How Leaders Foster Self-Managing Team Effectiveness: Design Choices versus Hands-on
Coaching
Author(s): Ruth Wageman
Reviewed work(s):
Source: Organization Science, Vol. 12, No. 5 (Sep. - Oct., 2001), pp. 559-577
Published by: INFORMS
Stable URL: http://www.jstor.org/stable/3085999 .
Accessed: 30/04/2012 20:59
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.
INFORMS is collaborating with JSTOR to digitize, preserve and extend access to Organization Science.
http://www.jstor.org
How
Leaders
Team
Self-Managing
Foster
Effectiveness: Design
Hands-on
Choices
Versus
Coaching
Ruth Wageman
Dartmouth College, The Amos Tuck School, 100 Tuck Hall, Hanover, New Hampshire 03755-9023
ruth.wageman@dartmouth.edu
Abstract
This multi-methodfield studyexaminesthe relativeeffects of
two kinds of leaderbehaviors-design choices and hands-on
coaching-on the effectivenessof self-managingteams.Findings show thathow leadersdesign theirteams and the quality
of their hands-on coaching both influence team selfmanagement,the qualityof memberrelationships,andmember
satisfaction,but only leaders'design activitiesaffectteamtask
performance.Moreover,design and coachinginteract,so that
well-designedteams are helpedmoreby effective coachingand underminedless by ineffectivecoaching-than arepoorly
designedteams.
(Team Effectiveness; Team Leadership; Self-Managing Teams; Team
Coaching)
Considerable research has examined the effects of implementing self-managing teams on team performance outcomes and member satisfaction (e.g., Cohen and Ledford
1994, Cordery et al. 1991, Cummings and Griggs 1977,
Goodman et al. 1988, Jackson et al. 1994, Macy et al.
1991, Wall et al. 1986). The difficulties of fostering selfmanagement teams-particularly in organizations with
histories of individualistic, manager-directedwork-have
been well documented.Attemptsto create self-management
teams have often resulted in poor performance, individualistic behavior, and avoidance of the decision making
necessary for effective, cooperative teamwork (Cohen
and Ledford 1994, Cordery et al. 1991, Cummings and
Griggs 1977, Hackman 1998). These difficulties have
been attributedto deficits in the motivation and ability of
managers to create the conditions that foster selfmanagement (Goliembiewski 1995, Hut and Molleman
1998), as well as to resistance from team members in
taking on self-management (Balkema and Mollerman
1999, Wellins et al. 1991). Much less is known about the
kinds of leader activities that surmount these difficulties
1047-7039/01/1205/0559/$05.00
1526-5455 electronic ISSN
to create self-managing teams that both perform well and
sustain their self-managing character over time. This paper investigates the separate and joint effects of two quite
differentkinds of leader activities-team design and handson coaching-on the degree to which teams become selfmanaging and on their performance effectiveness.
The domain of the research is restricted to selfmanaging teams that have a designated team leader
(sometimes called a manager) who is not a regular member of the team. Excluded are teams that have no authority
for managing their own performance processes (that is,
whose only responsibility is to follow procedures specified by others in pursuit of those others' objectives), and
teams with no designated leader or whose leadership is
exclusively informal.
Team Self-Management and Effectiveness
Four general functions need to be accomplished whenever work is performed in a purposive organization
(Hackman 1987). First, a person or group must actually
execute the work. Second, a person or group must monitor
and manage work processes, initiating changes in pace
or procedure as needed. Third, a person or group must
structure the performing unit and its context, setting up
the task of the unit, staffing it, and arranging for organizational resources and supports. And fourth, a person or
group must specify the goals or objectives that are to be
accomplished.
A self-managing team, by definition, has authority and
accountability for the first two functions-executing and
managing the work-but within a structure and toward
purposes set by others. Thus, a team's level of formal
authority determines whether or not it falls within the
present domain-that is, whether it is a "self-managing
team." The degree to which self-managing team members
actively use their authority to manage their work processes, however, varies from team to team and is a key
ORGANIZATIONSCIENCE, ? 2001 INFORMS
Vol. 12, No. 5, September-October
2001, 559-577
RUTH WAGEMAN
How Leaders Foster Self-Managing Team Effectiveness
dependent variable in this research. Specifically, three behavioral indicators of self-management identified by
Hackman (1986) are examined: (1) the degree to which
team members take collective responsibility for the outcomes of their work; (2) the degree to which the team
monitors its own performance, actively seeking data
about how well it is doing; and (3) the degree to which
the team manages its own performance, making alterations in work strategies when circumstances change or
feedback indicates that a new approach may be needed.
Self-management is a behavioral process, and it is entirely possible for a team to be highly self-managing but
relatively ineffective, or to be manager-led and to be
highly effective. Team effectiveness is therefore also assessed for each team studied. Effectiveness is defined as
having three components: (1) task performance-the degree to which the team's product or service meets the
needs of those who use it; (2) group process-the degree
to which members interact in ways that allow the team to
work increasingly well together over time; and (3) individual satisfaction-the degree to which the group experience, on balance, is more satisfying than frustrating
to team members (Hackman 1990).
Team Leaders' Actions
The two kinds of leader activities explored in this research flow directly from how authority is partitionedfor
self-managing teams. Self-managing teams, as defined
here, do not have the authority to set or alter their purposes, structures, or organizational contexts. One type of
leader activity, therefore, is to establish those features
(that is, to design the team) in a way that fosters selfmanagement and performance effectiveness. On the other
hand, self-managing teams do have the authority to monitor and manage, as well as to execute, their work. A second potentially valuable leader activity, therefore, is to
provide hands-on coaching that helps a team manage itself and its work well. These two kinds of leader activities
are explicated in detail next.
Design Choices. When a leader designs a team, he or
she always uses some model that specifies how the team
ought to be set up and what organizational resources and
supports ought to be provided. Sometimes the leader's
model is explicit, and its implementation is deliberate;
other times the model is implicit, and implementation is
relatively mindless. Sometimes the leader is proactive,
exercising influence with peers or senior managers when
he or she does not have sufficient authority to create the
desired design; other times the leaderjust accepts existing
organizational conventions and arrangements.And sometimes the design the leader comes up with is a good one,
in accord with what is known about the conditions that
560
foster team effectiveness; other times the leader's model
may be significantly flawed.
A growing body of evidence specifies the structuraland
contextual features that promote team effectiveness. In
the present research, I draw upon both that literatureand
Hackman's (1987) conceptual model of work-team effectiveness to identify four general conditions that, when
present, foster self-managing team effectiveness.
(1) A real team. Organizations sometimes attempt to
achieve the purported benefits of teamwork by, for example, merely calling people who have similar job responsiblities a "team." Real teams are defined for present
purposes as bounded social systems with clear membership that is reasonably stable over time, thereby providing
the capability for members to behave as a collective. This
design condition, although often violated in practice, is
the prerequisite for all the rest.
(2) Clear direction. This is the degree to which the purposes of a team are stated clearly, few enough in number
to be memorable to both team members and leaders, and
focused on the ends to be achieved rather than on the
details of the means to be used in pursuing those ends
(Atkinson 1958, Bennis and Nanus 1985, Cohen et al.
1996, Locke et al. 1981, Walton 1985, Zander 1971).
(3) An enabling team structure. Team structure includes five basic design features: appropriateteam sizeno larger than the minimum required to accomplish the
work (Druskat 1995, Hackman and Vidmar 1970, Jackson
1992, Steiner 1972); optimal skill diversity, with substantial heterogeneity of task-relevant skills among members-but not so much that members have trouble coordinating their efforts (Ancona and Caldwell 1992,
Campion et al. 1993, Druskat 1995, Goodman and Shah
1992, Jackson 1992); task interdependence, such that
members are dependent upon one another to accomplish
the collective work of the team (Campion et al. 1993,
Wageman 1995); challenging task goals with "stretch"
performance targets (Cohen and Ledford 1994, Hackman
and Oldham 1980, Saavedra et al. 1993); articulatedstrategy norms, which legitimize and support active strategizing and long-term planning by the team, ratherthan the
mindless or reactive execution of the work (Hackman et
al. 1976, Woolley in press).
(4) A supportive organizational context. A supportive
context provides (a) a reward system that recognizes and
rewards excellent team performance (Cohen et al. 1996,
Hackman 1987, Rosenbaum et al. 1980, Shea and Guzzo
1987, Wageman 1995), (b) an information system that
provides information members require to competently
plan their collective work (Abramis 1990, Goodman
1986, Hackman 1987), (c) an education system that is
available to provide training or technical consultation for
ORGANIZATIONSCIENCE/Vol. 12, No.
2001
5, September-October
RUTH WAGEMAN
How Leaders Foster Self-Managing Team Effectiveness
any aspects of the work that members are not themselves
competent to handle (Klaus and Glaser 1970, Liang et al.
1995, Prince et al. 1992, Hackman 1987, Salas et al.
1993), and (d) the mundane material resources that are
needed to carry out the work (Hackman 1987).
The four general conditions summarized above are defined in this research by a total of eleven specific design
features: real team, clear direction, appropriatesize, skill
diversity, task interdependence, challenging task objectives, core strategy norms, team excellence recognized/
rewarded, information for planning available, training/
technical consultation available, and material resources
available (see Table 1 for a summary of conceptual definitions of these features). While Hackman's (1987) conceptual model of work team effectiveness was derived
from extensive empirical work and is heavily cited in the
teams literature, the present research represents the first
direct test of the full model. Here, I predict that to the
degree that team leaders have put these features in place
(or arrangedfor others to do so), self-managing teams will
do better than when these features are absent.
