (e.g,, Golembiewski and McConkie, 1975; Kramer, 1999). Trust is defined as the expectation that others' future actions will be favorable to one's
interests, such that one is willing to be vulnerable to those
actions (Mayer, Davis, and Schoorman, 1995; Robinson,
1996). Team psychological safety involves but goes beyond
interpersonal trust; it describes a team climate characterized
by interpersonal trust and mutual respect in which people
are comfortable being themseives.
For team psychoiogicai safety to be a group-ievel construct,
it must characterize the team rather than individuai members
354/ASQ, June 1999
Psychological Safety
of the team, and team members must hold similar perceptions of it. Previous researchers have studied the similarity
of beliefs in social systems such as organizations and work
groups (for reviews, see Klimoski and Mohammed, 1994;
Walsh, 1995). Perceptions of psychoiogicai safety, like other
such beliefs, should converge in a team, both because team
members are subject to the same set of structural influences and because these perceptions develop out of salient
shared experiences. For example, most members of a team
wiil conclude that making a mistake does not lead to rejection when they have had team experiences in which appreciation and interest are expressed in response to discussion
of their own and others' mistakes.
Team psychoiogicai safety should facilitate learning behavior
in work teams because it alleviates excessive concern about
others' reactions to actions that have the potential for embarrassment or threat, which learning behaviors often have.
For example, team members may be unwilling to bring up
errors that couid hefp the team make subsequent changes
because they are concerned about being seen as incompetent, which allows them to ignore or discount the negative
consequences of their silence for team performance, in contrast, if they respect and feei respected by other team members and fee! confident that team members wili not hold the
error against them, the benefits of speaking up are iikely to
be given more weight. Support for the centraiity of interpersonal inferences in groups is found in research on distributive justice, which shows that peopie are more focused on
relational than instrumental considerations in their assessments of allocation decisions made by authority figures;
people are very attentive to the tone and quality of social
processes and are more willing to comply with these when
they feel valued (Tyier and Lind, 1992). Argyris and Schon
(1978) made a connection between interpersonal threat and
learning when they posited that interpersonaliy threatening
issues impede learning behavior, but they did not address
the possibility that dyads or groups may differ in their tacit
beliefs about interpersonal threat, thereby giving rise to different levels of learning. In contrast, I propose that psychological safety varies from team to team, such that otherwise
interpersonaliy threatening learning behavior can occur if the
team has a sufficiently safe environment:
Hypothesis 2 (H2): Team psychoiogicai safety is positively associated with iearning behavior in organizational work teams.
Psychological safety does not play a direct role in the team's
satisfying customers' needs, the core element of performance; rather, it facilitates the team's taking appropriate actions to accomplish its work. Thus, learning behavior should
mediate the effects of team psychological safety on performance outcomes:
Hypothesis 3 (H3): Team learning behavior mediates between
team psychological safety and team performance.
Team Efficacy and Team Learning
Building on earlier work on the role of self-efficacy in enhancing individual performance (Bandura, 1982), a body of
research has established group efficacy as a group-level phe355/ASQ, June 1999
nomenon (e.g., Guzzo et al., 1993) and also reported a relationship between group efficacy and performance (Lindsley,
Brass, and Thomas, 1995; Gibson, 1996). This work has not
specified mechanisms through which shared perceptions of
efficacy lead to good performance, and one possibility is that
efficacy fosters team members' confidence, which promotes
learning behavior and helps accomplish desired team goals:
Hypothesis 4 (H4): Team efficacy is positively associated with
team learning behavior.
Team members deciding whether to reveal errors they have
made are likeiy to be motivated to speak up if two conditions are satisfied: first, they believe they wili not be rejected (team psychological safety) and, second, they believe
that the team is capable of using this new information to
generate useful results (team efficacy). Team psychoiogicai
safety and team efficacy are thus complementary shared
beliefs, one pertaining to interpersonal threat and the other
characterizing the team's potential to perform. Team efficacy
thus shouid supplement team psychoiogicai safety's positive
effect on team iearning:
Hypothesis 5 (H5): Team efficacy is positively associated with
team learning behavior, controlling for the effects of team psychological safety.
