Content uploaded by Ann Yan Zhang
Author content
All content in this area was uploaded by Ann Yan Zhang on Jun 28, 2016
Content may be subject to copyright.
PARADOXICAL LEADER BEHAVIORS IN PEOPLE
MANAGEMENT: ANTECEDENTS AND CONSEQUENCES
YAN ZHANG
Peking University
DAVID A. WALDMAN
Arizona State University
YU-LAN HAN
Shanghai University of Finance and Economics
XIAO-BEI LI
East China University of Science and Technology
As organizational environments become increasingly dynamic, complex, and compet-
itive, leaders are likely to face intensified contradictory, or seemingly paradoxical,
demands. We develop the construct of “paradoxical leader behavior” in people man-
agement, which refers to seemingly competing, yet interrelated, behaviors to meet
structural and follower demands simultaneously and over time. In Study 1, we de-
velop a measure of paradoxical leader behavior in people management using five
samples from China. Confirmatory factor analyses support a multidimensional mea-
sure of paradoxical leader behavior with five dimensions: (1) combining self-centered-
ness with other-centeredness; (2) maintaining both distance and closeness; (3) treating
subordinates uniformly, while allowing individualization; (4) enforcing work require-
ments, while allowing flexibility; and (5) maintaining decision control, while allowing
autonomy. In Study 2, we examine the antecedents and consequences of paradoxical
leader behavior in people management with a field sample of 76 supervisors and 516
subordinates from 6 firms. We find that the extent to which supervisors engage in
holistic thinking and have integrative complexity is positively related to their para-
doxical behavior in managing people, which, in turn, is associated with increased
proficiency, adaptivity, and proactivity among subordinates.
The style of leaders should be similarly gentle and
soft, but also persistent and powerful.
(Lee, Han, Byron, & Fan, 2008: 93)
Leaders inevitably confront contradictions, and
resulting tensions, when managing people (Evans,
2000; Handy, 1994; Smith & Lewis, 2011). For ex-
ample, they are expected to treat followers uni-
formly and consistently, while considering indi-
vidual needs and sometimes making exceptions;
they must also maintain control by enforcing organ-
izational rules and procedures, while allowing em-
ployee flexibility. The ability to respond to such
paradoxical challenges is essential to effective peo-
ple management.
Western literature has previously considered the
management of paradoxes in organizational set-
tings (Lewis, 2000; Smith & Lewis, 2011; Smith &
Tushman, 2005), including strategies for resolving
tensions between incremental learning versus rad-
ical change and, to a lesser extent, micro issues
such as maintaining control versus empowering
followers. However, the common quest is to under-
stand when leaders might emphasize one pole of
This research was supported financially by grants from
the National Science Foundation of China (#71372022,
#71032001, and #91224008) and the Guanghua Leader-
ship Institute (GLI10-19RR). We thank guest editor Xiao-
Ping Chen, the three anonymous reviewers, and all of the
editors of the special issue for their constructive com-
ments. We also thank Y. R. Chen, M. B. Brewer, Y. T. Lee,
Y. Y. Hong, C. T. Tadmor, S. N. Liou, P. P. Li, K. Wood-
cock, Y.-H. Xie, and Z. Zhang for their help in the re-
search process. Correspondence regarding this article
should be addressed to Yan Zhang (annyan.zhang@pku.
edu.cn) or to Yu-Lan Han (han.yulan@mail.shufe.edu.
cn).
538
娀Academy of Management Journal
2015, Vol. 58, No. 2, 538–566.
http://dx.doi.org/10.5465/amj.2012.0995
Copyright of the Academy of Management, all rights reserved. Contents may not be copied, emailed, posted to a listserv, or otherwise transmitted without the copyright holder’s express
written permission. Users may print, download, or email articles for individual use only.
behavior rather than another—that is, when they
might fail to harmonize or integrate inherent para-
doxes (Clegg, Cuhna, & Cuhna, 2002; Peng & Nis-
bett, 1999). Typically, situational or contingency
theories of leadership emphasize, with mixed sup-
port (e.g., Smith & Lewis, 2011), an “either–or”
strategy (“A” or “B”), such as being directive or
participative, depending on the situation. Such the-
orists view A and B as separate and potentially
contradictory; they emphasize statically matching
leader behaviors with specific conditions to
achieve effectiveness. The contingent approach in-
dicates that choosing between competing tensions
is a “necessary evil.” We suggest a contrasting,
alternative reality: that choosing between appar-
ently competing demands may enhance short-term
performance, but, if they are to sustain long-term
effectiveness, leaders must accept and harmonize
paradoxes simultaneously (Smith & Lewis, 2011).
Thus, paradoxes represent opportunities for, rather
than threats to, leader effectiveness.
Eastern literature and philosophy provide in-
sights into the nature of paradoxes and effective
ways of dealing with workplace tensions or contra-
dictions. Historically, Chinese culture and the East-
ern mindset suggest that all universal phenomena
have competing tendencies (Chen, 2008), and that
societies and organizations naturally embrace op-
posites (Chen, 2002). Rather than being “either–or,”
all things, including problems and challenges, are
interrelated as “both–and.” Opposites coexist har-
moniously and interdependently to form a con-
tinuously changing and transforming totality. In
short, the Eastern approach to paradoxes is to
embrace, integrate, and transcend apparent oppo-
sites, in contrast with predominant Western ana-
lytical thinking, in which information is pro-
cessed by breaking the whole into parts (Peng &
Nisbett, 1999).
Both Western and Eastern literature traditions
have considered the relevance of paradoxes to or-
ganizational settings. Nevertheless, to the best of
our knowledge, little research has explored specific
leader behaviors in dealing with paradoxical situ-
ations, except for several exploratory studies that
treated competing leader behaviors or roles inde-
pendently (Denison, Hooijberg, & Quinn, 1995; Kai-
ser, Lindberg, & Craig, 2007; Lawrence, Lenk, &
Quinn, 2009). However, differentiating opposites
and dealing with them separately fails to capture
the essence of paradoxes—that is, that two oppo-
sites coexist and should be dealt with as a pair. To
address organizational paradoxes effectively, lead-
ers should adopt a “both–and” strategy over time.
We base our theory on Eastern thought to under-
stand the essence of paradoxes and to explore lead-
ers’ behavioral efforts to work through organization-
al paradoxes that they are likely to face—that is, to
address contradictory forces simultaneously and
over time. Correspondingly, we propose the con-
cept of “paradoxical leader behavior” (PLB) to de-
scribe leader behaviors that are seemingly compet-
ing, yet interrelated, to meet competing workplace
demands simultaneously and over time. In essence,
PLBs are dynamic and synergistic approaches to
contradictions in organizational management. For
more effective organizational functioning and de-
velopment, leaders may use paradoxical behaviors
to handle organizational paradoxes.
Our overall goal is to determine how PLB may
contribute to an understanding of leadership effec-
tiveness in managing people, beyond what we have
learned from more traditional models of effective
leadership. We first conceptualize PLB in dealing
with common paradoxes in organizations pertain-
ing to people management. Next, we develop a
measure and test a nomological network of PLB in
people management, including cognitive and con-
textual antecedents, as well as consequences.
Our findings contribute to theory and practice in
several ways. First, we use a paradoxical lens to
examine leader behavior. Although prior literature
has considered organizational paradoxes (e.g.,
Lewis, 2000), few researchers have tried to delin-
eate the nature of leader behavior in dealing with
paradoxes, especially in the management of people.
In this paper, we present a conceptualization of
PLB that may generate new directions in leadership
research. Second, we develop an empirical ap-
proach to study PLB. We use a survey design that
involves the simultaneous assessment of seeming
contradictions. Third, we bridge Eastern and West-
ern thinking regarding paradox-based phenomena
in organizations. Following a geocentric approach
to theory building (Li, Leung, Chen, & Luo, 2012),
we integrate Eastern yin–yang philosophy and
Western-based literature to broaden our under-
standing of paradoxes and effective approaches to
them. Aligned with our emphasis on Eastern think-
ing and traditions, we conducted our research in
China. While Eastern culture may provide an ideal,
initial context for examining approaches to para-
doxes, our methodology and findings can provide
insights for researchers and organizational leaders
in numerous geographical and cultural contexts.
2015 539Zhang, Waldman, Han, and Li
THEORETICAL DEVELOPMENT OF
PARADOXICAL LEADER BEHAVIOR
IN PEOPLE MANAGEMENT
We begin by describing the nature of paradoxes
in organizations, as portrayed largely in literature
emanating from Western scholars. We then exam-
ine Eastern philosophy and recent work that can
help to explain paradoxical cognition and associ-
ated leader behaviors. Next, we specifically con-
ceptualize PLBs involved in managing people. Fi-
nally, we use our theory to suggest a nomological
network of antecedents and outcomes of these
behaviors.
Paradoxes in Organizational Settings
Organizational functioning inherently involves
tensions, competing demands, and even contradic-
tions (Poole & Van de Ven, 1989). Since the 1980s,
Western scholars have increasingly emphasized ac-
tions for handling contradictions (e.g., Bobko,
1985; Denison et al., 1995; Sundaramurthy &
Lewis, 2003), systematically conceptualizing ten-
sions as “organizational paradoxes” (Cameron &
Quinn, 1999; Smith & Lewis, 2011; Yukl & Leps-
inger, 2004).
A “paradox” entails “contradictory yet interre-
lated elements that exist simultaneously and per-
sist over time. Such elements seem logical when
considered in isolation but irrational, inconsistent,
and even absurd when juxtaposed” (Smith & Lewis,
2011: 386). A distinguishing characteristic is the
simultaneous presence of two seemingly mutually
exclusive assumptions or conditions. Taken indi-
vidually, each is incontestably true or logical; com-
bined, they seem inconsistent and incompatible
(Quinn & Cameron, 1988). Yet dualities coexist in
organizational settings, and may be synergistic and
interrelated within a larger system (Cameron &
Quinn, 1999).
Organizational paradoxes arise at macro and mi-
cro levels. Research has predominantly focused on
macro-level paradoxes, such as tensions between
organizational exploitation and exploration (Vera &
Crossan, 2004), and conflicting demands of internal
and external stakeholders (Margolis & Walsh,
2003). Paradoxical challenges at micro levels, par-
ticularly in people management, have received less
attention. For instance, organizations expect super-
visors to depersonalize subordinates so that they
behave uniformly, while subordinates expect su-
pervisors to treat them as unique individuals
(Brewer & Gardner, 1996). Organizations expect su-
pervisors to control work processes and produc-
tion, while subordinates expect them to grant dis-
cretion in pursuing tasks. Those challenges can
influence effectiveness in managing people. Ac-
cordingly, we address supervisory-level leader be-
havior regarding such paradoxical challenges.
Western scholars have considered opposing com-
ponents of paradoxes to develop leadership ap-
proaches to manage them. These include behav-
ioral complexity (Denison et al., 1995; Hooijberg,
1996; Lawrence et al., 2009) and flexible leadership
(Kaiser et al., 2007). Other Western scholars have
sought strategies to connect two poles of a paradox.
Temporal separation is an example, whereby one
pole of a paradoxical tension is emphasized at one
time and the opposite pole at another. Spatial sep-
aration is another example: one organizational unit
emphasizes one pole, while another unit empha-
sizes the other. Those and other strategies suggest a
nonbipolar view of paradoxes that considers oppo-
site viewpoints together and integrates them over
time (Bobko, 1985; Van de Ven & Poole, 1988).
Under such frameworks, “the role of leadership [in
dealing with paradoxes] is to support opposing
forces and harness the constant tension between
them, enabling the system to not only survive but
[also] continuously improve” (Smith & Lewis,
2011: 386).
Eastern Philosophy and Approaches to
Paradoxes
Those nonbipolar responses to paradoxes align
with the basic roots of yin–yang philosophy, which
has the potential to broaden our understanding of
paradoxes and the leader behavior related to them.
Yin–yang philosophy views the world as being ho-
listic, dynamic, and dialectical, through which “all
universal phenomena are shaped by the integration
of two opposite cosmic energies, namely Yin and
Yang” (Fang, 2012: 31; Li, 1998). Yin represents
“female” energy; yang represents “male” energy:
both forces operate universally and integratively,
dynamically shaping all universal phenomena and
generating constant change (Fang, 2010).
Yin–yang philosophy suggests that, although
paradox dualities appear to oppose, negate, and
separate from each other, they are actually interde-
pendent and complementary, mutually composing
a harmonious whole (Chen, 2002; Fang, 2005). Par-
adoxical contradictions often mask the recognition
that conflicting poles are simultaneous and holis-
540 AprilAcademy of Management Journal
tic. More importantly, space and time weaken the
seemingly contradictory nature of dualities, while
strengthening their complementary nature. Duali-
ties do not absolutely separate from each other;
instead, they connect interdependently in a larger
system (Chen, 2002; Fang, 2010, 2012).
Yin–yang perspectives regarding paradoxes
differ from Western analytical approaches that es-
pouse contingency perspectives. Analytical ap-
proaches embrace “either–or” rather than “both–
and” thinking (Chen, 2002; cf. Li, 2012), isolate the
two sides of paradoxes, presume that one side is
right while the other is wrong (Peng & Nisbett,
1999), and deny the holistic nature of contradic-
tions. Yin–yang perspectives also differ from West-
ern dialectical approaches that try to reconcile op-
posing perspectives (Fang, 2012; Peng & Nisbett,
1999) while de-emphasizing connections and inter-
dependence between dualities and the possibility
of coexistence in a larger system (Fang, 2012).
For example, the “loose–tight” principle (Sagie,
1997) reflects a paradox of control versus auton-
omy, suggesting that managers should have tight
rules or procedures but also be willing to bend
them. A Western “either–or” orientation suggests
that tight control and autonomy are incompatible,
and that managers might adopt either approach.
However, a yin–yang, “both–and” interpretation is
that managers might actually maintain long-term
control by continuously granting employees the
discretion to bend the rules. Thus, yin–yang phi-
losophy suggests that dualities do not absolutely
separate; instead, paradoxes are integrated parts of
a larger whole.
Yin–yang philosophy fundamentally affects indi-
vidual cognition and behavior toward paradox
(Chen, Xie, & Chang, 2011). Paradoxical mindsets
cognitively perceive that both paradoxical poles
contain truth (Peng & Nisbett, 1999). They may
seem opposed but are actually inseparable and con-
nected in time and space (Peng & Nisbett, 1999),
and should be considered and accepted simultane-
ously over time. Indeed, the long-term perspective
inherent in yin–yang philosophy “allows events to
be put into the context of a greater whole, and
emphasizes connections instead of isolated mo-
ments” (Chen, 2002: 186).
Addressing Paradoxes in People Management
“All organization is founded on paradox: on the
one hand it contains free, creative, independent
human subjects; on the other hand the relation
between these subjects aspires to be one of organi-
zation, order and control” (Clegg et al., 2002: 483).
Thus, leaders will confront ongoing, competing de-
mands to meet organizations’ structural needs and
followers’ individual needs (Clegg et al., 2002;
Lewis, 2000; Poole & Van de Ven, 1989). Organiza-
tional formalities convey structural demands re-
quiring supervisors to maintain effective organiza-
tional stability and functioning. The hierarchical
structure gives them central, authoritarian, and
controlling roles at higher positions, and assigns
employees homogeneous subordinate roles at
lower positions. Nevertheless, followers have dif-
ferent individual needs and purposes, and are
likely to demand individual free will in directing
their own actions (Poole & Van de Ven, 1989). They
expect to influence others, to make decisions, and
to work in their own way. As socially interacting
individuals, they may also feel equal to supervisors
and desire to form close social relationships with
them. Therefore, supervisors are expected to meet
both structural and follower demands.
