Académique Documents
Professionnel Documents
Culture Documents
11-2006
Mary Uhl-Bien
University of Nebraska-Lincoln, mbien2@unl.edu
Russ Marion
Clemson University, marion2@clemson.edu
Anson Seers
Virginia Commonwealth University, aseers@vcu.edu
Lichtenstein, Benyamin B.; Uhl-Bien, Mary; Marion, Russ; Seers, Anson; Orton, James Douglas; and Schreiber, Craig, "Complexity
leadership theory: An interactive perspective on leading in complex adaptive systems" (2006). Management Department Faculty
Publications. 8.
http://digitalcommons.unl.edu/managementfacpub/8
This Article is brought to you for free and open access by the Management Department at DigitalCommons@University of Nebraska - Lincoln. It has
been accepted for inclusion in Management Department Faculty Publications by an authorized administrator of DigitalCommons@University of
Nebraska - Lincoln.
Authors
Benyamin B. Lichtenstein, Mary Uhl-Bien, Russ Marion, Anson Seers, James Douglas Orton, and Craig
Schreiber
Copyright © 2006 Institute for the Study of Coherence & Emergence; published by Lawrence Erlbaum
Associates. Used by permission.
Complexity leadership theory: An interactive perspective on leading in complex adaptive systems
E:CO Issue Vol. 8 No. 4 2006 pp. 2–12
Practitioner
Traditional, hierarchical views of leadership are et al., 2005). There is also a growing realization that
less and less useful given the complexities of our effective leadership does not necessarily reside
modern world. Leadership theory must transi- within the leader’s symbolic, motivational, or char-
tion to new perspectives that account for the ismatic actions.
complex adaptive needs of organizations. In this If leadership is not “in” a leader or “done
paper, we propose that leadership (as opposed by” a leader, however, how are we to insightfully
to leaders) can be seen as a complex dynamic conceive exactly what constitutes leadership and
process that emerges in the interactive “spaces from where it originates? A novel approach for
between” people and ideas. That is, leadership answering these questions is grounded in com-
is a dynamic that transcends the capabilities of plexity science, namely the notion that leader-
individuals alone; it is the product of interaction, ship is an emergent event, an outcome of relational
tension, and exchange rules governing changes interactions among agents. In this view, lead-
in perceptions and understanding. We label this ership is more than a skill, an exchange, or a
a dynamic of adaptive leadership, and we show symbol – leadership emerges through dynamic
how this dynamic provides important insights interactions (Bradbury and Lichtenstein, 2000).
about the nature of leadership and its outcomes “Complexity leadership theory” investigates the
in organizational fields. We define a leadership role of leadership in expediting those processes
event as a perceived segment of action whose in organizations through which interdependent
meaning is created by the interactions of actors actions among many individuals combine into
involved in producing it, and we present a set of a collective venture (Drath, 2001; Meyer et al.,
innovative methods for capturing and analyzing 2005).
these contextually driven processes. We provide Founding the approach of this paper on
theoretical and practical implications of these complexity theory per se moves us to a whole-
ideas for organizational behavior and organiza- systems view and thus away from the more tra-
tion and management theory. ditional approaches that focus on variables and
component parts. Instead, we will focus on:
A
Introduction
s twenty-first-century management contin- • Expanding the locus of leadership from the
ues to emphasize decentralized organizing isolated, role-based actions of individuals to
structures and co-evolutionary ecologies the innovative, contextual interactions that
of firms, institutions, and markets, there is a grow- occur across an entire social system;
ing recognition that traditional top-down theories • Extending current theory and practice by
of leadership are at best overly simplistic (Osborn focusing on micro-strategic leadership actions
et al., 2002). That is, leading-edge theorists and the across all organizational levels and across orga-
leaders they inform are questioning the assump- nizational boundaries;
tion that the essence of leadership rests within the • Increasing the relevance and accuracy of
character or the characteristic behaviors of effec- leadership theory by exploring how leadership
tive supervisors (Seers, 2004). Worse, the notion outcomes are based on complex interactions,
that a leader exogenously “acts on” organizations rather than “independent” variables;
in order to achieve the leader’s objectives may be
misguided in the presence of the insight that orga- • Highlighting the relational foundations of
nizations are highly complex and nonlinear (Meyer change in emerging organizational fields,
L
leadership theory emergent learnings, capabilities, innovations, and
eadership study, indeed society in general, is adaptability. Importantly, such elaborations are
infatuated with leaders – people who occupy products of interactions among agents, rather than
some elevated status or position and to whom being “caused” by the specific acts of individuals
we often ascribe some form of “greatness” (Gronn, described as leaders.