Team Leader Coaching and Context Support as
Antecedents of Team Psychological Safety
A set of structural featuresconsisting of a clear compelling
team goal, an enabling team design (including context support such as adequate resources, information, and rewards),
along with team leader behaviors such as coaching and direction settinghave been shown to increase team effectiveness (Hackman, 1987; Wageman, 1998). These structural
features provide a starting point for examining antecedents
of team psychoiogicai safety. The extent of context support
experienced by a team shouid be positively associated with
team psychoiogicai safety because access to resources and
information is likely to reduce insecurity and defensiveness
in a team. Team leader coaching is also likeiy to be an important influence on team psychoiogicai safety. A team leader's behavior is panicularly salient; team members are likeiy
to attend to each other's actions and responses but to be
particularly aware of the behavior of the leader (Tyler and
Und, 1992). If the leader is supportive, coaching-oriented,
and has non-defensive responses to questions and challenges, members are ^keiy to conclude that the team constitutes a safe environment. In contrast, if team leaders act in
authoritarian or punitive ways, team members may be reluctant to engage in the interpersonal risk involved in learning
behaviors such as discussing errors, as was the case in the
study of hospital teams mentioned above (Edmondson,
1996). Furthermore, team ieaders themseives can engage in
learning behaviors, demonstrating the appropriateness of and
iack of punishment for such risks.
Hypothesis 6 (H6): Team leader coaching and context support are
positively associated with team psychoiogicai safety.
Through enhancing psychological safety, team ieader coaching and context support are iikely to facilitate team iearning.
356/ASQ, June 1999
Psy^ological Safety
Team
Team
Beliefs
Behaviors
Outcomes
TEAM LEARNING
TEAM
STRUCTURES
Context support
Team leader
coaching
TEAM
PERFORMANCE
TEAM SAFETY
TEAM EFFICACY
Seeking feedback,
discussing errors,
seeking information - *
and feedback from
customers and others
METHODS
To test the hypotheses in the team iearning modei, i studied
reai work teams in an organization that has a variety of team
types, using a combination of qualitative and quantitative
methods to investigate and measure the constructs in the
modei. Preliminary observation and interviews In the organization suggested that there was considerable variation in the
extent to which teams engaged in \earning behavior, making
it a good site in which to explore the phenonnenon and to
investigate factors associated with team learning.
Research Site and Sample
"Office Design incorporated" (ODD, a manufacturer of office
furniture with approximately 5,000 employees and a reputation for product and management innovation, provided the
research site for this study. Teams in this company, implemented in 1979 to promote employee participation and
cross-functional collaboration, consisted of four types. Most
were functional teams, made up of managers or supervisors
and direct reports, and these included sales teams, management teams, and manufacturing teams; this type of team
existed within and supported the work of a single functional
department. Although encompassing dyadic reporting relationships, functional teams had shared goals, and members
were interdependent in reaching them. As with other teams
at ODI, they aiso typically had some training in teamwork.
Second, ODi had a growing number of self-managed teams
in both manufacturing and sales; these teams consisted of
peers from the same function. The third type was time-iimited cross-functionai product deveiopment teams, and the
fourth was time-limited cross-functional project teams, convened to work on other projects that involved muitipie departments. The company was willing to participate in this
research to obtain feedback on how well its teams were
working.
My primary contact at ODI was a manager in an internal organization development group who worked closely with me
to facilitate data collection. She scheduled interviews and
meetings, recruited teams to participate in the study, and
identified recipients of the work of each of these teams. As
ODI did not have a central roster of all work teams, she distributed a memo to managers throughout the company describing the goal of the study (to assess team effectiveness
at ODI) and asking for lists of teams in their area. This
yielded a list of 53 teams, encompassing differences in organization ievei, department, type, size, self- versus leadermanaged, and tenure or team age. At the time of survey
data collection, the oldest team had been together for about
seven years, and the newest had been in place for four
months; both the oldest and newest teams were production
teams. These 53 teams included 34 functionai teams (in
saies, manufacturing, and staff services such as information
technology and accounting), nine self-managed teams (in
manufacturing and sales), five cross-functional product development teams, and three cross-functionai project teams. As
the purpose of the study was to test a theoretical model
rather than to describe properties of this particular organiza358/ASQ, June 1999
Psychological Safety
Tabie 1
Construct Development from Preliminary Qualitativtt Data*
Constructs
Positive form
Negative form
Beliefs about the team interpersonal context (inferred from informant quotes)
Members of this team respect each
other's abilities.
