Distributed leadership as a unit of analysis


This article proposes a new unit of analysis in the study of leadership. As an alternative to the
current focus, which is primarily on the deeds of individual leaders, the article proposes distributed
leadership. The article shows how conventional constructs of leadership have difficulty accommodating
changes in the division of labor in the workplace, especially, new patterns of interdependence and
coordination which have given rise to distributed practice. A number of forms of distributed leadership
are then outlined, in particular, three varieties of concertive action in which a key defining criterion is
conjoint agency. These forms provide the basis for a taxonomy of distributed leadership and a review
of examples in the literature. The article concludes with a consideration of some implications of the
adoption of a revised unit of analysis, particularly for recent work on levels of analysis and for future
research into leadership as a process.

  1. Introduction
    In the social sciences, the study of leadership has long been dominated by a conception of
    focused leadership. At the heart of this understanding is a strong commitment to a unit of
    analysis consisting of a solo or stand-alone leader. This claim is confirmed by Rost’s (1993)
    recent extensive review of the literature. Rost (p. 70) noted, for example, that despite signs of
    an incipient challenge to mainstream analyses of leadership by the end of the 1970s, more

than 130 books published in the next decade reinforced the conventional, orthodox message
that ‘‘leadership is basically doing what the leader wants done.’’ This continued endorsement
of the traditional focus of study is odd, given other important efforts to recontour the entire
field of leadership. While surprisingly little work has been undertaken on units of analysis,
significant developments have been made with levels of analysis (see Dansereau &
Yammarino, 1998b). An important contribution here has been Hunt’s work on extended
multilevel leadership (e.g., Hunt, 1991; Hunt & Ropo, 1995) which has built on Jaques’
theory of stratified systems (e.g., Jaques & Clement, 1995). Even within these extended level
analyses, however, the idea of focused leadership continues to find favour with the majority
of commentators.
The argument of this article will be that students and practitioners of leadership would be
better served by a more expanded unit of analysis, based on a revised conception of
leadership. As an alternative to focused leadership, I argue for a unit of analysis which
encompasses patterns or varieties of distributed leadership. To that end, I outline a framework
for understanding distributed organizational leadership and a taxonomy for classifying
varieties of distributed patterns, based on a range of constituent elements identified in
research studies.
Rost’s pessimism about the 1980s was succeeded by a decade which bore witness to a
surge of interest in a range of distributed phenomena among organizational theorists and
researchers. Commentators drew attention to distributed decision-making (Committee on
Human Factors, 1990), distributed cognition (Hutchins, 1996), and shared or dispersed
leadership (Bryman, 1996, pp. 283–284; Miller, 1998, p. 4). Yukl (1999, pp. 292–293), a
critic of what he terms the heroic leader paradigm, has noted of distributed leadership that it:
does not require an individual who can perform all of the essential leadership functions,
only a set of people who can collectively perform them. Some leadership functions (e.g.,
making important decisions) may be shared by several members of a group, some
leadership functions may be allocated to individual members, and a particular leadership
function may be performed by different people at different times. The leadership actions of
any individual leader are much less important than the collective leadership provided by
members of the organization.
Apart from this kind of comment which expresses a realisation of the existence of
distributed leadership, and which implies its potential advantages from the perspective of
overall organizational capability, there is a dearth of extended, analytical discussions of the
concept (although see Gronn, 2000; Spillane, Halverson, & Diamond, 2000a, 2000b). This
paucity of coverage is equally as puzzling as the lack of discussion of units of analysis, for the
idea of dispersed or distributed leadership is not new. Indeed, it was in the 1950s that the
Australian leadership theorist, the late C.A. Gibb (1913–1994), first raised the possibility of
leadership displaying a distributed pattern. Gibb’s remarks about leadership are the point of
departure for my argument. The purpose of this article is not to reaffirm his work, nor to
account for the curious neglect of his suggestion by commentators until its revival by Brown
(1989) and Brown and Hosking (1986). Rather, the aim is to develop the idea of distributed
leadership as a unit of analysis to which Gibb made brief mention in his entry on leadership in

the Handbook of Social Psychology (Gibb, 1954) and repeated in the second edition of the
Handbook (Gibb, 1969).
The discussion is organized as follows. First, I consider a number of problems concerned
with some current conceptions of leaders and leadership. Here I show that these stem mainly
from the inability of orthodox dichotomies and dualisms to adequately represent leadership
practice. Second, I distinguish between two forms of distributed leadership, numerical action
and concertive action. As part of concertive action, I then discuss the three main patterns
which are the focus of this article: spontaneous collaboration, intuitive working relations, and
institutionalised practices. Third, these three forms of distributed leadership are positioned in
relation to the prevailing assumptions in the field, and it is suggested how and why a
distributed perspective provides a helpful perspective on practice.
The main argument will be that a distributed understanding is well aligned with the
processes through which work is currently articulated as part of an emerging and everchanging
division of labor, due to task differentiation and reintegration. In particular, I show
how new workplace imperatives are generating qualitatively different forms of interdependence
between organizational personnel and that these have stimulated the adoption of
distributed modes of work coordination. Fourth, I provide an indicative taxonomy of
examples of concertive action, following which I discuss a range of distributed synergies
currently experienced in the workplace. Fifth, I discuss two implications of distributed
leadership. Finally, in the Discussion section, a number of issues are considered concerning
levels and units of analysis, and future research.

  1. The trouble with existing dualisms
    Criticism of prevailing conceptions of leadership units of analysis is palpable and
    increasing (see Hunt, 1999). At least three sets of concerns among commentators can be
    distinguished. In each case, there is dissatisfaction with the two sacrosanct binaries or
    dualisms, which for so long have defined the focus of the field: leader–followers and
    leadership–followership. First, there are concerns with leader-centrism. Second, there is
    dissatisfaction with individually conceived leadership. A third criticism is antileadership,
    which views the very idea of leadership as anathema. These objections are dealt with in turn.
    Due to limitations of space, the summary and discussion of the various arguments concerning
    the basic leadership constructs is condensed. In the next section, the article develops a fourth
    criticism, which is that the retention of a unit of analysis incorporating the two conventional
    binaries precludes accurate analysis and understanding of leadership practice, in particular,
    the actual divisions of leadership labor which prevail in different contexts.
    Concerns with leader-centrism take a number of forms, most of which are attempts to
    mitigate or eliminate the subordinate and dependent status inherent in what it means to be a
    follower. First, some commentators have sought to revitalise the notion of followership by
    purging it of an image of the obedient docility and asserting the complementarity of leader–
    followership (e.g., Berg, 1998). Second, leadership has been retained as a category, but with
    followership abolished by default through the adoption of a mechanism of leader rotation, as

at Moosewood, a restaurant collective (Vanderslice, 1988). This kind of turn-taking approach
to leadership is one way of countering claims that when leadership–followership is
institutionalised it yields learned helplessness for those relegated to the residual category
of follower (Gemmill & Oakley, 1992). A third response has been to reject outright the cliche´
that there can be no leaders without followers by dispensing entirely with the notion of
followership (Miller, 1998). The final, contrasting, response has been to take the cliche´ at face
value and to see leaders as literally dependent on followers, and to concentrate exclusively on
the behavior of the latter (e.g., Meindl, 1993, 1995). This strategy has been met with both
unease (e.g., Dansereau & Yammarino, 1998a, pp. xxxiv–xxxv) and accusations of bias.
Thus, although Ehrlich (1998, p. 309) concedes that ‘‘prior research on leadership has tipped
the scale in favor of leaders,’’ this is not a reason to ‘‘retreat toward followers, however
enticing this path may be.’’
Dissatisfaction with individualism has focused mainly on normative aspects of the
transformational and charismatic approaches to leadership. While some critics consider
these two approaches together, perhaps in recognition of their conceptual overlap, one
distinctive feature of transformational leadership for its proponents, in contrast to charismatic
leadership, is that the leadership of teams can be collectively transformational
(Sivasubramaniam, Murry, Avolio, & Jung, 2002). One concern with both approaches
has been with the value attached to the concentration of influence in individual leaders.
Keeley (1998, p. 125), for example, attacks Bass’s model of transformational leadership as
culturally at odds with James Madison’s view that ‘‘inspired leadership can do as much
harm as good’’ in a democracy, whereas transactional leadership is better suited to a system
founded on a separation of powers doctrine. Bass’s (1998, p. 175) response is that the
checks and balances required by power separation serve to promote gridlock, which can
only be cleared away by the best of leadership which is ‘‘both transformational and
transactional.’’ A second broad concern has been that the paradigm of individual
transformational, charismatic, and visionary leadership represents a retreat to discredited
heroics. Here, two particular concerns can be identified. First, the paradigm may have run
its race. Thus, Miller (1998, pp. 9–12) inquires whether, in circumstances in which an
increased incidence of abstention, withdrawal, and weakened commitment is detectable
among organization members, time is up for the leader with vision. Members have
switched off in such circumstances which are ill-conducive to vision-bearing, larger-thanlife
individuals, for a vision is ‘‘usually the product of a collective effort, not the creation of
a single, exceptional leader’’ (Yukl, 1999, p. 298). Second, there is concern at the
unconstrained scope for effectiveness attributed to the agency of individual leaders. Yet
‘‘it makes no sense,’’ says Beyer (1999, p. 311) (and see Gronn, 1995, pp. 20–21), ‘‘to
assume that all or even most of people’s behaviors are caused by something some kind of
leader does.’’
Antileadership arguments appeared in the mid-1970s. These began with Miner’s (1975, p.
200) assertion that leadership had ‘‘outlived its usefulness,’’ and culminated in Argyris’s
(1979) abandonment of the field for action science. Argyris claimed that leadership research
was mainly additive, rather than cumulative, and that for it to be useful leadership knowledge
had to connect with practitioners’ theories-in-use. Calder (1977) gave coherence to anti leader