HYPOTHESIS 1. Well-designed teams exhibit more self-
management and are more effective than teams whose
designs are flawed.
Hands-on Coaching. Both scholars and practitioners
who write about self-managing team effectiveness focus
a great deal of attention on what leaders do in their dayto-day interactions with team members-that is, they focus on hands-on coaching (e.g., Barry 1991, Bass 1957,
Berkowitz 1953, Fiedler 1958, Jackson 1953, Likert
1958, Lippitt 1940, Manz 1986, Manz and Sims 1987,
Schlesinger et al. 1960). Coaching refers to direct interaction with the team that is intended to shape team processes to produce good performance. Pervading the literatureon team coaching is the view that leader coaching
behaviors can directly affect team members' engagement
with their task, their ability to work through interpersonal
problems that may be impeding progress, and the degree
to which members accept collective responsibility for
performance outcomes.
Some research evidence supports the view that coaching can improve both the quality of group processes
(Kaplan 1979, Schein 1988) and the level of member satisfaction (Cohen et al. 1996, Yukl 1989). Findings regarding the effects of coaching on performance outcomes, however, are, at best, mixed. Some studies have
found that operant-based coaching does facilitate team
performance (Komacki et al. 1989, Smoll and Smith
1989). Operant coaching refers to the direct reinforcement (generally through positive feedback) either of particular task behaviors when exhibited by the team members or of good performance directly. Operant studies of
ORGANIZATION
SCIENCE/VOl. 12, No. 5, September-October 2001
coaching often do not specify what team processes are
most beneficial for performance, however; one study of
operant coaching in which coaches attempted to affect
team performance through positive reinforcement of coordinating behavior failed to find any relationship between such coaching and performance outcomes (Komacki et al. 1989).
Further, Kaplan (1979) reviewed research on the effects of process consultation-a form of coaching aimed
specifically at improving the interpersonal relations of
team members-on team task performance and found no
studies to supportthe hypothesis that process consultation
improves performance (see also Woodman and Sherwood
1980). More recently, Cohen et al. (1996) found that "encouraging behavior" (providing feedback intended to enhance team motivation) from supervisors was negatively
associated with team performance as assessed both by
managers and by customers, and Beekun (1989) found
that self-managing teams that had no coaches significantly outperformed those that did.
In sum, existing research evidence suggests that leaders' coaching in some circumstances fosters team selfmanagement, the quality of members' interpersonal relationships, and member satisfaction with the team and
its work. But coaching alone (that is, without reference
to the quality of a team's design) may make little or even
a negative difference in how well a team actually performs. Moreover, it remains an open question as to just
what "competent" coaching is. Rather than adopt an a
priori conceptualization of coaching behavior, this research captured all nontrivial interactions between teams
and their leaders, grouped those interactions into substantive categories, and then empirically assessed the relationships between those emergent categories and measures of team self-management and effectiveness.
HYPOTHESIS 2. Teams that receive coaching
exhibit
more self-management, higher quality interpersonal relationships, and higher member satisfaction-but not
higher task performance-than do teams that receive no
coaching at all.
The Interaction of Team Design and Hands-on Coaching. There is reason to believe that design and coaching
may interact in shaping how, and how well, a selfmanaging team performs. Specifically, Hackman (1987)
has suggested that a leader's influence comes mainly from
his or her design choices, with his or her coaching activities making a difference only at the margins through
small adjustments in what is an already well-determined
trajectory. In this view, leaders have the opportunity to
coach a team to higher levels of self-management and
superior performance only when the team is relatively
well designed. If design conditions are stacked in favor
561
RUTH WAGEMAN
Table 1
How Leaders Foster Self-Managing Team Effectiveness
Coding of Team Design Variables
1. A real team
Definition:
Data from:
Trueif:
2. Cleardirection
Definition:
Data from:
Trueif:
Grouphas clear membershipand is stable over time.
Team interview,managerinterview,organizationalarchives
(1) Team membershipis clear (truefor all teams studied).
(2) Team membershipis stable-one or no membershipchanges in past year.
Statementof purpose forthe team is clear and is about ends and not means.
Team interview,manager interview
(1) Bothteam and managercan articulatedirection.
(2) Threeor fewer objectives are stated.
(3) Is about end states, not means.
3. Enablingstructure
A. Appropriateteam size
Definition:
Numberof members is adequate forthe workand no larger.
Data from:
Managerinterview,organizationalarchives
Trueif:
(1) Size is between 4 and 7 (fromfindingsof Wageman1995).
B. Optimalskilldiversity
Definition:
Team is composed of memberswithsubstantialheterogeneityof task-relevantskillsamong members.
Data from:
Team interview,team survey
Trueif:
(1) Skilldiversityis withinone scale pointof surveyoptimum.
C. Task interdependence
Definition:
Team has a group task such that members must worktogetherto accomplish it.
Data from:
Managerinterview
Trueif:
(1) Allfive of the followingtask elements are present:
(a) collective responsibilityfor customers, (b) team responsibilityfor partsexpenses, (c) collectivelydesigned
maintenancepractices, (d) frequentrequiredteam meetings, and (e) cross-trainingof membersfordifferent
machinetypes (fordetails, see Wageman1995).
D. Challengingtask goals
Definition:
Team has objective performancetargets thatare "stretch"goals.
Data from:
Team interview,team survey
Trueif:
(1) Presentobjectives requireperformanceexceeding previouslevels, withspecific deadlines and harddata about
whetheror not objectives are accomplished.
E. Core strategy norms
Definition:
Articulatednormsthat representexpectationsof strategic planning.
Data from:
Team interview,managerinterview,team survey
Trueif:
(1) Storieselicited fromteam and managerabout expectationsfor how team should respond to a change or
unfamiliarprobleminclude(a) conveninga problem-solvingmeeting,and/or(b) initiatingchange in work
practices, and/or(c) solicitingoutside assistance.
(2) The score is above midpoint on all four survey items assessing the team's normative expectations of members,
regarding(a) experimentingwithnew workprocedures,(b) seeking best practices fromotherteams, (c)
initiatingactions to solve problems,and (d) discussing what differentmembers have to contributeto the work.
4. Supportiveorganizationalcontext
A. Grouprewardsystem
Definition:
Group-levelexcellence is recognized and rewarded.
Data from:
Managerinterview,organizationalarchives
Trueif:
(1) 80%or more of availablerewardsare contingenton the team ratherthan individualperformance.
562
ORGANIZATION
SCIENCE/Vol. 12, No. 5, September-October 2001
RUTH WAGEMAN How LeadersFosterSelf-ManagingTeamEffectiveness
Table 1 (cont'd.)
Coding of Team Design Variables
B. Availableinformation
Definition:
Information
system providesthe data thatthe group needs to manage its work.
Datafrom:
Team interview,managerinterview
Trueif:
(1) Task relevantinformation
(e.g., about trends in machineperformance,customerfeedback, etc.) is given to
team.
is received once a week or morefrequently.
(2) Information
C. Availableeducation
Definition:
Educationalsystem providestrainingand technicalconsultation.
Datafrom:
Team is interview,managerinterview
Trueif:
(1) Team is trainedin both technicaland interpersonalskills.
(2) Technicaland interpersonaltrainingis readilyavailableto team.
D. Materialresources
Definition:
Groupis providedwiththe basic materialresources it needs to accomplishthe work.
Data from:
Team interview,managerinterview
Trueif:
(1) No materialresources are needed forthe workreportedas unavailableby eitherteam or manager.
of good performance, then the team coach can help the
team exploit its favorable circumstances. If the team is
poorly designed, on the other hand, attempts to foster
team effectiveness through hands-on coaching may be futile or even backfire. In such cases, the flawed design may
create dysfunctional processes so strong and preoccupying that coaching interventions risk merely adding to the
disarray (Wood 1990). Because the present research setting has significant variation both in leaders' design
choices and in their coaching behaviors, it offers a unique
opportunity to test this as yet untested line of reasoning.
HYPOTHESIS
3. Leaders' design activities and handson coaching interact in affecting team self-management
and effectiveness, with coaching having a greaterpositive
impactfor well-designed teams thanfor poorly designed
teams.
Method
Thirty-four self-managing teams, split between consistently superb performers and consistently poor performers, participatedin the research. Multiple measures of the
design features of the teams, team leaders' behaviors, and
level of team self-management were obtained using both
structuredinterviews with the teams and their leaders and
surveys completed by all team members. Quantitative
measures of team performance were obtained from organizational archives.
Research Site
The research requiredthat there be variation in the design
of teams. It was possible to locate an organization in
which a number of design elements of teams were known,
ORGANIZATION
SCIENCE/Vol. 12, No. 5, September-October 2001
based on prior research (Wageman 1995), to vary
widely-namely, the U.S. Customer Services Division of
Xerox Corporation. The service organization is divided
into nine geographical areas that are in turn subdivided
into districts. Each district consists of five to ten subdistricts, formed either on the basis of geography or type of
machines serviced. Each subdistrict is headed by a field
manager to whom 20 to 30 technicians report. The technicians are organized into work teams of between three
and nine individuals. The basic task of the teams is to
respond to customer calls about machine breakdowns and
to initiate visits to customer sites for preventive maintenance.
Team Selection Process
Teams were nominated for inclusion in the research by
field and district managers. Managers in 12 districts were
asked to nominate both superb teams and ineffective
teams in their districts. Superb teams were defined as
those that (1) consistently meet the needs of their customers, (2) have been operating increasingly well over time,
and (3) whose members are engaged in and satisfied with
their work. Ineffective teams were defined as those that
(1) frequently fail to meet the needs of their customers,
(2) have been operating increasingly poorly over time,
and (3) whose members are alienated from or dissatisfied
with their work. A clerical assistant sorted nominations
by district and identified the teams within each district
that met two criteria for inclusion: First, that the team had
received at least three manager nominations (of a possible
nine, on average) as either superb or ineffective, and second, that no team be included that received nominations
in both categories (no teams, in fact, did receive conflicting nominations).