Based on yin–yang philosophy, or paradoxical
cognition, PLBs are structurally and individually
ambidextrous. The two sides of behaviors coexist,
like yin and yang, depending on and complement-
ing one another to jointly support leader effective-
ness in people management. To characterize the
two sides of behavior, we use “both–and” terminol-
ogy to describe five behavioral dimensions: (1)
combining self-centeredness with other-centered-
ness; (2) maintaining both distance and closeness;
(3) treating subordinates uniformly, while allowing
individualization; (4) enforcing work requirements,
while allowing flexibility; and (5) maintaining de-
cision control, while allowing autonomy.
Combining self-centeredness with other-cen-
teredness. A structural orientation implies that
leaders are the center of influence, while individ-
ual consideration implies that leaders have con-
cerns or deference for others. Concern for others
has been touted as the more ethical and moral
stance (Jones, Felps, & Bigley, 2007), while self-
centeredness, arrogance, and a lack of concern for
others has been disdained as a mark of narcissism
(Galvin, Waldman, & Balthazard, 2010; Judge, Pic-
colo, & Kosalka, 2009). Some researchers have
called for organizations to broaden their hierar-
chical, formal leadership focus to include other-
centered leadership that embraces shared or
collective leadership principles (e.g., Denis, Lan-
gley, & Sergi, 2012).
2015 541Zhang, Waldman, Han, and Li
Nevertheless, leaders have the potential for
harmonizing self-centeredness versus other-cen-
teredness. For example, leaders who are highly
self-confident and desire to be the center of atten-
tion—termed “productive narcissists” (Maccoby,
2004)—might simultaneously show humility and
recognition of others’ value (Rosenthal & Pittinsky,
2006). Although it is important for group members
to undertake leadership roles (e.g., Day, Gronn, &
Salas, 2004; Pearce & Conger, 2003), hierarchically
based leadership still plays an important role:
“Shared leadership is not mutually exclusive to
other leadership forms and behaviors, but can be
engaged in simultaneously with other approaches
such as vertical leadership” (Hoch & Dulebohn,
2013: 117). In short, paradoxically oriented leaders
may be able to maintain their central influence,
while simultaneously sharing recognition and lead-
ership with followers.
Maintaining both distance and closeness. Lead-
ers assign vertical structural relationships to define
their distance from followers in status, rank, au-
thority, and power (Antonakis & Atwater, 2002).
However, adherence to follower demands inher-
ently involves minimizing status distinctions,
combined with a degree of close interpersonal re-
lationships (Antonakis & Atwater, 2002)—that is,
supervisors do not emphasize status differences,
remain aloof, or avoid forming interpersonal con-
nections. Indeed, how leader behaviors affect fol-
lowers and how followers evaluate the behaviors
depend on the balance between leader–follower
closeness and distance (Howell, Neufeld, & Avolio,
2005). Followers may perceive distant leaders as
being more prototypical, authorized, and charis-
matic (Shamir, 1995; Yagil, 1998). Yet leader–mem-
ber exchange (LMX) suggests that close relation-
ships enhance followers’ positive work attitudes
and behaviors (e.g., Gerstner & Day, 1997). How-
ever, closeness can be detrimental, causing leaders
to avoid conflicts and difficult decisions regarding
followers (Yukl, 2010). Leader–member closeness
may also destroy followers’ charismatic attribu-
tions toward leaders (Galvin et al., 2010; Shamir,
1995). Thus, leaders are challenged by the need to
maintain distance while simultaneously forming
interpersonal bonds.
To handle the hierarchical distance and interper-
sonal closeness paradox, supervisors “do not take
employees merely as subordinates” (supervisor in-
terviewed by authors); rather, they maintain hier-
archical distinctions in dealing with work issues,
while simultaneously forming close interpersonal
bonds. For example, a supervisor whom we inter-
viewed demonstrated his hierarchical position by
criticizing the work of a young accountant, bringing
her to tears seven times in six months. However,
when she transferred to another unit, she embraced
him and cried because she felt that he had treated
her with “tough love” while ultimately looking out
for her interests.
Treating subordinates uniformly while allow-
ing individualization. To value uniformity as a key
principle for treating people based on their mem-
bership in a social group, leaders may assign sub-
ordinates to homogeneous positions with identical
privileges, rights, and status without displaying fa-
voritism (Lewis, 2000). However, such uniformity
may depersonalize them and deprive them of
unique individual identity (Brewer, 1991;
Kreiner, Hollensbe, & Sheep, 2006). Some estab-
lished leadership theories suggest treating subor-
dinates uniquely or personally, such as by means
of the individualized consideration inherent to
transformational leadership (Bass & Avolio,
1995) or LMX differentiation within groups
(Boies & Howell, 2006).
Leaders could harmonize uniformity and indi-
vidualization. Our interviews revealed that one
leader gave similarly priced gifts every company
anniversary to thank subordinates for their work for
the group and the company, but selected each gift
to fit the individual subordinate. Another leader
gave subordinates tasks of similar scope and diffi-
culty, but allocated different parts of the work
based on the individual’s skills or interest in
the task.
Two dimensions of control and empowerment.
The loose–tight principle (Sagie, 1997) mentioned
earlier is another paradox juxtaposing leader con-
trol with empowerment. Nevertheless, research has
considered it within alternative associations, such
as control and flexibility (e.g., Lewis, 2000; Smith &
Lewis, 2011), control and autonomy (e.g., Feldman,
1989; Lewis, 2000), discipline and empowerment
(e.g., Lewis, 2000), and authority and democracy
(Lewis, 2000), indicating two different aspects:
control and flexibility in terms of behavior, and
control and autonomy in decision making. Indeed,
Ouchi (1978) classified control into two categories:
“behavior control,” which uses discipline and re-
quirements to regulate subordinate behaviors; and
“output control,” which uses authority in decision
making to ensure subordinate work outcomes.
Thus, we propose two paradoxes: (1) enforcing
work requirements while allowing flexibility (rele-
542 AprilAcademy of Management Journal
vant to behavioral control), and (2) maintaining
decision control while allowing autonomy (rele-
vant to output control). In other words, leaders can
control subordinate behavior and decision making
in work processes while giving employees discre-
tion to act flexibly and autonomously.
The leadership literature has typically framed
those paradoxes as “either–or” situational leader-
ship phenomena. Some situations might call for
strict control of behaviors and decisions (e.g.,
Vroom & Jago, 2007). For example, transactional
leadership emphasizes behavioral control, while
autocratic leadership emphasizes decision control.
Other situations might call for empowering leader-
ship allowing autonomy and avoiding micro man-
agement (Srivastava, Bartol, & Locke, 2006). Con-
versely, leaders engaging in paradoxical behavior
attempt to integrate or harmonize inherent tensions
concerning control and empowerment over time.
Our interviews revealed an example of enforcing
work requirements while allowing flexibility in
which supervisors established overly challenging
goals to push subordinates to perform beyond ex-
pectations. When enforcing the goals, however,
they understood that subordinates encountered
real difficulties along with situational changes;
consequently, supervisors provided additional
support and often loosened goal requirements.
Southwest Airlines former CEO Herb Kelleher
(1997: 20) also revealed an approach involving de-
cision control, combined with autonomy: “I’ve
never had control and I never wanted it. If you
create an environment where the people truly par-
ticipate, you don’t need control.” Kelleher was still
concerned with authority and with creating desired
procedures and policies, but also saw the impor-
tance of granting autonomy.
In summary, PLB simultaneously and dynami-
cally adheres to structural and follower demands in
managing people over time. We now turn to the
antecedents and outcomes of PLB.
NOMOLOGICAL NETWORK OF PARADOXICAL
LEADER BEHAVIOR IN PEOPLE
MANAGEMENT
We propose a nomological network that follows
from our conceptualization of PLBs in managing
people. This network delineates personal cognitive
and contextual antecedents and follower outcomes
(see Figure 1).
Based on yin–yang philosophy, effective PLB is
likely to have a “both–and” cognitive basis (Chen,
2008; Chen et al., 2011). First, a likely antecedent of
PLB is holistic thinking, which constitutes the yin–
yang essence by assuming that everything is inte-
grated, including contradictory demands (Peng &
Nisbett, 1999). Second, integrative complexity is
likely to be an important cognitive process for dif-
ferentiating “either–or” issues and for integrating
understandings of, and approaches to, contradic-
tions (e.g., Denison et al., 1995; Tetlock, Peterson, &
Berry, 1993). In short, holistic thinking and integra-
tive complexity are key cognitive antecedents
of PLB.
We further posit that organizational structure
determines PLBs (Capelli & Sherer, 1991; Pillai &
Meindl, 1998). Under simple and stable contexts,
paradoxes are less salient or intensified than un-
der conditions of change or uncertainty (Smith &
Lewis, 2011). Thus, we expect PLBs to be more
likely under organic structures featuring greater
tension than under mechanistic structures.
FIGURE 1
Nomological Network of Paradoxical Leader Behavior in People Management
H2
H4c
H4b
H4a
H1
H3
Organic vs.
mechanistic structure
Integrative
complexity
Holistic
thinking
Paradoxical
leader behavior in
people management
Proficient behavior
Adaptive behavior
Proactive behavior
2015 543Zhang, Waldman, Han, and Li
We suggest that PLB is related to how followers
handle work roles that evolve dynamically in re-
sponse to changing conditions and demands (Grif-
fin, Neal, & Parker, 2007). Leaders who demon-
strate acceptance of contradictions and the ability
to explore coexisting possibilities show followers
how to be open, learning oriented, and flexible
toward external challenges, and further build a
bounded discretionary environment for them. Fol-
lowers who are exposed to PLB will work more
proficiently, adaptively, and proactively.
Cognitive Antecedents of Paradoxical Leader
Behavior in People Management
“Holistic thinking” attends to “relationships be-
tween a focal object and the field and explaining
and predicting events on the basis of such relation-
ships” (Nisbett, Peng, Choi, & Norenzayan, 2001:
293). The locus of attention is the field, or the
“whole picture,” rather than individual elements
(Choi, Koo, & Choi, 2007). Holistic thinkers view
elements in the universe as continuous, intercon-
nected, and interpenetrated (Koo & Choi, 2005;
Masuda & Nisbett, 2001; Nisbett et al., 2001; Peng &
Nisbett, 1999).
Such cognitive thinking affects leader behavior
(Lee, 2000). Holistic-thinking leaders view both as-
pects of paradoxes as being true (Choi & Nisbett,
2000). Through holistic thinking, leaders handle
paradoxes by accepting seeming contradictions in
paradoxes, connecting and integrating them into a
larger system, and finding possibilities for dynamic
coexistence. In dealing with specific paradoxes in
managing people, they will consider and integrate
both organizational and subordinate requirements.
For example, a holistic-thinking leader might avoid
aloofness or disrespect, but retain the psychologi-
cal or status distance needed to require followers to
undertake a heavy workload or responsibilities for
organizational success. Simultaneously, the leader
can be close to followers and deal with their indi-
vidual needs and difficulties in completing their
work. In short:
Hypothesis 1. The extent to which a leader
engages in holistic thinking will be positively
associated with the leader’s paradoxical be-
havior in managing people.
Another predictor of paradoxical behavior is “in-
tegrative complexity”—that is, the capacity and
willingness to acknowledge the legitimacy of com-
peting perspectives on the same issue (namely,
evaluative differentiation) and to forge conceptual
links among perspectives (namely, conceptual in-
tegration) (Suedfeld, Tetlock, & Streufert, 1992;
Tadmor, Tetlock, & Peng, 2009; Tetlock et al.,
1993). Evaluative differentiation requires dialecti-
cal reasoning (Tetlock, Armor, & Peterson, 1994).
Conceptual integration requires integrative reason-
ing to explain why reasonable people view the
same events differently and specify how to balance
between conflicting values or perspectives (Tetlock
et al., 1994).
Integratively complex leaders accept divergent
perspectives and are open to possible contradictory
information (Tetlock et al., 1993; Wong, Ormiston,
& Tetlock, 2011). Being sensitive to external cues
such as value conflicts and accountability inconsis-
tency, they can cope better with conflicting de-
mands (Tadmor, Galinsky, & Maddux, 2012). For
instance, they are likely to identify both structural
and follower demands in organizational contexts.
Moreover, integratively complex leaders seek inte-
grative solutions. They search for new approaches
to combining and integrating divergent perspec-
tives (Tadmor et al., 2012). To meet both structural
and follower requirements, integratively complex
leaders make decisions as formal leaders, while
giving followers voice regarding alternative solu-
tions, thereby allowing others to lead and be recog-
nized for contributions. In contrast, integratively
simple leaders may ignore seemingly irrelevant or
contradictory information or perspectives (Tetlock
et al., 1993). They emphasize one side of a paradox
and overlook the other. In sum:
Hypothesis 2. The extent to which a leader
displays integrative complexity will be posi-
tively associated with the leader’s paradoxical
behavior in managing people.
Contextual Antecedent of Paradoxical Leader
Behavior in People Management
Paradoxical leader behaviors may also depend on
organizational structure (Capelli & Sherer, 1991;
Pawar & Eastman, 1997; Pillai & Meindl, 1998;
Shamir & Howell, 1999), characterized by power,
reporting relationships, and rules or policies per-
taining to employee behaviors, decision making,
and communication (Donaldson, 1996; Ford & Slo-
cum, 1977).
Under a mechanistic structure, higher-level au-
thorities clearly hold decision-making power, com-
munications follow restricted channels, and rules
544 AprilAcademy of Management Journal
and regulations are rigid and uniform (Ambrose &
Schminke, 2003). Such tight contexts are more
likely to strengthen structural requirements and
less likely to consider follower demands. Supervi-
sors adhere to embedded, formalized rules, such as
treating subordinates uniformly, and maintaining
psychosocial and status distance. They will have
little room to bend rules for individualized con-
siderations. Furthermore, the structure prohibits
supervisors’ discretion to develop unconven-
tional approaches to work problems (Courtright,
Fairhurst, & Rogers, 1989; Pillai, 1995; Porter &
McLaughlin, 2006).
Under an organic structure, authority is decen-
tralized, communication is open and flexible, and
formal rules are less emphasized. Such loose con-
texts may make paradoxes more salient (Smith &
Lewis, 2011). Supervisors should initiate structure
for group and organizational functioning; at the
same time, they should consider subordinates’ in-
dividualized needs. For example, they can empha-
size uniform or consistent treatment of followers
from a structural perspective, while allowing indi-
vidualized treatment for meeting diverse follower
demands or needs. Moreover, organic structures
offer supervisors opportunities to be less con-
strained and more discretionary in responding to
environments (Pillai, 1995; Porter & McLaughlin,
2006). Thus, under an organic structure, PLB is
more likely to emerge. In short:
Hypothesis 3. The extent to which a leader
exhibits paradoxical behavior will be associ-
ated with organizational structure: under a
more organic structure, paradoxical behavior
in managing people will be more prevalent;
under a more mechanistic structure, paradox-
ical behavior in managing people will be less
prevalent.
Subordinate Work Role Performance as a
Consequence of Paradoxical Leader Behavior
in People Management
Work roles involve performance responsibilities
(Murphy & Jackson, 1999), including proficient,
adaptive, and proactive behaviors. “Task profi-
ciency” indicates how well employees meet work
role expectations and prescribed requirements
(Griffin et al., 2007). “Adaptive behavior” is “the
extent to which an individual adapts to changes in
a work system or work roles” (Griffin et al., 2007:
329), and involves handling crises, work stress,
uncertainties, and emergencies (Pulakos, Arad,
Donovan, & Plamondon, 2000). “Proactive behav-
ior” is “the extent to which the individual takes
self-directed action to anticipate or initiate
change in the work system or work roles” (Griffin
et al., 2007: 329), and includes acting in advance
and being change oriented (Parker, Williams, &
Turner, 2006). The three aspects of work role
performance are important in increasingly dy-
namic environments.