ship. He tried to reorient the object of research by replacing leadership, which he saw as a lay
label for everyday interpersonal influence, with the construct of attribution. Calder (p. 190)
claimed that ‘‘inferences about leadership are made only from differences in behavior which
fit expectations of how leaders typically behave.’’ For Pfeffer (1977, p. 110), these
attributions of causality were reinforced by the effects of selection and inauguration
processes, and symbols and ceremonies, but could be reduced were leaders to be chosen
‘‘by using a random number table.’’ Leader causality was also queried by Kerr and Jermier
(1978) who argued that other factors (e.g., employee self-motivation to perform, work group
norms and characteristics inherent in the work itself) could substitute for leadership as equally
plausible causal explanations of organizational outcomes. Two decades later, Jermier and
Kerr (1997, p. 97) maintained that their substitutes claim had still not been properly addressed
by the field.
These concerns with individualism have been met with reassertions of individualism.
Initially, there was a rebuttal of antileadership. The decentralisation and fragmentation of
leadership throughout many social groups, according to Gardner (1984), impaired a society’s
capacity to deal with transcendent questions. What was needed were leaders to induce social
unity and purpose.
More recently, the concerns have been both accommodated and marginalised. Conscious
of a groundswell of support for ideas such as self-leadership and leader substitutes, as well as
shared or distributed leadership, Shamir (1999, p. 51) dichotomises focused leadership and
instances of so-called leaderless collective action. The former he labels ‘‘strong’’—or
‘‘disproportionate social influence in which the party that exerts greater influence on others
(the leader) can be identified’’—whereas the latter (i.e., distributed leadership) he deems
‘‘weak.’’ In recognition that numerous workplaces are being reshaped by computer-supported
cooperative work (CSCW) systems, and imperatives such as flexibilisation and career
boundarylessness, Shamir’s (p. 64) response is to call for a new charisma. He seeks charisma
minus ‘‘the perception of the leader as extraordinary or the attribution of super-human
qualities to the leader,’’ but with:
strong enough referent power to explain how leaders perform the integrative functions, offer
answers to ‘why’ questions, increase members’ internalized commitment, and provide the
psychological safety needed in times of change, all these without the support of permanent
strong cultures, or other ‘substitutes’.
By contrast, an alternative way forward has been summarised by Miller (1998, p. 22) as a
switch from devising normative models of organization to ‘‘a full appreciation of what it [i.e.,
organisation] is and how it works at present.’’
The argument of this article is consistent with this sentiment and with Miller’s (1998, p.
18) suggestion to dispense with the category of followership, so that ‘‘organization becomes a
process of negotiation between leaders.’’ Gibb proposed two forms of leadership: distributed
and focused. Instead of dualisms, then, which create hard and fast, either–or categories, a
useful option is to interpret Gibb’s two suggestions as end points of a continuum or a duality
of possibilities. The remainder of this article concentrates on the distributed polarity of this
continuum and anchors the discussion in the division of labor.

  1. Distributing leadership
    The main difficulty created by orthodox formulations such as leader–followers and
    leadership–followership is that they prescribe, rather than describe, a division of labor. This
    prescription poses methodological hazards for researchers, however, particularly when actual
    divisions of labor are constantly changing. Schools, for example, are known to rely
    increasingly on teams in order to cope with the intensification of school administrators’
    work (Grace, 1995) that has accompanied the global-wide trend to site-based devolution.
    Such adaptive responses are less accommodated by conventional binaries and highlight the
    need for new focal units which connect with actual work practices.
    3.1. Division of labor
    The idea of the division of labor means the totality of the tasks, and the technological
    capability (i.e., tools and knowledge) for the completion of those tasks by workers.
    Changes in the division of labor occur with the addition of new tasks and new task
    requirements, and with the adoption of new technologies. Together, these elements
    constitute the technical side of the division of labor. There is also a social form of the
    division of labor. This is evident when individuals and groups decide, on the basis of their
    values and interests, the preferred arrangement or configuration of tasks (e.g., their
    scheduling, physical alignment, available technology). These relations are a key source
    of social and organizational power.
    Inherent in the division of labor is a dialectical imperative encompassing both fragmentation
    and fusion. As tasks proliferate, are modified qualitatively, or become redundant due to
    external factors in the task environment, the technical side of the division of labor
    differentiates itself through a process of task specialisation. The rate and form of task
    differentiation are partly determined by decisions about appropriate and desirable ways of
    reorganizing and configuring the work, both through task integration and reorganization of
    the necessary labor (Sayer & Walker, 1992, pp. 15–17). This duality of differentiation–
    integration inherent in a division of labor is the source of emerging new forms of role
    interdependence and coordination which have resulted in distributed patterns of leadership.
    3.2. Definition of leadership
    In keeping with the thrust of Calder’s (1977) argument about inferred causation, leadership
    is defined in this article as a status ascribed to one individual, an aggregate of separate
    individuals, sets of small numbers of individuals acting in concert or larger plural-member
    organizational units. The basis of this ascription is the influence attributed voluntarily by
    organization members to one or the other of these focal units. The basis of the attribution of
    legitimate influence by the attributing agents may be either direct experience, through firsthand
    engagement with the particular focal unit, or vicarious experience, and thus reputed,
    presumed or imagined. The scope of attributed influence encompasses the workplace-related
    activities defined by the employment contracts which operate in particular contexts. These

activities may be confined to one of the domains in Hunt and Ropo’s (1995, p. 392)
multilevel leadership model (i.e., systems, organizational and direct) or cut across two or
three domains.
The individuals or multiperson units to whom influence is attributed include, potentially,
all organization members, not just managerial role incumbents. Managers may be leaders but
not necessarily by virtue of being managers, for management denotes an authority, rather than
an influence, relationship (Coleman, 1990, pp. 65–81). Finally, the duration of the attributed
influence may be short or long term.
3.3. Distributed leadership as numerical action
If focused leadership means that only one individual is attributed with the status of leader,
an additive or numerical view of distributed leadership means the aggregated leadership of an
organization is dispersed among some, many, or maybe all of the members. This additive
understanding does not privilege the work of particular individuals or categories of persons,
nor is there a presumption about which individual’s behavior carries more weight with
colleagues. On the other hand, numerical or multiple leadership allows for the possibility that
all organization members may be leaders at some stage (Wenger, 2000, p. 231). In this way,
says Miller (1998, p. 4):
a telephone operator, a receptionist, a salesperson, and a chief executive is each from their
different positions representing the system to the outside world and reflecting pictures of the
outside world back into the system.
This multiple sense is the most common understanding invoked in the growing number of
references to distributed leadership in the literature. It is also the distributed leadership which
Gibb had in mind, although he thought quantifying it in practice by drawing ‘‘an arbitrary line
on a frequency continuum’’ (Gibb, 1958, p. 103) was problematic.
3.4. Distributed leadership as concertive action
Distributed leadership in its numerical sense may be seen as the sum of its parts (i.e., the
sum of the attributed influence), but there is also a holistic way of construing it. Gibb’s
(1969, p. 215) two passing mentions of holism were to ‘‘the multiplicity or pattern of group
functions’’ of leaders and to the many roles constituting the ‘‘leadership complex.’’ In this
second sense of distribution, the conduct which comprises the unit of analysis is concertive
action, rather than aggregated, individual acts. At least three forms of concertive action may
be attributed with leadership. First, there are collaborative modes of engagement which
arise spontaneously in the workplace. Second, there is the intuitive understanding that
develops as part of close working relations among colleagues. Third, there is a variety of
structural relations and institutionalised arrangements which constitute attempts to regularise
distributed action. These modes of concertive action are the focus of the remainder of
the discussion.

3.4.1. Spontaneous collaboration
Spillane, Halverson, & Diamond (2000b, p. 6, original emphasis) refer to leadership as
distributed practice. This idea means that leadership is evident in the interaction of many
leaders, so that ‘‘leaders’ practice is stretched over the social and situational contexts of the
school; it is not simply a function of what a school principal, or indeed any other individual
leader, knows and does.’’ The concertively aligned conduct which exemplifies this stretching is
evident in a number of ways and in the accomplishment of numerous tasks. These may be
regular and anticipated (e.g., budget meetings, staff appraisals) or unanticipated (e.g., crises,
major problems), and they vary in scale, complexity, and scope.
One way is when sets of two or three individuals with differing skills and abilities, perhaps
from across different organizational levels, pool their expertise and regularise their conduct to
solve a problem, after which they may disband. These occasions provide opportunities for
brief bursts of synergy which may be the extent of the engagement or the trigger for ongoing
collaboration. Burns (1996, p. 1) cites the example of a person who, ‘‘because of certain
motivations of her own combined with a certain self-confidence, takes the first step toward
change, out of a state of equilibrium in the web [of relations]’’ and triggers interactions which
may later crystallise into a routine.
3.4.2. Intuitive working relations
In the second instance, intuitive understandings are known to emerge over time when two
or more organization members rely on each other and develop a close working relationship.
In this instance, leadership is manifest in the shared role space encompassed by their
partnership. It is the working partnership as a focal unit which is attributed with leadership by
colleagues, and the partners are aware of themselves as co-leaders. Fondas and Stewart’s
(1994) notion of a role set, which encompasses the dynamic interplay of the role perceptions
and expectations of set members, is helpful here. Shared roles emerge when set members
capitalise on their opportunities for reliance on others (e.g., by balancing each other’s skill
gaps) or because they are constrained to do so (e.g., due to overlapping role responsibilities).
Intuitive working relations are analogous to intimate interpersonal relations (e.g., successful
marriages and friendships), and two or more members act as a joint working unit within an
implicit framework of understanding. Gabarro (1978, p. 294) (see also Gabarro, 1987) found
that the influence of one person on another in such relations was ‘‘very much dependent on
how much that person was trusted by the other.’’
3.4.3. Institutionalised practices
The third concertive form of distributed leadership can be seen in the tendency to
institutionalise formal structures. Commentators acknowledge that structural relations in
organizations are formalised either by design or by adaptation. In the former case, new
structures may be mandated. Dissatisfaction with existing arrangements can often stimulate
the search for new design elements. Greenleaf (1977, p. 62) (see also Miller, 1998, p. 22)
provides the example of primus inter pares. This is a leadership group headed by a first
among equals, or a leadership team ‘‘of equals with a primus,’’ instead of the hierarchical
system of ‘‘the lone chief atop a pyramidal structure’’ typical of many organizations.