563
RUTH WAGEMAN
How Leaders Foster Self-Managing Team Effectiveness
The assistant gave me a list of teams that met the nomination criteria, with categories removed, so that I could
remain blind to condition. I then selected a final sample
of 33 teams, of which 18 were superb teams and 15 were
ineffective teams. For efficiency of data collection, I included only teams from the seven districts where at least
three teams met the nomination criteria. These districts
were located in a variety of urban, suburban, and rural
settings across the nation.
Data Collection
Three research associates (graduate students) and I collected and coded the data. At the time of entry, the researchers knew only which teams in the district they
would be studying, and not their nomination category.
Two researchers collected data for each team, spending
two hours or more conducting a structuredteam interview
that elicited collective descriptions of significant events
in the team's life, the team's structureand organizational
context, and typical daily interactions with the team's
manager. Interviews with managers provided furtherdata
about team history, structure, and context, as well as
about the manager's own view of his or her interactions
with the team. The interview questions, adapted from
Hackman (1982), asked for straightforwarddescriptions
of various organizational features, not for interviewees'
opinions about whether a team was well or poorly designed. The three research associates and I test-coded the
first six interviews, and discrepancies were discussed and
resolved to increase interraterreliability for the remaining
coding. All subsequent interviews were coded by two
members of the research team.
Team members also completed a 108-item survey that
assessed team design, quality of group interaction processes, perceived task interdependence, use of problemsolving tools, and member satisfaction with the team and
its work. The survey was sent to members by mail and
collected at the team interview. Each survey was given a
unique code to identify the respondent's team, and teams
sealed their completed surveys together in an envelope
before handing them to the researchers. Thus, individual
team member responses were not identifiable either to the
team's leader, the team members, or the researchers. Response rate to the survey was 92%, and I had at least
three-fourths of the team members' responses from each
team in the research. Finally, organizational records were
consulted for information about team membership, distribution of financial rewards, and objective team performance.
Measures
Data from multiple sources using different data collection
methodologies was drawn upon in constructing all measures. The measure development strategy, measure descriptions, and psychometric data are reported below; the
564
team survey, team and manager interview protocols, and
details of coding procedures can be obtained from the
author.
Team Design. The 11 theory-specified design features
previously described were coded from the team and manager interviews, supplemented by data obtained from the
team survey and from organizational archives. Each of
the design features was coded dichotomously as either
being in place or not.' Coding procedures were highly
conservative, with a feature coded as "present" only if
there was convergence about its presence across multiple
indicators. In the event of a disagreement between the
codes assigned from team and manager interview data,
the team's survey score for the variable in question was
consulted. In the six cases that fell into that category, the
design feature was considered to be present for the team
only if its survey score was above the mean, and if descriptive data from either the manager or the team interview unambiguously indicated that the feature was, in
fact, in place. Intraclass correlations were computed for
all team-level variables to ensure that they could appropriately be aggregated ; ICCs for survey measures of the
team design features ranged from 0.18 to 0.41, all significant at p < 0.05. Overall quality of team design was computed as the total number of design conditions in place
for a team, from zero (no conditions present) to 11 (all
conditions present).2Coding conventions are summarized
in Table 1, and descriptive statistics and inter-raterreliabilities of design variables are included in Table 2.
Correlations among the individual design conditions
from the dichotomous coding range from -0.14 to 0.48,
with the highest being between task goals and strategy
norms, and between clear direction and task interdependence (see Table 6). These correlations probably are not
attributable to confounds introduced by data collection
methods because each condition was coded independently from descriptive accounts of work practices rather
than from self-report ratings. It is unlikely, for example,
that team and managerial descriptions of the amount of
interdependence in specific task elements would be spuriously affected by whether the team also had a clear and
engaging direction. Two explanations for these correlations are more likely. First, teams may not achieve excellence unless many of the design features are simultaneously present; consequently, the selection process may
have yielded only groups with either many or few of the
design features in place. Second, managers who create
favorable designs for their teams may tend to put multiple
features in place ratherthan just one or two. This pattern
of cooccurence of design features is consistent with
Hackman (1990).
ORGANIZATIONSCIENCE/Vol. 12, No. 5, September-October 2001
RUTH WAGEMAN
Table 2
How Leaders Foster Self-Managing Team Effectiveness
Team Design Descriptive Statistics
Data Source
GroupInterview
ManagerInterview
Team Designttt
M
Direction
0.61
Optimaldiversity
Appropriatesize
Stabilityof membership
Task interdependence 0.56
0.47
Task goals
0.62
Core strategynorms
0.63
Grouprewards
0.33
Information
supports
0.71
Educationsupports
0.70
Materialresources
Survey
r withother
SD IRR* design features** M
r withother
SD IRR* design features** M
0.50 0.94
0.37
0.54
0.28
0.48
0.34
0.51
0.49
0.31
0.52
0.92
0.84
1.00
1.00
0.20
0.08
0.22
-0.01
0.50
0.50
0.41
0.49
0.48
0.46
0.50
0.29
0.04
0.11
0.20
0.06
0.12
0.20
0.56
0.55
0.52
0.29
0.57
0.73
0.39
0.51
0.48
0.50
0.50
0.51
0.90
0.86
1.00
1.00
0.95
0.94
0.07
0.41
0.20
-0.08
0.09
0.13
1.00
0.89
0.91
1.00
1.00
0.99
0.98
SD
r within
r withother
scale (o)t design featurestt
4.96 1.12 0.44(0.77)
2.04 0.98
5.41 1.32
2.40
4.09
3.69
4.99
5.30
5.19
5.30
0.51
1.13
0.61
1.06
1.00
1.05
1.04
0.29(0.61)
0.58(0.78)
0.40(0.87)
0.32(0.69)
0.49(0.83)
0.49(0.80)
0.46(0.77)
0.21
0.09
0.17
0.11
0.33
0.19
0.25
0.26
0.22
0.30
reliabilities
*Interrater
**Average correlations with all other team design variables measured using same source (manager or team interview)
tAverage correlationof items withinscale and (internalconsistency reliability,i.e., Cronbach'salpha)
ttAverage correlationswithitems fromotherscales withinthe same survey section
tttFor overallteam design, which combines both across all design dimensionsand across sources, M = 6.12, SD = 1.39, and IRR= 0.90.
Leader Coaching. Coaching behavior was coded from
several questions asked in both the team and the manager
interviews. The team versions of these questions were:
"How often is your manager with the whole team? With
individuals? How does s/he use time with the team versus
with individuals? How frequently does s/he do those
things?" Rather than use an a priori coding scheme, I
allowed categories of behaviors to emerge from the behavioral descriptions that leaders and teams provided in
response to these questions.
First, each behavior mentioned in the transcriptsof the
team and managerial interviews was marked. These behaviors were then sorted into related categories-e.g.,
"showing us performance shortfalls in reports"and "telling the team when reliability is slipping" were categorized as "identifying the team's problems." Ten such behavior categories were identified. Four that had little or
no variance-that is, for which all managers engaged in
the behavior or only one or two managers engaged in the
behavior-were dropped.3
Each behavior category was then coded for frequency
on a scale from 1 = low to 3 = high for each team leader,
based on team and manager descriptions of how often that
behavior occurred. If a particularbehavior category was
not identified either by the team or the leader as something the manager did, it was coded as low. If both the
ORGANIZATION SCIENCE/Vol.
team and the leader identified a behavior as something
the leader did at least once per month, it was coded as
high. If the behavior was identified by the leader or the
team but not both, or if it was identified as happening less
frequently than once per month, it was coded as moderate.
The final behavior categories used in the analyses are
listed below; their means, standarddeviations, and reliabilities are presented in Table 3.
(1) Providing informal rewards and other cues that the
group-as-a-whole is responsible for managing itself (e.g.,
rewarding the group for solving a problem, spending
more time in interaction with the group-as-a-whole than
with individual members).
(2) Broadening the group's repertoire of problemsolving skills throughappropriateproblem-solvingconsultation (e.g., teaching the group to use a problem-solving
process; facilitating problem-solving discussions).
(3) Dealing with interpersonal problems in the team
throughteam process consultation (e.g., bringing up intrateam conflicts for discussion).
(4) Signaling that individuals (or the manager her/himself) are mainly responsible for managing the team's
work (e.g., by spending more time with individuals than
with the group; by running group meetings rather than
teaching the group how to run its own meetings).
(5) Intervening in the task (e.g., monitoring call rates
2001
12, No. 5, September-October
565
RUTH WAGEMAN
How Leaders Foster Self-Managing Team Effectiveness
and asking a specific member to take a particular call;
dealing directly with a team's customer without involving
the team).
(6) Identifying the team's problems (e.g., pointing out
overexpenditures on parts; showing the group a downward trend in its response time).
Even though these dimensions of coaching behavior
were derived entirely from what teams and their leaders
in this organization identified as common coaching behaviors, they bear a strong resemblance to others in the
literatureon self-management-and especially to those of
Manz and Sims (1987). These authors also identify the
"leader behaviors" of positive feedback, problem-solving
consultation, intervening as a "foreman," and process
consultation. Thus, this method of assessing leader coaching appears to be both ecologically valid for this research
setting and consistent with other research on the coaching
of self-managing teams.