We suggest that PLB may enhance follower work
role performance in two ways. First, leaders may
serve as role models (Graen & Scandura, 1987;
Shamir, House, & Arthur, 1993) to show employees
how to accept and embrace contradictions in com-
plex environments (Fang, 2005). Subordinates
learn to be similarly open to challenges and possi-
bilities by expanding their holistic understanding
of work requirements, and identifying emerging job
demands and opportunities for responding to
changing environments (Detert & Burris, 2007).
Consequently, followers will be more likely to
work proficiently and adaptively. By further ex-
ploring the ways in which to connect and integrate
competing structural and follower demands, lead-
ers model how to develop persistent learning pro-
cesses for using fresh perspectives to examine
seemingly incompatible challenges and pursue
continuous improvement beyond the status quo.
They also show followers how to be flexible when
adjusting conventional approaches, or even radi-
cally exploiting unconventional approaches, so
that followers perceive it to be normal to solve
work issues proactively (Parker et al., 2006). In
short, role modeling shows followers how to be
open, learning oriented, and flexible to work profi-
ciently, adaptively, and proactively.
Second, PLB will create conjoined bounded
and discretionary work environments. To build
bounded environments, leaders use the locus of
influence to initiate the hierarchical structure; they
maintain distance and treat followers uniformly to
clarify structural roles; they establish high work
requirements and preserve decision control to im-
plement structural roles. Bounded environments
stress norms and standards whereby followers un-
derstand their work roles and responsibilities. At
the same time, PLB gives followers discretion and
individuality within the structure, allows them to
be the focus of influence to maintain their dignity
and confidence (Bligh, Pearce, & Kohles, 2006),
considers their social and individual needs to uti-
lize their personal strengths and capabilities, and
2015 545Zhang, Waldman, Han, and Li
empowers them to increase their self-motivation
(Srivastava et al., 2006). When leaders combine
those practices, followers know clearly what to do
and how to do it. They can also approach work
flexibly and autonomously, believing that they can
bend the rules and make mistakes, and thus will-
ingly work proactively to achieve proficiency
(Parker et al., 2006). Under such environments, fol-
lowers learn to meet complex and even ambivalent
requirements, thus becoming more adaptive (de
Jong & de Ruyter, 2004). If leaders emphasize only
structural aspects, followers may become nonpro-
active, strict rule-keepers, unable to adjust to task
changes and eventually suffering decreased profi-
ciency. If leaders emphasize only individuality and
discretion, the ensuing chaos may further damage
proficiency. Thus, both sides of leader behavior are
interdependent, jointly affecting follower behav-
iors. In short, through PLB, leaders embrace ambi-
dextrous requirements, engendering more profi-
cient, adaptive, and proactive follower behavior:
Hypothesis 4. Paradoxical leader behavior in
managing people will be positively associated
with followers’ (a) task proficiency, (b) adap-
tive behavior, and (c) proactive behavior.
STUDY 1
Item Generation, Exploratory Factor Analysis,
and Item Reduction
We generated 26 items for PLB based on: our
conceptualization; relevant literature concerning
various leadership measurements, such as humble
leadership (Owens, Johnson, & Mitchell, 2013), in-
dividualized consideration (Bass & Avolio, 1995),
and empowering leadership (Srivastava et al.,
2006); and 86 incidents described in interviews
with 11 male and 17 female supervisors and sub-
ordinates, averaging 31 years old. Sample incidents
included: “[supervisors] should not treat employ-
ees only as subordinates . . . despite hierarchical
differences, supervisors should treat subordinates
as friends”; “as long as subordinates ensure higher
work efficiency, supervisors should give them ap-
propriate discretion”; and “[the supervisor] has
high work criteria . . . but is also congenial, allow-
ing [subordinates] to make mistakes and learn and
grow from those mistakes.”
Among the 26 items, 6 items pertained to com-
bining self-centeredness with other-centeredness
(SO); 5 items pertained to maintaining both dis-
tance and closeness (DC); 5 items pertained to treat-
ing subordinates uniformly while allowing individ-
ualization (UI); 5 items pertained to enforcing work
requirements while allowing flexibility (RF); and 5
items pertained to maintaining decision control
while allowing autonomy (CA).
To ensure content validity, two organizational
behavior experts worded the items independently.
A third expert compared the lists for similarities
and differences; combined, revised, or deleted the
items; and returned them to the two experts, who
re-revised the items independently and then for-
warded them to the third expert.
After several rounds, all three experts agreed on a
final version of 26 items. They shared this version
with three groups of graduate-level organizational
behavior students, each group comprising five to
six participants, to determine whether additional
revision was necessary. A fourth group of five stu-
dents sorted the 26 items into the five dimensions
of PLB independently, and the percentage of cor-
rect assignments for each item reached 100%.
We administered a survey with the 26 PLB items
to two independent samples, Samples 1 and 2, to
explore the factor structure of PLB. Table 1 shows
Study 1’s sample characteristics and purposes.
We asked respondents to evaluate how frequently
each statement fit their current or past immediate
supervisor, using a five-point Likert-type scale
(“0” ⫽not at all, “4” ⫽alot).
Exploratory factor analysis results. In Sam-
ple 1, we found a general factor structure of five
distinct factors, but with nine cross-loading items.
The two organizational behavior experts revised
the cross-loading items independently, and then
discussed and re-revised them until they agreed
that each reflected a single dimensional meaning.
We used Sample 2 to conduct an additional ex-
ploratory factor analysis (EFA) on the revised 26
items. After deleting four cross-loading items, the
EFA results demonstrated five suitable factors: (1)
UI, 5 items; (2) SO, 5 items; (3) CA, 4 items; (4) RF,
4 items; and (5) DC, 4 items.
Eigenvalues of the five factors ranged from 1.20
to 6.44, with 65.6% of total variance explained.
Table 2 shows items, EFA factor loadings, percent-
age of variance explained, and factor reliabilities.
Confirmatory Factor Analysis and Item Design
Assessment
We define our measurement model as a “reflec-
tive first-order, reflective second-order model” (Jar-
vis, MacKenzie, & Podsakoff, 2003: 205), because
546 AprilAcademy of Management Journal
items are reflective indicators of each respective
dimension and the dimensions (first-order factors)
are reflective indicators of the latent second-order
construct: PLB. We used Sample 3 for confirma-
tory factor analysis (CFA). To examine the factor
structure of PLB, we first compared the fit of three
factor structures. The first structure was a one-fac-
tor model in which all 22 items indicated one PLB
factor; the second was a first-order, five-factor
model in which items were allowed to load onto
their respective factors; the third was a second-
order factor model in which items were loaded
onto their respective factors and the five factors
were loaded onto a second-order, latent PLB factor.
We further compared the second-order factor
model with alternative first-order, three-factor and
first-order, two-factor models.
As our conceptual description of PLB suggested
and as the items in Table 2 show, item design for
assessing PLB may need to shift fundamentally
from the traditional assumption that choices in-
volve a bipolar paradox (Bobko, 1985). Instead, rat-
ers should evaluate how extensively leaders em-
brace both bipolar actions, and recognize that both
behavioral orientations toward the two poles of a
paradox are inseparable and interdependent, con-
sistent with our conceptualization of paradoxes:
that opposites coexist and may be embraced simul-
taneously (Fang, 2005). Thus, we used a double-
barreled item design approach to capture leaders’
paradoxical behaviors. Only leaders who exhibited
two seemingly divergent behaviors would be rated
high on each item of PLB. Although such item
construction has been challenged for being poten-
tially ambiguous and engendering different rater
interpretations (see Kaiser & Craig, 2005), “double-
barreled items are not inherently poor” (Brutus &
Facteau, 2003: 323); they can perform favorably
compared with items targeting a single behavior.
Thus, we purport that double-barreled item design
may be appropriate for scale development for PLB.
To assess the appropriateness of our item design,
we compared it with alternative procedures in pre-
dicting two criterion variables: identification with
the leader, capturing subordinates’ personal attach-
ment to supervisors (Zhang & Chen, 2013); and job
engagement, capturing physical, cognitive, and
emotional energy in active work performance
(Rich, LePine, & Crawford, 2010). Both variables
reflect people management effectiveness derived
from structural and individual behavior. First,
each PLB item includes two opposite poles, so we
examined for an order effect in the sequence of
the two meanings. We compared our original ver-
sion (Cronbach’s alpha ⫽.95) with the alterna-
tive version differing only in the order of the two
meanings, which we call “the measurement ver-
sion with a reversed order of meaning” (Cron-
bach’s alpha ⫽.95).
Second, we tested our measure against alterna-
tive measurement approaches that combined inde-
pendent solutions to each pole of respective para-
TABLE 1
Characteristics and Purposes of Samples for Study 1
a
Sample
Size % Male
Mean Age
(SD)
Mean Work
Tenure (SD) Purpose
Sample 1 228 52.2% 30.7 (7.3) 8.2 (7.5) EFA
Sample 2 204 32.2% 28.6 (4.1) 6.0 (4.2) EFA
Sample 3 193 20.3% 28.2 (3.7) 6.1 (3.7) CFA
Subsample 1 87 23.0% 28.4 (3.8) 6.2 (4.0) Item assessment: order effects
Subsample 2 106 18.1% 28.1 (3.6) 6.0 (3.4) Item assessment: comparison between alternative
“additive” and “multiplicative” models
Sample 4, Time 1 186 26.7% 28.9 (4.2) 7.0 (4.1) Test–retest reliability; discriminant and
convergent validation using behavioral
complexities and managerial flexibility
Sample 4, Time 2 172 26.7% 28.9 (4.2) 7.0 (4.1)
Sample 4, Time 3
b
116 23.4% 29.2 (4.3) 7.4 (4.3)
Sample 5
c
478 35.0% 30.2 (6.5) 5.2 (6.1) Discriminant validation using five alternative
leadership measures
a
We recruited respondents from a continuing education program of a large, northern Chinese university. All were full-time employees
from various companies, had college experience, and attended weekend classes for three years to earn bachelor’s degrees.
b
The nonrespondent subsample was similar to the respondent subsample in age (t⫽.76, p⫽.45) and years’ tenure (t⫽1.19,
p⫽.20), but differed in gender (
2
⫽5.18, p⬍.05). However, gender was not correlated significantly with PLB dimensions, indicating
that Time 3 of Sample 4 had a nonsignificant nonresponse bias.
c
We collected Sample 5 data from five manufacturing and service firms in China.
2015 547Zhang, Waldman, Han, and Li
doxes. One is based on an “additive” model
(Lubatkin, Simsek, Ling, & Veiga, 2006) that aggre-
gates the two independent behavioral solutions to
the two opposite poles of paradoxes. The other
approach reflects a “multiplicative” model (Gibson
& Birkinshaw, 2004; Lubatkin et al., 2006), based
TABLE 2
Exploratory Factor Analysis Results of the Paradoxical Leader Behavior in People Management Scale
Factors and items EFA Loadings CFA Loadings
a
1. Treating subordinates uniformly while allowing
individualization (UI)
.79 .85 .77
(a) Uses a fair approach to treat all subordinates
uniformly, but also treats them as individuals.
.80 .02 .23 .11 .16 .88 .76 .89
(b) Puts all subordinates on an equal footing, but
considers their individual traits or personalities.
.79 .00 .22 .04 .20 .84 .77 .85
(c) Communicates with subordinates uniformly without
discrimination, but varies his or her communication
styles depending on their individual characteristics or
needs.
.79 .14 .20 .10 .18 .83 .79 .87
(d) Manages subordinates uniformly, but considers their
individualized needs.
.73 ⫺.05 .08 .19 .05 .83 .81 .82
(e) Assigns equal workloads, but considers individual
strengths and capabilities to handle different tasks.
.66 .01 .25 .22 .17 .87 .80 .83
2. Combining self-centeredness with other-centeredness (SO) .63 .74 .79
(a) Shows a desire to lead, but allows others to share the
leadership role.
⫺.02 .84 .03 .07 ⫺.10 .73 .65 .77
(b) Likes to be the center of attention, but allows others to
share the spotlight as well.
.05 .82 .00 .11 ⫺.07 .77 .75 .64
(c) Insists on getting respect, but also shows respect
toward others.
⫺.18 .76 .05 ⫺.07 .19 .57 .66 .70
(d) Has a high self-opinion, but shows awareness of
personal imperfection and the value of other people.
.05 .72 ⫺.07 .12 .15 .73 .72 .75
(e) Is confident regarding personal ideas and beliefs, but
acknowledges that he or she can learn from others.
.24 .50 .13 .08 .15 .78 .75 .79
3. Maintaining decision control while allowing autonomy (CA) .77 .79 .71
(a) Controls important work issues, but allows
subordinates to handle details.
.17 ⫺.01 .80 .15 .18 .86 .79 .88
(b) Makes final decisions for subordinates, but allows
subordinates to control specific work processes.
.12 .00 .77 .11 .02 .82 .82 .81
(c) Makes decisions about big issues, but delegates lesser
issues to subordinates.
.25 .09 .75 .14 .10 .84 .74 .78
(d) Maintains overall control, but gives subordinates
appropriate autonomy.
.38 .05 .71 .15 .13 .88 .84 .80
4. Enforcing work requirements while allowing flexibility
(RF)
.83 .75 .73
(a) Stresses conformity in task performance, but allows for
exceptions.
.09 .17 .08 .85 .13 .80 .77 .90
(b) Clarifies work requirements, but does not micro-
manage work.
.19 .03 .17 .75 .21 .84 .78 .87
(c) Is highly demanding regarding work performance, but
is not hypercritical.
.19 .03 .10 .72 .29 .90 .81 .88
(d) Has high requirements, but allows subordinates to
make mistakes.
.12 .09 .19 .72 .14 .83 .71 .85
5. Maintaining both distance and closeness (DC) .86 .83 .75
(a) Recognizes the distinction between supervisors and
subordinates, but does not act superior in the
leadership role.
.19 ⫺.01 .26 .14 .77 .85 .71 .86
(b) Keeps distance from subordinates, but does not remain
aloof.
.01 .11 .12 .22 .71 .74 .56 .81
(c) Maintains position differences, but upholds
subordinates’ dignity.
.25 .10 .07 .22 .70 .82 .82 .84
(d) Maintains distance from subordinates at work, but is
also amiable toward them.
.39 .10 ⫺.03 .21 .67 .87 .80 .82
% variance explained 30.4 13.4 8.7 7.2 5.7
Reliability .87 .81 .84 .83 .78
Note: The extraction method for EFA is principal component analysis. The rotation method is varimax, with Kaiser normalization. Bold
figures are the highest factor loadings of indicators on factors from EFA.
a
The standardized CFA loadings in the right three columns are from Samples 3, 4, and 5.
548 AprilAcademy of Management Journal
on the product between the two independent solu-
tions to the two opposite poles. Thus, we split our
double-barreled items into separate items in terms
of meaning and examined whether the original
measure was better than indices based on these
models. (For example, we split “manages subordi-
nates uniformly while still considering their indi-
vidualized needs” into two items: “manages subor-
dinates uniformly,” and “considers subordinates’
individualized needs”). We used a five-point Lik-
ert-type scale (“0” ⫽notatall, “4” ⫽a lot).
1
We call the index derived from the additive
model the “additive-model index” (e.g., aggregat-
ing the score on “manages subordinates uni-
formly” and the score on “considers subordi-
nates’ individualized needs”), and the index from
the multiplicative model the “multiplicative-
model index” (e.g., multiplying the score on
“manages subordinates uniformly” by the score
on “considers subordinates’ individualized
needs”).
We used a six-item scale (Mael & Ashforth, 1992)
to measure identification with the leader (Cron-
bach’s alpha ⫽.82), and a nine-item scale (Rich et
al., 2010) to measure job engagement (Cronbach’s
alpha ⫽.96).