Greenleaf illustrated his point with the example of the office of the university president,
which appears to be a one-person chief but is not in practice. ‘‘With regard to the essential
work of the university—teaching students—the president is not the chief,’’ but the
establishment of a ‘‘council of equals,’’ Greenleaf (1977, p. 77, original emphasis) said,
would remove any ambiguity.
In the latter case, new elements may be grafted onto existing arrangements or managers
may try to regularise informal relations. An example would be the temporary task force or
team cited earlier, which is a concertive mechanism for pooling distributed capacity. A
derivative of it may be incorporated into an organization’s formal framework of governance.
Regardless of how and why practices are institutionalised, concertively acting units can be the
focus of colleagues’ attributions of leadership.
3.5. Conjoint agency
These three forms of concertive action represent successive stages in a process of
institutionalisation. In each case, the agents constituting the membership of the units act
conjointly. Conjoint agency means that agents synchronise their actions by having regard to
their own plans, those of their peers, and their sense of unit membership. Conjoint agency
entails at least two processual components. The first is purely internal to the concertive unit.
This is the experience of synergy, in which, as Follett (1973, p. 162) says, each unit member
‘‘calls out something from the other, releases something, frees something, opens the way for the
expression of latent capacities and possibilities.’’ The second is reciprocal influence, which is
both internal and external in its effects. Reciprocity denotes the influence of two or more parties
on one another and it occurs in a manner akin to a virtuous cycle or zigzagging spiral. Here, A
influences B and C, and is influenced in turn by them (i.e., AaB, AaC, and also BaC) with
each person subsequently bearing the accumulated effects of successive phases of influence, as
they begin influencing one another once again. The internal relationship of the conjoint agents
is one of reciprocal influence. And in their relations with their organizational peers, who
attribute leadership to them as focal units, conjoint agents both influence colleagues and are
influenced in return. Such reciprocal or collective influence, for example, is central to the idea
of ‘‘team leadership’’ (Sivasubramaniam, Murry, Avolio, & Jung, 2002, p. 68).
This conjoint agency, which is the defining attribute of concertive action, differs
analytically and contractually from Coleman’s (1990) idea of conjoint authority. In conjoint
authority relations, individuals unilaterally transfer the right of control over their actions with
the expectation of achieving benefits. The transfer is undertaken voluntarily on the
presumption of common interests between an individual and the new controller, so that in
a conjoint authority relation, ‘‘the superordinate’s directives implement the subordinate’s
interests’’ (Coleman, 1990, p. 74). An example is the decision to join a trade union. Conjoint
agents, by contrast, do not necessarily decide to act concertively to pursue common
interests—although their interests may be furthered by coordinated effort—and their
collaboration entails no alienation of rights and exchange of authority.
To the extent that conjoint agency is contractual, the contract is a psychological bond
(through synergy) which strengthens a coincidence of effort, goals, and resources in the

pursuit of mutually agreed ends. On the other hand, the work of conjoint agents is conducted
within a framework of authority relations secured by an employment contract. Because this
legal foundation defines agents’ formal roles, it affects conjoint agency relations by shaping
the synergies experienced (see below).

  1. Distributed leadership properties
    The principal change in the technical side of the division of labor, which has produced an
    increased incidence of conjoint agency, is networked computing. On the one hand, networked
    computing creates complex, information-rich, working environments which prioritise computation,
    scanning, and search routines as firms and human service organizations reposition
    themselves in fast-paced, fiercely competitive markets. On the other hand, networked
    computing bridges traditional barriers to simultaneous collaboration (e.g., distance, different
    time zones, geographic separation) between separate organizational units (Committee on
    Human Factors, 1990, p. 5): hence, the notion of CSCW. Given that it exacerbates the volume
    and scope of fragmented and dispersed knowledge noted by Hayek (1945, p. 519), increased
    environmental complexity provides the impetus to devise alternative modes of articulating the
    flow of work, in particular, the redefinition and reintegration of tasks.
    Work articulation is part of the more inclusive and generic problem of work design and has
    been defined by Strauss (1988, p. 164, original emphases) as ‘‘the overall process of putting
    all the work elements together and keeping them together.’’ This process entails ‘‘the
    specifics of putting together tasks, task sequences, task clusters—even aligning larger units
    such as lines of work and subprojects—in the service of work flow.’’ It is during work
    rearticulation that new forms of interdependence and coordination emerge.
    4.1. Interdependence
    To be dependent is to be constrained from autonomous task execution, with interdependence
    meaning reciprocal dependence between two or more organization members. Interdependence
    is manifest in two main ways. First, members’ responsibilities may overlap.
    Second, their responsibilities may be complementary.
    Overlapping interdependent role-related conduct occurs due to mutual needs for information
    and support (Stewart, 1991b, p. 128). A consequence of role overlap is redundant effort.
    Heller and Firestone (1995, p. 66) found that in eight US elementary schools, leadership was
    displayed by a number of people ‘‘sometimes in a jointly coordinated manner and sometimes
    with relatively little communication.’’ A virtue of redundancy, however, is that it provides
    mutual reinforcement, for ‘‘the more of each [leadership function] that is done, the better, and
    doing one helps accomplish others’’ (Heller & Firestone, 1995, p. 83). A second advantage of
    role overlap is that it reduces the likelihood of decision errors, because when two or more
    people share roles they tend to cross-check each other’s performance.
    Complementary interdependent role behavior was a strong feature of the working relations
    between three members of an executive role constellation in a medical psychiatric hospital

(Hodgson, Levinson, & Zaleznik, 1965). Here, complementarity was advantageous because
it enabled the interdependent executives to capitalise on the range of their individual
strengths. Role complementarity operates at two levels, the material and the emotional.
First, role set members rework the physical differentiation of tasks to create a pooled
resource of skills and attributes. By capitalising on their particular competencies, each person
performs specialised labor in a concerted approach to task accomplishment. An added
advantage of specialisation within a role set is that, while members rely on the strengths of
their peers, they enhance their lesser skills through frequent shared talk and observation of
each other in a range of venues. Second, because each person shares the effects of successful
and unsuccessful collaborative effort, role set members also experience common emotions.
Such negotiated working relationships cement the trust conducive to a nonthreatening
emotional climate and peer support.
The significance of both forms of interdependence is twofold. First, the more comprehensive
the breadth of the task interdependence among conjoint agents, the greater the
difficulty, as Simon (1991, p. 33) notes, of measuring ‘‘their separate contributions to the
achievement of organizational goals.’’ Second, the greater the extent of organization-wide
interdependence, the greater the density of the reserves of overall leadership capability. And
density, notes March (1984, p. 29), is what ‘‘makes an organization function well.’’
4.2. Coordination
Coordination means ‘‘managing dependencies between activities’’ (Malone & Crowston,
1994, p. 90) and encompasses the design, elaboration, allocation, oversight, and monitoring
of the performance of an organization’s technical core. The particular coordination mechanisms
utilised, singly or in combination, vary with the interdependencies and activities to be
managed, and the extent of their routinisation. Coordination arrangements include the
personnel, resources, materials, trajectories, tasks, and output required to complete activities.
The range of coordination mechanisms includes scheduling (e.g., synchronising of practices),
sequencing (e.g., task alignment), planning, bidding (e.g., by internal cost centres), standardising
(e.g., units of resource, quality control), information management, and consultation
and communication.
Coordination patterns in diverse communities of practice change. At ‘‘Alinsu,’’ a medical
insurance firm, for example, coordination of the interdependent work of claims technicians and
claims processors evolved in search of a requisite mode of engagement. Initially, technicians
working with processors were thought to prejudice consistency of technical advice, but their
subsequent segregation isolated technicians and caused work overload (due to indiscriminate
advice referrals). Clustering of both groups in a common work area eventually balanced
consultation with uniform information provision (Wenger, 1999, pp. 116–117).
Work coordination may be either explicit, as at Alinsu, or implicit. Explicit coordination
mechanisms are stipulated in the duty statements based on managers’ employment contracts.
In reality, however, a considerable amount of work coordination is implicit and informal.
Thus, ‘‘secretaries on research projects are commonly regarded as invaluable, if usually
unsung, heroic coordinators’’ (Strauss, 1988, p. 169). Implicit coordination occurs because

role definitions misidentify or fail to anticipate the nature and scope of the exigencies of work
performance. An analysis of a job’s history would reveal the normalisation of implicit
coordination as a tacit component of routine work. Such normality fuels the common
impression that things get done automatically, with the irony that ‘‘the better the work is
done, the less visible it is to those who benefit from it’’ (Suchman, 1995, p. 58). Despite their
invisibility and gendered subordination in serving the career interests of (mostly male)
managers, secretaries’ relations with managers illustrate role interdependence, because
secretaries are extensions of a boss’s working capacity (Golding, 1986, pp. 101–102).
Just as the conventional discourse of office work glosses these realities of secretarial
practice, so the rhetoric of leader–followership obscures pressures conducive to distribution.
The problem of coordination, where the bottom line in expanding open-ended information
environments is Hayekian dispersal, is less one of hierarchical consolidation of knowledge,
than one of ‘‘those ‘lower down’ finding more and more ways of getting connected and
interrelating the knowledge each one has’’ (Tsoukas, 1996, p. 22).