Psychological Safety
Table 1 {continuedi
Construct Development from Preliminary Qualitative Data*
Constructs
Positive form
Negative form
"This team gets stuck,, ,, It's hard to
get a decision. The dynamics are that
the conversation gets shut down,"
(Management team 2)
Experimenting
NPD = new product development. Text in italics became the basis of a new survey item.
Psydiological Safety
Table 2
Chronbach's Alpha and lntercoirelations between Group-level Survey Variables*
Variable
1, Context support
2, Team leader coaching
3, Team psychoiogicai
safety
4, Team efficacy
5, Team learning behavior
6, Team performance
7, Internal motivation
8, Job involvement
9, Team tenure
{m veatsl
10, Average company
tenure (in years)
11, Team learning
(observer rated)
12, Team performance
(observer rated)
Mean
S.D.
4.78
3,77
-97
-81
.65
,69
.80
5.25
5.07
4,67
5,10
6,11
3,30
1-03
1.07
,93
1,03
-68
1,69
,70
,70
,68
,60
(,03)
-,16
2,40
1,70
(-,061
9,00
6,70
3,48
4,95
,63
,50
,63
,45
{-,06)
--22
.82
,50
,80
,72
,15
{-,07)
.63
.50
-50
(-,02)
--26
.78
,71
,12
-,09
.76
.33
(-,01)
.64
,31
.34
-,26
-,15
-,16
-,09
i.O5)
ton
,33
--31
,26
,15
,17
,14
,77
,49
-.48
,60
,52
,60
,34
-,16
( -,02)
-,21
,30
.84
1,29
,48
-,50
,47
,43
,52
,36
-,11
-,12
-,21
-22
,81
{,06) ( -,01)
10
11
12
t
,16
.87
* Chronbach's alpha coefficients are presented on the diagnonal. Correlations in parentheses not significant at p < ,05; all other
correlations are significant at p < ,05,
t Only 1 survey item.
Psychological Safety
Table 3
Analysis of Variance and Intraclass Correlation Coefficients for
Group-Level Scales
Team survey variables
ICC
Context support
Team leader coaching
4.80
4.88
<,001
<,001
.29
,30
6,98
5,70
<.OO1
<.OO1
,39
.34
5.79
6.02
<.OO1
<.OO1
.27
,35
Internal motivation*
Job involvement*
1,13
1.25
.07
,06
,03
,04
M50,135)
ICC
2.27
2,90
<,001
<,001
,19
.21
Two variables that are conceptually individual-level variables are included for
purposes of comparison, to demonstrate the contrast between these results
and those for the variables from the same survey that are conceptually grouplevel. One-way ANOVA shows these two variables are not significantly different across teams, in contrast to the group-levei variables, which are significant
to the p < .0001 level.