2
In addition to the original instru-
ment, we included the measurement version with a
reversed order of meaning to assess PLB and the
criterion variables in the survey for subsample 1 of
Sample 3. We included our instrument, the crite-
rion variables, and the PLB items separated accord-
ing to meanings in the survey for subsample 2 of
Sample 3. To avoid the sequential effect, we
placed the original measure in the first section,
criterion variables second, and the alternative
leader behavior measure third in half of the sur-
veys. We placed the alternative measure first and
the original items last in the other half of the
surveys.
Confirmatory factor analysis results. The upper
section in Table 3 shows fit statistics for the three
structural models of Sample 3. The worst-fitting
model was the one-factor model, as demonstrated
by the relatively poor fit indices.
3
The second-order
and first-order factor models were a better fit than
the one-factor model. The second-order factor
model was not significantly better than the first-
order five-factor model, based on the chi-square
difference (⌬
2
[5] ⫽9.32, p⬎.05). Indeed, the
two models are mathematically equivalent (Bollen,
1989). The relatively good fit of the first-order fac-
tor model may derive from the five dimensional
measures, which conceptually address different
paradoxes and have significant, but relatively low,
relationships. Nevertheless, they share variance, re-
flected in a latent higher-order factor: leader behav-
ior simultaneously directed toward competing
structural and follower demands. Thus, the second-
order model is preferred, because it allows the co-
variation among first-order factors by accounting
for corrected errors that are common in first-order
CFA (Gerbing & Anderson, 1984).
The factor loadings of the second-order factor
PLB model are presented in the far right column of
Table 2. Item loadings on the five factors ranged
from .57 to .90. Loadings of the leadership factors
on the second-order PLB factor vary from .63
to .86. Correlation coefficients among five dimen-
sions were from .41 to .68.
4
Cronbach’s alphas for
each dimension were above .84. Average variances
extracted (AVE) for Sample 3 (in Table 3) reached
the criterion of .5, suggesting that variances cap-
tured by the construct are larger than variances
resulting from measurement error (Fornell &
Larcker, 1981). The square roots of AVE (from .72
to .81) were larger than the correlation coefficients
among the five factors (maximum value was .68),
demonstrating the discriminant validity of the five-
factor structure (Fornell & Larcker, 1981).
Results of order effect. In subsample 1 of Sam-
ple 3, our original measure of PLB correlated
highly with the measurement version with a re-
versed order of meaning (r⫽.88, p⬍.001),
suggesting convergent validity of the original
measure. The two measurement versions corre-
lated similarly with identification with the leader
(r⫽.69 vs. r⫽.63, t⫽1.475, p⫽.18) and
job engagement (r⫽.67 vs. r⫽.67, t⫽⫺.08,
p⫽.90), suggesting no significant order effect in
1
We used Likert-type scaling—one normative ap-
proach to measurement. We used a non-normative for-
mat—ipsative measure—in another sample. The re-
sults of comparing the original measurement with the
ipsative measurement showed that the original ex-
plained additional variance beyond the ipsative
measurement.
2
We followed the translation-back translation ap-
proach (Brislin, 1980) to generate a Chinese version of
these and all other measures used in this research.
3
Because of space limitations, we provide only a short
summary of our construct validation results.
4
Correlations and reliabilities of PLB dimensions are
available on request.
2015 549Zhang, Waldman, Han, and Li
terms of the sequence of the two meanings in
PLB items.
Results of comparing alternative forms. In sub-
sample 2 of Sample 3, the original PLB measure
correlated highly with the additive-model index
(r⫽.68, p⬍.001), suggesting convergent validity
of the original measure. The original measure and
the additive-model index each correlated with
leader identification (r⫽.51, p⬍.001; r⫽.30,
p⬍.001) and job engagement (r⫽.52, p⬍.001;
r⫽.41, p⬍.001). We regressed the two criterion
variables on our original measure and the additive-
model index, controlling for employee age, gender,
education, work tenure, and job position. The orig-
inal measure of PLB significantly predicted iden-
tification with the leader (

⫽.44, p⬍.01) and
job engagement (

⫽.39, p⬍.01), whereas the
effect of the additive-model index was insignifi-
cant (

⫽.14, p⫽.24;

⫽.21, p⫽.11).
These results demonstrate that our original mea-
sure is superior to the additive-model index. The
original measure of PLB correlated with the mul-
tiplicative-model index (r⫽.23, p⬍.05). The
multiplicative-model index did not correlate
with leader identification (r⫽⫺.03, ns)orjob
engagement (r⫽.04, ns). In the regular regres-
sion including control variables, the original
measure significantly predicted identification
with the leader (

⫽.56, p⬍.001) and job en-
gagement (

⫽.57, p⬍.001), whereas the
multiplicative-model index was insignificant
(

⫽⫺.08, p⫽.36;

⫽⫺.15, p⫽.11). These
results demonstrate that our measure is superior
to the multiplicative-model index. In sum, our
original measurement of PLB explained variance
in criterion variables better than the alternative
measurement indices.
Test–Retest Reliability, and Discriminant and
Convergent Validation
We used Sample 4 to further validate PLB. We
measured PLB twice for test–retest reliability, and
used two measures of behavioral complexity and
one measure of managerial flexibility (i.e., capaci-
ties for handling paradoxes) to evaluate its discrim-
inant validity. We asked respondents to rate their
current immediate supervisors’ PLB and their own
demographic information in a weekend morning
class (Time 1, 186 respondents), to rate behavioral
complexities and managerial flexibility in the after-
noon class of the same day (Time 2, 172 respon-
dents), and to rate PLB again four weeks later
(Time 3, 116 respondents).
TABLE 3
Confirmatory Factor Analysis Results of Paradoxical Leader Behavior in People Management
Model
2
df ⌬
2
CFI GFI SRMR RMSEA AVEs
Models used in Study 1, Sample 3
a
Second-order factor model 399.00 204 .97 .84 .05 .07 AVE
CA
⫽.72
AVE
RF
⫽.71
AVE
UI
⫽.72
AVE
DC
⫽.67
AVE
SO
⫽.51
First-order, five-factor model 390.32 199 9.32(5) .97 .82 .05 .07
One-factor model 1,713.92 209 1,314.92(5)* .88 .55 .10 .19
First-order, three-factor model (CA ⫹RF, UI ⫹DC, SO) 1,262.15 206 863.15(2)* .92 .63 .08 .16
First-order, three-factor model (CA ⫹DC, UI ⫹RF, SO) 1,323.07 206 924.07(2)* .91 .61 .09 .17
First-order, three-factor model (CA ⫹UI, DC ⫹RF, SO) 1,134.90 206 735.90(2)* .92 .65 .10 .15
First-order, two-factor model (CA ⫹RF ⫹SO, UI ⫹DC) 1,513.56 208 1,114.56(4)* .90 .58 .10 .18
First-order, two-factor model (CA ⫹RF, DC ⫹UI ⫹SO) 1,604.06 208 1,205.06(4)* .89 .57 .10 .19
Models used in Study 1, Sample 5
b
Second-order factor model 515.21 204 .99 .91 .04 .06 AVE
CA
⫽.67
AVE
RF
⫽.77
AVE
UI
⫽.70
AVE
DC
⫽.69
AVE
SO
⫽.53
First-order, five-factor model 504.99 199 10.22(5) .99 .91 .04 .06
One-factor model 4,524.24 209 4,009.03(5)
*
.86 .54 .11 .21
First-order, three-factor model (CA ⫹RF, UI ⫹DC, SO) 2,954.43 206 2,439.22(2)
*
.91 .64 .12 .17
First-order, three-factor model (CA ⫹DC, UI ⫹RF, SO) 3,351.10 206 2,835.89(2)* .90 .61 .10 .18
First-order, three-factor model (CA ⫹UI, DC ⫹RF, SO) 2,898.53 206 2,383.31(2)* .91 .64 .11 .17
First-order, two-factor model (CA ⫹RF ⫹SO, UI ⫹DC) 3,726.53 208 3,211.32(4)* .89 .58 .11 .19
First-order, two-factor model (CA ⫹RF, DC ⫹UI ⫹SO) 3,998.60 208 3,483.39(4)* .89 .57 .13 .20
Note: We also conducted first-order, four-factor model analysis in each sample; the five-factor model showed better fit than the
four-factor models. Because of space limitations, we do not report the specific results here.
a
n⫽193.
b
n⫽478.
*p⬍.001 (two-tailed)
550 AprilAcademy of Management Journal
We used a four-dimensional, 36-item scale (Law-
rence et al., 2009) and four-dimensional, 16-item
scale (Denison et al., 1995) to measure behavioral
complexity, and adopted a five-item scale (Kaiser et
al., 2007) to assess management flexibility.
5
The
alpha coefficients were above .80 for these mea-
sures. The alpha coefficients PLB dimensions
were DC ⫽.82, RF ⫽.85, UI ⫽.89, CA ⫽.88,
SO ⫽.83 (Time 1), and .84, .90, .91, .90, and .85
(Time 2).
We used a field sample (Sample 5) to further
validate our measure of PLB. We assessed subordi-
nate-rated leadership behaviors and demographic
information at Time 1, and measured subordi-
nates’ self-rated attitudes (affective commitment,
leave intention, leader effectiveness) and supervi-
sor-rated subordinates’ work behavior (in-role task
performance; task-oriented organizational citizen-
ship behavior (OCB)) after two weeks.
First, we examined the factor structure of PLB.
Second, we tested the discriminant validity of PLB
dimensions using alternative leadership mea-
sures—that is, humble leadership (HL) (i.e., how it
differs from the SO dimension), LMX (i.e., how it
differs from the DC and UI dimensions), and trans-
actional leadership (TAL) (i.e., how it differs from
the CA and RF dimensions) in CFA. We also differ-
entiated PLB from transformational and paternalis-
tic leadership to validate the overall construct.
Third, we demonstrated predictive validity in cor-
relations between PLB and criterion variables in-
cluding affective commitment, leave intentions,
in-role and extra-role task performance (task citi-
zenship behavior—Farh, Hackett, & Liang, 2007),
and the subjective assessment on leader effective-
ness. We also conducted usefulness analysis
(Judge, Erez, Bono, & Thoresen, 2003) to demon-
strate “incremental validity” (Judge et al., 2003:
309) of PLB beyond alternative forms of leadership.
Using hierarchical regression, we entered each al-
ternative form of leadership first to predict the cri-
terion variables. Then, we added PLB to equations
to ascertain the increase in variance accounted for.
We then compared these results to the reverse sit-
uation, whereby we entered PLB on the first step
and each of the alternative forms of leadership in
the second step. Except for our PLB measures, we
used established scales to assess the above leader-
ship behaviors and subordinate attitudes and be-
haviors.
6
Their reliabilities ranged from .76 to .95.
Validation results of Sample 4. We used the
dimensional scores of PLB and behavioral com-
plexities and item scores of management flexibility
to discriminate PLB from the three alternative mea-
sures in factor analyses. We ran three two-factor
models and three corresponding one-factor models
(Table 4). The model comparison results demon-
strate that PLB differs from the three measures,
suggesting its high discriminant validity. The AVE
results show that the square roots of AVE scores
(Table 4) were greater than the correlations be-
tween PLB and alternative measures, further show-
ing discriminant validity (Fornell & Larcker, 1981).
Paradoxical leader behavior was significantly cor-
related with Lawrence et al.’s (2009) behavioral
complexity (r⫽.53, p⬍.000), with Denison et
al.’s (1995) measure (r⫽.49, p⬍.000), and with
managerial flexibility (r⫽.55, p⬍.000), demon-
strating high convergent validity. The correlation
between the Time 1 and Time 2 PLB measure-
ments was .69, demonstrating high test–retest
reliability.
Validation results of Sample 5. Following the
procedure in Sample 3 CFA analyses, we used
item scores to compare the three factor models.
Similarly to the Sample 3 CFA results showing
that the first-order five-factor model was insignifi-
cantly better than the second-order factor model
(lower part of Table 3; ⌬
2
[5] ⫽10.22, p⬎.05), we
prefer the second-order factor model. Item loadings
on the five factors and loadings of the factors on the
second-order, PLB factor are listed in Table 2. Cor-
relation coefficients among the five factors ranged
from .48 to .57.
7
Cronbach’s alphas for factors
were above .86. The AVEs for all factors (Table 3)
were above the criteria of .5. The square roots of
5
Because of space limitations, we do not report exam-
ple items.
6
We used 9 items to measure HL (Owens et al., 2013),
7 items for LMX (Scandura & Graen, 1984), 5 items for
TAL (Waldman, Ramirez, House, & Puranam, 2001), 20
items for transformational leadership (Bass & Avolio,
1995), 15 items for paternalistic leadership (Cheng,
Chou, & Farh, 2000), 6 items for affective commitment
(Meyer, Allen, & Smith, 1993), a Chinese-version 4-item
scale for leave intentions (Wang, Law, & Chen, 2002), 11
items for in-role task performance (Tsui, Pearce, Porter, &
Tripoli, 1997), 5 items for task OCB (Farh et al., 2007),
and 4 items for leader effectiveness (Rodan & Galu-
nic, 2004).
7
Correlations and reliabilities of PLB dimensions are
available on request.
2015 551Zhang, Waldman, Han, and Li
AVE (from .73 to .87) were larger than the correla-
tion coefficients among the five factors (maximum
value was .57), demonstrating the discriminant va-
lidity of the five-factor structure (Fornell &
Larcker, 1981).
As CFA results show in Table 4, the SO dimen-
sion differs from HL; the UI and DC dimensions
differ from LMX; the CA and RF dimensions differ
from TAL. Furthermore, additional CFA results
and AVE tests demonstrated that PLB was distinc-
tive from alternative leadership measures, includ-
ing transformational and paternalistic leadership
(see Table 4).
As Table 5 shows, PLB was strongly associated
with criterion variables, showing its criterion va-
lidity. As Table 6 shows, usefulness analysis re-
sults indicated that, except for regressing task OCB
on PLB and paternalistic leadership, PLB signifi-
cantly contributed to predicting the five criteria
beyond alternative leadership behaviors.