  1. Towards a taxonomy of distributed leadership
    Examples of intuitive working relations and institutionalised practices are presented in a
    taxonomy of distributed leadership (Fig. 1).1
    5.1. Status of the taxonomy
    It is important to be clear about the status of the illustrations as examples of distributed units
    of analysis. While the discussion to this point has emphasised the importance of leadership as
    an attributed status, not all of the authors of the work cited considered their subject matter from
    within a leadership framework and, of those who did, few defined leadership in attribution
    terms. In a handful of cases (certainly Denis, Lamothe, & Langley, 2001; Denis, Langley, &
    Cazale, 1996; Gronn, 1999; Hodgson, Levinson, & Zaleznik, 1965; Powell, 1997; and almost
    certainly, George & George, 1964; Newton & Levinson, 1973), the evidence provided by the
    authors suggests that it was the holistic units of analysis (i.e., sets of concertively acting
    agents) which were ascribed with leadership, not just individuals. Further, the evidence of
    internal and external reciprocal influence is unevenly reported across the case studies. For
    these reasons, the later discussion of internal relations among the conjoint agents is restricted
    to synergy.
    There are three other features to note. First, the taxonomy distinguishes between jointly
    authored work performed by bodily co-present agents in close physical proximity, and
    collaborating agents dispersed across a work site or over a number of sites. The former is
    defined as co-performed work and the second as collectively performed work. Important
    means of facilitating collective work performance include video-conferencing and electronic

group mailing. As yet, however, the full implications of the new group sizes and boundaries,
and the patterns of influence and identity norms facilitated by synchronous and asynchronous
electronic systems are unclear (although see Finholt & Sproull, 1991).
Second, in the examples cited, the collaborating agents are mostly individuals but
occasionally individuals and groups in coalition. Third, in each part of the taxonomy, two
criteria are used to distinguish the focal units. The first is membership size. To enhance the finegrained
analysis of interdependence, the second of the Committee on Human Factors’ (1990,
pp. 38–47) distinctions between two-member and multimember distributed work systems has
been subdivided into three-, four-, and five or more-member units. While the exact numerical
threshold in face-to-face groupings at which the intensity of the interpersonal emotions and
norms changes character is unclear, additions and subtractions affect the interplay of small
numbers significantly. Further, numerical increases trigger concerns with membership identity.
While the aggregated skills and values-base are enlarged and diversified, increased membership
means more energy is expended on the maintenance of a sense of collective definition. The
second criterion is the location of the unit’s membership in relation to a range of factors

including time, place, distance, and culture. These factors either constrain or enable particular
modes of coordination among unit members and their capacity to articulate work.
5.2. Other features of the taxonomy
There are some important points to note regarding the sample. First, the selection is
arbitrary, rather than exhaustive, in that the examples are known by the author to illustrate
aspects of the actual negotiated division of leadership labor between agents. In the coperformance
part of the taxonomy, for example, the possibilities opened up by the fourmember
and five or more-member cells are numerous and include such important membership
units as committees and teams. Apart from one or two cases, however, these have been
excluded, mainly due to widely varying membership sizes.
Second, in respect of levels of analysis, the examples are skewed towards Hunt and Ropo’s
(1995) system and organizational domains and formal incumbencies. Finally, the line
between intuitive and institutionalised action is acknowledged as unclear, and is partly
dependent for its validity on assumptions of the agents and their peers about contextualised
working practices. Moreover, the distinction elides the matter of structural transition and the
formalising of emergent conjoint agency relations.
With these caveats, the division of labor between individuals and sets of agents
summarised in the taxonomy is as follows:
Co-performance—intuitive working relations:
 part-time board chairs and full-time CEOs: Israeli companies (Chityayat, 1985), and UK
district health authorities (Stewart, 1991a, 1991b)
 heads of state and informal advisers: US President Woodrow Wilson and Colonel
Edward Mandell House (George & George, 1964)
 heads of government and deputies: former Australian Prime Minister, Gough Whitlam,
and Deputy Prime Minister, Lance Barnard (Powell, 1997)
 full-time CEOs and COOs, and sports coaches and deputies: US corporations and sports
teams (Heenan & Bennis, 1999)
 medical administrators: a US psychiatric teaching hospital (Hodgson, Levinson, &
Zaleznik, 1965)

Co-performance—institutionalised practices:
 co-principals: an Australian Catholic secondary school (Doyle & Myers, 1999)
 dual control clergyman presidents and headmasters: colonial Australian Methodist
schools (Zainu’ddin, 1981)
 musicians: UK string quartets (Murnighan & Conlon, 1991)
 health professionals: relations between a management group and clinical research teams
in a US psychiatric hospital ward (Newton & Levinson, 1973)
 scientists and technicians: research laboratory of 17th-century English chemist, Robert
Boyle (Shapin, 1989)

 worker cooperative members: a US restaurant collective (Vanderslice, 1988)
 school principals and senior management teams: English secondary schools (Hall &
Wallace, 1996; Wallace & Hall, 1994)
Collective performance—intuitive working relations:
 school heads and heads of campus: an Australian multicampus boys boarding school
(Gronn, 1999)
 university governors, presidents, managers, and faculty: US universities and colleges
(Birnbaum, 1992)
 boards, CEOs, medical councils, and health care professionals: Canadian provincial
hospitals (Denis, Langley, & Cazale, 1996; Denis, Lamothe, & Langley, 2001)
Collective performance—institutionalised practices:
 members of social movement organizations: UK women’s centres (Brown, 1989; Brown
& Hosking, 1986)
5.3. Accomplishing conjoint agency
The two distinguishing features of conjoint agency cited earlier were interpersonal synergy
and reciprocal influence. While the varying purposes of the bulk of the 21 studies cited
preclude meaningful conclusions about reciprocal influence, two main types of synergies are
worthy of comment: formal and informal. Formal synergies are based on role incumbency
while informal synergies are anchored on personal relations (e.g., friendship). Formal
synergies include four subtypes: cross-hierarchy, trusteeship, parity of relations, and separation
of powers (Fig. 2). Summary detail is used to contrast the dynamic features of each type.

5.3.1. Cross-hierarchy
Cross-hierarchical synergies entail negotiation of role boundaries. Agents negotiate their
role boundaries either by blurring or by expanding them.
Role blurring occurred in the role constellation of three senior hospital psychiatrists
depicted by Hodgson, Levinson, & Zaleznik (1965, p. xii). This executive triumvirate
operated as ‘‘a relatively integrated whole’’ and displayed role–task specialisation, differentiation,
and complementarity. Complementary specialisation allowed each man to act as he
preferred and as he was best equipped, within a jointly agreed-upon framework of activities in
pursuit of the interests of the hospital. Extensive role blurring meant that ‘‘each member of
the system could, in part, impose his own personality on the system,’’ but there was ‘‘a limit
to the extent of any one individual’s control’’ of that system (Hodgson, Levinson, & Zaleznik,
1965, p. 287).
Four factors are likely to account for the depth of trust in role constellations and couples:
shared values, complementary temperaments, requisite psychological space, and previous
experience of collaboration (Gronn, 1999, pp. 54–57). These factors were evident in the
‘‘odd couple’’ formed by J.R. Darling (headmaster) and E.H. Montgomery (master in
charge), in the foundation of the Timbertop campus of the Geelong Grammar School in
Australia in the early-1950s. Despite the fact that they were separated by 200 miles, and
(given the initial absence of a telephone connection) relied mostly on the postal service for
their communications, the two men evolved a close working relationship in which
Montgomery made on-site decisions within a broad framework reflecting his chief’s wishes.
Darling depended on Montgomery for the success of his scheme, which he described as ‘‘the
very apple of my eye’’ (Gronn, 1999, p. 51). Where one or more of these four factors is
absent, couples fail to develop this level of trust and merely ‘‘bring out the worst in each
other’’ (Krantz, 1989, p. 164).
By contrast, boundary expansion requires the preparedness of organizational superiors to
include junior colleagues within the locus of their authority. Senior management teams
(SMTs) in UK secondary schools, for example, are established entirely at the discretion, and
on the initiative, of school heads (Wallace & Hall, 1994). Because the contractual authority
relationship here is disjoint, rather than conjoint, expansion of the locus of authority may be
anxiety-inducing for subordinates and superiors. Unlike conjoint authority, in disjoint
authority relations, a subordinate’s interest is realised when compensation is paid, rather
than through a superior’s directives (Coleman, 1990, p. 74). The UK heads’ dilemma was
that they exercised sole authority for overall operations, and responsibility for student
learning and the work of teachers, yet, within a policy framework of local management, they
depended on SMTs to perform the work. Trust between team members was only possible
through mutual consent, with the result that heads were vulnerable and dependent on their
Thus, while it was ‘‘up to the head to provide conditions which encourage all members to
contribute fully,’’ the interchangeability of team leadership and followership roles meant that
‘‘other members must also facilitate the teamwork process.’’ The strength of teamwork
depended on each member’s capacity to cope with these dualities (Hall & Wallace, 1996,
p. 304). Teaming through boundary expansion is a potentially risky strategy: teams that

succeed bolster a superior’s credibility, but failing teams strain an organization’s authority
relations. For these reasons, SMTs in schools were akin to marriages, in which ‘‘there were
high and low points and sometimes divorce hovered uncomfortably in the wings’’ (Hall &
Wallace, 1996, p. 300).
5.3.2. Friendship
Synergies grounded in friendships are noncontractual, and therefore entail neither conjoint
nor disjoint authority. Depending on the degree of calculation, friendships capitalise on the
advantages of mutual attraction and compatible personal attributes.
Career- and work-based friendships are common. They cut across or are absorbed into
organization members’ role relations. A friendship which became the basis for professional
working relations was Woodrow Wilson’s attachment to Colonel House. House was an
inveterate schemer, political kingmaker, and playmaker. The two men hit it off at their first
meeting in late-1911, but throughout their friendship House studiously avoided Wilson’s
offers of formal positions. Instead, he noted in his diary how he ‘‘much preferred remaining
free to advise the President about matters in general.’’
The need to shun official positions was essential, House once confided, ‘‘if he wanted to
retain his personal influence over Wilson’’ (George & George, 1964, p. 110). His goal was to
be part of his president’s ‘‘personal sphere of power’’ yet, even though he studied Wilson’s
character closely, he once said that he ‘‘could never really understand him’’ (George &
George, 1964, p. 124). As well as mutually shared affection, the two men’s friendship was
fuelled by self-interest and ambition on both sides. A grounding in some of the factors
associated with trust in couples and constellations would appear to be necessary if workrelated
friendship synergies are to be immune to such calculus.
5.3.3. Trusteeship
The essence of trusteeship, according to Greenleaf (1977, p. 103), is an oversight of
executive power, to check the ‘‘corrupting influence’’ of power on executives and to prevent
harm to ‘‘those affected by its use.’’ To that end, the ideal synergy would be one in which
trustees, and in particular, their board chairs, were proactive, rather than reactive, in their
stewardship of organizations.
Despite the ‘‘legal fiction’’ that boards are the agents of stockholders, ‘‘they are most often
the creatures of top management’’ (Coleman, 1990, p. 563), an assessment confirmed by
Chityayat’s (1985, p. 69) study of working relations between CEOs and chairs. In the public
sector, however, the situation may approximate more closely the ideal depicted by Greenleaf. In
public health in the UK, the most common pattern of relations between board chairs and district
general managers (DGMs) was found to be ‘‘mutual dependence,’’ analogous once again to a
marriage (Stewart, 1991a, p. 518). The creative tension in their engagement stemmed from the
chairs’ need for information, and the DGMs’ reliance on their chairs for the conduct of the
health authorities and the interpretation of the chair’s role. Both individuals shared the
leadership of public health in their districts. As with the negotiated division of labor—both
psychological and task-related—in Hodgson, Levinson, & Zaleznik’s (1965) constellation and
Gronn’s (1999) couple, the strength of the DGM–chair synergy was anchored in the interplay