To test hypotheses relating team shared beliefs, iearning behavior, and performance, I conducted a series of regression
analyses, using customers' ratings of team performance as
the dependent variable and measures obtained from team
members as regressors. Because respondents belonging to
the same team are not independent, i performed regression
anaiyses on the group-levei data set (N = 51) to avoid vioiating the regression assumption of independence. The resuits
are shown in tabie 4. First, regressing team learning (selfreported) on team performance (observer-rated) reveals that
learning behavior is a significant predictor of team perfor365/ASQ, June 1999
Psychoiogicai Safety
Table 4
Regression Models of Observer-assessed Team Performance (/Vs 51)
Modei
Variable
Constant
Team learning behavior
Context support
Team leader coaching
Team psychological safety
Team efficacy
Adiusted fl-squared
(1)
(2)
(3)
(4)
1,31
,80-
,44
.40
.22
,45
.87
1.41
(5)
2.04"
,43
,57
(6)
1,48
(7)
2,32"
,75,93,67-
.26
,27
,26
,21
,60,17
.22
,23
(5)
(6)
(7}
1,41-
1.94-
.42
,51"
-,05
,28
,14
.66
(1)
(2)
(3)
(4}
,69
,76-
2,45-
,41
,70
,11
1,06
,45,63
.23
,63
,52,31
,52
,69,45
,75"
,38
,99*
.46-
1.5.38-
.35
,26
,48
,35
,22*
,40
1.27*
1,53-
.27*
.23
.26
.40"
,23
1,73"
,46,21
.53
.33"
,21
-,07
,12
.36
Self-report learning
behavior
B
FC
Independent variable
.46
5.26
<.OO1
,33
,76
9.16
<,001
,63
.33
,56
1.89
3-82
.06
<.OO1
,52
,33
.56
1,89
3.82
,06
<,001
.52
,29
.09
.12
2,46
.66
.78
.02
.51
,21
,33
,51
,24
,14
4,56
1.83
1.00
<,001
,07
,32
,66
Psychological Safety
Table 7
Tests of Team Efficacy as a Mediator between Coaching, Context Support, and Leaming
Observer-assessed
learning behavior
Self-report learning
behavior
ff
.38
4,27
<,001
.26
,01
,77
,02
,67
06
5,01
.21
444
,95
<,001
,83
<,001
Independent variable
Team efficacy
.45
3.97
<,001
,01
,77
,26
.20
.06
5.01
2.05
1,46
,95
<.OO1
,05
,15
,23
,49
,27
,49
,52
rr = .OJ
Independent variable
Fratio
Team type
Team membership*
Context support
Team deader coaching
Team tenure
R3,51)
f(50,427)
F{1,427)
fll,427)
Fi:i,427)
=
=
=
=
=
2,02
3.25
26.83
39.81
0.10
Team
Team
Team
Team
Team
F13,51) =
fl50,427) =
F(1,427) =
F(1.427)=
F(1,427) =
2.21
2.64
42.21
10.52
0.22
type
membership*
psychoiogicai safety
efficacy
tenure
.12
<.OO1
<,001
<.OO1
,74
,10
<,001
<.OO1
<,001
,49
* Team membership is the categorical variable identifying each team. The result that team membership accounts for
significant variance in team psychoiogicai safety or in team teaming behavior indicates that variance is attributable to
unexplained effects of belonging to the same team.
369/ASQ, June 1999
Psychological Safety
S-5 o
CD " n
>
F
= E
o
idn
CO
XI
D)
*-' ,E!? CJ
^ "ci . 2
c
'c
ro g p XJ
5
.5
-5
-o
O O)S
CD
hro
"03
CL
O ro ro ro
C X .c ^ cn
Xl in , =
<^
>
?r^
CO =
"D ro c
X) 2. n
ro
ro 10
a
ro
,c D
E "o
"o
: C CD ro -Cl -,
"5
! , .c ro 0)
rol
omer
CO
?>
to
XI
E
CD ;
X)
D *.*
pro
CO
""~
X!
ent
_
i
I
CD
= J
ro
in
CO d)
^ S^tf
CD
"S
CD
O)
o
u
>.
w
rent
|OAUi
'&
cn
"(5
E,
CO
co
ro
E .c
ro
CD
CO
t^
"
;. o
TO
a>
t-
.c
S5 5
u
03 -
D. ^ o c
O
>
CD "S
-T n
CJ
03
P 03
o E c
u
E
Q
CD
03 a
*-
>
. c ^ o
-O
>-
o5
,E S E
o
">
TO
ro
ro i^ <
J:
d)
Xl
X)
w
CD
"D
_m
E
E
CQ
E
CD
il
CJ -
IO
_^^
03
.c
XI
1^
*^ '-' t^ i^ =
cy
T3) O ro c (0 . ?
5cT3cDOT),t;t
"O
ro
'c
x:
-^ ^
CO
O>
_ 03 p
CO
E E _ro S o
o ^ c
cn
cn Xl
E '5 u
o
CD
to
D.