Self-Managing Behaviors. The level of team selfmanagement, a key dependent variable in the research,
was coded from the team interview. Three theoryspecified components of self-management (collective responsibility for work outcomes, monitoring own performance, and managing own performance) were coded
separately (as high, medium, or low) and then averaged
to form an overall measure of a team self-management
(Cronbach's alpha = 0.94). The behaviors coded for each
component, along with means, standard deviations, and
intercoder reliabilities, are shown in Table 4.
Groups that scored high on collective responsibility (n
= 14) were those that described multiple instances of
initiative on a customer's or team member's behalf and
reported no instances of members avoiding responsibility-such as one group whose members came in on days
Table 3
Leader Coaching Behavior Descriptive Statistics
Data Source
Managerand GroupInterviews
Coaching behaviors
Cues/rewardsfor self-management
Problem-solvingconsultation
Process consultation
Negative signals
Interventionin the task
Identifyingteam problems
M
1.82
1.48
2.41
1.93
1.55
1.38
SD
r with
IRR* others**
1.49 0.96
1.63 0.86
1.02 0.91
1.04 0.94
1.06 0.92
1.09 0.98
0.15
-0.20
0.13
-0.06
0.13
0.06
*Interrater
reliabilities
**Averagecorrelationswithall othercoaching behaviorsmeasured
566
off to take calls when the call rate became too high for
remaining members to handle. Groups that scored low in
responsibility (n = 12) described multiple instances of
members avoiding responsibility-such as a team that
was unable get its members to take calls from particularly
difficult customers. Groups that described instances of
both high and low responsibility were coded as moderate
(n = 8).
Measurements of active monitoring were coded principally from responses to the question, "What kinds of
data does your group receive about its performance?And
how do you use these data, if at all?" Groups that were
coded high on monitoring (n = 12) actively sought (or
generated) data about their performance-such as one
team that requested regular reports about its 20 poorest
performing machines to identify common causes of problems. Groups that scored low on monitoring (n = 16)
ignored or refused performance-relevant data-such as
one whose members reportedthat company reports about
machine reliability were routinely discarded unread.
Groups moderate in monitoring (n = 6) reportedthatthey
did sometimes use company-provided data, but never
sought additional information.
Finally, measurements of managing own performance
were coded mainly from responses to three interview
questions: (1) "Describe a time when your group deliberately took time out from its regular work to discuss
better ways of operating;"(2) "Describe a time when your
group should have taken time out but did not do so;" and
(3) "Describe a time when you felt you could not act as
a team." Groups coded as high on managing (n = 11)
described multiple instances of autonomously convening
problem-solving meetings and altering their performance
strategies-such as a team that redesigned standardmaintenance practices to fix a machine reliability problem.
Groups coded as low (n = 10) tended to let chronic problems ride unaddressed. Groups coded as moderate on
managing (n = 13) were those that neglected some ongoing problems but altered their performance strategies
to deal with other problems or opportunities. One group,
for example, had redesigned its call-queuing strategy to
improve response time, but had not addressed (and had
no active plans to address) a long-standing reliability
problem.
Team Effectiveness. Team effectiveness was assessed
using (1) archival data about group performance, (2) survey measures of the degree to which members interact in
ways that over time increase their ability to work together,
and (3) survey measures of individual work satisfaction.
Performance data were collected from company records
for the six months prior to and six months after field data
ORGANIZATION
SCIENCE/Vol. 12, No. 5, September-October
2001
RUTH WAGEMAN
Table 4
How Leaders Foster Self-Managing Team Effectiveness
Measures of Level of Team Self-Management
Components
Collective
Responsibility
M
SD
IRR
r with
other
SMBs
2.07
0.98
0.94
0.78
MonitoringOwn
Performance
1.88
0.98
0.95
0.85
ManagingOwn
Performance
2.03
0.95
0.92
0.85
BehaviorsCoded
(1) Tooktime and initiativeto help othermembersor to solve customer
problemsnot in the group's purview.
(2) Avoidedworkthatwas formallythe team's responsibility(reverse
scored).
(1) Sought or generated data about aspects of team performance.
(2) Ignoredor discarded data about aspectsof team performance(reverse
scored).
(1) Alteredperformancestrategies to solveproblemwithoutmanagerial
directive.
(2) Had long-termproblemstillnot addressed (reverse scored).
(3) Had never takentime fromregularworkto discuss betterways of
operating(reversescored).
CompositeMeasure
1.99
0.75
0.92
collection.4 These 12 monthly assessments of team performance were averaged to generate summary measures
for the following five dimensions.
(1) Customer satisfaction with service, derived from a
customer survey distributed periodically by the organization. Survey ratings range from 1 ("very dissatisfied")
to 5 ("very satisfied"). The customer service measure was
not computed for the 11% of teams that had received
fewer than 10 surveys during the 12-month assessment
period.
(2) Parts expense, computed as the expected cost of
parts (based on machine type and level of usage) divided
by actual expenditures for parts.
(3) Response time, computed as the percentage of calls
taken by the team within specified time limits. This measure reflects how quickly a team gets to customers who
have malfunctioning machines.
(4) Repair time, computed as the amount of repairtime
a machine is expected to require, based on historical data,
divided by the actual number of minutes spent repairing
it. Repair time assesses technical proficiency, how
quickly teams can fix problems.
(5) Machine reliability, computed as the number of repair calls expected on a machine, based on historical data,
divided by the actual number of repair calls. Machine
reliability assesses how well the machines are maintained.
An overall performance measure was generated by tallying the number of these five dimensions for which a
team scored above Xerox norms. This measure ranges
from 0 (low performing on all criteria) to 5 (high performing on all criteria).
ORGANIZATIONSCIENCE/Vol. 12, No. 5,
2001
September-October
Quality of group process was assessed from survey
data, using seven items adapted from Allmendinger et al.
(1992). Team members rated, on a scale ranging from 1
("strongly disagree") to 7 ("strongly agree") their agreement with statements describing positive and negative
group processes (e.g., "Every time someone tries to
straighten out a work group member whose behavior is
not acceptable, things seem to get worse ratherthan better"). Scores on the seven items were combined into a
composite measure of process quality for each team.
Individual work satisfaction also was assessed from
survey data, using three job satisfaction items from the
Job Diagnostic Survey (Hackman and Oldham 1974).
Each item (e.g., "Generally speaking, I am very satisfied
with this job") used a seven-point agree-disagree scale.
Scores on the three items were combined into a composite
measure of individual work satisfaction for each team.
Psychometric data for measures of team performance,
quality of group process, and individual work satisfaction
are presented in Table 5. All variances and internal consistency reliabilities are acceptable. In addition, to ensure
that the composite scores are meaningful group-level
measures, intraclass correlations were computed for both
the quality of group process and the work satisfaction
measures; these coefficients were 0.23 and 0.29, respectively, both significant at p < 0.01.
Intercorrelations among all measures used in the research-team design, leader coaching, self-management,
and team effectiveness-are presented in Table 6.
567
RUTH WAGEMAN
How Leaders Foster Self-Managing Team Effectiveness
Results
Table 5
To assess the separate andjoint effects of the six measures
of leader coaching, a preliminary analysis regressed each
of the three measures of team self-management(collective
responsibility, monitoring own performance, and managing own performance) on overall quality of team design, the six coaching measures, and their interactions,
with all variables entered simultaneously. Two types of
coaching (providing cues and informal rewards for selfmanaging behaviors, and problem-solving consultation)
contributedpositively and significantlyto self-management,
whereas two other types of coaching (identifying team
problems, and leader task intervention) contributednegatively to self-management. Consistent with previous findings, leaders' interpersonal process consultations had no
significant effects. Therefore, the two coaching measures
with positive effects on self-management were combined
into a measure of "positive coaching," (alpha = 0.77)
and the two measures with negative effects were combined into a measure of "negative coaching" (alpha =
0.56). Principle components analysis confirmed that these
two summary indices represented two orthogonal dimensions (Eigen values >1.0). Further, preliminary inspection of patterns across the six leader coaching behaviors
in interaction with design suggested that there were two
distinct interaction patterns for these indices. Consequently, these summary measures are used in all subsequent analyses.
The main analyses, in which dependent variables are
regressed on team design, positive coaching, negative
coaching, and the two-way interactions of positive and
negative coaching with team design are presented in Table 7. According to the adjusted r-squares, the design and
coaching variables together account for 77% of the variation in overall self-management, 39% of the variance in
objective team performance, 50% of the variance in the
quality of member relationships, and 38% of the variance
in individual work satisfaction. These are large effects,
especially given that the predictors and the dependent
variables were based on data from different sources and
collected using different methods.
Effects of Design Conditions
Hypothesis 1, that well designed teams exhibit more selfmanagement and are more effective than teams whose
designs are flawed, is supported. The summary measure
of team design contributed positively and significantly to
overall level of team self-management (adjusted r-square
= 0.42), to objective group performance (adj. r-square
= 0.37), and to quality of group process (adj. r-square
= 0.12), but not to individual work satisfaction (p =
0.13). Regression of the 11 specific design features on
568
Team Effectiveness Descriptive Statistics
Data Source
Variable
ArchivalData
r with
otherperf.