of perceptions and expectations within the role set. In particular, in DGM–chair partnerships,
the success of working relations depended on whether one of the two conceived of the
relationship ‘‘in that way’’ and ‘‘which of them should undertake certain kinds of work’’
(Stewart, 1991a, p. 525).
5.3.4. Parity of relations
Two alternatives to role sharing by crossing hierarchical boundaries are to dispense with
hierarchies (parity of relations) or to establish multiple competing institutional structures
(separation of powers). Each form generates its own unique synergy.
Three studies in the taxonomy illustrate different dimensions of, and tensions inherent in,
the distributed leadership of different sized membership groups governed by parity. As group
membership numbers increase, group incentive structures rely on greater amounts of zeal to
enforce norms (Coleman, 1990), guaranteeing action in concert. A musical string quartet, the
smallest of these units, is a self-governing interdependent work group whose work ‘‘is done
only as a unit’’ in which members use ‘‘each other’s outputs as their own inputs’’ (Murnighan
& Conlon, 1991, p. 165). The incentive to perform as a unit is that members share a common
interest in excellence and fidelity to a composition while striving for a distinctive interpretation
of a musical score.
Performance tensions emerge from coping with two paradoxes. The initial paradox is the
first violinist’s duality of role as a musician who provides a lead, as the publicly acknowledged
ensemble leader, but who, likewise, is just one fourth of a quartet. Another paradox
concerns the second fiddle, who may be equally proficient but is subordinate in status to the
first violinist and also only one fourth of a quartet. Next in ascending membership size is the
Moosewood collective. This 15-year-old example of conjoint authority comprised 18
members who severed the link between leadership and leaders, by institutionalising turntaking,
and rotating all tasks and responsibilities. In order to abolish followership and to be
‘‘leaderful,’’ thereby institutionalising leadership without bosses, policy and operational
decisions at Moosewood were made consensually (Vanderslice, 1988, p. 685). The durability
of collective self-management, however, is really tested in large social movements which
strive for ‘‘enough order, but not too much’’ (Brown & Hosking, 1986, p. 73), in seeking to
avoid the twin evils of elitism and structurelessness. Yet the rejection of fixed roles, adoption
of minimal structures, encouragement of the participation of all, resort to rotating meeting
chairs, and achievement of solidarity through networking entailed by walking this fine line
still require skilled performers. Sometimes a sufficiency of these emerges, but on other
occasions it does not. The vitality and longevity of such social movements are dependent on a
successful combination of spontaneous collaboration and the numerical version of distributed
leadership distinguished at the outset of this article.
5.3.5. Separation of powers
With the exception of friendship, the synergies considered so far have been the
products of either vertical or horizontal systems of authority. A different kind of synergy
is in evidence, however, when authority is segmented, as in a separation of powers

The segmentation, rather than the concentration, of authority creates a ‘‘pluralistic domain’’
(Denis, Lamothe, & Langley, 2001, p. 809) of multiple agents, pursuing different objectives in
fluid relationships. This situation generates qualitatively different kinds of tensions. These are
evident in boundary disputes between separate authorities over jurisdictional ambiguities and in
the alliances pursued by different sets of agents. An example of the former is the conflict
concerned with the competing priorities of research and care which arose between a psychiatric
ward management group and three clinical teams in a hospital ward, and the question of whether
the ideologies of research and care were even compatible (Newton & Levinson, 1973).
An instance of the latter tension is found in the numerous sources of leadership in
universities, and in the power balances that emerge periodically between different constituencies
within an at times unwieldy university structure (e.g., departments, faculties, colleges,
professorial boards, senates, councils, and offices of presidents). One result, Birnbaum’s
(1992, p. 124) research found, was that ‘‘institutions could improve even when their
presidents were not considered particularly effective.’’ An important long-term consequence
of a separation of power arrangement is that strategic change becomes ‘‘sporadic and
unpredictable,’’ as Denis, Lamothe, & Langley (2001, p. 810) discovered in an investigation
of Canadian health care. With provincial authority for overall health care delivery shared
between CEOs, hospital boards, physician-elected medical councils, and a range of other
clinical health professionals, significant change depended on ‘‘a tightly knit leadership group
that can act in concert’’ (Denis, Lamothe, & Langley, 2001, p. 817). A decade’s research,
however, revealed shifting institution-based patterns of collective leadership of varying
strength by different combinations of actors, resulting in sporadic achievement of change.

  1. Implications of distributed leadership
    The argument of this article has been that leadership would be better served by understandings
    more closely connected to the realities of workplace practice. To that end, in
    developing Gibb’s suggestion of distributed leadership, the discussion has concentrated on
    the division of labor and extended the existing unit of analysis to include the leadership of
    agents acting conjointly. Two implications follow.
    6.1. Leadership and management
    The first is to caution against hasty assumptions that managers are automatically leaders or
    that only managers lead. This implication is especially important for those commentators
    who utilise interchangeable leader–follower and superior–subordinate dyads. Among
    proponents of multilevel analysis, for example (see below), the attribution of leadership as
    a status contingent upon others’ judgments is acknowledged explicitly in some discussions of
    superior–subordinate relations (e.g., Dansereau et al., 1995, p. 441), whereas in others (e.g.,
    Klein & House, 1995, p. 190), terms like ‘‘leader’’ and ‘‘head’’ (a managerial role) are
    conflated. But if, as has been suggested, leadership is attributed influence, and potentially
    attributable to any individual or concertive unit then, despite the preponderance of

managerial formations in the taxonomy, managers are no different from those whom they
manage, for they too have to be attributed with leadership status.
First and foremost, managers are legal agents of owners or employers for, unlike
leadership, management is defined by an employment contract. Employment contracts
enshrine disjoint authority relations in which employees accept the authority of an
employer’s agent and obey, in return for guaranteed remuneration (Coleman, 1990, pp.
79–81). Generally, employees will accept managerial directives provided these are within
their ‘‘zone of indifference’’ (Barnard, 1982, p. 167). If, however, they can be convinced to
regard their managers as leaders, employees are not only likely to comply but also to make
value-added commitments. They may, in fact, take little convincing of leadership (Calder,
1977, p. 191):
The fire chief who goes with a group of firemen into a burning building probably will be
perceived as a leader even if going into a building is his duty.
In the articulation of work practice, the conflation of management and leadership confuses
the division of rights with the division of labor. The authority to prescribe a division of labor
is not the same as the actual conduct which comprises that division of labor. Employees may
expect their managers to lead, but as for any other organization member, commentators need
to adduce evidence of leadership attributions.
6.2. Leadership and training
Many factors contribute to successful learning from training including, presumably, a
working familiarity with the possibilities of conjoint agency and concertive action. Stewart’s
(1991b) findings about role sharing in the UK health sector illustrate the point. Role sharing
between DGMs and board chairs prompted Stewart (p. 136) to interpret a managerial job, not
as a slot filled by an incumbent performing prescribed duties, but as ‘‘flexible space that is
only partially held by any one job holder.’’ Stewart chose flexibility because job boundaries
were shifting, and were partially occupied because ‘‘there is more work that can be done in
many management jobs than any individual will have the time, perception, interest and
sometimes the ability to do.’’
Role sharing was an antidote to managers’ selective attention and it occurred ‘‘because two
or more individuals can make good use of their particular interests so that together they make
a more effective contribution than they would separately.’’ Training with a view to equipping
individuals to cope with flexible space would require ‘‘a strategic view of what needs doing
and the ability to develop complementary relationships,’’ but its inclusion in programs may
not yet be extensive. In schooling, for example, where accreditation standards for school
leaders are increasingly popular, standards are framed as expectations of individual jobholders.
One set of standards, for example, says: ‘‘A school administrator is an educational leader
who promotes the success of all students by. . .,’’ and thereby perpetuates the conflation of
leadership and management. Ironically, the commitment to individual job-holding in this
instance occurs despite a public espousal of distributed leadership by the standards sponsoring
agency (Gronn, in press).