2002). The Western mindset about leaders seems A complex systems perspective introduces
ruled by assumptions that leaders have some innate a new leadership “logic” to leadership theory and
capacity to plan futures, arrive at rational and cor- research by understanding leadership in terms of
rect decisions (Bluedorn, 2002), and control social an emergent event rather than a person. A com-
outcomes (Meindl et al., 1985). plexity view suggests a form of “distributed” lead-
A new mindset is beginning to emerge, ership (Brown and Gioia, 2002; Gronn, 2002) that
however, which recognizes that social processes does not lie in a person but rather in an interactive
are too complex and “messy” to be attributed to a dynamic, within which any particular person will
single individual or pre-planned streams of events participate as leader or a follower at different times
(Finkelstein, 2002; Marion and Uhl-Bien, 2001). As and for different purposes. It is not limited to a for-
Finkelstein (2002: 77) put it: mal managerial role, but rather emerges in the sys-
temic interactions between heterogeneous agents
“I understand that as researchers we need to simplify (Marion and Uhl-Bien, 2001, 2003). Therefore,
very complex processes to study them carefully, but complexity leadership includes a descriptive
what are we left with when we remove the messiness, analysis examining the conditions and dynamic
the back-and-forth, the reality?” processes of these interactions and the emergent
phenomena that they call forth:
Although the complexity leadership
approach redirects emphasis away from the indi- “There is a growing sense that effective organization
vidual as leader, it does not in any way diminish change has its own dynamic, a process that cannot
the importance of leadership as an organizational simply follow strategic shifts and that is longer and
phenomenon; rather, it recognizes that leadership subtler than can be managed by any single leader. It
transcends the individual by being fundamentally a is generated by the insights of many people trying to
system phenomenon (Marion and Uhl-Bien, 2001, improve the whole, and it accumulates, as it were, over
2003; Uhl-Bien et al., 2004; Hazy, 2006). Drawing long periods.” (Heckscher, 1994: 24)
from complexity science (Marion, 1999), complexity
leadership theory offers a new perspective for leader In other words, “leaders” in the formal sense can
ship research by considering leadership within the enable the conditions within which the process
framework of the idea of a complex adaptive system occurs, but they are not the direct source of change.
(CAS). In such systems, relationships are not pri- A key contribution of a complexity leader
marily defined hierarchically, as they are in bureau- ship theory is that it provides an integrative
cratic systems, but rather by interactions among theoretical framework for explaining interactive
heterogeneous agents and across agent networks.