TABLE 4
Confirmatory Factor Analysis Results for Scale Validation of Paradoxical Leader Behavior in People Management
2
df ⌬
2
CFI GFI SRMR RMSEA AVEs
Models used to discriminate PLB from alternative
measures (Sample 4
a
)
Two-factor model: PLB and Lawrence et al.’s (2009)
behavioral complexity (LBC)
89.9 26 .95 .90 .07 .11 AVE
PLB
⫽.51
One-factor model: two factors merged 269.4 27 179.5 (1)* .85 .76 .11 .22 AVE
LBC
⫽.72
Two-factor model: PLB and Hooijberg’s (1996)
behavioral complexity (HBC)
65.1 26 .96 .92 .07 .09 AVE
PLB
⫽.51
One-factor model: two factors merged 198.8 27 133.7 (1)* .86 .79 .11 .19 AVE
HBC
⫽.52
Two-factor model: PLB and managerial flexibility
(MF)
50.7 34 .99 .94 .04 .05 AVE
PLB
⫽.51
One-factor model: two factors merged 253.4 35 202.7 (1)* .90 .77 .09 .19 AVE
MF
⫽.56
Models used to validate PLB dimensions
(Sample 5
b
)
Two-factor model: SO dimension and humble
leadership (HL)
372.91 73 .97 .90 .06 .09 AVE
HL
⫽.58
AVE
SO
⫽.52
One-factor model: two factors merged 1,134.88 74 761.97 (1)* .92 .75 .09 .17
Three-factor model: UI dimension, DC dimension,
and LMX
413.10 101 .98 .90 .05 .08 AVE
LMX
⫽.54
AVE
UI
⫽.73
AVE
DI
⫽.89
Two-factor model: UI dimension merged with LMX
factor
1,889.69 103 1,476.59 (2)* .92 .67 .10 .19
Two-factor model: DC dimension merged with LMX
factor
1,696.51 103 1,283.41 (2)* .92 .69 .10 .18
One-factor model: three factors merged 2,782.28 104 2,369.18 (3)* .86 .58 .12 .23
Three-factor model: CA dimension, RF dimension,
transactional leadership (TAL)
168.71 62 .98 .95 .03 .06 AVE
TAL
⫽.52
AVE
CA
⫽.67
AVE
RF
⫽.76
Two-factor model: RF dimension merged with TAL
factor
1,279.84 64 1,111.13 (2)* .86 .71 .14 .20
Two-factor model: CA dimension merged with TAL
factor
1,270.13 64 1,101.42 (2)* .87 .71 .13 .20
One-factor model: three factors merged 2,454.13 65 2,285.42 (3)* .74 .56 .17 .28
Models used to discriminate PLB from alternative
leaderships (Sample 5)
Two-factor model: PLB and humble leadership (HL) 303.19 70 .98 .92 .04 .08 AVE
PLB
⫽.51
AVE
HL
⫽.57
One-factor model: two factors merged 519.76 71 216.57 (1)* .96 .87 .06 .12
Two-factor model: PLB and LMX 245.17 52 .97 .92 .05 .09 AVE
PLB
⫽.51
AVE
LMX
⫽.53
One-factor model: two factors merged 567.75 53 322.58 (1)* .94 .83 .07 .14
Two-factor model: PLB and TAL 110.16 34 .98 .96 .05 .07 AVE
PLB
⫽.51
AVE
TAL
⫽.53
One-factor model: two factors merged 735.23 35 625.07 (1)* .87 .76 .10 .21
Two-factor model: PLB and transformational
leadership (TFL)
87.00 25 .98 .96 .03 .07 AVE
PLB
⫽.51
One-factor model: two factors merged 283.00 26 196.00 (1)* .95 .88 .06 .14 AVE
TFL
⫽.60
Four-factor model: PLB, three dimensions of
paternalistic leadership
813.08 164 .95 .85 .07 .09 AVE
PLB
⫽.51
AVE
BEN
⫽.67
AVE
MOR
⫽.76
AVE
AUT
⫽.41
Three-factor model 1: PLB and benevolence (BEN)
merged
1,886.87 167 1,073.79 (3)* .91 .72 .11 .13
Three-factor model 2: PLB and morality (MOR)
merged
1,462.59 167 649.51 (3) * .91 .77 .10 .15
Three-factor model 3: PLB and authoritarianism
(AUT) merged
1,585.14 167 772.06 (3) * .91 .75 .11 .13
One-factor model: four factors merged 3,999.88 170 3,186.80 (6)* .79 .54 .14 .22
a
n⫽188.
b
n⫽478.
*p⬍.001 (two-tailed)
552 AprilAcademy of Management Journal
TABLE 5
Correlations of Sample 5 in Study 1
Mean SD 12345678910111213
1. PLB in people management 2.84 0.54 .94
2. Transformational leadership 2.71 0.58 .65** .95
3. Morality of paternalistic
leadership
2.97 0.71 .60** .65** .94
4. Benevolence of paternalistic
leadership
2.07 0.79 .50** .64** .49** .91
5. Authoritarianism of paternalistic
leadership
2.02 0.63 ⫺.14** ⫺.02 ⫺.03 .01 .77
6. Humble leadership 2.68 0.63 .69** .71** .65** .60** ⫺.13** .93
7. Leader-member exchange 2.51 0.62 .64** .68** .58** .60** ⫺.14** .72** .89
8. Transactional leadership 2.57 0.64 .51** .48** .34** .41** .03 .51** .47** .83
9. Organizational commitment 4.55 1.00 .37** .36** .28** .29** ⫺.21** .39** .40** .27** .91
10. Leave intention 1.99 1.05 ⫺.30** ⫺.28** ⫺.23** ⫺.16** .14** ⫺.26** ⫺.21** ⫺.14** ⫺.56** .93
11. Task performance 4.25 0.54 .20** .17** .18** .08
†
⫺.03 .14** .15** .05 .02 ⫺.02 .89
12. Task OCB 4.20 0.60 .20** .20** .22** .12* ⫺.05 .20** .16** .07 .09
†
⫺.05 .62** .78
13. Leader effectiveness 3.87 0.85 .28** .28** .16** .22** ⫺.21** .31** .31** .18** .50** ⫺.22** .08
†
.10* .76
Note: n ⫽478. Reliability coefficients are reported in bold italic along the diagonal.
†
p⬍.10
*p⬍.05
** p⬍.01 (two-tailed)
TABLE 6
Usefulness Analyses of Paradoxical Leader Behavior in People Management Compared to Alternative Leadership
Measures in Study 1
Organizational
Commitment Leave Intention Task Performance Task OCB Leader Effectiveness
1. Transformational leadership (1) .36
***
.28
***
.17
***
.20
***
.28
***
2. PLB (1) .04
***
.04
**
.04
**
.02
*
.03
**
1. PLB (1) .37
***
.30
***
.20
***
.20
***
.28
***
2. Transformational leadership (1) .03
***
.02
**
.01 .02
*
.03
**
1. Three dimensions of paternalistic
leadership (3)
.39
***
.28
***
.18
**
.22
***
.31
***
2. PLB (1) .04
***
.05
***
.04
**
.02
†
.04
***
1. PLB (1) .37
***
.30
***
.20
***
.20
***
.28
***
2. Three dimensions of paternalistic
leadership (3)
.06
***
.03
*
.02 .04
*
.07
***
1. Humble leadership (1) .39
***
.26
***
.14
**
.20
***
.31
***
2. PLB (1) .03
**
.05
***
.06
**
.02
*
.01
*
1. PLB (1) .37
***
.30
***
.20
***
.20
***
.28
***
2. Humble leadership (1) .04
***
.01
†
.00 .02
†
.05
***
1. Leader-member exchange (1) .39
***
.21
***
.15
**
.16
***
.31
***
2. PLB (1) .03
***
.09
***
.06
**
.05
**
.02
*
1. PLB (1) .37
***
.30
***
.20
***
.20
***
.28
***
2. Leader-member exchange (1) .05
***
.00 .00 .01 .05
***
1. Transactional leadership (1) .27
***
.14
**
.05 .07 .18
***
2. PLB (1) .11
***
.15
***
.16
***
.14
***
.10
***
1. PLB (1) .37
***
.30
***
.20
***
.20
***
.28
***
2. Transactional leadership (1) .01
*
.00 .01 .01 .00
Note: n ⫽478. The number of variables entered into the regression on each step is provided in parentheses. Table entries are multiple
correlations (multiple R). Numbers in second step are change in multiple correlations (⌬R).
†
p⬍.10
*p⬍.05
** p⬍.01
*** p⬍.001 (two-tailed)
2015 553Zhang, Waldman, Han, and Li
Discussion
In Study 1, we developed and validated a five-
dimensional measure of PLBs pertaining to peo-
ple management at the supervisory level. Using
five Chinese samples, we conducted EFA and
CFA, and assessed test–retest reliability, and con-
vergent, discriminant, and predictive validities.
Overall, our analyses demonstrated that the mea-
sure is valid in the Chinese context. Although
Western thinking is believed to be more analyti-
cal, while Chinese thinking is thought to be more
synthetic and integrative (Chen, 2002), Western-
ers may also understand and act paradoxically
(Fang, 2012; see Kelleher, 1997). For example,
the Lego Group asked its managers to navigate 11
paradoxes, such as “to be able to build a close
relationship with one’s staff, and to keep a suit-
able distance,” and “to be able to lead, and to
hold oneself in the background” (Evans, 2000:
75). Thus, PLB may not necessarily be exclusive
to the Chinese context, but may rather be a set of
leader behaviors relevant to, and manifested in,
multiple cultural contexts.
Furthermore, we assessed the appropriateness of
using double-barreled items. Clear and consistent
results of EFAs and CFAs reveal that the item de-
sign is understandable and acceptable. Empirical
analyses also suggest that this approach is better
than the alternative measurement indices, which
are functions of the two independent, behavioral
strategies toward two competing poles. Moreover,
our approach to item design best captures the fun-
damental assumption of viewing paradoxes as a
nonbipolar, “both–and” strategy for dealing with
competing demands simultaneously.
The simultaneous bipolar nature of our measure-
ment approach reveals the interconnections of two
poles. In a given paradox, the two single behaviors
do not provide meaning simply by being com-
bined, as in the multiplicative model; rather, one
pole depends on the other, and they equally con-
tribute to paradox resolution—that is, “one side of
the paradox may influence the conditions under
which the other will operate” (Van de Ven & Poole,
1988: 24). For example, a leader places high re-
quirements on subordinates, but also allows them
to make mistakes. Following high requirements is
the condition under which subordinates are al-
lowed to make mistakes. Separate measurement
might show high requirements, suggesting overly
strict leadership, and also show allowance for mis-
takes, potentially suggesting leadership lacking
standards. Separate measurement of those leader
behaviors (i.e., the “additive” model) would thus
fail to show bipolar consideration. Therefore, high
levels of PLB do not necessarily mean high levels
of both sides of behavior, but instead imply con-
necting and embracing two paradoxical poles con-
sistently over time.
Paradoxical leader behavior uses time and situa-
tion to separate dualities and to transform dualities
from contradictions into complementarities. Such
behavior emphasizes the persistent coexistence of du-
alities to achieve dynamic equilibrium in the long
term (Smith & Lewis, 2011). Our PLB measure dem-
onstrates how (a) two poles of paradoxes can be con-
nected and complementary, aligning with the holistic
nature of paradoxes, and (b) two sides in paradoxes
are relatively, rather than absolutely, stable, aligning
with the dynamic nature of paradoxes.
STUDY 2
Sample and Procedures
We conducted this study on site in six privately
owned technical or engineering services companies
in China. In total, 90 supervisors and 607 lowest-
level subordinates participated. We first asked hu-
man resources (HR) managers from each firm to
provide a list of work units, with names of super-
visors and subordinates. We coded the question-
naires and envelopes to identify the direct relation-
ships between them. We pasted name tags
corresponding to individuals’ codes on the enve-
lopes and asked them to tear off the tags on receipt.
This approach helped us to distribute the enve-
lopes to the correct participants, and ensured the
structural relationships and confidentiality among
respondents and informants. An HR manager in
each company helped us to collect questionnaires,
which respondents completed and returned in
sealed envelopes.
Of 607 subordinates, 599 responded and 588
were valid.
8
Of 90 supervisors, 76 responses were
valid. Supervisors rated 516 subordinates on their
work behaviors. They rated all of their subordinates
when they had no more than 12 subordinates, or
else they rated 12 of their subordinates who were
selected randomly by one of the authors. Individu-
als in the subordinate sample averaged 30 years old
(SD ⫽8.23), with an average 2.86 years’ firm ten-
8
Invalid data refer to ratings that are uniform across
items or show a zigzag pattern.
554 AprilAcademy of Management Journal
ure (SD ⫽2.19); 53% were male and 72% had
college educations. Participants in the supervisor
sample averaged 35 years old (SD ⫽8.03) and had
6.96 years’ firm tenure (SD ⫽3.00); 64% were
male and 84% had college educations.
Measures
We collected data twice. At Time 1, subordi-
nates rated leadership behaviors, and supervisors
rated their own holistic thinking and integrative
complexity, as well as organizational structure. Af-
ter three weeks, supervisors rated subordinates’
proficient, adaptive, and proactive work role be-
haviors. The scales used a six-point Likert-type
scale (“1” ⫽strongly disagree, “6” ⫽strongly
agree), with the exception of leader behavior mea-
sures, which used a five-point scale (“0” ⫽not at
all, “4” ⫽a lot).
Paradoxical leader behavior in people man-
agement. We used the 22 items developed in
Study 1 to measure PLB. Reliabilities for dimen-
sions ranged from .80 to .88. We completed AVE
tests and showed the discriminant validity of the
five-factor structure (Fornell & Larcker, 1981).
9
The second-order factor model showed good
fit (
2
[204] ⫽880.73; comparative fit index
(CFI) ⫽.96; goodness of fit index (GFI) ⫽
.89; standardized root mean square residual
(SRMR) ⫽.07; root mean square error of approx-
imation (RMSEA) ⫽.08), so we averaged scores
across dimensions to form an overall measure
of PLB.
Holistic thinking. We used six items from a scale
to measure locus of attention of holistic thinking
(Choi et al., 2007); for example, “It is more impor-
tant to pay attention to the whole than its parts,”
and “It is more important to pay attention to the
whole context rather than the details.” The alpha
coefficient was .70.
Integrative complexity. We used the traditional
method of content analyzing open-ended re-
sponses and the method of item ratings. We fol-
lowed Baker-Brown, Ballard, Bluck, de Vries,
Suedfeld, and Tetlock (1992) to code integrative
complexity by focusing on the structure of
thought rather than on its content. Low differen-
tiation indicates a tendency to focus on only one
theme in the analysis, or on simple, one-dimen-
sional rules for interpreting events or making
choices. High differentiation indicates that recog-
nizing and accepting alternative perspectives or
different dimensions are legitimate and valid.
High integration indicates conceptual links for
differentiated elements. We adopted a five-point
assessment (“1” ⫽absence of both differentia-
tion and integration, “3” ⫽presence of differen-
tiation, but absence of integration, “5” ⫽pres-
ence of both differentiation and integration). Two
trained coders independently assessed the inte-
grative complexity of responses to each of two
open-ended questions. The two questions as-
sessed generalized complexity by probing per-
ceptions of work: “Some people feel that organi-
zations waste too much time listening to different
points of view and opinions during group meet-
ings. Others feel they don’t spend enough time.
How do you feel? What do you think should be
the right balance?” (Tadmor et al., 2009), and
“Some feel that a work group should comprise
different people. Others feel that a work group
should comprise similar people. How do you
feel? What do you think should be the right
balance?”
Their interrater agreements on the two questions
were .85 and .80, respectively. After discussions,
the two coders agreed on all cases. We aggregated
the two scores on the two questions to formulate a
coded measure of integrative complexity.
Following Wong et al. (2011), we generated a
two-dimensional measure for integrative complex-
ity: five items for differentiation and six for inte-
gration, based on the Group Dynamics Q-sort mea-
sure (Peterson, Owens, Tetlock, Fan, & Martorana,
1998; Wong et al., 2011). Sample items for differen-
tiation include “I believe in the value of dissent” and
“I understand how there can always be divergent
viewpoints on certain issues.” Sample items for in-
tegration included “I believe that trade-offs can
be avoided when making decisions” and “When
there are different perspectives on an issue, I
often point out the common areas of overlap that
may serve to bridge these differences.” The reli-
ability coefficients were .76 and .73, respec-
tively. The second-order factor model showed
good fit (
2
[42] ⫽56.25, CFI ⫽.94, GFI ⫽.88,
SRMR ⫽.08, RMSEA ⫽.07). Thus, we aver-
aged scores across the two dimensions to form an
item measure of integrative complexity.
9
AVEs were UI ⫽.67, SO ⫽.54, CA ⫽.59,
RF ⫽.70, and DC ⫽.60. The square roots of AVE
(from .73 to .84) were larger than the correlation coeffi-
cients among the five factors (maximum value .49).
2015 555Zhang, Waldman, Han, and Li
Organizational structure. We followed Ambrose
and Schminke (2003) and Covin and Slevin (1989),
and used the seven-item scale of Khandwalla
(1976–77). Supervisors rated the degree to which
their organizations reflected mechanistic or organic
characteristics (“1” ⫽Tight formal control of most
operations by means of sophisticated control and
information systems, “7” ⫽Loose, informal con-
trol; heavy dependence on informal relationships
and norm of cooperation for getting things done).
Higher values represented a more organic structure
(Cronbach’s alpha ⫽.79).