Q.
t-
Dl
C
'c
in
iiti
c
CB
Tea
UIO
(S
Quail
>
table
r-
to
o -^
to 5.
c S-03 S S
8 "3 -p E 5 1"
= 1 o
D.E
III
III
3
n
372/ASQ,June 1999
o I S ^KH
liillil
o i2
Psychological Safety
JOI
CO
.c
.13
rni
DJ
C
CO
0
c
CO
x:
CO
5
XI
CO
berl
r s ^"
c c
CJ
ro
til
c
03
ro
D0
E 0
0
0
ffi tli to ^ .
*- -S -c jr -,
Ec
CC
E
03
E
0
0
c XI
O Q
>
,'5
" ro ^
.? ^ >
ro cn o
to
a.
0
CO
iE
o^
i ^ 03
c ,^ o
E
o
ro.o
E .9 I
"^ O
X! 01 O
E.E.-
..roc Q.
Cfl . C CO
c!- E H '
w u . c cn
. 0 ro
T:
CO
'0
E
CO
0
0
E
03
<.X
(-
5 E
0
ro
"ro 0
Q.
Q.
CO
0 0
CD
CO
C 2
CD
p ro
03
CT
sop
15
CD
E
ro b
03
a.
0
0
aboi
1CD
ro
_ ro
0
sic
0)
.^
nni
CO
ader
nizat
txpi
(O
tea
cn
ro
CD
ro c
c
0
03
ro
a
Q.
to
X
c
0
U
CD
CO
c
c
V t:
03
X) a:
if
lac
EX
a ro
a.
Z)
CO
deqi
CO
u
ro U
.2
c
_ro
c/
cr
O 03 U! Q
X) '
rzz E
Psychological Safety
ries lend weight to the premise that learning behavior in sociai settings is risky but can be mitigated by a team's tolerance of imperfection and error. This appeared to be a
tolerance (or lack of tolerance) that was understood by all
team members.
The results of the study supported the proposition that team
psychological safety affects learning behavior, which in turn
affects team performance. Quantitative analyses provided
consistent support for six of the eight hypotheses. This included support for two mediating relationships: learning behavior appears to mediate between team psychoiogicai
safety and team performance, and team psychoiogicai safety
appears to mediate the effects of context support and team
leader coaching on learning behavior. Data from team observers on team performance, independent of other data
sources, strengthen these resuits.
Two hypotheses^that team efficacy would be associated
with learning behavior when controlling for team psychological safety and that team efficacy mediates the effect of context support and leader behavior on team learningwere not
supported. This outcome may in fact strengthen the core
argument in this paperthat engaging in learning behavior in
a team is highly dependent on team psychological
safetyby suggesting that team members' confidence that
they will not be punished for a well-intentioned interpersonal
risk enables learning behavior in a way that team efficacy, or
confidence that the team is capable of doing its work, does
not. In contrast to the uneven results for efficacy, one of the
most striking resuits is the degree to which the proposed
relationship between team psychoiogicai safety and team
learning behavior received consistent empirical support
across severai analyses and independent measures. The implication of this result is that people's beliefs about how others will respond if they engage in behavior for which the outcome is uncertain affects their willingness to take
interpersonai risks. Because beliefs about team efficacy are
unrelated to this central interpersonai concern, it may be
iess important for learning behavior. Thus, the theoretical
premise that iies at the core of the team learning modei
does not appear to require the suppiementary effects of
team efficacy. Moreover, the conclusion that team psychoiogicai safety fosters team iearning behavior is both consistent with existing organizational iearning theory and has a
certain degree of face validity; that is, the juxtaposition of
team members' descriptions of the interpersonal context in
their team with their stories of learning behavior is not a surprising one.
Quantitative and qualitative resuits both suggested that context support accounts for variance in iearning behavior but,
aiso, that it provides an incompiete explanation. The quantitative data demonstrated a positive association between
context support and psychoiogicai safety, and the qualitative
data allowed isolation of specific cases from within this general trend that suggested different ways reai teams handle
the absence or presence of enabling design conditions. For
example, the Stain Team tacked context support and yet
was learning-oriented. The Publications Team, despite having
a simitar set-up and level of context support as two other
376/ASQ, June 1999
Psychological Safety
publications teams, showed substantially less learning behavior than they did. Thus, a focus on just these two teams
might suggest that context support and other features of
team design account for iittle variance in learning behavior.
In contrast, the seven high- and low-learning teams studied
provided more data and do suggest an important role for
team design in team iearning.