Team performance
Customer satisfaction
Parts expenses
Response time
Repair time
Machine reliability
Overall group performance
M
SD
measures*
4.31
106.82
86.67
104.03
101.90
2.21
1.35
22.49
9.59
11.19
11.81
1.38
0.16
0.30
0.35
0.01
0.22
0.19
Survey
M
r within
rwith
SD scale (alpha) otherscales**
Qualityof group process 4.51 1.33 0.51 (0.85)
5.06 1.25 0.59 (0.75)
Membersatisfaction
0.25
0.28
*Averagecorrelationswithall othervariableswithinthe same category
**Averagecorrelationswithitems fromotherscales withinthe same
surveysection
overall self-management show that the major effects of
design are due to clear direction (t = 2.64, p < 0.05),
task interdependence (t = 2.51, p < 0.05), group rewards
(t = 2.31,p
< 0.05), and strategy norms (t = 2.01,p
<
0.05).
Regression of the individual design features on task
performance show that this relationship is principally due
to clear direction, task interdependence, and group rewards (t values of 2.67, 2.56, and 2.15, respectively, all
p < 0.05). A parallel analysis for quality of group process
shows that this relationship is mainly due to the effects
of clear direction (t = 2.67, p < 0.05), group rewards
(t = 2.35, p < 0.05), strategy norms (t = 2.14, p < 0.05),
and challenging task objectives (t = 1.80, p = 0.07).
Chi-square analyses were conducted assessing the relation of the dichotomous selection criterion (effective vs.
ineffective) on key dependent variables and are shown in
Table 8. These analyses confirm the regression findings.
Effects of Leader Coaching
Hypothesis 2 states that well-coached teams exhibit more
self-management, higher quality interpersonal relationships, and higher member satisfaction-but not higher
ORGANIZATIONSCIENCE/Vol. 12, No. 5, September-October 2001
o0
0
z
0
CO
Z
n
m
Table 6
Correlations Among Design Factors, Coaching Behaviors, and Dependent Measures*
1
p
C)
("I
D
nt)
(,/
_t
L~
Team design
1. Direction
2. Optimal diversity
3. Appropriate size
4. Stability of memb.
5. Task interdependence
6.
7.
8.
9.
10.
11.
Task goals
Core strategy norms
Group rewards
Available information
Available training
Material resources
Coaching behaviors
12. Cues/rewards for s-m
13. Problem-solving cons.
14. Process consultation
15. Negative signals
16. Intervention in the task
17. Identifying problems
0.08
-0.03
0.11
0.46
-0.04
0.12
0.05
0.22
0.37
-0.03
0.34
2
3
4
5
6
7
8
9
10
11
12
0.10
0.33
-0.13
0.21
0.11
0.00
0.10
0.41
-0.12
0.22
-0.05
0.06
-0.01
-0.07
0.06
0.26
0.04
-0.14
-0.14
-0.12
-0.11
0.02
-0.01
-0.03
0.19
0.22
-0.11
-0.02
0.11
0.48
0.30
-0.03
-0.14
0.03
0.19
0.17
0.01
0.27
-0.07
0.01
0.32
0.17
0.36
0.11
0.23
0.33
-0.02
-0.15
-0.06
-0.16
-0.04
0.07
-0.08
-0.45
0.50
0.32
0.02
0.08
0.03
-0.27
-0.02
0.22
0.09
0.11
0.00
-0.12
0.12
-0.01
0.27
0.35
0.06
0.09
0.01
-0.23
0.43
0.53
0.26
-0.20
0.02
0.23
0.46
0.61
0.30
-0.25
-0.14
0.16
0.04
0.30
0.23
-0.31
-0.07
-0.04
0.19
0.18
-0.26
-0.08
-0.02
0.09
-0.09
0.18
0.25
0.21
-0.02
0.11
0.32
0.24
0.08
-0.08
-0.02
0.02
0.62
0.34
-0.05
-0.06
-0.12
0.22
0.36
0.41
0.29
0.56
0.09
0.43
0.01
0.08
0.06
0.04
0.25
-0.02
0.10
0.11
0.10
0.12
0.12
0.19
0.12
-0.12
-0.02
0.06
-0.01
0.00
0.15
0.17
0.11
0.38
0.44
0.47
0.44
0.43
0.25
0.41
0.09
0.26
0.21
0.19
0.16
0.31
0.40
0.36
0.49
0.57
0.56
0.45
0.70
0.24
0.55
0.40
0.57
0.52
0.52
0.33
0.52
0.08
0.31
0.19
0.23
0.22
0.16
0.29
0.26
0.23
0.11
0.19
0.05
0.05
0.04
0.51
0.23
0.30
0.48
0.41
0.28
0.31
0.32
0.40
0.21
0.35
0.07
0.47
0.27
13
-0.40
-0.49
-0.43
-0.18
14
15
16
0.36
-0.03
0.13
0.02
0.21
0.1
-0.18
0.11
0.12
0.07
0.17
0.05
0.06
-0.11
0.02
-0.11
-0.32
-0.34
-0.34
-0.14
-0.22
-0.44
-0.42
-0.32
-0.36
-0.07
-0.37
measures
Dependent
18. Collective responsibility
19. Monitoring performance
20. Managing performance
21. Overall self-management
22. Overall performance
23. Quality of process
24. Member satisfaction
*AIIcorrelations are based on group-level data. Correlations greater than (0.37) are significant at p < 0.05.
O\
11D
.021
0.28
0.58
0.55
0.20
0.40
0.51
RUTH WAGEMAN
How Leaders Foster Self-Managing Team Effectiveness
task performance-than do teams that receive poor
coaching. This hypothesis also is supported. Positive
coaching contributes positively and significantly to overall level of team self-management (r-square = 0.08) and
to quality of group process (r-square = 0.18), but not to
either group performance (p = 0.75) or work satisfaction
(p = 0.87). Negative coaching contributes negatively to
overall self-management (r-square = 0.11) and to work
satisfaction (r-square = 0.30), but not to either group
performance (p = 0.37) or quality of group process (p
Table 7
OverallSelf-Management
IndependentVariable
= 0.36).
Interaction of Design and Coaching
So far we have seen that team design conditions are
strongly and positively related to level of team selfmanagement, how well teams perform, and the quality of
teams' interpersonal processes. By contrast, the findings
for leaders' coaching behaviors differ in that (a) level of
self-management is relatively weakly predicted by coaching (average r-squares of 0.46 and 0.06 for design and
coaching, respectively), (b) negative coaching is the only
predictor to influence member satisfaction, and (c) neither
positive nor negative coaching is related to objective team
performance. These findings corroborate Hackman's
(1990) qualitative findings exploring similar relationships.
We turn now to Hypothesis 3, which states that leaders'
design activities and hands-on coaching interact in affecting team self-management and effectiveness. Specifically, it is predicted that competent coaching has a
greater positive impact for well-designed teams than for
poorly designed teams. The interaction terms reported in
Tables 7 and 8 show that this hypothesis is also supported
for level of self-management and for quality of group
process, but not for objective task performance or member work satisfaction.
For level of team self-management, the interaction between design and coaching has a markedly different form
for positive and negative coaching behaviors. As is seen
in Figure 1, leaders' positive coaching behavior enhances
team self-management more when teams are well designed than when they are not. Figure 2, by contrast,
shows that negative coaching behavior undermines team
self-management more for poorly designed than for welldesigned teams.
Findings for measures of team interpersonalprocesses
exactly parallel those for level of self-management. Positive coaching has a stronger positive effect on process
quality in well-designed groups than in poorly designed
groups, and ineffective coaching undermines the interpersonal processes of poorly designed teams more than
those of well-designed teams.
570
Effects of Design and Coaching on Self-Managing
Behavior, Group Performance, Group Process, and
MemberSatisfaction*
Team design
Positivecoaching
Negative coaching
Design x positivecoaching
Design x negative coaching
B
4.492
0.523
-0.353
.606
- 0.687
Groupperformance
IndependentVariable
Team design
Positivecoaching
Negative coaching
Design x positivecoaching
Design x negative coaching
B
0.601
0.129
-0.238
0.712
-1.37
Qualityof group process
IndependentVariable
Team design
Positivecoaching
Negative coaching
Design x positivecoaching
Design x negative coaching
B
0.305
0.437
-0.138
0.603
-0.805
MemberSatisfaction
IndependentVariable
Team design
Positivecoaching
Negative coaching
Design x positivecoaching
Design x negative coaching
B
0.183
0.033
-0.441
0.015
-0.598
OverallAdj.R2 = 0.77
adj. rt
square
p
4.09
2.24
-2.42
1.99
- 2.33
0.00
0.01
0.02
0.05
0.03
0.42
0.08
0.11
0.06
0.09
OverallAdj.R2 = 0.39
adj. rt
square
p
2.59
0.32
-0.91
0.73
- 1.53
0.01
0.75
0.37
0.47
0.13
0.37
0.00
0.00
0.00
0.02
OverallAdj.R2 = 0.50
adj. rt
p
square
2.33
3.08
-0.94
1.88
-2.47
0.03
0.01
0.36
0.05
0.02
0.12
0.18
0.00
0.07
0.13
OverallAdj.R2 = 0.38
adj. rt
p
square
1.00
0.17
-2.71
0.03
- 1.37
0.13
0.87
0.01
0.97
0.18
0.02
0.00
0.30
0.00
0.04
For all analyses, n = 33
In sum, the effects of leaders' coaching behaviors depend substantially on how well they have designed their
teams: Effective coaching helps well-designed teams
more than poorly designed teams, and ineffective coaching undermines poorly designed teams more than welldesigned teams.
Discussion
The data presented here demonstrate both strengths and
weaknesses for addressing team leader activities and their
ORGANIZATIONSCIENCE/Vol.