Subordinate work role performance. We used
three three-item scales (Griffin et al., 2007) to mea-
sure proficient behavior (e.g., “carries out the core
parts of his or her job well”), adaptive behavior
(e.g., “adapts well to changes in core tasks”), and
proactive behavior (e.g., “initiates better ways of
doing core tasks”). The reliabilities were .79, .74,
and .77, respectively. The three-factor, CFA
model (
2
[24] ⫽109.58, CFI ⫽.98, GFI ⫽.96,
SRMR ⫽.04, RMSEA ⫽.08) had better fit than
the one-factor model (
2
[27] ⫽220.10, CFI ⫽.96,
GFI ⫽.92, SRMR ⫽.05, RMSEA ⫽.11; ⌬
2
[3] ⫽
110.52, p⬍.001), suggesting that the three measures
were separate constructs.
Control variables. We controlled for two types
of variable. First, we controlled for supervisors’ and
subordinates’ demographic characteristics that
might influence their behaviors, including age,
gender, and tenure as a supervisor (or tenure under
the current supervisor). To avoid the decrease of
degrees of freedom, we included subordinates’ de-
mographics for regressions involving their work
role behavior and included supervisors’ demo-
graphics for regressions involving their PLB. Sec-
ond, we controlled for power distance and rela-
tional orientation. Power distance emphasizes
hierarchical structure, while relational orientation
emphasizes harmonious and close social relation-
ship. Individuals with high power distance and
relational orientation are “required to maintain
close and well-structured relationships with oth-
ers” (Masuda & Nisbett, 2001: 923). Thus, the two
cultural values may influence supervisors’ re-
sponses toward the structural and subordinate de-
mands, and subordinates’ acceptance of PLB. We
used six items to measure power distance (e.g., “It
is frequently necessary for a manager to use author-
ity and power when dealing with subordinates”
(Dorfman & Howell, 1988)) and seven items to mea-
sure relational orientation (e.g., “It is important for
me to maintain social relations with others” (Vos,
van der Zee, & Buunk, 2012)). Reliabilities
were .82 and .83, respectively, for the subordinate
sample, and .86 and .82, respectively, for the su-
pervisor sample.
Analyses
Before testing the hypotheses, we aggregated PLB
from the individual, subordinate level to the super-
visory level, because PLB is theoretically meaning-
ful as a supervisory- or group-level phenomenon.
To test the appropriateness of the aggregation
(Bliese, 2000), we calculated the interrater agree-
ment coefficient (R
wg
) and intraclass correlation
coefficients (ICC1, ICC2) (Bliese, 2000; James, De-
maree, & Wolf, 1984). The median R
wg
was .92, the
ICC1 value was .29, and the ICC2 value was .74,
suggesting appropriateness of aggregation. We also
aggregated the organizational structure to the firm
level. The median R
wg
was .83, the ICC1 value
was .32, and the ICC2 value was .85. Those aggre-
gation indices were acceptable.
We used hierarchical linear modeling (HLM) to
conduct analyses, because the data were nested in
terms of supervisors and firms. More specifically,
we used two-level HLM (organization level and
supervisor level) to analyze the effects of holistic
thinking, integrative complexity, and organic
structure on PLB, and used three-level HLM (or-
ganization level, supervisor level, and subordi-
nate level) to analyze the effects of PLB on sub-
ordinate behaviors. We group mean-centered all
individual-level (level 1) variables and grand
mean-centered interaction variables (Hofmann &
Gavin, 1998) before entering them into the equa-
tions. We also computed pseudo R
2
(which rep-
resents the percentage of the total variance in the
dependent variable accounted for by the added
predictors) to show effect sizes for multilevel
analyses (Snijders & Bosker, 1999; Zhang, Wald-
man, & Wang, 2012).
Results and Discussion
Table 7 shows correlations at both supervisory
and subordinate levels. At the supervisory level,
PLB correlates positively with holistic thinking
(r⫽.31, p⬍.01), the coded measure of integra-
tive complexity (r⫽.27, p⬍.05), the item
measure of complexity (r⫽.28, p⬍.05), and
organic structure (r⫽.25, p⬍.05). At the sub-
ordinate level, PLB correlates positively with
subordinate proficient (r⫽.25, p⬍.01), adap-
556 AprilAcademy of Management Journal
tive (r⫽.34, p⬍.01), and proactive behavior
(r⫽.34, p⬍.01).
In the HLM analyses, we first estimated a fully
unconditional, intercept-only model for PLB and
subordinate behavior to examine within-group
and between-group variability. We found signif-
icant between-firm variance in groups’ PLB
(
00
⫽.02,
2
[5] ⫽12.09, p⫽.03) and subor-
dinate work role behavior (proficient behavior:
00
⫽.17,
2
[5] ⫽17.39, p⬍.005; adaptive be-
havior:
00
⫽.17,
2
[5] ⫽27.63, p⬍.001; pro-
active behavior:
00
⫽.15,
2
[5] ⫽40.25,
p⬍.001). We also found significant between-
group and within-group variances for subordi-
nate behavior (proficient behavior:
2
⫽.30,
2
[70] ⫽255.24, p⬍.001; adaptive behavior:
2
⫽.24,
2
[70] ⫽192.17, p⬍.001; proactive
behavior:
2
⫽.31,
2
[70] ⫽155.04, p⬍.001).
Those variances demonstrate the data’s nested
nature and justified our use of multilevel
analyses.
Next, we introduced all subordinate-, supervi-
sory- or group-, and firm-level control variables
(Step 1), followed by the predictor variable(s)
(Step 2). Tables 8 and 9 show the unstandardized
coefficients, as well as the explained variances
for each step. Holistic thinking and integrative
complexity (coded measure) predicted PLB
(Step 2 in Table 8:
␥
⫽.19, p⬍.01;
␥
⫽0.15,
p⬍.05). Pseudo R
2
was .23. Thus, Hypotheses 1
and 2 were supported. Organic structure was insig-
nificant (
␥
⫽.12, p⫽.12), thus rejecting Hypothe-
sis 3. Paradoxical leader behavior significantly pre-
dicted subordinate proficient behavior (Step 2 in
Table 9:
␥
⫽.35, p⬍.05), adaptive behavior
(
␥
⫽.43, p⬍.01), and proactive behavior (
␥
⫽.40,
p⬍.01). Pseudo R
2
values were .09, .10, and .09,
supporting Hypotheses 4a– 4c.
TABLE 7
Means, Standard Deviations, and Correlation Coefficients of Variables for HLM
Supervisory-Level Variables Mean SD 1234 5 678910
1. Age 34.87 8.03
2. Gender
a
1.36 0.48 .03
3. Work tenure as a supervisor 4.13 2.47 .37** ⫺.18
4. Power distance 3.32 1.18 ⫺.06 ⫺.15 ⫺.03 .86
5. Relational orientation 4.37 0.96 ⫺.11 ⫺.09 .08 .08 .82
6. Holistic thinking 4.45 0.68 ⫺.11 .02 ⫺.19 .30** .07 .70
7. Integrative complexity (coded measure) 2.29 0.82 ⫺.05 .10 ⫺.19 ⫺.21 .03 ⫺.06 —
8. Integrative complexity (item measure) 4.59 0.61 .00 .21 ⫺.10 ⫺.18 .30** .23
†
.22
†
.84
9. Organic structure
b
4.19 1.13 ⫺.18 ⫺.13 .15 .13 .23* .19 .02 .03 .79
10. PLB in people management 2.55 0.43 ⫺.18 .05 ⫺.17 ⫺.16 ⫺.00 .31** .27* .28* .25* .94
TABLE 7
(continued)
Subordinate-Level Variables Mean SD 123456789
1. Age 29.49 8.25
2. Gender 1.47 0.50 ⫺.11**
3. Work tenure under the current supervisor 2.25 1.67 .46** ⫺.06
4. Power distance 3.20 1.13 ⫺.06 ⫺.04 .03 .82
5. Relational orientation 4.64 0.81 ⫺.12** .06 ⫺.06 .16** .83
6. PLB 2.52 0.41 ⫺.26** .08 ⫺.14** ⫺.03 .13** .91
7. Proficient behavior 4.50 0.73 ⫺.15** ⫺.12** ⫺.05 .03 .14** .25** .79
8. Adaptive behavior 4.43 0.71 ⫺.36** ⫺.02 ⫺.07 .08 .13** .34** .66** .74
9. Proactive behavior 4.38 0.79 ⫺.33** ⫺.06 ⫺.14** .14** .13** .34** .57** .68** .77
Note: n (supervisory level) ⫽76; n(subordinate level) ⫽516. Reliability coefficients are reported in bold italic along the diagonal.
a
“1” ⫽male, “2” ⫽female.
b
Organic structure is a firm-level variable.
†
p⬍.10
*p⬍.05
** p⬍.01 (two-tailed)
2015 557Zhang, Waldman, Han, and Li
One unexpected finding is that organizational
structure was an insignificant antecedent of PLB,
most likely because supervisors did the rating.
Their demographic and value-orientation variables
may have affected the role of supervisor-rated, or-
ganizational structure in predicting supervisory
paradoxical behavior, since organic structure was
significant in HLM analyses with the control vari-
ables excluded. Another reason might be that the
small, firm-level sample size yielded unstable re-
sults. Previous literature has emphasized that or-
ganizational contexts are important to the emer-
gence of specific leadership (e.g., Shamir & Howell,
1999). Future researchers might use HR managers
to rate firm-level variables, or they might collect
data from more firms.
In sum, we used a limited set of variables to
examine a nomological network of PLB in Study 2.
Future research could expand the nomological net-
work to further validate PLB. For example, beyond
holistic thinking and integrative complexity as an-
tecedents, future researchers might include con-
structs such as open-mindedness, learning orienta-
tion, and cultural values, including uncertainty
avoidance and long-term orientation, as well as
contextual factors such as environmental uncer-
tainty and complexity, type of work, job character-
istics, and firm and industry types. Leaders who
engage in paradoxical behavior may affect not only
follower work-role behaviors, but also group-level
states and outcomes, such as group efficiency and
innovation.
GENERAL DISCUSSION
Based on yin–yang philosophy, we conceptual-
ized the construct of “paradoxical leader behavior”
and examined how effectively supervisory leaders
who engage in paradoxical behavior deal with par-
adoxical challenges in managing people. We used
five independent samples to validate a new mea-
sure of PLB in people management. Results provide
robust evidence of our measure’s construct validity
and predictive capability beyond established West-
ern and Eastern leadership behaviors. We further
tested a nomological network model for PLB, and
confirmed that holistic thinking and integrative
complexity are significant predictors. As expected,
the ultimate display of PLB has a strong cognitive
basis. In addition, we found evidence connecting
PLB to follower work role proficiency, adaptivity,
and proactivity.
Theoretical Contributions
Our study provides new directions in studying
organizational paradoxes and addressing effective
leadership practices. Although organizational par-
adoxes have been considered since the 1980s, they
have been categorized largely as macro-level stra-
tegic challenges (e.g., Smith & Lewis, 2011). We
contribute theoretically by showing that paradoxes
TABLE 8
HLM Regression Results Predicting Paradoxical
Leader Behavior in People Management
Dependent Variable PLB
Step 0: Null model
- Intercept 2.55 (.06)***
Variance of the dependent variable .178
-
2
.161
-
.017
Step 1: Control variables
a
Supervisory-level variables:
Age .00 (.01)
Gender .06 (.10)
Work tenure as a supervisor ⫺.03 (.03)
Power distance ⫺.11 (.04)*
Relational orientation ⫺.03 (.05)
Variance unexplained in the dependent variable .173
-
2
.155
-
.017
Pseudo R
2b
.03
Step 2: Independent variables
Supervisory-level variables:
Holistic thinking .19 (.06)**
Integrative complexity
c
.15 (.05)*
Firm-level variables:
Organic structure .12 (.06)
Variance unexplained in the dependent variable .137
-
2
.133
-
.004
Pseudo R
2
.23
Note: n (firm level) ⫽6; n(supervisory level) ⫽76.
a
Following an anonymous reviewer’s suggestion, we ana-
lyzed the models without control variables. Holistic thinking
(
␥
⫽.17 (.05), p⫽.002), integrative complexity (
␥
⫽.16
(.04), p⫽.001), and organic structure (
␥
⫽.13 (.04),
p⫽.039) significantly predict PLB.
b
Pseudo R
2
⫽(Variance of the dependent variable in null
model ⫺Variance of the dependent variable in the current
model) ⫺Variance of the dependent variable in null model
(Zhang et al., 2012).
c
We report the results of using the coded measure of integra-
tive complexity. Using the item measure of integrative complex-
ity showed a similar pattern to using the coded measure of
integrative complexity.
*p⬍.05
** p⬍.01
*** p⬍.001 (two-tailed)
558 AprilAcademy of Management Journal
are also relevant for leaders facing people manage-
ment issues at micro levels—that is, that leaders
face paradoxes inherent in structural and follower
demands.
We further deviate from the current leadership
literature in showing how leaders may approach
paradoxes when they manage employees. Leaders
who use contingency-based action (e.g., Smith &
Lewis, 2011) emphasize either pole of a paradox
separately, depending on the situation. However,
in line with Eastern yin–yang philosophy and some
Western-based approaches to paradoxes (e.g.,
Lüscher & Lewis, 2008; Smith & Tushman, 2005),
we diverge from contingency-based action to show
that effective PLB reflects a “both–and” strategy
that behaviorally accepts and integrates competing
demands simultaneously over time.
Our framework addresses the increasing com-
plexities in organizational and environmental con-
texts. Preeminent leadership models over the last
century may be effective for relatively simple and
stable contexts, but complex and dynamic environ-
ments call for new leadership paradigms (Uhl-Bien,
Marion, & McKelvey, 2007). Structural and fol-
lower demands tend to breed paradoxes requiring
more than simple, linear, and static management
TABLE 9
HLM Regression Results Predicting Subordinate Behavior
Dependent Variables Proficient Behavior Adaptive Behavior Proactive Behavior
Step 0: Null model
Intercept 4.46 (.10)*** 4.38 (.12)*** 4.35 (.14)***
Variance of the dependent variable .472 .409 .465
-
2
.302 .244 .313
-
.170 .165 .152
Step 1: Control variablesa
Subordinate-level variables:
Age .01 (.00)** ⫺.01 (.00) ⫺.00 (.00)
Gender ⫺.00 (.06) .07 (.06) ⫺.02 (.07)
Work tenure under the current supervisor .05 (.02)** .06 (.02)** .01 (.02)
Power distance .01 (.03) .03 (.03) .06 (.03)
Relational orientation .08 (.03)* .07 (.03)* .04 (.04)
Supervisory-level variables:
Holistic thinking ⫺.11 (.09) ⫺.07 (.09) ⫺.07 (.08)
Integrative complexity .09 (.07) .10 (.07) .15 (.07)*
Firm-level variable:
Organic structure .25 (.08)* .30 (.09)* .30 (.13)
†
Variance unexplained in the dependent variable .451 .393 .445
-
2
.285 .231 .309
-
.166 .162 .115
Pseudo R
2
.04 .04 .04
Step 2: Independent variable
Supervisory-level variable:
PLB in people management
b
.35 (.14)* .43 (.13)** .40 (.13)**
Variance unexplained in the dependent variable .431 .370 .424
-
2
.285 .231 .309
-
.146 .139 .115
Pseudo R
2
.09 .10 .09
Note: n (firm level) ⫽6; n(supervisory level) ⫽76; n(subordinate level) ⫽516.
a
Following an anonymous reviewer’s suggestion, we analyzed the models without control variables. PLB had significant effects on
proficient (
␥
⫽.32 (.13), p⫽.015), adaptive (
␥
⫽.39 (.12), p⫽.003), and proactive (
␥
⫽.38 (.12), p⫽.003) behaviors.
b
Following an anonymous reviewer’s suggestion, we included group-mean-centered PLB at the subordinate level in the HLM. The
results of the leader behavior measure at the group level were consistent with the results of models that did not involve the leader behavior
variable at the subordinate level. Moreover, leader behavior at the subordinate level did not have significant effects on outcomes. We
omitted them in this table because we conceptually frame PLB at the group level.