First, the four high- and three low-learning teams differed
markedly by team type. Functional teams were overrepresented in the iow-learning teams (two of three), and product
development, project, and self-managed teams made up
three of the four high-learning teams. The two groups also
differed somewhat on whether they exhibited key eiements
of a weil-designed team (cf. Hackman, 1987). The low-learning teams' tasks at the time of the study tended to iack interdependence; for example, in the Publications Team, each
member had his or her "own assignment"; other team
members could be used as resources, but as a design for
teamwork, the arrangement was suboptimal. But the fact
that the two other publications teams in the survey sample
had higher levels of learning demonstrates that the degree
of task interdependence can be modified through learning
behavior. Similarly, context support was adequate for the
three low-learning teams and inadequate for two of the four
high-learning teams, also illustrating that it is possible for
teams to overcome limitations in their context through learning behavior. These few cases thus provided evidence that
high-learning teams could overcome obstacles they faced in
their initial set-up; a lack of structural support did not seal
their fate. The Stain Team overcame personnel limits that
repeatedly depleted their ranks, and the Fusion Team (described in table 9) overcame time and staffing constraints to
push energetically forward on its shared project. In contrast,
low-learning teams, such as the Publications and Help Desk
teams, appeared vulnerable to a self-sealing pattern of members having private concerns about the team environment,
which led to withholding relevant thoughts and actions and
made it difficult to escape the low-learning condition. These
cases suggest an asymmetry, in which high-learning teams
can confront and work with design and other constraints to
improve their situation, while low-learning teams are far
more likely to get stuck and be unable to alter their situation
without intervention.
An integrative perspective that mirrors and reinforces the
results of the quantitative data can be articulated from the
seven cases; in this, team psychological safety is a mechanism that mediates between effective team design and
learning behavior. Effective team leader coaching and context support, such as access to information and resources,
appear to contribute to, but not to fuliy shape, an environment in which team members can deveiop shared beliefs
that well-intentioned interpersonal risks wili not be punished,
and these beliefs enable team members to take proactive
learning-oriented action, which in turn fosters effective performance. Quantitative results also suggest that team psychoiogicai safety mediates between team structures (context
support and coaching) and the behaviorai outcome of team
iearning. These findings have important implications for theo377/ASQ, June 1999
Psychological Safety
The cross-sectional survey design prevented a demonstration of causality and also limited my ability to explore dynamic issues. The theoretical model also leaves out the dynamic interaction that is likely to take piace in which team
psychological safety facilitates taking the risks of learning
behavior, which, when unpunished by the team, further reinforces team psychological safety. A teann's history includes
events that demonstrate to members that interpersonal risk
is or is not worthwhile, and thus both psychoiogica! safety
and learning may be influenced as much or more by the cumulative effects of interpreting these events as by initial design features. Some evidence of the effects of history could
be seen in the Stain Team, where informants contrasted
present conditions with those under an earlier leader. Nonetheless, how shared beliefs are created gradually in teams
over time as a consequence of minor events and subtle interactions cannot be assessed in this study, nor can whether
self-reinforcing cycles or spirals exist. Given the inherently
dynamic nature of learning, this snapshot approach provides
an incomplete picture. Issues of how team psychological
safety develops over time and how team iearning behavior
might alter undesirable structural factors warrant careful consideration and future research.
Finally, conducting the study in a single company imposed
limitations, suggesting caution in drawing conclusions for
teams in other organizations. Although there was considerable diversity across teams in work context, organization
levei, education, and tenure, the sample may not be representative of the full spectrum of possible teams in work organizations. Moreover, with 51 teams, the sample size is
small for multivariate analyses. The inclusion of four team
types is both a strength and a weakness of the study. On
one hand, unlike studies that include only one type of team,
such as sales teams or production teams, the findings cannot be said to be merely a function of the nature of the
team task. On the other hand, this inclusion also brought in
more variables than could be thoroughly tested with oniy 51
teams. Larger studies could strengthen the validity of the
findings.
CONCLUSION
Structural and interpersonal factors have been viewed in the
literature as alternative explanations for team effectiveness.