12, No. 5, September-October 2001
RUTH WAGEMAN
Table 8
How Leaders Foster Self-Managing Team Effectiveness
Figure 1
Relationships of Design Conditions with Team
Effectiveness
NominationCategory
High 3.0
Well-designedteams
Superb
Ineffective
17
1
1
14
Task interdependence
15
Yes
3
No
13
Grouprewards
Yes
No
15
3
4
11
8.92
16
8.74
<0.01
2
5
10
16
2
7
8
4.77
<0.05
Challengingtask goals
13
Yes
5
No
5
10
2.99
=0.08
9
9
7
8
1.87
Information
Yes
No
8
10
5
10
0.160
n.s.
ct
Memberdiversity
Yes
No
4
11
0.099
0.099 n.s.
n.s.
0
5
13
Design Condition
Cleardirection
Yes
No
Strategynorms
Yes
No
Materialresources
Yes
No
AppropriateSize
Yes
No
InteractiveEffects of Qualityof Team Design
and Positive Leader Coaching on Team SelfManagement
X2
p
c
2
19.90
<0.01
S
Moderate 2.0
?
12.46
<.01
I
Poorly designed teams
40
<0.01
Low
1.0
1.0
Low
n.s
2.0
Moderate
3.0
High
Positive Leader Coaching
Figure 2
InteractiveEffects of Qualityof Team Design
and Negative Leader Coaching on Team SelfManagement
High 3.0
Well-designedteams
|
S
Moderate 2.0
Poorly designed teams
Low
RealTeam
Yes
No
7
11
5
10
Education
Yes
No
12
6
9
6
0.082
n.s.
2.0
Moderate
3.0
High
Negative Leader Coaching
0
n.s.
Note. Cell contents are the numbersof superb vs. ineffectiveteams
for which each design conditionwas in place vs. not in place. x2
Reportedare continuity-adjusted.
effects on self-managing
1.0
Low
teams. Among the strengths is
the fact is that the measures of design features are derived
from coded descriptions of actual organizational features,
rather than from team-member ratings of design quality.
This method of data collection is quite rare in the literature, and significantly adds to the trustworthinessof these
SCIENCE/Vol.12, No. 5, September-October 2001
ORGANIZATION
measures. In addition, independent data sources provided
assessments of team performance,quality of interpersonal
processes, and team member satisfaction (archival and
survey sources, respectively). As a consequence, the findings of this research were protected from problems pervasive in teams research: Namely, relations between assessments of team effectiveness and assessments of
design features and leader coaching are potentially attributable to overall positivity or negativity biases on the part
of respondents. Moreover, the findings present the first
direct test of Hackman's (1986) model of work team effectiveness, and provide confirmation for the model.
This research has several weaknesses, as well, that call
571
RUTH WAGEMAN
How Leaders Foster Self-Managing Team Effectiveness
for caution in the interpretationof findings. First, the sample of teams was small, and thus interpretationsof findings, especially analyses involving the numerous design
conditions entered separately, must be drawn with caution. Also, measures of the frequency of effective and
ineffective coaching behaviors may not fully capture the
overall quality of coaching, by leaders, that teams receive.
Finally, the selection process in this research necessarily
produced teams that tended to be extremely effective or
extremely ineffective, limiting the opportunity to explore
curvilinear effects, and also limiting the degree to which
conclusions might be drawn about teams that are only
moderately effective on all criteria.
Nonetheless, the findings of this research have implications for three issues about the dynamics of selfmanaging teams: (1) the degree to which self-managing
behaviors by team members mediate the effects of design
and coaching; (2) how team design, leader behavior, selfmanagement, and team performance can evolve into a
system whose parts are mutually reinforcing; and (3) the
conditions under which team leaders can have the greatest
influence on team effectiveness. Below, I draw upon the
patternof findings in this research to explore each of these
questions in turn.
Self-Managing Behaviors as Mediators
Self-management may-or may not-mediate the relationship between team design and coaching, on the one
hand, and team effectiveness on the other. It could be, for
example, that team design and leader coaching are consequential only to the extent that they affect the degree
to which members take on high levels of responsibility
for the work, engage in monitoring performance over
time, and develop high-quality work strategies. Alternatively, design and coaching may influence team effectiveness only partly (or not at all) through the team's selfmanaging behavior. For example, team composition, a
design feature, may have positive effects both on how
members work together and on the team's capability in
actually executing the work (Ancona and Caldwell 1992,
Druskat 1995, Goodman and Shah 1992). In this case,
performance effects associated with composition would
be due in part to the effects of composition on team selfmanagement and in part to the quality of members' task
execution.
The present data allows an exploratory test of the degree to which self-managing behavior does mediate the
impact of design and coaching on team effectiveness. According to Baron and Kenny (1986), three conditions
must be met to demonstrate mediation: (1) The independent variables (in this case, team design and leader coaching) must significantly predict the proposed mediator
572
(self-managing behavior); (2) the mediator must significantly predict the dependent variable (performance,
group process, and satisfaction); and (3) when the effects
of the independent variable and mediator are entered simultaneously, the contribution of the independent variable should drop substantially for partial mediation and
to nonsignificance for full mediation.
The first of these three conditions is met (see Table 7).
The strong relationship between team self-management
and team effectiveness shown in Table 9 (Model 1) establishes that the second condition is also met. To assess the
third condition, team design and leader coaching are entered into the regression simultaneously with level of selfmanagement. Comparison of Model 2 with Model 1 in
Table 9 shows that the relative contributions of the predictors do change substantiallywhen this is done. For team
performance, the contribution of self-management becomes nonsignificant, with only team design predicting
performance.This shows that team design influences performance independentlyof its effects on self-management.
For quality of group process, however, self-management
remains a significant predictor-but coaching does not.5
This patternsuggests that the effects of leader coaching on
quality of group process are mediated by their effects on
self-management,in contrastto team design, which affects
process quality independentlyof its influence on team selfmanagement. Finally, for work satisfaction, level of selfmanagement becomes nonsignificant, but negative leader
coaching remains marginally significant. This patternsuggests that ineffective coaching undermines work satisfaction in part throughits tendency to reduce team members'
self-management-but that ineffective coaching also undermines work satisfaction directly.
In sum, self-managing behaviors do appear to mediate
the effects of team design and coaching to some extentbut differently for the three components of effectiveness.
Self-management fully mediates the relationshipbetween
coaching and the quality of members' interpersonalprocesses; it accounts partially for the effect of coaching on
work satisfaction; but it plays no mediating role at all in
the relationship between team design and objective group
performance. Indeed, self-management never fully accounts for the effects of quality of team design on any of
the three components of effectiveness. This finding suggests that overall quality of team design is equally important for teams, regardless of the level of behavioral
self-management.
Dynamic Relations Among Design, Coaching,
Process, and Performance
The present findings may call into question conventional
understanding about the relationships among leader
ORGANIZATIONSCIENCE/Vol. 12,
No. 5, September-October
2001
RUTH WAGEMAN How LeadersFosterSelf-ManagingTeamEffectiveness
Table 9
Test of Self-Managementas a Mediatorof the Effects of Team Design and Leader Coaching Behavior on Team
Effectiveness
Model2
Model1
OverallAdj.R2 = 0.19
Team performance
B
t
p
adj. r-square
Independentvariable
B
t
0.795
2.90
0.01
0.19
Self-management
Team design
Positiveleader behavior
Negative leader behavior
Qualityof group process
0.224
1.691
0.246
-0.318
0.41
2.03
0.49
-0.57
B
t
Independentvariable
Self-management
OverallAdj.R2 = 0.52
Qualityof group process
B
t
p
adj. r-square
IndependentVariable
0.723
5.45
0.00
0.52
Self-management
Team design
Positiveleader behavior
Negative leader behavior
Membersatisfaction
Independentvariable
Self-management
OverallAdj.R2 = 0.24
Membersatisfaction
0.717
0.289
0.134
0.055
3.47
1.50
0.99
0.49
p
adj. r-square
0.01
0.68
0.39
0.05
0.02
0.63
0.34
0.02
OverallAdj. R2 = 0.57
p
adj. r-square
0.33
0.00
0.09
0.04
0.04
0.33
0.01
0.63
OverallAdj. R2 = 0.44
B
t
p
adj. r-square
Independentvariable
B
t
p
adj. r-square
0.559
3.02
0.01
0.24
Self-management
Team design
Positiveleader behavior
Negative leader behavior
0.146
0.521
0.009
-0.295
0.55
1.97
0.04
-1.97
0.59
0.05
0.96
0.06
0.01
0.17
0.00
0.13
Independentvariable
Self-management
OverallAdj.R2 = 0.44
Team performance
*Forall analyses, n = 33.
coaching behavior, team design, team processes, and
team performance effectiveness.
Team and Leader Behavior. When teams in this study
were well designed, effective coaches tended to have a
more positive influence on team processes than they did
when teams were poorly designed. Moreover, welldesigned teams appeared more robust-that is, ineffective coaching behavior did not undermine them nearly as
much as it undermined teams with flawed designs. Thus,
the impact of leaders' coaching on their teams is conditioned by the way in which they set the team up in the
first place.
However, not only does the impact of leader coaching
behavior depend upon team design, but particularleader
coaching behaviors may also be elicited by quality of
team design. Recall that when leaders pointed out work
problems to a team and when leaders intervened in the
task, teams were less likely to manage their own performance. These patterns raise the question of whose behavior is influencing whose. One possibility-the one that
ORGANIZATIONSCIENCE/Vol. 12, No. 5, September-October 2001
predominates in the team leadership literature-is that the
coaching behavior of the leader drives the self-managing
behavior of the team. Thus, in the examples cited above,
the leaders' behaviors may prompt team members to attribute authority not to the team, but to managers, and
thereafter they may take responsibility only for task execution, leaving the managing to managers.