  1. Discussion
    The preceding discussion has some important consequences for the development of theory,
    particularly with respect to the reliance of commentators on units and levels of analysis, and
    for future research.
    7.1. Levels and units
    These two terms designate organizational phenomena associated with forms of organizational
    interaction and conduct. The earlier suggestion was that in leadership, theoretical work
    on units lagged behind similar developmental work on levels of analysis. Consideration is
    now given to some aspects of the relation between levels and units. Although the literature on
    levels in organizational behavior and leadership is extensive, space permits attention to just a
    few of the more salient points.
    In most discussions of levels, the authors’ overriding concerns are conceptual and
    methodological, whereas with units the key issues are ontological. The idea of a level tends
    to be used in one of two senses by leadership commentators, both of which are based on
    hierarchical representations of organizations. First, there is Hunt’s (1991) and Hunt and
    Ropo’s (1995) multiple organizational level leadership model which divides leadership
    functions into three ascending strata of complexity of demands on leaders. Here, for the
    purpose of representing the totality of leadership processes, ‘‘level’’ is based on Jaques’ (e.g.,
    Jaques & Clement, 1995) notion of time span of discretion, with each level encompassing
    broadly grouped organizational functions.
    Second, there are multilevel approaches whose purpose is to disaggregate the construct
    ‘‘organization’’ into sublevels (typically, individual, dyad, group, etc.) for purposes of data
    procurement and aggregation, measurement of leader effects, and application of findings.
    In each case, the agent of influence is an individual leader, although Hunt (1991, p. 37)
    and Hunt and Ropo (1998, p. 356) make allowances for collective formations to be
    subsumed under ‘‘the leader’’ designation, and levels are articulated within the conventional
    leader–follower dualism—i.e., the assumption is that there is one leader (or
    ‘‘superior’’) and a number of followers. In the second approach, however, the recipients
    of influence, designated by the construct ‘‘followers’’ (although sometimes by ‘‘subordinates’’),
    are disaggregated according to various demographic (e.g., age, gender), identity
    (e.g., self-concept, efficacy, values), positional (e.g., distance from the leader), perceptual,
    and other measures.
    The aim is to ascertain the degree of independent, heterogeneous, and homogeneous
    responses to leader stimuli on a number of measures (e.g., the extent of individualised
    leadership), the level and extent (although not the mechanics of the process) of the diffusion
    of leader effects, and to draw conclusions about the varying quality of leader–follower or
    superior–subordinate relationships.
    Conceptually, the idea of ‘‘level’’ satisfies two purposes. First, it partitions a theoretical
    construct (i.e., organization) into subcomponents for the purposes of reference and representation.
    Second, levels specify target applications of theory. In some discussions (e.g.,

House, Rousseau, & Thomas-Hunt, 1995, pp. 73–74; Klein, Dansereau, & Hall, 1994),
however, levels are treated as synonymous with units, and the commonly invoked ascending
series of individual–dyad–group–organization is used in reference to units of analysis as
well as levels. At least three senses of ‘‘unit’’ are confounded in the literature: unit of
measurement, unit of work, and unit of analysis. As units, ascending designations like
individual–dyad–group–organization have, at best, analytical rather than entitive status for,
strictly speaking, these are measurement units (i.e., abstract nested parts and wholes which
are the targets of research instrumentation). By contrast, the notion of units of work refers to
real-world objects which are the outcome of the processes of differentiation and reintegration
that are integral to particular contextualised divisions of labor (e.g., the HRM department of
the Melbourne branch of X bank). In this sense, units of work are ‘‘task-oriented subsystems’’
(Miller, 1959, p. 247); i.e., managerially defined and sanctioned sets of operations which
reflect the transition from simple to complex processes of production. In short, these are
creatures of the division of rights distinguished earlier.
The idea of a unit of analysis is different again. Here, unit refers to the idea of a bounded
set of elements comprising the entity which is the focus of research. Contrary to Freeman’s
(1978, p. 337) observation, however, units of analysis need not be ‘‘unambiguously separable
from each other and from their environments in time and space,’’ for the criteria for unit
inclusiveness may be fluid. In contrast with this traditional static view, Mandelblit and
Zachar’s (1998, p. 230, original emphasis) notion of a dynamic unit in cognitive science as
‘‘more flexible, context-dependent, and has fuzzier boundaries based on a general pattern of
correlation across its elements,’’ is helpful.
This dynamic understanding encompasses the overall unity or holism created by combining
the constituent properties while simultaneously recognising the emergent nature of those
properties and the unit. In respect of the division of labor, therefore, the idea of the unit of
analysis, dynamically defined, includes not only the officially defined units of work noted by
Miller (1959), but the totality of the unofficial and informal work practices, including those
understood tacitly but rendered invisible as part of the normalisation of work practices which
was referred to earlier in the article. For leadership, the implications of a dynamic
understanding of the unit of analysis include a view of leadership as less the property of
individuals and more as the contextualised outcome of interactive, rather than unidirectional
causal, processes (Hosking & Morley, 1991, pp. 239–261). The fluidity of the spontaneous
collaboration and intuitive working relations distinguished earlier, along with the continuity
of workplace relations implied by the idea of institutionalised practices, accords with this
processual understanding of leadership. But what are the implications for research of a
process view of leadership?
7.2. Future research
In a recent review of historical developments, Hunt (1999, p. 139) suggests that the field of
leadership is currently in stage two (‘‘concept evaluation/augmentation’’) of a three-stage
concept evolutionary model, with a number of new approaches in the first stage (‘‘concept
introduction/elaboration’’), one of which is the topic of this article: ‘‘collective, distributive,

dispersed or relational leadership.’’ In addition to Hosking and Morley, Hunt (1999, p. 138) is
one of a growing chorus of voices (e.g., Conger, 1998a, 1998b; Parry, 1998) bemoaning the
dearth of processual studies of leadership.
With the possibility of distributed leadership first raised more than four decades ago, a
revised unit of analysis is long overdue. Although recent workplace restructuring has been the
trigger for a reawakened interest in distributed practice, a number of studies in the taxonomy
predate these developments, either in their subject matter or in their publication date. A key
question, therefore, is whether distributed leadership is new or whether commentators are
beginning to recognise a phenomenon which has always existed? On this point, Hodgson,
Levinson, & Zaleznik (1965, pp. 391–392) raised the possibility that distribution may be
inherent in the very nature of what it means to organize when they noted that a number of factors
facilitate ‘‘the emergence of interlocking roles.’’ These factors include the assignment of
employees ‘‘to formally defined aggregates of individuals that prompt the interlocking of
roles,’’ required contacts with workplace peers, and the tendency for employees to gravitate to
compatible colleagues. While the question of newness is moot and provides fuel for
speculation, the comments of Hodgson, Levinson, & Zaleznik suggest that role interlocks
among co-performing and collectively performing sets of agents constitute the key focus of
research into leadership processes.
As regards process, the literature in the taxonomy raises more questions than it answers.
With a few notable exceptions, evidence of the dynamics of role performance among the
conjoint agents depicted in the studies, and between the agents and their colleagues within
and across levels, and of the social construction processes by which concertively performing
agents might have been attributed with leadership, is patchy.
At best, the discussion provided evidence of some of the necessary and sufficient
conditions to be satisfied to facilitate productive synergies among agents. On the other hand,
the absence of these data offers an exciting window of opportunity for qualitative,
longitudinal field studies designed to capture ‘‘the richness and nuances of contextual
variables and their influence on leadership’’ (Conger, 1998b, p. 81). This kind of research
would inquire, inter alia, into the wider environmental and organizational circumstances
governing the genesis and development of forms of distributed leadership, and the factors
conducive to their duration and continuity. In respect of levels, it would also seek to account
for the location of forms of distributed leadership and the frequency of their occurrence, and it
would contour the scope and breadth of the properties they embody. Finally, from the
conventional perspective of leadership as causal agency, and to the extent that informants’
implicit theories construct the role of forms of distributed leadership as causally potent, such
research would endeavour to assess the influence and impact of distributed forms on shortand
long-term organizational performance.
On the other hand, to take a processual approach to leadership means to acknowledge that
organization is as much a structural outcome of action as a vehicle for it, and that leadership is
but one of a number of structuring reactions to flows of environmental stimuli. Indeed, other
factors may substitute for leadership in the accomplishment of organization (Jermier & Kerr,
1997; Kerr & Jermier, 1978). To the extent that leadership does play a structuring role,
however, the focus of field work would be on ‘‘influential ‘acts of organizing’’’ which

‘‘contribute to the structuring of interactions and relationships, activities and sentiments’’
(Hosking, 1988, p. 147).
An understandable reaction to what Hosking (1988, p. 150) termed top-down, ‘‘physicalist’’
views of organizations, when attending to the contribution of leadership to organizing, might be
to counteract a structural emphasis by according open-ended degrees of freedom and choice to
actors in socially constructing organization through their leadership. Such a reaction would be
unfortunate, for structure, as Archer (1995, pp. 71–75) has noted, is always paramount in the
duality of its interplay through time with agency. Inevitably, structures constrain the work of
agents for structural outcomes mostly outlive their particular creators to preexist as contexts for
succeeding generations of agents. Equally, however, structures are enabling and, as the intended
and unintended outcomes of the accumulated deeds of numerous previous cohorts of
structuring agents, they are both reproducible and transformable. In short, in the duality of
structure and agency, ‘‘agential power is always restricted to remaking, whether this be
reproducing or transforming our social inheritance’’ (Archer, 1995, p. 72).
While at one level, therefore, organization is evident in numerous cognitively derived
representational media and tools—e.g., plans, budgets, forecasts, routines—which are the
artifacts of collective learning and the products of organization members’ collaborative efforts
at problem-solving, at another level, organization is constituted by structured role relations.
Yet, there is always the potential for existing sets of structured role relations to be redefined as
a result of the culturally informed activities of combinations of agents. If so, then a key
element in research into distributed leadership has to be structural uptake, or work which, as
Conger (1998a, p. 80) suggests, ‘‘effectively documents the actual process from leader
behavior to ongoing norm.’’