This study supported, instead, an integrative perspective, in
which structural and interpersonal characteristics both infiuence learning and performance in teams. In particular, the
results showed that psychological safety is a mechanism
that helps explain how previousiy studied structural factors,
such as context support and team ieader coaching, influence
behavioral and performance outcomes. Future team research
has much to gain by investigating how structural and interpersonal factors are interreiated rather than which is more
important. To do this, it is essentiai to study real work
teams. There was some evidence in this study that a team's
history matters in shaping psychoiogicai safety. Shared beliefs about how others will react are estabiished over time;
these cannot take shape in the iaboratory in a meaningful
way. Moreover, for the risks of learning to be salient, the
379/ASQ, June 1999
interpersonal consequences must matteras they do in ongoing work relationships. Studying learning in laboratory
groups is therefore likely to miss an essential source of variance. Beyond the need to study rea! groups, longitudinal research could help to develop an understanding of how psychological safety develops or erodes with changes in
membership, leadership, or context.
in this study, my focus on (earning behavior and its accompanying risk made the interpersonai context especially saiient; however, the need for iearning in work teams is likely
to become increasingly critical as organizational change and
complexity intensify. Fast-paced work environments require
learning behavior to make sense of what is happening as
well as to take action. With the promise of more uncertainty,
more change, and less job security in future organizations,
teams are in a position to provide an important source of
psychoiogicai safety for individuais at work. The need to ask
questions, seek help, and tolerate mistakes in the face of
uncertaintywhile team members and other colleagues
watchis probably more prevalent in companies today than
in those in which earlier team studies were conducted. This
may partially account for the empirical support I found for
the role of psychological safety in promoting performance in
these work teams; however, it also suggests that psychological safety and ways to promote it will be increasingly relevant for future research on work teams.
REFERENCES
Alderfar, Clayton P.
1987 "An tntergroup perspective on
organizational behavior," In
J, W. Lorsch {ed.), Handbook
of Organizational Behavior:
190-222. Engiewood Cliffs,
NJ: Prentice-Hall,
Alderfer, Clayton P., and D, L.
Brown
1972 "Designing an empathic questionnaire for organizational research." Journal of Applied
Psychoiogy, 56: 456-468,
Ancona, Deborah G., and David F.
Caldwell
1992 "Bridging the boundary: External activity and performance in
organizational teams," Administrative Science Quarterly, 37:
634-655.
Argote, Linda, Deborah Gruenfeld,
and Charles Naquin
1999 "Group learning in organizations." In M. E. Turner (ed.).
Groups at Work: Advances in
Theory and Research. New
York: Erlbaum (forthcoming),
Argyris, Chris
^982 Reasoning, Leaming and Action: Individual and Organizational. San Francisco: JosseyBass.
Brown, Roger
1990 "Politeness theory: Exemplar
and exemplary," In I. Rock
(ed.), The Legacy of Soiomon
Asch: Essays in Cognition and
Social Psychology: 23-37,
Hillsdaie, NJ: Erlbaum,
Campbell, D. T., and D. W. FIske
1959 "Convergent and discriminant
validation by the multitraitmuitimethod matrix," In D. N.
Jackson and S, Messick
(eds.). Problems in Human
Assessment, New York:
McGraw-Hill,
Campion, Michael A., Gina J.
Medsker, and A. Catherine Higgs
1993 "Relations between work
group characteristics and effectiveness: Implications for
designing effective work
groups." Personnel Psychology, 46: 823-850.
Cohen, Susan G., and Gerald E.
Ledford
1994 "The effectiveness of selfmanaging teams: A quasi-experiment." Human Relations,
47: 13-43,
Dewey, John
1922 Human Nature and Conduct,
New York: Holt,
1938 Logic: The Theory of Inquiry.
New York: Holt.
Psychological Safety
Dutton, Jane
1993 "The making of organizational
opportunities: An interpretive
pathway to organizational
change," In L, L. Cummings
and B. M. Staw (eds.). Research in Organizational Behavior, 15: 195-226. Greenwich, CT: JAI Press.
Edmondson, Amy C.