An alternative explanation is that leaders monitor the
performance and manage the tasks of poorly designed
teams because members themselves are not doing so.
Leaders may respond to low team self-management and
poor performance by monitoring team performance
closely, by increasing their own interventions in the work
of their teams, and by providing fewer cues and rewards
for team self-management. By contrast, well-designed
teams are highly likely to take on management functions
themselves, making it unnecessary for the leader to do
so. Thus, the degree of effective self-management by
teams, itself influenced by the design of the team, may
be shaping the coaching behavior of team leaders. This
573
RUTH WAGEMAN
How Leaders Foster Self-Managing Team Effectiveness
possibility, which is consistent with other research on
how subordinate behavior can shape leader style (e.g.,
Farris and Lim 1969, Lowin and Craig 1968), merits additional investigation in the specific case of selfmanaging teams.
Design, Self-Management, and Performance. Over
time, design factors, team self-management, and team
performance may become so interdependentthat they set
in motion a self-reinforcing spiral. In the present research
setting, team design is largely in the hands of the teams'
immediate managers. These managers could redesign rewards, alter tasks, articulate direction, and provide resources to teams at their discretion. Many team leaders
did so-and the better the design conditions they provided, the more their teams were self-managing and the
better they performed. But leaders' decisions to provide
better design conditions may themselves be influenced by
prior team performance. For example, teams well designed enough to perform adequately are more likely to
be given additional authority over their work, more support resources, and/or more challenging goals (e.g.,
Ancona and Caldwell 1992). By contrast, teams with few
support conditions tend not to use the authority they do
have-nor do they perform well. Leaders may be understandably reluctant to bestow even more resources on
those teams, even though that might be just what is
needed to remedy their performance problems.
Thus, design, self-management, and performance may
operate as self-reinforcing spirals, wherein already welldesigned teams manage themselves effectively, receive
even more organizational support, and thus become better
self-managing performing units over time (Lindsley et al.
1995). By contrast, poorly designed teams that manage
themselves ineffectively may receive from their leaders
fewer of the very supports and resources that contribute
to team effectiveness-and even risk having withdrawn
some of the positive design features that they presently
enjoy.
Under What Conditions Do Leaders Affect Team
Outcomes?
The findings of this research suggest that team behavior
and performance may be most affected by structural,
technological, and contextual factors-factors that often
lie beyond team leaders' direct control. These exogenous
factors can significantly constrain the variance in how
teams are managed within organizations and within single
industries. For example, Hackman (1993) shows how the
factors that most strongly influence the behavior of aircraft flightdeck crews are themselves shaped by three exogenous factors: standard cockpit technology (determined by engineers at corporations that design and
574
manufactureaircraft), standardoperating procedures (determined by regulatory agencies), and the deeply rooted
and highly individualistic culture of flying. Thus, environmental and institutional forces (DiMaggio and Powell
1983, Meyer and Rowan 1977) determine the structures
that are proximal to the team (such as the team's purpose,
the design of its work, how rewards are allocated, and so
on), which in turn shape actual team behavior.
Because the majorinfluences on team behavior and performance often have little variation within any given organization, single-organization (or even single-industry)
investigations of influences on team outcomes can explain
relatively little variance in how, and how well, teams operate. And because the effects of leaders' day-to-day
coaching of teams-indeed, coaching behavior itselfmay depend to a considerable extent on how those teams
are designed, leader coaching will appear to have only a
modest influence on the overall trajectoryof a team.
This view suggests two fruitful avenues for future investigation-both at the organizational level-for understanding how and when leaders influence team selfmanagement and effectiveness. In the present research
setting, front-line managers had real impact because they
had a great deal of discretion in determining their teams'
basic structure, how rewards and information were distributed, and overall team direction. This setting, then, is
an exception to the usual strong exogenous influences on
team design features. In this exception lie two leads to
researchable questions about when team leaders can really make a difference.
The first feature of this organization was an unusual
degree of authority invested in front-line managers due
to decentralization. Individual districts were in the process of becoming profit centers. Within each district, both
district and field managers had the authorityto alter major
design features of their business units. For example, while
constraints existed on total funds available for rewards,
how those rewards were distributed to technicians as autonomously determined by field managers-and often
differently by different field managers. Moreover, it was
left to the managers to communicate to their teams their
role in the new units, allowing them to shape their teams'
direction.
Second, a number of design features came under the
control of team leaders because the organization was undergoing major change. The organizational structuresand
systems that ordinarily operate as tightly interconnected
and inert components of an organization were therefore
open to change. For example, service territorieshad to be
redetermined as some districts acquired new geography.
How territories were defined and staffed directly affects
ORGANIZATIONSCIENCE/Vol.12, No. 5, September-October 2001
RUTH WAGEMAN
How Leaders Foster Self-Managing Team Effectiveness
team composition-especially the number of group members and the mix of skills among them. Changes of staffing and territoryresponsibilities also created demands for
training in some teams-demands that team leaders could
decide to accept or to deny.
These two organizational features-decentralization
and major structural change-represent two conditions
under which team leaders have the latitude to make real
differences in team effectiveness. They are also the same
conditions under which researchers are likely to find real
variation in team design features. A focus on leaders'
behavior-both as team designers and as coaches-in circumstances in which leaders have such latitude may be
more fruitful in the long term than continuing to search
for the best kinds of day-to-day styles for leaders to use
in interaction with their teams.
Acknowledgments
The authoris gratefulto Ann Burress,CarltonCrowley, David Hyatt, and
Cathy Sirett for assistance in collecting and coding the data, and also to
David for helping to design the interview protocols. Many thanksto Tom
Ruddy and to the teams and managers at Xerox who made this research
possible. Thanks especially to RichardHackmanfor invaluablecomments
on earlierversions of this work.
Endnotes
1Supplementaryanalyses conducted on several variables that could be
construed as continua (e.g., group rewards and strategy norms) showed
that treating them as dichotomous variables did not misrepresent their
effects. In no case, for example, were curvilinear effects obtained for
the continuous versions of these variables.
2This straightforward additive model assumes that all elements of design are equally important. A fine-grained analysis of the differential
effects of individual design features on the dependent variables is reported in the results section of the paper.
3The behaviors not analyzed were "providing organization-level information to the team," and "attending team meetings," which all managers did at least monthly; and "overriding group decisions" and "producing computer analyses of performance data," which very few
managers did.
4The use of averages of multiple performance indicators across such a
substantial time range also serves to defend the findings of this research
against validity threats from regression to the mean.
5Although self-managing behavior and quality of interpersonal processes are highly correlated, this relationship cannot be accounted for
by method variance because the two constructs were assessed using
different methods. Moreover, they operate differently in the mediation
analyses, suggesting that the relationship is a substantive one.
References
Abramis, D. J. 1990. Semiconductor manufacturing team. J. R. Hackman, ed. Groups That Work and Those That Don't. Jossey-Bass,
San Francisco, CA, 449-470.
Allmendinger, J., J. R. Hackman, L. Kowal-Wolk, E. V. Lehman. 1992.
Methods and measures for the cross-national study of symphony
ORGANIZATIONSCIENCE/Vol. 12, No. 5, September-October 2001
orchestras. Report no. 4, Cross-National Study of Symphony Orchestras, Harvard University, Cambridge, MA.
Ancona, D. G., D. F. Caldwell. 1992. Demography and design: Predictors of new product team performance. Organ. Sci. 3, 321341.
Atkinson, J. W. 1958. Towards experimental analysis of human motivation in terms of motives, expectancies, and incentives. J. W.
Atkinson, ed. Motives in Fantasy, Action, and Society. Van Nostrand, Princeton, NJ.
Balkema, A., E. Molleman. 1999. Barriers to the development of selforganizing teams. J. Managerial Psych. 14 134-149.
Baron, R. M., D. A. Kenny. 1986. The moderator-mediator variable
distinction in social psychological research: Conceptual, strategic
and statistical considerations. J. Personality and Soc. Psych. 51
1173-1182.
Barry, D. 1991. Managing the bossless team: Lessons in distributed
leadership. Organ. Dynamics 20 31-47.
Bass, B. M. 1957. Behavior in Groups. Louisiana State University,
Baton Rouge, LA.
Beekun, R. I. 1989. Assessing the effectiveness of sociotechnical interventions: Antidote or fad? Human Relations 47 877-897.
Berkowitz, L. 1953. Sharing leadership in small, decision-making
groups. J. Abnormal and Soc. Psych. 48 231-238.
Bennis, W., B. Nanus. 1985. Leaders: The Strategies for Taking
Charge. Harper & Row, New York.
Campion, M. A., G. J. Medsker, A. C. Higgs. 1993. Relations between
work group characteristics and effectiveness: Implications for designing effective work groups. Personnel Psych. 46 823-850.
Cohen, S. G., G. E. Ledford, Jr. 1994. The effectiveness of selfmanaging teams: A quasi-experiment. Human Relations 47 1343.
G. Spreitzer, G. E. Ledford Jr. 1996. Developing effective self-,
managing work teams in service organizations. Working paper.
Cordery, J. L., W. S. Mueller, L. M. Smith. 1991. Attitudinal and behavioral effects of autonomous group working: A longitudinal
field study. Acad. Management J. 34 464-476.
Cummings, T. G., W. H. Griggs. 1977. Workers' reactions to autonomous work groups: The impact of control. Organization and Admin. Sci. 7 87-100.
DiMaggio, P. J., W. W. Powell. 1983. The iron cage revisited: Institutional isomorphism and collective rationality in organizational
fields. Amer. Sociological Rev. 48 147-160.