  1. Conclusion
    The argument of this article has been that, in the face of a dissonance between traditional
    constructs of leadership and the reality of the division of labor in leadership practice, the field
    of leadership would be more profitably served by a revised unit of analysis. Despite
    protestations that ‘‘there is no such thing as a detached, free-standing leader–follower role
    relationship,’’ and that leadership is ‘‘but one part of the total accountability being discharged’’
    by managers (Jaques & Clement, 1995, pp. 5, 39), however, the dominant unit of
    analysis continues to be the solo or stand-alone leader. But an unfortunate by-product of the
    presumption of measurable, causal influence exercised by solo leaders, as in currently popular
    transformational and charismatic theories, for example, especially the latter, is that ‘‘the
    organizational processes that ultimately determine effectiveness are seldom described in any
    detail’’ (Beyer, 1999, p. 325 and see Yukl, 1999, p. 288).
    This article has suggested some ways in which the neglect of these processes might be
    rectified. In particular, the discussion concentrated on leadership rather than followership and
    revised the unit of analysis to include forms of distributed practice. Distributed leadership was
    shown to encompass not merely the structuring influence of numerous individuals, but three
    forms of concertively patterned and reproduced activity-based conduct, each representing

‘‘contribute to the structuring of interactions and relationships, activities and sentiments’’
(Hosking, 1988, p. 147).
An understandable reaction to what Hosking (1988, p. 150) termed top-down, ‘‘physicalist’’
views of organizations, when attending to the contribution of leadership to organizing, might be
to counteract a structural emphasis by according open-ended degrees of freedom and choice to
actors in socially constructing organization through their leadership. Such a reaction would be
unfortunate, for structure, as Archer (1995, pp. 71–75) has noted, is always paramount in the
duality of its interplay through time with agency. Inevitably, structures constrain the work of
agents for structural outcomes mostly outlive their particular creators to preexist as contexts for
succeeding generations of agents. Equally, however, structures are enabling and, as the intended
and unintended outcomes of the accumulated deeds of numerous previous cohorts of
structuring agents, they are both reproducible and transformable. In short, in the duality of
structure and agency, ‘‘agential power is always restricted to remaking, whether this be
reproducing or transforming our social inheritance’’ (Archer, 1995, p. 72).
While at one level, therefore, organization is evident in numerous cognitively derived
representational media and tools—e.g., plans, budgets, forecasts, routines—which are the
artifacts of collective learning and the products of organization members’ collaborative efforts
at problem-solving, at another level, organization is constituted by structured role relations.
Yet, there is always the potential for existing sets of structured role relations to be redefined as
a result of the culturally informed activities of combinations of agents. If so, then a key
element in research into distributed leadership has to be structural uptake, or work which, as
Conger (1998a, p. 80) suggests, ‘‘effectively documents the actual process from leader
behavior to ongoing norm.’’

  1. Conclusion
    The argument of this article has been that, in the face of a dissonance between traditional
    constructs of leadership and the reality of the division of labor in leadership practice, the field
    of leadership would be more profitably served by a revised unit of analysis. Despite
    protestations that ‘‘there is no such thing as a detached, free-standing leader–follower role
    relationship,’’ and that leadership is ‘‘but one part of the total accountability being discharged’’
    by managers (Jaques & Clement, 1995, pp. 5, 39), however, the dominant unit of
    analysis continues to be the solo or stand-alone leader. But an unfortunate by-product of the
    presumption of measurable, causal influence exercised by solo leaders, as in currently popular
    transformational and charismatic theories, for example, especially the latter, is that ‘‘the
    organizational processes that ultimately determine effectiveness are seldom described in any
    detail’’ (Beyer, 1999, p. 325 and see Yukl, 1999, p. 288).
    This article has suggested some ways in which the neglect of these processes might be
    rectified. In particular, the discussion concentrated on leadership rather than followership and
    revised the unit of analysis to include forms of distributed practice. Distributed leadership was
    shown to encompass not merely the structuring influence of numerous individuals, but three
    forms of concertively patterned and reproduced activity-based conduct, each representing.

A body of indicative evidence of distributed forms of
leadership was reviewed. This review was confined to numerically small work groupings,
and over 20 studies were used to provide a preliminary taxonomy of distributed leadership
and to illustrate the constituent elements of distributed units, in particular, the core property of
conjoint agency. Finally, some of the key assumptions informing a program of qualitative
research into distributed work arrangements were discussed. Such research should either
confirm or invalidate a range of potential co-performed or collectively performed synergies
evident in forms of distributed leadership which include geographically dispersed couples,
co-leaders and partners, triadic role constellations, and rotating leader systems. Moreover,
research should advance understanding of the circumstances and factors which facilitate or
impede participants’ perceptions, acceptance, and expectations of distributed arrangements,
and provide evidence of the nature and extent of workplace interdependence and reciprocities.
In short, rather than continuing to rely on language which presumes a division of labor a
priori, when distributed leadership becomes the unit of analysis, productive outcomes of
research for the scholarly community are likely to include a revised discourse which
accurately reflects the evolving division of labor in the workplace and analyses which are
realistically aligned with practice.
Archer, M. (1995). Realist social theory: the morphogenetic approach. Cambridge: Cambridge University Press.
Argyris, C. (1979). How normal science methodology makes leadership research less additive and less applicable.
In J. G. Hunt, & L. L. Larson (Eds.), Crosscurrents in leadership (pp. 47–73). Carbondale, IL: Southern
Illinois University Press.
Barnard, C. I. (1982). The functions of the executive. Cambridge, MA: Harvard University Press.
Bass, B. M. (1998). The ethics of transformational leadership. In J. B. Ciulla (Ed.), Ethics: the heart of leadership
( pp. 169–192). Westport, CT: Quorum Books.
Berg, D. (1998). Resurrecting the muse: followership in organizations. In E. B. Klein, F. Gabelnick, & P. Herr
(Eds.), The psychodynamics of leadership (pp. 27–52). Madison, CT: Psychosocial Press.
Beyer, J. M. (1999). Taming and promoting charisma to change organizations. Leadership Quarterly, 10 (2),
Birnbaum, R. (1992). How academic leadership works: understanding success and failure in the college presidency.
San Francisco: Jossey-Bass.
Brown, M. H. (1989). Organizing activity in the women’s movement: an example of distributed leadership. In B.
Klandermans (Ed.), International social movement research, vol. 2 (pp. 225–240). Greenwich, CT: JAI Press.
Brown, M. H., & Hosking, D.-M. (1986). Distributed leadership and skilled performance as successful organization
in social movements. Human Relations, 39(1), 65–79.
Bryman, A. (1996). Leadership in organizations. In S. R. Clegg, C. Hardy, & W. Nord (Eds.), Handbook of
organization studies (pp. 276–292). London: Sage.
Burns, J. M. (1996). Empowerment for change: a conceptual working paper. The James MacGregor Burns Academy
of Leadership, University of Maryland. http://www.academy.umd.edu/scholarship/casl/klspocs/jburn_p1.htm.
Calder, B. J. (1977). An attribution theory of leadership. In B. M. Staw, & G. R. Salancik (Eds.), New directions in
organizational behavior (pp. 179–204). Chicago, IL: St. Clair.
Chityayat, G. (1985). Working relationships between the chairman of the boards of directors and the CEO.
Management International Review, 25(3), 65–70.

Coleman, J. S. (1990). Foundations of social theory. Cambridge, MA: Harvard University Press.
Committee on Human Factors (1990). Distributed decision making: report of a workshop. Washington: National
Academy Press.
Conger, J. A. (1998a). Qualitative research as the cornerstone methodology for understanding leadership. Leadership
Quarterly, 9(1), 107–121.
Conger, J. A. (1998b). The folly of knowing an elephant by its tail: why the leadership field needs multiple levels
of analysis. In F. Dansereau, & F. J. Yammarino (Eds.), Leadership: the multi-level approaches: Part B.
Contemporary and alternative (pp. 77–83). Stamford, CT: JAI Press.
Dansereau, F., Alutto, J. A., Nachman, S. A., Al-kelabi, S. A., Yammarino, F. J., Newman, J., Naughton, T. J., Lee,
S., Markham, S. E., Dumas, M., Kim, K., & Keller, T. (1995). Individualized leadership: a new multi-level
approach. Leadership Quarterly, 6(3), 413–450.
Dansereau, F., & Yammarino, F. J. (1998a). Introduction and overview. In F. Dansereau, & F. J. Yammarino
(Eds.), Leadership: the multi-level approaches: Part A. Classical and new wave (pp. xxv–xliii). Stamford,
CT: JAI Press.
Dansereau, F., & Yammarino, F. J. (1998b). A multi-level leadership mosaic: one way to put the pieces together. In
F. Dansereau, & F. J. Yammarino (Eds.), Leadership: the multi-level approaches: Part B. Contemporary and
alternative (pp. 327–349). Stamford, CT: JAI Press.
Denis, J.-L., Lamothe, L., & Langley, A. (2001). The dynamics of collective leadership and strategic change in
pluralistic organizations. Academy of Management Journal, 44(4), 809–837.
Denis, J.-L., Langley, A., & Cazale, L. (1996). Leadership and strategic change under ambiguity. Organization
Studies, 17(4), 673–699.
Doyle, M., & Myers, V. (1999). Co-principalship: a different approach to school leadership. Learning Matters, 4(2),
Ehrlich, S. B. (1998). Leader-centric and follower-centric research at multiple levels of analysis: toward a balanced
perspective. In F. Dansereau, & F. J. Yammarino (Eds.), Leadership: the multi-level approaches: Part B.
Contemporary and alternative (pp. 303–309). Stamford, CT: JAI Press.
Finholt, T., & Sproull, L. S. (1991). Electronic groups at work. Organization Science, 1(1), 41–64.
Follett, M. P. (1973). In E. M. Fox, & L. Urwick (Eds.), Dynamic administration: the collected papers of Mary
Parker Follett. London: Pitman and Sons.
Fondas, N., & Stewart, R. (1994). Enactment in managerial jobs: a role analysis. Journal of Management Studies,
31(1), 83–103.
Freeman, J. H. (1978). The unit of analysis in organizational research. In M. W. Meyer & Associates (Eds.),
Environments and organizations (pp. 335–351). San Francisco: Jossey-Bass.
Gabarro, J. J. (1978). The development of trust, influence, and expectations. In A. G. Athos, J. J. Gabarro (Eds.),
Interpersonal behavior: communication and understanding in relationships (pp. 290–303). Englewood Cliffs,
NJ: Prentice-Hall.
Gabarro, J. J. (1987). The development of working relationships. In J. Galeghar, R. E. Kraut, & C. Egido (Eds.),
Intellectual teamwork: social and technological foundations of co-operative work (pp. 79–110). Hillsdale, NJ:
Lawrence Erlbaum.
Gardner, J. W. (1984). The anti-leadership vaccine. In W. E. Rosenbach, & R. L. Taylor (Eds.), Contemporary
issues in leadership (pp. 318–325). Boulder, CO: Westview Press.
Gemmill, G., & Oakley, J. (1992). Leadership: an alienating social myth. Human Relations, 45(2), 113–129.
George, A. L., & George, J. L. (1964). Woodrow Wilson and Colonel House: a personality study. New York:
Dover Publications.
Gibb, C. A. (1954). Leadership. In G. Lindzey (Ed.), Handbook of social psychology, vol. 2 (pp. 877–917).
Reading, MA: Addison-Wesley.
Gibb, C. A. (1958). An interactional view of the emergence of leadership. Australian Journal of Psychology, 10
(1), 101–110.
Gibb, C. A. (1969). Leadership. In G. Lindzey, & E. Aronson (Eds.), The handbook of social psychology, vol. 4
(2nd ed.) ( pp. 205–283). Reading, MA: Addison-Wesley.