†
p⬍.10
*p⬍.05
** p⬍.01
*** p⬍.001 (two-tailed)
2015 559Zhang, Waldman, Han, and Li
approaches. Paradoxically oriented leaders may
position themselves more effectively in complex
environments. By responding to paradoxical forces,
they dynamically coexist with environmental com-
plexities over time. In short, a PLB framework may
broaden our perspective regarding leadership mod-
els in complex and dynamic environments.
Lastly, cognitive thinking is integral to PLBs.
Such behaviors may appear contradictory to man-
agers who do not embrace holistic thinking or who
are less integratively complex. Managers who have
such a “both–and” cognitive basis instead view
seemingly contradictory behaviors as integrative.
Although a number of studies stress cognitive
differences between East Asian cultures and
North American cultures (Nisbett et al., 2001),
holistic thinking and integrative complexity are
universal in that they are important information-
processing approaches (Fific´ & Townsend, 2010;
Koo, Han, & Kim, 2002). Thus, Western managers
can be holistic, integrative, and effective in man-
aging paradoxes.
Limitations and Future Research
Future research will benefit from considering
several limitations of our study, along with our
suggestions for advancing PLB research. First, fu-
ture research should target PLB in Western contexts
(Zhang, Chen, Chen, & Ang, 2014). We examined
PLB in a Chinese context, so our findings may
largely reflect Chinese-specific ways of integrating
opposites and achieving positive subordinate out-
comes. Future research might address how well
Western leaders display the same leader behav-
iors and achieve similar outcomes. Scholars
might conduct cross-cultural research to explore
whether PLB is effective in other cultural con-
texts and whether leaders within those contexts
display somewhat different behaviors in manag-
ing paradoxes.
Second, we focused on a theoretically identi-
fied set of paradoxical leader behaviors in people
management. Future research might expand par-
adoxical leader behavior to other management
domains. For example, strategic leaders may
need to display paradoxical behavior in manag-
ing stakeholders to reconcile the interests and
demands of seemingly disparate stakeholder
groups, such as owners, employees, suppliers,
customers, and community. In globalization
management, executive leaders often confront
conflicting demands from their headquarters and
host countries. To succeed in China, for example,
paradoxical leaders may “understand the market
but work with the states...adapt to local condi-
tions, but implement global standards...drive
costs down but maintain quality” (Paine, 2010:
104–105).
Third, PLB could potentially cause some nega-
tive consequences. Behaving paradoxically may tax
leaders’ cognitive resources, because they must si-
multaneously consider multiple competing de-
mands and seek ways in which to integrate those
demands effectively. Consequently, paradoxically
oriented leaders may experience more psychologi-
cal stress. Meanwhile, PLB effects on subordinates
may depend somewhat on whether subordinates
themselves endorse yin–yang perspectives. Subor-
dinates embracing yin–yang perspectives are likely
to appreciate PLB, but those who have “black–
white,” “either–or” views may be uncomfortable
about following the paradoxical leader (DeRue &
Ashford, 2010). Therefore, future research could
explore the potential downsides of PLB.
Fourth, our research used a paradox perspective
to understand the competing demands that super-
visory leaders face in people management. That
perspective may help us to better understand some
workplace conflicts. For example, when employees
undertake their work roles, they may confront com-
peting demands from peers, customers, and their
own family members. Instead of making “either–
or” judgments, the paradox approach suggests ho-
listically understanding and integrating tensions
between interpersonal cooperation and competi-
tion, between customers and firms, and between
work and family. Moreover, we may be better po-
sitioned to integrate seeming contradictions in
other workplace phenomena. For example, a par-
adox lens may help managers to recognize and
manage the simultaneous value and potential
dysfunction of risk taking, diversity, turnover,
organizational culture, and mergers and acquisi-
tions. In short, research should determine how a
paradox perspective may allow a more compre-
hensive understanding of a range of work issues
and phenomena.
Fifth, we offer a new approach to the assessment
of paradoxical behavior. Our double-barreled item
design abandons the bipolar assumptions of tradi-
tional psychometric scales (Bobko, 1985), and ap-
propriately captures the fundamental “both–and”
nature of paradoxes. Although we suggest that our
approach is likely to advance research, we also
recommend continued scrutiny and testing, per-
560 AprilAcademy of Management Journal
haps comparing our method with other approaches
for examining PLB.
Finally, PLB explained relatively small, addi-
tional variance over the alternate leadership con-
structs that we tested. Future researchers should
further examine the incremental validity of PLB.
The validation samples in Study 1 included eve-
ning students from a single university, which may
have biased our findings. Moreover, the field sam-
ple respondents were predominantly college edu-
cated and thus represent a professional work pop-
ulation in China. Future research could explore
whether the findings generalize to other work pop-
ulations, such as factory or service workers. Addi-
tionally, the relationship between PLB, subordinate
commitment, and leave intention in Study 1
should be interpreted cautiously because of poten-
tial common method effects.
Practical Implications and Conclusion
Paradoxical leader behavior may enlighten prac-
titioners. International firms may particularly feel
the effects of uncertainties and complexities when
operating in global or novel national contexts. A
paradox-based lens can help managers to better
understand how to deal with increasing uncertain-
ties that often involve competing possibilities in
terms of managerial actions. Not surprisingly, deal-
ing with paradoxes is becoming an increasingly
important skill for leaders at various organizational
levels (e.g., Clegg et al., 2002; Evans, 2000; Smith &
Tushman, 2005).
Management and organization theories have
lacked a framework for showing how paradox per-
spectives might provide insights for leadership
practice (Clegg et al., 2002). Instead, leaders have
been encouraged to think situationally and to
choose among behavioral alternatives, depending
on variables such as aspects of subordinates or
organizational contexts (Vroom & Jago, 2007). Such
contingency thinking allows leaders to deal with
short-term status quo demands, but they should
also be trained to consider long-term behavioral
strategies for handling seemingly disparate, para-
doxical poles. In short, we suggest that leaders
should be trained to handle both short-term contin-
gencies and long-term perspectives, focusing
largely on paradoxical thinking and actions.
In conclusion, by connecting paradox and lead-
ership theories, we provide a new lens for iden-
tifying, diagnosing, and resolving organizational
issues. Furthermore, by bridging Eastern and
Western thinking in forming such connections,
we contribute uniquely to management research
and practice.
REFERENCES
Ambrose, M. L., & Schminke, M. 2003. Organization
structure as a moderator of the relationship between
procedural justice, interactional justice, perceived
organizational support, and supervisory trust. Jour-
nal of Applied Psychology, 88: 295–305.
Antonakis, J., & Atwater, L. 2002. Leader distance: A
review and a proposed theory. Leadership Quar-
terly, 13: 673–704.
Baker-Brown, G., Ballard, E. J., Bluck, S., de Vries, B.,
Suedfeld, P., & Tetlock, P. E. 1992. The conceptual/
integrative complexity scoring manual. In C. P.
Smith (Ed.), Motivation and personality: Handbook
of thematic content analysis: 401–418. Cambridge:
Cambridge University Press.
Bass, B. M., & Avolio, B. J. 1995. MLQ multifactor lead-
ership questionnaire for research. Redwood City,
CA: Mindgarden.
Bliese, P. 2000. Within-group agreement, nonindepen-
dence, and reliability. In K. Klein & S. Kozlowski
(Eds.), Multi-level theory, research, and methods in
organizations: 349–381. San Francisco: Jossey-Bass.
Bligh, M. C., Pearce, C. L., & Kohles, J. C. 2006. The
importance of self- and shared leadership in team-
based knowledge work: A meso-level model of lead-
ership dynamics. Journal of Managerial Psychol-
ogy, 21: 296–318.
Bobko, P. 1985. Removing assumptions of bipolarity: To-
wards variation and circularity. Academy of Man-
agement Review, 10: 99–108.
Boies, K., & Howell, J. M. 2006. Leader–member ex-
change in teams: An examination of the interaction
between relationship differentiation and mean LMX
in explaining team-level outcomes. Leadership
Quarterly, 17: 246–257.
Bollen, K. A. 1989. Structural equations with latent
variables. New York: Wiley.
Brewer, M. B. 1991. The social self: On being the same
and different at the same time. Personality and So-
cial Psychology Bulletin, 17: 475–482.
Brewer, M. B., & Gardner, W. 1996. Who is this “we”?
Levels of collective identity and self representations.
Journal of Personality and Social Psychology, 71:
83–93.
Brislin, R. W. 1980. Translation and content analysis of
oral and written material. In H. C. Triandis & J. W.
Berry (Eds.), Handbook of cross-cultural psychol-
2015 561Zhang, Waldman, Han, and Li
ogy: Methodology, vol. 2: 389–444. Boston, MA:
Allyn & Bacon.
Brutus, S., & Facteau, J. 2003. Short, simple, and specific:
The influence of item design characteristics in multi-
source assessment contexts. International Journal
of Selection and Assessment, 11: 313–325.
Cameron, K. S., & Quinn, R. E. 1999. Diagnosing and
changing organizational culture. Reading, MA: Ad-
dison Wesley.
Capelli, P., & Sherer, P. 1991. The missing role of context
in OB: The need for a meso-level approach. Re-
search in Organizational Behavior, 13: 55–110.
Chen, M.-J. 2002. Transcending paradox: The Chinese
middle way perspective. Asia Pacific Journal of
Management, 19: 179–199.
Chen, M.-J. 2008. Reconceptualizing the competition–
cooperation relationship: A trans-paradox perspec-
tive. Journal of Management Inquiry, 17: 288–304.
Chen, X.-P., Xie, X., & Chang, S. 2011. Cooperative and
competitive orientation among Chinese people:
Scale development and validation. Management
and Organization Review, 7: 353–379.
Cheng, B. S., Chou, L. F., & Farh, J. L. 2000. A triad model
of paternalistic leadership: Constructs and measure-
ment. Indigenous Psychological Research in Chi-
nese Societies, 14: 3–64 [In Chinese].
Choi, I., & Nisbett, R. E. 2000. Cultural psychology of
surprise: Holistic theories and recognition of contra-
diction. Journal of Personality and Social Psychol-
ogy, 79: 890–905.
Choi, I., Koo, M., & Choi, J. A. 2007. Individual differ-
ences in analytic versus holistic thinking. Personal-
ity and Social Psychology Bulletin, 33: 691–705.
Clegg, S. R., Cuhna, J. V., & Cuhna, M. P. 2002. Manage-
ment paradoxes: A relational view. Human Rela-
tions, 55: 483–503.
Courtright, J. A., Fairhurst, G. T., & Rogers, L. E. 1989.
Interaction patterns in organic and mechanistic sys-
tem. Academy of Management Journal, 32: 773–
802.
Covin, J. G., & Slevin, D. P. 1989. Strategic management
of small firms in hostile and benign environments.
Strategic Management Journal, 10: 75–87.
Day, D. V., Gronn, P., & Salas, E. 2004. Leadership capac-
ity in teams. Leadership Quarterly, 15: 857–880.
De Jong, A., & de Ruyter, K. 2004. Adaptive versus pro-
active behavior in service recovery: The role of self-
managing teams. Decision Sciences, 35: 457–491.
Denis, J., Langley, A., & Sergi, V. 2012. Leadership in the
plural. Academy of Management Annals, 6: 211–
283.
Denison, D. R., Hooijberg, R., & Quinn, R. E. 1995. Para-
dox and performance: Toward a theory of behavioral
complexity in managerial leadership. Organization
Science, 6: 524–540.
DeRue, D. S., & Ashford, S. J. 2010. Who will lead and
who will follow? A social process of leadership iden-
tity construction in organizations. Academy of Man-
agement Review, 35: 627–647.
Detert, J. R., & Burris, E. R. 2007. Leadership behavior
and employee voice: Is the door really open? Acad-
emy of Management Journal, 50: 869–884.
Donaldson, L. 1996. Four positivist organization theory.
Thousand Oaks, CA: Sage.
Dorfman, P. W., & Howell, J. P. 1988. Dimensions of
national culture and effective leadership patterns:
Hofstede revisited. Advances in International Com-
parative Management, 3: 127–150.
Evans, P. A. L. 2000. The dualistic leader: Thriving on
paradox. In S. Chowdhury (Ed.), Management 21C:
New visions for the new millennium: 66–82. New
York, NY/London, UK: Prentice Hall/Financial
Times.
Fang, T. 2005. From “onion” to “ocean” paradox and
change in national cultures. International Studies
of Management and Organization, 35: 71–90.
Fang, T. 2010. Asian management research needs more
self-confidence: Reflection on Hofstede (2007) and
beyond. Asia Pacific Journal of Management, 27:
155–170.
Fang, T. 2012. Yin yang: A new perspective on culture.
Management and Organization Review, 8: 25–50.
Farh, J. L., Hackett, R. D., & Liang, J. 2007. Individual-
level cultural values as moderators of perceived or-
ganizational support–employee outcome relation-
ships in China: Comparing the effects of power
distance and traditionality. Academy of Manage-
ment Journal, 50: 715–729.
Feldman, S. P. 1989. The broken wheel: The inseparabil-
ity of autonomy and control in innovation within
organizations. Journal of Management Studies, 26:
83–102.
Fific´ , M., & Townsend, J. T. 2010. Information-processing
alternatives to holistic perception: Identifying the
mechanisms of secondary-level holism within a cat-
egorization paradigm. Journal of Experimental Psy-
chology, 36: 1290–1313.
Ford, J. D., & Slocum, J. W. 1977. Size, technology, envi-
ronment and the structure of organizations. Acad-
emy of Management Review, 2: 561–575.
Fornell, C., & Larcker, D. F. 1981. Evaluating structural
equation models with unobservable variables and
measurement error. Journal of Marketing Research,
18: 39–50.
562 AprilAcademy of Management Journal
Galvin, B. M., Waldman, D. A., & Balthazard, P. 2010.
Visionary communication qualities as mediators of
the relationship between narcissism and attributions
of leader charisma. Personnel Psychology, 63: 509–
537.
Gerbing, D. W., & Anderson, J. C. 1984. On the meaning
of within-factor correlated measurement errors.
Journal of Consumer Research, 11: 572–580.
Gerstner, C. R., & Day, D. V. 1997. Meta-analytic review
of leader–member exchange theory: Correlates and
construct issues. Journal of Applied Psychology, 82:
827–844.
Gibson, C. B., & Birkinshaw, J. 2004. The antecedents,
consequences, and mediating role of organizational
ambidexterity. Academy of Management Journal,
47: 209–226.
Graen, G. B., & Scandura, T. 1987. Toward a psychology
of dyadic organizing. Research in Organizational
Behavior, 9: 175–208.
Griffin, M. A., Neal, A., & Parker, S. K. 2007. A new
model of work role performance: Positive behavior
in uncertain and interdependent contexts. Academy
of Management Journal, 50: 327–347.
Handy, C. 1994. The age of paradox. Cambridge, MA:
Harvard Business School Press.
Hoch, J. E., & Dulebohn, J. H. 2013. Shared leadership in
enterprise resource planning and human resource
management system implementation. Human Re-
source Management Review, 23: 114–125.
Hofmann, D. A., & Gavin, M. B. 1998. Centering decisions
in hierarchical linear models: Implications for re-
search in organizations. Journal of Management,
24: 623–641.
Hooijberg, R. 1996. A multidirectional approach toward
leadership: An extension of the concept of behav-
ioral complexity. Human Relations, 49: 917–946.
Howell, J. M., Neufeld, D. J., & Avolio, B. J. 2005. Exam-
ining the relationship of leadership and physical
distance with business unit performance. Leader-
ship Quarterly, 16: 273–285.