Druskat, V. U. 1995. A team competency study of self-managed manufacturing teams. Doctoral dissertation, Boston University, Boston,
MA.
Fiedler, F. E. 1958. Leader Attitudes and Group Effectiveness. University of Illinois Press, Urbana, IL.
Farris, G. F., F. G. Lim. 1969. Effects of performance on leadership,
cohesiveness, influence, satisfaction, and subsequent performance. J. Appl. Psych. 53 490-497.
Goliembiewski, R. T. 1995. Managing diversity in organizations. University of Alabama Press, Tuscaloosa, AL.
Goodman, P. 1986. Leading groups in organizations. Designing Effective Workgroups. Jossey-Bass, San Francisco, CA.
, R. Devadas, T. L. Hughson. 1988. Groups and productivity: Analyzing the effectiveness of self-managing teams. J. P. Campbell,
575
RUTH WAGEMAN
How Leaders Foster Self-Managing Team Effectiveness
R. J. Campbell, and Associates, eds. Productivity in Organizations. Jossey-Bass, San Francisco, CA.
, S. Shah. 1992. Familiarity and work group outcomes. S. Worchel, W. Wood., and J. Simpson, eds. Group Process and Productivity. Sage, London, U.K.
Hackman, J. R. 1982. A Set of Methods for Research on Work Teams.
Technical report no. 1, Research Program on Group Effectiveness, Yale School of Organization and Management, Yale University New Haven, CT.
- . 1985. Doing research that makes a difference. E. E. Lawler,
A. M. Mohrman, S. A. Mohrman, G. E. Ledford, and T. G. Cummings, eds. Doing Research That Is Usefulfor Theory and Practice. Jossey-Bass, San Francisco, CA.
- . 1986. Psychology of self-management in organizations. M. S.
Pollack and R. O. Perloff, eds. Psychology and Work:Productivity Change and Employment. Amer. Psych. Assoc.
- . 1987. The design of work teams. J. W. Lorsch, ed. Handbook of
Organizational Behavior. Prentice-Hall, Englewood Cliffs, N.J.
- . 1990. Groups That Work(and Those That Don't): Creating Conditions for Effective Teamwork. Jossey-Bass, San Francisco, CA.
-, G. R. Oldham. 1974. The job diagnostic survey: An instrument
for the diagnosis of jobs and the evaluation of job redesign projects. JSAS Catalogue of Selected Documents in Psych. 4 148.
-- ,
. 1980. WorkRedesign. Addison-Wesley, Reading, MA.
, N. J. Vidmar. 1970. Effects of size and task type on group performance and member reactions. Sociometry 33 37-54.
- , K. R. Brousseau, J. A. Weiss. 1976. The interaction of task design and group performance strategies in determining group effectiveness. Organizational Behavior and Human Performance
16 350-365.
.1993. Teams, leaders, and organizations: New directions for
crew-oriented flight training. E. L. Wiener, B. G. Kanki, R. L.
Helmreich, eds. Cockpit Resource Management. Academic Press,
Orlando, FL.
. 1998. Why teams don't work. R. S. Tindale, J. Edwards, E. J.
-Posavac, eds. Applications of Theory and Research on Groups to
Social Issues. Plenum, NY.
Hut, J. A., E. Molleman. 1998. Empowerment and team development.
Team Performance Management J. 4 53-66.
Jackson, J. M. 1953. The effect of changing the leadership of small
groups. Human Relations, 6 25-44.
Jackson, P. R., S. Mullarkey, S. Parker. 1994. The implementation of
high-involvement work teams: A four-phase longitudinal study.
Paper presented at BPS Occupational Psych. Conf. Birmingham,
U.K.
Jackson, S. 1992. Team composition in organizations. S. Worchel, W.
Wood, and J. Simpson, eds. Group Process and Productivity.
Sage, London, U.K.
Kaplan, R. E. 1979. The conspicuous absence of evidence that process
consultation enhances team task performance. J. Appl. Behavioral
Sci. 15 346-360.
Komacki, J. L., M. L. Deselles, E. D. Bowman. 1989. Definitely not a
breeze: Extending an operant model of effective supervision to
teams. J. Appl. Psych. 74 522-529.
Klaus, D. J., R. Glaser. 1970. Reinforcement determinants of team
proficiency. Organ. Behavior and Human Performance 5 33-67.
Liang, D. W., R. Moreland, L. Argote. 1995. Group vs. individual
576
training and group performance: The mediating role of transactive
memory. Personality and Soc. Psych. Bull. 2, 384-393.
Likert, R. 1958. Effective supervision: An adaptive and relative process. Personnel Psych. 11 317-332.
Lindsley, D. H., D. J. Brass, J. B. Thomas. 1995. Efficacy-performance
spirals: A multilevel perspective. Acad. Management Rev. 20
645-678.
Lippitt, R. 1940. An experimental study of the effect of democratic
and authoritariangroup atmospheres. University of Iowa Studies
in Child Welfare 16 43-95.
Locke, E. A., L. M. Soari, K. N. Shaw, G. D. Latham. 1981. Goal
setting and task performance: 1969-1980. Psych. Bull. 90 125152.
Lowin, B., J. R. Craig. 1968. The influence of level of performance on
managerial style: An experimental object-lesson in the ambiguity
of correlational data. Organ. Behavior and Human Performance
3 440-458.
Macy, B. A., P. D. Bliese, J. J. Norton. 1991. Organizational change
and work innovation: A meta-analysis of 131 North American
field experiments-1961-1990. Annual Convention of the Acad.
Management, Miami, FL.
Manz, C. C. 1986. Self-leadership: Toward an expanded theory of selfinfluence processes in organizations. Acad. Management Rev. 11
585-600.
, H. P. Sims. 1987. Leading workers to lead themselves: The external leadership of self-managing work teams. Admin. Sci. Quart.
32 106-128.
Meyer, J. W., B. Rowan. 1977. Institutionalized organizations: Formal
structureas myth and ceremony. Amer. J. Sociology 83 340-363.
Prince, C., T. R. Chidester, J. A. Cannon-Bowers, C. S. Bowers. 1992.
Aircrew coordination: Achieving teamwork in the cockpit. R. W.
Swezey and E. Salas, eds. Teams: Their Training and Performance. Ablex, Norwood, NJ.
Rosenbaum, M. E., D. L. Moore, J. L. Cotton, M. S. Cook, R. A. Hieser,
M. N. Shovar, M. J. Gray. 1980. Group productivity and process:
Pure and mixed reward structures and task interdependence. J.
Personality Soc. Psych. 39 626-642.
Saavedra, R., P. C. Earley, L. Van Dyne. 1993. Complex interdependence in task-performing groups. J. Appl. Psych. 78 61-72.
Salancik, G. R., J. Pfeffer. 1977. Constraints on administratordiscretion: The limited influence of mayors on city budgets. Urban Affairs Quart. 12 475-498.
Salas, E., J. A. Cannon-Bowers, E. L. Blickensderfer. 1993. Team performance and training research: Emerging principles. J. Washington Acad. Sci. 83 81-106.
Schein, E. H. (1988). Process Consultation. Addison-Wesley, Reading,
MA.
Schlesinger, L., J. M. Jackson, J. Butman. 1960. Leader-member interaction in management committees. J. Abnormal and Soc.
Psych. 61 360-364.
Shea, G. P., R. A. Guzzo. 1987. Groups as human resources. G. R.
Ferris and K. M. Rowlands, eds. Research in Personnel and Human Resources Management, Vol. 5. JAI Press, Greenwich, CT
323-356.
Smoll, F. L., R. E. Smith. 1989. Leader behaviors in sport:A theoretical
model and research paradigm. J. Appl. Soc. Psych. 19 1522-1551.
ORGANIZATION
SCIENCE/Vol. 12, No. 5, September-October 2001
RUTH WAGEMAN
How Leaders Foster Self-Managing Team Effectiveness
Steiner, I. D. 1972. Group Process and Productivity. Academic Press,
New York.
andgroupeffectiveness.Admin.
Wageman,R. 1995. Interdependence
Sci. Quart. 40 145-180.
Wall,T. D., N. J. Kemp,P. R. Jackson,C. W. Clegg. 1986.Outcomes
of autonomousworkgroups:A long-termfieldexperiment.Acad.
Management J. 29 280-304.
of
Walton,R. E. 1985. Fromcontrolto commitment:Transformation
workforcemanagementstrategiesin the United States. K. B.
Clark,R. H. Hayes, and C. Lorenz,eds. The UneasyAlliance:
Managing the Productivity-Technology Dilemma. Harvard Busi-
ness School Press,Boston,MA.
Wellins,R. S., W. C. Byham,J. M. Wilson. 1991.EmpoweredTeams:
CreatingSelf-ManagingWorkingGroupsand the Improvement
of Productivityand Participation.Jossey-Bass,San Francisco,
CA.
Wood, J. 1990. J. R. Hackman, ed. Groups that Work (and Those That
Don't): Creating Conditions for Effective Teamwork. Jossey-
Bass, San Francisco,CA.
Woodman,R. W., J. J. Sherwood.1980.Theroleof teamdevelopment
in organizational
effectiveness:A criticalreview.Psych.Bull.88
166-186.
Wooley, A. W. 2001. Effects of interventiontype and timingon task
performance. J. Appl. Behavioral Sci.
Yukl,G. 1989.Managerialleadership:A reviewof theoryandresearch.
J. Management 15 251-289.
Zander, A. 1971. Motives and Goals in Groups. Academic Press, New
York.
Accepted by M. Scott Poole.
ORGANIZATION
SCIENCE/Vol. 12, No. 5, September-October 2001
577