Golding, J. (1986). Some problems in the concept of secretary. International Studies of Management and Organization,
18(1), 94–111.
Grace, G. (1995). School leadership: beyond education management—an essay in policy scholarship. London:
Greenleaf, R. (1977). Servant leadership: a journey into the nature of legitimate power and greatness. New York:
Paulist Press.
Gronn, P. (1995). Greatness re-visited: the current obsession with transformational leadership. Leading and
Managing, 1(1), 14–27.
Gronn, P. (1999). Substituting for leadership: the neglected role of the leadership couple. Leadership Quarterly, 10
(1), 41–62.
Gronn, P. (2000). Distributed properties: a new architecture for leadership. Educational Management and Administration,
28(3), 317–338.
Gronn, P. (2002). Designer-leadership: the emerging global adoption of preparation standards. Journal of School
Leadership, 12(3) (in press).
Hall, V., & Wallace, M. (1996). Let the team take the strain: lessons from research into senior management teams
in secondary schools. Schools Organisation, 16(3), 297–308.
Hayek, F. A. (1945). The use of knowledge in society. American Economic Review, 35(4), 519–530.
Heenan, D. A., & Bennis, W. (1999). Co-leaders: the power of great partnerships. New York: Wiley.
Heller, M. F., & Firestone, W. A. (1995). Who’s in charge here?: sources of leadership for change in eight schools.
Elementary School Journal, 96(1), 65–86.
Hodgson, R. C., Levinson, D. J., & Zaleznik, A. (1965). The executive role constellation: an analysis of personality
and role relations in management. Boston: Graduate School of Business Administration, Harvard University.
Hosking, D.-M. (1988). Organizing, leadership and skilful process. Journal of Management Studies, 25(2),
Hosking, D.-M., & Morley, I. E. (1991). A social psychology of organizing: people, contexts and processes. New
York: Harvester/Wheatsheaf.
House, R., Rousseau, D. M., & Thomas-Hunt, M. (1995). The meso paradigm: a framework for the integration of
micro and macro organizational behavior. In L. L. Cummings, & B. M. Staw (Eds.), Research in organizational
behavior, vol. 17 (pp. 71–114). Greenwich, CT: JAI Press.
Hunt, J. G. (1991). Leadership: a new synthesis. Newbury Park, CA: Sage.
Hunt, J. G. (1999). Transformational/charismatic leadership’s transformation of the field: an historical essay.
Leadership Quarterly, 10(2), 129–144.
Hunt, J. G., & Ropo, A. (1995). Multi-level leadership: grounded theory and mainstream theory applied to the case
of General Motors. Leadership Quarterly, 6(3), 379–412.
Hunt, J. G., & Ropo, A. (1998). Grounded theory and mainstream theory in multi-level leadership: beyond the
foothills. In F. Dansereau, & F. J. Yammarino (Eds.), Leadership: the multi-level approaches: Part A. Classical
and new wave (pp. 353–362). Stamford, CT: JAI Press.
Hutchins, E. (1996). Cognition in the wild. Cambridge, MA: MIT Press.
Jaques, E., & Clement, S. D. (1995). Executive leadership: a practical guide to managing complexity. Oxford:
Basil Blackwell.
Jermier, J., & Kerr, S. (1997). ‘‘Substitutes for leadership: their meaning and measurement’’—contextual recollections
and current observations. Leadership Quarterly, 8(1), 95–101.
Keeley, M. (1998). The trouble with transformational leadership: toward a federalist ethic for organizations. In
J. B. Ciulla (Ed.), Ethics: the heart of leadership ( pp. 111–144). Westport, CT: Quorum Books.
Kerr, S., & Jermier, J. (1978). Substitutes for leadership: their meaning and measurement. Organization and
Human Performance, 22, 374–403.
Klein, K. J., Dansereau, F., & Hall, R. J. (1994). Levels issues in theory development, data collection and analysis.
Academy of Management Review, 19(2), 195–229.
Klein, K. J., & House, R. J. (1995). On fire: charismatic leadership and levels of analysis. Leadership
Quarterly, 6(2), 183–198.

Krantz, J. (1989). The managerial couple: superior–subordinate relationships as a unit of analysis. Human
Resource Management, 28, 161–175.
Malone, T. W., & Crowston, K. (1994). The interdisciplinary study of coordination. ACM Computing Surveys,
26 (1), 87–119.
Mandelblit, N., & Zachar, O. (1998). The notion of dynamic unit: conceptual developments in cognitive science.
Cognitive Science, 22(2), 229–268.
March, J. G. (1984). How we talk and how we act: administrative theory and administrative life. In T. J. Corbally,
& J. E. Corbally (Eds.), Leadership and organizational culture: new perspectives on administrative theory and
practice (pp. 18–35). Chicago: University of Illinois Press.
Meindl, J. R. (1993). Reinventing leadership: a radical, social psychological approach. In J. K. Murnighan (Ed.),
Social psychology in organizations: advances in theory and research (pp. 89–118). Englewood Cliffs, NJ:
Meindl, J. R. (1995). The romance of leadership as a follower-centric theory: a social constructionist approach.
Leadership Quarterly, 6(3), 329–341.
Miller, E. J. (1959). Technology, territory, and time: the internal differentiation of complex production systems.
Human Relations, 12(3), 243–272.
Miller, E. J. (1998). The leader with the vision: is time running out? In E. B. Klein, F. Gabelnick, & P. Herr (Eds.),
The psychodynamics of leadership (pp. 3–25). Madison, CT: Psychosocial Press.
Miner, J. B. (1975). The uncertain future of the leadership concept: an overview. In J. G. Hunt, L. L. Larson
(Eds.), Leadership frontiers (pp. 197–208). Kent, OH: Kent State University Press.
Murnighan, J. K., & Conlon, D. (1991). The dynamics of intense work groups: a study of British string quartets.
Administrative Science Quarterly, 36(2), 165–186.
Newton, P. M., & Levinson, D. J. (1973). The work group within the organization: a sociopsychological approach.
Psychiatry, 36, 115–142.
Parry, K. W. (1998). Grounded theory and social process: a new direction for leadership research. Leadership
Quarterly, 9(1), 85–105.
Pfeffer, J. (1977). The ambiguity of leadership. Academy of Management Review, 2, 104–114.
Powell, M. (1997). The Whitlam labor government: Barnard and Whitlam: a significant historical dyad. Australian
Journal of Politics and History, 43, 183–199.
Rost, J. C. (1993). Leadership for the twenty-first century. Westport, CT: Praeger.
Sayer, A., & Walker, R. (1992). The new social economy: reworking the division of labor. London: Blackwell.
Shamir, B. (1999). Leadership in boundaryless organizations: disposable or indispensable? European Journal of
Work and Organizational Psychology, 8(1), 49–71.
Shapin, S. (1989). The invisible technician. American Scientist, 77(6), 554–563.
Simon, H. (1991). Organizations and markets. Journal of Economic Perspectives, 5(2), 25–44.
Sivasubramaniam, N., Murry, W. D., Avolio, B., & Jung, D. I. (2002). A longitudinal model of the effects of
team leadership and group potency on group performance. Group and Organization Management, 27 (1),
Spillane, J. P., Halverson, R., & Diamond, J. B. (2000a). Investigating school leadership practice: a distributed
perspective. Educational Researcher, 30(3), 23–28.
Spillane, J. P., Halverson, R., & Diamond, J. B. (2000b). Towards a theory of leadership practice: a distributed
perspective. Working Paper, Institute for Policy Research, Northwestern University.
Stewart, R. (1991a). Chairmen and chief executives: an exploration of their relationship. Journal of Management
Studies, 28(5), 511–527.
Stewart, R. (1991b). Role sharing at the top: a neglected aspect of studies of managerial behaviour. In S. Carlson
(Ed.), Executive behaviour (reprinted with contributions from Henry Mintzberg and Rosemary Stewart)
( pp. 120–136). Uppsala: Studia Oeconomiae Negotiorum.
Strauss, A. L. (1988). The articulation of project work: an organizational process. Sociological Quarterly, 29(2),
Suchman, L. (1995). Making work visible. Communications of the ACM, 38(9), 56–64.

Tsoukas, H. (1996). The firm as a distributed knowledge system: A constructionist approach. (Winter special
issue) Strategic Management Journal, 17, 11–25.
Vanderslice, V. J. (1988). Separating leadership from followers: an assessment of the effect of leader and follower
roles in organizations. Human Relations, 41(9), 677–696.
Wallace, M., & Hall, V. (1994). Inside the SMT: teamwork in secondary school management. London: Paul
Wenger, E. (1999). Communities of practice: learning, meaning, and identity. Cambridge: Cambridge University
Wenger, E. (2000). Communities of practice and social learning systems. Organization, 7(2), 225–246.
Yukl, G. (1999). An evaluation of conceptual weaknesses in transformational and charismatic leadership theories.
Leadership Quarterly, 10(2), 285–305.
Zainu’ddin, A. T. (1981). The career of a colonial schoolmaster: Frank Wheen esq. B.A. (1857–1933). In S. Murray-
Smith (Ed.), Melbourne studies in education 1981 ( pp. 60–97). Melbourne: Melbourne University Press.