A
adaptive leadership
ongoing cycle of events. Events are the observable ccording to most complexity researchers,
nodes in these cycles; multiple cycles may interact agent interactions are governed by rules
directly or they may be tangential. Allport’s theory and mechanisms for changing rules. One
provides a powerful precedent to complexity sci- fundamental form of rule change occurs when
ence in affirming that longitudinal analyses of interactions in leadership events produce a new
interaction events should replace cross-sectional identity (e.g., Gioia et al., 2000). According to the
G
exploring and analyzing leadership events
traditional models of leadership, and from most iven our interest in exploring the events
complexity models of agent rule following. Most that generate leadership, we have identi-
simulation researchers suggest that agents are gov- fied several methods that can be used to
erned by a selfish rule (Bonabeau and Meyer, 2001). measure and analyze specific leadership events
For example, Nowak et al., (1995) show cellular over time, as well as the interrelationships that
automata simulations in which selfish behaviors of enact them. Specifically we are interested in “epi-
agents may, under certain circumstances, generate sodes” of leadership, and on the interactions that
cooperative behaviors across an interactive system. are bracketed into those events. Since interactive
In contrast, complexity leadership theory develops dynamics are processes that take place over time,
a more nuanced view of how rules are used and how we need methods that attend to the longitudinal
they can change through interactions over time. and dynamic nature of interactive events and the
relationships that construct them.
Tension as a driver of adaptive leadership Focusing on events as the prime unit of
A second driver of innovation in adaptive leader- analysis means more than applying new methods
ship events occurs when the interactions between in order to analyze cross-sectional data on individ-
agents spark tension that leads to adaptive change. ual characteristics. Instead, measuring “the space
According to complexity leadership theory, when between” involves:
agents interact they may experience tension in the
form of pressures on and challenges to their per- • Identifying and bracketing the events, episodes,
sonal knowledge base (Carley and Hill, 2001). Such and interactions of interest;
challenges to agent schema can, under the right
• Capturing these events or interactions as data
enabling conditions, foster realignment of agents’
in a systematic way;
cognitive maps to resonate better with the new
information. That is, agents realign their schema in • Gathering individual/agent level data that
order to accommodate and thus mitigate disagree- describe interaction cues received over time;
ment (Kauffman, 1993; Marion and Uhl-Bien, • Modeling these data in ways that highlight
2001). their longitudinal and relational qualities;
These tension-related accommodations • Analyzing these data in terms of their relational
often generate completely new information; that
Lichtenstein, Uhl-Bien, Marion, Seers, Orton, & Schreiber
qualities and longitudinal dynamics. tion and surveys in the laboratory (Guastello et al.,
Identifying and bracketing events need not be com- 2005) or in the field (Schreiber and Carley, 2005).
plicated, depending on the nature of the organiz- To fully understand leadership events, however,
ing processes one is examining. A paradigmatic it is also necessary to know how these leadership
case is Barley’s (1986) examination of interactions cues or triggers are perceived by the individual
between radiologists and technicians during a agents who must make particular choices or take
period of dramatic technological change. In that specific actions (Hazy, 2006). Because events
case, the context of these interactions was defined unfold over time, the data set must be longitudinal,
(bracketed) by a radiological procedure, which in to capture how these qualities change over time, as
our case would be the event within its nexus of rela- well as cross-sectional, to understand agents’ per-
tionships. Less common, but no less interesting, is ceptions and qualities at specific moments in time.
the research on organizational meetings or special Accessing and gathering this type of data is challeng-
events, in which the crucial episodes are defined ing; fortunately, software tools and other techniques
a priori by the research interest, and the bracket- enable detailed data gathering at regular intervals in
ing of these processes is structurally produced organizations (www.leadershipscience.com; Amabile
and distinguished by the organizational members et al., 2005). Once gathered, the data can be used as
themselves. inputs to computational models as described below
Capturing events and interactions system- or for other quantitative, qualitative, or mixed
atically may result from the in-depth exploration method analytic techniques.