James, L. R., Demaree, R. G., & Wolf, G. 1984. Estimating
within-group interrater reliability with and without
response bias. Journal of Applied Psychology, 69:
85–98.
Jarvis, C., MacKenzie, S., & Podsakoff, P. 2003. A critical
review of construct indicators and measurement
model misspecification in marketing and consumer
research. Journal of Consumer Research, 30: 199–
218.
Jones, T. M., Felps, W., & Bigley, G. A. 2007. Ethical
theory and stakeholder-related decisions: The role of
stakeholder culture. Academy of Management Re-
view, 32: 137–155.
Judge, T. A., Erez, A., Bono, J. E., & Thoresen, C. J. 2003.
The core self-evaluations scale: Development of a
measure. Personnel Psychology, 56: 303–331.
Judge, T. A., Piccolo, R. F., & Kosalka, T. 2009. The bright
and dark sides of leader traits: A review and theoret-
ical extension of the leader trait paradigm. Leader-
ship Quarterly, 20: 855–875.
Kaiser, R. B., & Craig, S. B. 2005. Building a better mouse-
trap: Item characteristics associated with discrepan-
cies in 360-degree feedback. Consulting Psychology
Journal: Research and Practice, 57: 235–245.
Kaiser, R. B., Lindberg, J. T., & Craig, S. B. 2007. Assess-
ing the flexibility of managers: A comparison of
methods. International Journal of Selection and
Assessment, 15: 40–55.
Kelleher, H. 1997. A culture of commitment. Leader to
Leader, 4: 20–24.
Khandwalla, P. N. 1976–77. Some top management
styles, their context and performance. Organization
and Administrative Sciences, 7: 21–51.
Koo, J., Han, J., & Kim, J. 2002. Integrative complexity of
South–North Korean correspondences: A time-series
analysis, 1984–1994. Journal of Conflict Resolu-
tion, 46: 286–304.
Koo, M., & Choi, I. 2005. Becoming a holistic thinker:
Training effect of oriental medicine on reasoning.
Personality and Social Psychology Bulletin, 31:
1264–1272.
Kreiner, G. E., Hollensbe, E. C., & Sheep, M. L. 2006.
Where is the “me” among the “we”? Identity work
and the search for optimal balance. Academy of
Management Journal, 49: 1031–1057.
Lawrence, K. A., Lenk, P., & Quinn, R. E. 2009. Behav-
ioral complexity in leadership: The psychometric
properties of a new instrument to measure behav-
ioral repertoire. Leadership Quarterly, 20: 87–102.
Lee, Y. T. 2000. What is missing in Chinese–Western
dialectical reasoning? American Psychologist, 55:
1065–1066.
Lee, Y. T., Han, A. G., Byron, T. K., & Fan, H. X. 2008.
Daoist leadership, theory and application. In C. C.
Chen & Y. T. Lee (Eds.), Leadership and manage-
ment in China: Philosophies, theories, and prac-
tices: 83–107. New York: Cambridge University
Press.
Lewis, M. W. 2000. Exploring paradox: Toward a more
comprehensive guide. Academy of Management
Review, 25: 760–776.
Li, P. P. 1998. Towards a geocentric framework of organ-
izational form: A holistic, dynamic and paradoxical
approach. Organization Studies, 19: 829–861.
Li, P. P. 2012. Toward an integrative framework of indig-
2015 563Zhang, Waldman, Han, and Li
enous research: The geocentric implications of yin–
yang balance. Asia Pacific Journal of Management,
29: 849–872.
Li, P. P., Leung, K., Chen, C. C., & Luo, J. 2012. Indige-
nous research on Chinese management: What and
how? Management and Organization Review, 8:
7–24.
Lubatkin, M. H., Simsek, Z., Ling, Y., & Veiga, J. F. 2006.
Ambidexterity and performance in small- to medi-
um-sized firms: The pivotal role of top management
team behavioral integration. Journal of Manage-
ment, 32: 646–672.
Lüscher, L. S., & Lewis, M. W. 2008. Organizational
change and managerial sensemaking: Working
through paradox. Academy of Management Jour-
nal, 51: 221–240.
Maccoby, M. 2004. Narcissistic leaders: The incredible
pros, the inevitable cons. Harvard Business Review,
82: 92–101.
Mael, F., & Ashforth, B. E. 1992. Alumni and their alma
mater: A partial test of the reformulated model of
organizational identification. Journal of Organiza-
tional Behavior, 13: 103–123.
Margolis, J. D., & Walsh, J. P. 2003. Misery loves compa-
nies: Rethinking social initiatives by business. Ad-
ministrative Science Quarterly, 48: 268–305.
Masuda, T., & Nisbett, R. E. 2001. Attending holistically
versus analytically: Comparing the context sensitiv-
ity of Japanese and Americans. Journal of Personal-
ity and Social Psychology, 81: 922–934.
Meyer, J. P., Allen, N. J., & Smith, C. A. 1993. Commit-
ment to organizations and occupations: Extension
and test of a three-component conceptualization.
Journal of Applied Psychology, 78: 538–551.
Murphy, P. R., & Jackson, S. E. 1999. Managing work-role
performance: Challenges for 21st century organiza-
tions and employees. In D. R. Ilgen & E. D. Pulakos
(Eds.), The changing nature of work performance:
325–365. San Francisco: Jossey-Bass.
Nisbett, R. E., Peng, K., Choi, I., & Norenzayan, A. 2001.
Culture and systems of thought: Holistic versus an-
alytic cognition. Psychological Review, 108: 291–
301.
Ouchi, W. G. 1978. The transmission of control through
organizational hierarchy. Academy of Management
Journal, 21: 173–192.
Owens, B. P., Johnson, M. D., & Mitchell, T. R. 2013.
Expressed humility in organizations: Implications
for performance, teams, and leadership. Organiza-
tion Science, 24: 1517–1538.
Paine, L. S. 2010. The China rules. Harvard Business
Review, 88: 103–108.
Parker, S. K., Williams, H. M., & Turner, N. 2006. Mod-
eling the antecedents of proactive behavior at work.
Journal of Applied Psychology, 91: 636–652.
Pawar, B. S., & Eastman, K. K. 1997. The nature and
implications of contextual influences on transforma-
tional leadership: A conceptual examination. Acad-
emy of Management Review, 22: 80–109.
Pearce, C. L., & Conger, J. A. 2003. All those years ago:
The historical underpinnings of shared leadership.
In C. L. Pearce & J. A. Conger (Eds.), Shared lead-
ership: Reframing the how’s and whys of leader-
ship: 1–18. Thousand Oaks, CA: Sage.
Peng, K., & Nisbett, R. E. 1999. Culture, dialectics, and
reasoning about contradiction. American Psycholo-
gist, 54: 741–754.
Peterson, R. S., Owens, P. D., Tetlock, P. E., Fan, E. T., &
Martorana, P. 1998. Group dynamics in top manage-
ment teams: Groupthink, vigilance, and alternative
models of organizational failure and success. Organ-
izational Behavior and Human Decision Processes,
73: 272–305.
Pillai, R. 1995. Context and charisma: The role of organic
structure, collectivism, and crisis in the emergence
of charismatic leadership. Academy of Manage-
ment Proceedings, 1995: 332–336.
Pillai, R., & Meindl, J. R. 1998. Context and charisma: A
“meso” level examination of the relationship of or-
ganic structure, collectivism, and crisis to charis-
matic leadership. Journal of Management, 24: 643–
671.
Poole, M. S., & Van de Ven, A. H. 1989. Using paradox to
build management and organization theories. Acad-
emy of Management Review, 14: 562–578.
Porter, L. W., & McLaughlin, G. B. 2006. Leadership and
the organizational context: Like the weather? Lead-
ership Quarterly, 17: 559–576.
Pulakos, E. D., Arad, S., Donovan, M. A., & Plamondon,
K. E. 2000. Adaptability in the workplace: Develop-
ment of a taxonomy of adaptive performance. Jour-
nal of Applied Psychology, 85: 612–624.
Quinn, R. E., & Cameron, K. S. 1988. Paradox and trans-
formation: Toward a theory of change in organi-
zation and management. Cambridge, MA: Ball-
inger.
Rich, B. L., LePine, J. A., & Crawford, E. R. 2010. Job
engagement: Antecedents and effects on job perfor-
mance. Academy of Management Journal, 53: 617–
635.
Rodan, S., & Galunic, C. 2004. More than network struc-
ture: How knowledge heterogeneity influences man-
agerial performance and innovativeness. Strategic
Management Journal, 25: 541–562.
564 AprilAcademy of Management Journal
Rosenthal, S. A., & Pittinsky, T. L. 2006. Narcissistic
leadership. Leadership Quarterly, 17: 617–633.
Sagie, A. 1997. Leader direction and employee participa-
tion in decision making: Contradictory or compati-
ble practices? Applied Psychology: An Interna-
tional Review, 46: 387–452.
Scandura, T. A., & Graen, G. B. 1984. Moderating effects
of initial leader–member exchange status on the ef-
fects of a leadership intervention. Journal of Ap-
plied Psychology, 69: 428–436.
Shamir, B. 1995. Social distance and charisma: Theoret-
ical notes and an exploratory study. Leadership
Quarterly, 6: 19–47.
Shamir, B., & Howell, J. M. 1999. Organizational and
contextual influences on the emergence and effec-
tiveness of charismatic leadership. Leadership
Quarterly, 10: 257–283.
Shamir, B., House, R. J., & Arthur, M. B. 1993. The mo-
tivational effects of charismatic leadership: A self-
concept based theory. Organization Science, 4:
577–594.
Smith, W. K., & Lewis, M. W. 2011. Toward a theory of
paradox: A dynamic equilibrium model of organiz-
ing. Academy of Management Review, 36: 381–
403.
Smith, W. K., & Tushman, M. L. 2005. Managing strategic
contradictions: A top management model for man-
aging innovation streams. Organization Science,
16: 522–536.
Snijders, T. A. B., & Bosker, R. J. 1999. Multilevel anal-
ysis: An introduction to basic and advanced mul-
tilevel modeling. Thousand Oaks, CA: Sage.
Srivastava, A., Bartol, K. M., & Locke, E. A. 2006. Em-
powering leadership in management teams: Effects
on knowledge sharing, efficacy, and performance.
Academy of Management Journal, 49: 1239–1251.
Suedfeld, P., Tetlock, P. E., & Streufert, S. 1992. Concep-
tual/integrative complexity. In C. P. Smith (Ed.), Mo-
tivation and personality: Handbook of thematic
content analysis: 393–400. Cambridge: Cambridge
University Press.
Sundaramurthy, C., & Lewis, M. 2003. Control and col-
laboration: Paradoxes of governance. Academy of
Management Review, 28: 397–415.
Tadmor, C. T., Galinsky, A. D., & Maddux, W. W. 2012.
Getting the most out of living abroad: Biculturalism
and integrative complexity as key drivers of creative
and professional success. Journal of Personality
and Social Psychology, 103: 520–542.
Tadmor, C. T., Tetlock, P. E., & Peng, K. P. 2009. Accul-
turation strategies and integrative complexity: The
cognitive implications of biculturalism. Journal of
Cross-cultural Psychology, 40: 105–139.
Tetlock, P. E., Armor, D., & Peterson, R. S. 1994. The
slavery debate in antebellum America: Cognitive
style, value conflict, and the limits of compromise.
Journal of Social Psychology, 66: 115–126.
Tetlock, P. E., Peterson, R. S., & Berry, J. M. 1993. Flat-
tering and unflattering personality portraits of inte-
gratively simple and complex managers. Journal of
Personality and Social Psychology, 64: 500–511.
Tsui, A. S., Pearce, J. L., Porter, L. W., & Tripoli, A. M.
1997. Alternative approaches to the employee–or-
ganization relationship: Does investment in employ-
ees pay off? Academy of Management Journal, 40:
1089–1121.
Uhl-Bien, M., Marion, R., & McKelvey, B. 2007. Complex-
ity leadership theory: Shifting leadership from the
industrial age to the knowledge era. Leadership
Quarterly, 18: 298–318.
Van de Ven, A. H., & Poole, M. S. 1988. Paradoxical
requirements for a theory of organizational change.
In R. Quinn & K. Cameron (Eds.), Paradox and
transformation: Toward a theory of change in or-
ganization and management: 19–80. New York:
HarperCollins.
Vera, D., & Crossan, M. 2004. Strategic leadership and
organizational learning. Academy of Management
Review, 29: 222–240.
Vos, M. W., van der Zee, K. I., & Buunk, A. P. 2012.
Individual differences in self-construal: The de-
velopment of a measure of personal, relational
and collective self. Unpublished manuscript.
Vroom, V. H., & Jago, A. G. 2007. The role of the situation
in leadership. American Psychologist, 62: 17–24.
Waldman, D. A., Ramirez, G. G., House, R. J., & Puranam,
P. 2001. Does leadership matter? CEO leadership
attributes and profitability under conditions of per-
ceived environmental uncertainty. Academy of
Management Journal, 44: 134–143.
Wang, H., Law, K. S., & Chen, G. 2002. A structural
equation model of the effects of multidimensional
leader–member exchange on task and contextual
performance. Paper presented at the Society of In-
dustrial and Organizational Psychology (SIOP) Con-
ference, Toronto, Canada.
Wong, E. M., Ormiston, M. E., & Tetlock, P. E. 2011. The
effects of top management team integrative complex-
ity and decentralized decision making on corporate
social performance. Academy of Management Jour-
nal, 54: 1207–1228.
Yagil, D. 1998. Charismatic leadership and organization-
al hierarchy: Attribution of charisma to close and
distant leaders. Leadership Quarterly, 9: 161–176.
Yukl, G. 2010. Leadership in organizations (7th ed.).
Upper Saddle River, NJ: Prentice Hall.
2015 565Zhang, Waldman, Han, and Li
Yukl, G., & Lepsinger, R. 2004. Flexible leadership: Cre-
ating value by balancing multiple challenges and
choices. San Francisco: Jossey-Bass.
Zhang, Y., & Chen, C. C. 2013. Developmental leadership
and organizational citizenship behavior: Mediating
effects of self-determination, supervisor identifica-
tion, and organizational identification. Leadership
Quarterly, 24: 534–543.
Zhang, Z., Chen, G. Z. X., Chen, Y., & Ang, S. 2014.
Business leadership in the Chinese context: Trends,
findings, and implications. Management and Or-
ganization Review, 10: 199–221.
Zhang, Z., Waldman, D. A., & Wang, Z. 2012. A multi-
level investigation of leader–member exchange, in-
formal leader emergence, and individual and team
performance. Personnel Psychology, 65: 49–78.
Yan Zhang (annyan.zhang@pku.edu.cn) is an associate
professor in the psychology department at Peking Uni-
versity. She received her PhD in organizational manage-
ment from Peking University. Her research focuses on
paradox management, leadership, team dynamics, and
cross-cultural management.
David A. Waldman (waldman@asu.edu) is a professor of
management in the W. P. Carey School of Business at
Arizona State University. He received his PhD from Col-
orado State University. His research interests include
leadership at multiple levels of analysis, including such
specific issues as responsible leadership and the neuro-
science of leadership. He is a fellow of the American
Psychological Association and of the Society for Indus-
trial and Organizational Psychology.
Yu-Lan Han (han.yulan@mail.shufe.edu.cn) is an associ-
ate professor in the College of Business at Shanghai Uni-
versity of Finance and Economics. She received her PhD
in organizational behavior from Peking University. Her
research focuses on managerial mindfulness, leadership,
and teams.
Xiao-Bei Li (xiaobeili@ecust.edu.cn) is an assistant pro-
fessor of organizational behavior at the School of Busi-
ness at East China University of Science and Technology.
She earned her PhD from the University of New South
Wales, Australia. Her current research interests include
human resource effectiveness, executive leadership, and
decision making.
566 AprilAcademy of Management Journal