1996 " Learning-fTom mistakes is
easier said than done: Group
and organizational influences
on the detection and correction of human error," Journal
of Applied Behavioral Science,
32: 5-32,
Edmondson, Amy C , and
Bertrand Moingeon
1998 "From organizational (earning
to the learning organization."
Management Learning, 29:
499-517.
Gibson, Christina B.
1996 "They do what they believe
they can: Group-efficacy beliefs and group performance
across tasks and cultures."
Academy of Management
Joumal, 42: 138-152,
Goffman, Erving
1955 "On face-work: An analysis of
ritual eiements in social interaction." Psychiatry, 18: 213231,
Golembiewski, Robert T., and
Mark McConkie
1975 "The centrality of interpersonal trust in group process."
In Cary L. Cooper (ed.). Theories of Group Process: 131181, London: Wiley,
Goodman, Paul, S. Devadas, and
T. L. Hughson
1988 "Groups and productivity:
Analyzing the effectiveness of
self-managing teams," In J, P.
Campbell, R. J. Campbell, and
Associates (eds.). Designing
Effective Work Groups: 295327. San Francisco: JosseyBass.
Goodman, Paul, E. Ravlin, and M.
Schminke
1987 "Understanding groups in organizations." In L. L, Cummings and B. M. Staw (eds.).
Research in Organizational
Behavior, 9: 121-173. Greenwich, CT: JAI Press,
Guzzo, R. A., P. R. Yost, R. J.
Campbell, and G. P. Shea
1993 "Potency in groups: Articulating a construct." British Journal of Social Psychology, 32:
87-106.
Hackman, J. Richard
1987 "The design of work teams."
In J. Lorsch (ed.). Handbook
of Organizational Behavior:
315-342, Engiewood Cliffs,
NJ: Prentice-Hall,
Walsh, James
1995 "Managerial and organizational
cognition: Notes from a trip
down memory lane." Organization Science, 6: 280-321.
Watkins, Victoria J., and Karen E.
Marsick
1993 Sculpting the Leaming Organization: Lessons in the Art of
Systemic Change. San Francisco: Jossey-Bass.
Pearson fl correlations
43
,29
,26
,22
.26
.30
,24
.26
,29
,15
,23
,27
Task design
1, The work that this team does makes a difference for the people who receive or use it,
2, The work we do on this team ;fse/f provides us with plenty of feedback about how weil the
team is performing,
3, Those who receive or use this team's output rarefy give us feedback about how well our
work meets their needs.
,25
Clear direction
1. It is clear what this team is supposed to accomplish,
2. This team spent time making sure every team member understands the team objectives.
3, The team has invested plenty of time to cianfy our goals.
Team composition
1, Most people in this team have the ability to soive the problems that come up in our work,
2, All members of this team have more than enough training and experience for the kind of
work they have to do.
3, Certain individuals in this team lack the special skiiis needed for good team work.
,41
,38
,65
,26
,34
.27
Team efficacy
,37
.43
,28
is readily available.
,36
.38
,28
.49
,56
.32
,41
,35
.45
,37
,34
,34
.45
.37
,42
,43
,37
.33
,48
.41
,39
,38
,47
,70
Psychological Safety
APPENDIX (continued)
,23
.28
,35
,41
,30
,23
.26
,31
,27
,29
,22
.38
,35
,41
,37
,47
,47
,37
,37
,25
,43
,38
41
,44
,40
26
,42
,36
,47
.26
,51
,33
,48
Team performance*
1. Recently, this team seems to be "slipping" a bit in its level of performance and
accomplishments,
2. Those who receive or use the work this team does often have complaints about
our work,
3. The quality of work provided by this team is improving over time,
4. Critical quality errors occur frequently in this team,
5. Others in the company who interact with this team often complain about how it functions.
Internal motivation*
1, My opinion of myself goes up when I do my job well.
2, 1 feel bad and unhappy when I discover that I have performed less weli than I should have
in my job.
3, I feel a great sense of personal satisfaction when I do my job well.
,30
Job Involvement*
1, I live, eat, and breathe my job.
Team performance*
1,
2,
3,
4,
,24
55
46
,39
,37
,47
,52
,36
.61
,37
,47
,38
,30
,36
.51
,36
.44
.49
.22
44
,30
,77
,62
,57
,64
,71
,53