of organizing processes. For example, management Modeling data in ways that highlight their
researchers have recognized the important role that longitudinal and relational qualities enables explo-
temporal events play in making progress (Brown ration of the complex and interrelated dynam-
and Eisenhardt, 1997) and catalyzing changes ics inherent in leadership events. A recent survey
(Gersick, 1994) in dynamic contexts. In those two of computer modeling approaches in leadership
studies, the researchers were able to identify tempo- research (Hazy, in press) identified several differ-
ral and event-based transitions that structured the ent techniques that have been used for this, includ-
development of the project/venture being studied. ing system dynamics modeling (Davis, 2005; Hazy,
A more formal approach was taken by Lichtenstein 2004; Jacobsen and House, 2001), discrete event
et al. (2006) in their discovery of an “emergence simulation (Jiang and Burton, 2002), agent-based
event” within a nascent entrepreneurial venture. modeling (Black and Oliver, 2004; Black et al., 2006;
Using grounded theory (Strauss and Corbin, 1993), Carley and Ren, 2001), network modeling such as
they coded bi-weekly interview data into four cat- the NK Model (Solow and Leenawong, 2003), and
egories, then transformed these codes into a quan- dynamical network analysis (Schreiber and Carley,
titative format (Van de Ven and Poole, 1990). Next, 2004a, 2005a). These techniques can be used to
they analyzed each of the time series’ using qual- explore the nonlinear relationships resident in the
ity control methods, which highlighted a dramatic data and to better understand the analytical impli-
change in one variable (Dooley and Van de Ven, cations of theory. Based on these synthetic results,
1999). Post-hoc heuristic tests confirmed the pres- computational analysis can pose research questions
ence of two distinct “epochs” (events) within these and identify hypotheses for empirical studies that
data.The interview corresponding to the specific might have otherwise gone unnoticed.
change point was more deeply examined; it became In addition, computational modeling can
the nexus of a series of changes that were explained be used to answer questions that are normative
as interdependent aspects of an emergence event. or plausible. Plausible questions ask “what might
Gathering individual/agent level data about be” and explore or go beyond what has trans-
the members’ interdependencies and the interaction pired (Burton, 2003). Computational models are
cues that they receive over time (traditionally, the particularly useful in respect to research on orga-
cues that “lead them”) is necessary for exploring nizational complexity, as real-world complex adap-
how leadership events diffuse through “the space tive systems do not lend themselves to controlled
between” the participants to influence a popula- experimentation. Through simulation, we can
tion. These data can be gathered through observa- explore the complex effects of explanatory vari-
B
science
that have been measured in traditional research, y looking for leadership as emerging endog-
such as individual traits or behaviors, new met- enously within interactions while being
rics must be identified that more fully capture the embedded within organizations, so-called
system dynamics. This is a work in progress and leaders are not assumed to be directing collective
offers opportunities for methodological research. action. There is no linear cause-and-effect relation-
In addition, new analytical techniques must be ship to discover. Instead, “leadership” becomes a
developed. term that is descriptive of certain social forces at
One such technique for rigorously under- play among actors, which may include a formal
standing these relational dynamics is dynamic net- leader. This view is consistent with Giddens’s (1984)
work analysis. The new dynamic network analysis duality of structure in that social structures produce
methodology combines techniques of social net- and in some sense lead collective action, while at
work analysis with multi-agent simulations (Carley, the same time being reproduced by those actions
2003). Dynamic network analysis represents socio- over time. By considering “leadership action” from
technical systems in terms of the complex relational an endogenous, time-dependent perspective, we
qualities that characterize the interdependencies of are better able to integrate the time dimension of
the system (Krackhardt and Carley, 1998). Also, social systems into organization theory, revealing
dynamic network analysis models dynamic changes a unique method for addressing Radcliffe-Brown’s
resulting from natural evolutionary processes such (1952) challenge to sociological theory: “How do
as learning (Carley and Hill, 2001) and strategic new types of social structure come into existence?”
intervention processes such as altering the set of (cited in Burrell and Morgan, 1979: 56).
individuals within a group (Schreiber and Carley, Complexity leadership theory begins to
2004b). Through the use of dynamic network address this issue by arguing that certain inter
analysis, the contextual nature of the network and actions in a social network will have a nonlinear
emergent structure and behavior, including leader- influence on future interactions within the net-
ship events, can be analyzed as well as the effects work. As such, leadership actions may be seen as
of emergence on outcomes such as performance, “field”-level effects that potentially catalyze the
innovation, and adaptability. emergence of new firms (Uhl-Bien et al., in press),
In addition, nonsimulation methods are proto-institutions (Maguire et al., 2004) or organi-
being perfected for developing rigorous longitu- zational fields (Chiles et al., 2004). “Field” is being
dinal analysis of critical events in emergence over used here in a cognate sense to that found in phys-
time. An exemplar is the study of events leading to ics; that is, a matrix underlying a social grouping
the emergence of the Branson, Missouri commu- whose influence reaches to all the actors within
nity (Chiles et al., 2004). Their data analysis meth- that “field.”
ods (see pp. 504–506) include grounded theory, Another application for complexity lead-
pattern matching, visual mapping, narrative tech- ership theory focuses on how leadership events
niques, temporal bracketing, and quantification may occur within and/or give rise to emergent
using an event count model analyzed through a nodes in a social network. Such an approach pres-
Poisson regression. This approach resulted in the ents a unique addition to research on networks,
identification of four specific eras of emergence by exploring how and when certain nodes may be
punctuated by a carefully defined series of events; highly leveraged within a collective social system.
moreover, the researchers were able to generalize Moreover, by exploring influential nodes in terms
from these events four drivers of organizational of leadership outcomes – rather than in terms of
emergence – fluctuation dynamics, positive feed- the individualized roles these nodes might repre-
Mary Uhl-Bien, Ph.D. (University of Cincinnati, James Douglas Orton (Ph.D., University of
1991) is the Howard Hawks Chair in Business Michigan) is an expert on strategy-losing and
Ethics and Leadership and the Associate Director strategy-remaking processes in the U.S. national
of the Gallup Leadership Institute at the University security community and other loosely coupled
of Nebraska-Lincoln. She has published articles networks. Dr. Orton is a senior faculty member
on leadership (e.g., relational leadership theory, at the George Washington University’s Executive
leader–member exchange, social exchange, and Leadership Doctoral Program. He teaches doc-
complexity leadership) in leading national and toral seminars on strategy, leadership, organization
international journals, including Academy of theory, loosely coupled systems, sensemaking pro-
Management Journal, Journal of Applied Psychology, cesses, high-reliability organizations, and the craft
Journal of Management, Human Relations, and of organizational scholarship.
The Leadership Quarterly. She is Senior Editor
of the Leadership Horizons Series published by Craig Schreiber recently earned his Ph.D. in
Information Age Publishing, and serves on the edi- Computation, Organizations and Society from
torial boards of The Leadership Quarterly and the Carnegie Mellon University. He was a member of
Academy of Management Journal. She has consulted the Center for Computational Analysis of Social and
with organizations including State Farm Insurance, Organizational Systems (CASOS) at the Institute
Walt Disney World, the U.S. Fish & Wildlife Service, for Software Research International in the School of
British Petroleum, and the General Accounting Computer Science. He is currently a research asso-
Office. ciate for the National Research Council. Previously
he has worked on research projects sponsored by
Russ Marion (Clemson University) is author of the National Science Foundation, NASA, the Office
The Edge of Organization (1999), Leadership in of Naval Research and Army Research Labs. His
Education (2001), and Leadership in Complex interests include strategic management, organiza-
Organizations (The Leadership Quarterly). tion and management theory, leadership, influence
Marion is currently co-editor of a special edition and power, organizational structure, organizational
on Complexity Leadership for The Leadership performance, organizational risk, organizational
Quarterly, and is co-editor of a volume of Leadership learning, knowledge management, information
Horizons: The Series. He co-organized workshops technology, computational organization science,
on complexity leadership at the Center for Creative social network analysis, dynamic network analysis,
Leadership and at George Washington University. and model validation.
Marion has presented on complexity leadership at
the India Institute of Technology, the Institute for
Management Development in Switzerland, and in
workshops on destructing complex movements at
the US Department of Defense.