Content uploaded by John Antonakis
Author content
All content in this area was uploaded by John Antonakis on Mar 27, 2016
Content may be subject to copyright.
Charisma: An ill-defined and ill-measured gift
John Antonakis
&
Nicolas Bastardoz
Department of Organizational Behavior
University of Lausanne
john.antonakis@unil.ch
Philippe Jacquart
EMLYON Business School
jacquart@em-lyon.com
Boas Shamir
Department of Sociology and Anthropology
The Hebrew University of Jerusalem
In press:
Annual Review of Organizational Psychology and Organizational Behavior
“Posted with permission from the Annual Review of Organizational Psychology and
Organizational Behavior, Volume 3 © 2016 by Annual Reviews, http://www.annualreviews.org.”
Note: Sadly, Boas passed away before this article could be completed. In consultation with the
editorial team at the Annual Reviews, we added Boas posthumously in the author list. Boas
significantly contributed to what we planned to write, he extensively commented on the coding
procedure we used, and we had several discussions with him about the content and direction of
the article. We thank Laurent Lehmann, Thomas von Ungern-Sternberg, and Christian Zehnder
for helpful comments they provided us during the development of this manuscript. We are also
very grateful to Manon Jaquerod and Sirio Lonati for their assistance in coding the articles.
Abstract
We take historical stock of charisma, tracing its origins and how it has been conceptualized in the
sociological and organizational sciences literatures. Although charisma has been intensely
studied, the concept is still not well understood and much of the research undertaken cannot
inform policy. We show that the major obstacles to advancing our understanding of charisma
have included issues with its definition, its confusion with transformational leadership, the use of
questionnaire measures, and that it has not been studied using correctly-specified causal models.
To help spawn a new genre of research in charisma, we use signaling theory to provide a general
definition of charisma, and make suggestions about how charisma should be conceptualized,
operationalized, and modeled. We also describe trends and patterns in articles we reviewed, using
co-citation as well as bibliometric analyses, and discuss the practical implications of our findings.
Keywords: charismatic leadership, signaling theory, endogeneity, causality, leader development.
Contents
Introduction ...................................................................................................................................... 1
A brief history of research on charisma ........................................................................................... 3
The foundations: Sociology and political science ........................................................................ 4
The edifice: Applied psychology and management ..................................................................... 8
Defining Charisma ......................................................................................................................... 13
Problems with current definitions ............................................................................................... 13
A solution to the definitional problem ........................................................................................ 16
Studying charisma: Challenges and recommendations .................................................................. 20
Design of studies ........................................................................................................................ 20
The “MLQ problem” and endogeneity ....................................................................................... 22
Operationalizing charisma: The way forward ............................................................................ 25
State of the science and future research directions......................................................................... 27
Managerial and cross-cultural implications ................................................................................... 32
Conclusions .................................................................................................................................... 34
Summary points .............................................................................................................................. 35
Related resources ............................................................................................................................ 35
References ...................................................................................................................................... 36
Figure 1: The core of the intellectual landscape of the socio-scientific study of charisma ........... 43
Figure 2: Publishing Trends: Number of articles using the term “charisma” or variants .............. 45
Table 1: Definitions of Charisma ................................................................................................... 46
Appendix ........................................................................................................................................ 47
1
Introduction
What is charisma? Most scholars and lay people have some sort of implicit notion of what
charisma is, the “you-know-it-when-you-see it” phenomenon. As it refers to leadership, many use
the term to refer to some sort of gift, charm, or alchemic ability—inaccessible to most—that
some leaders have making them able to federate followers around a cause. Yet, thinking of
charisma in such terms makes for a fuzzy concept to study. How can we measure such a gift? Is it
really a “gift,” does it depend on individual differences like personality or intelligence, is it a
context-triggered phenomenon, or a set of skills that can be developed? What is the nature of this
concept and does it have consequential outcomes?
Despite decades of research on the topic, answers to these questions are not so clear, and
this for a myriad of reasons that we will discuss in this article. Some have even gone on to
suggest that charisma is “illusionary . . .U.F.O. phenomenon,” a “black hole,” and a “social
delusion” (Gemmill & Oakley 1992, p. 119). Of course, there is a lot of good research that has
been done on charisma in various fields, which we will cover and critically discuss in our review.
However, how the concept is defined and has evolved, the theories of charisma that dominate the
literature, and the ways many researchers have studied and tested the concept have not helped
research advance by much.
Charisma has been the subject of much research since House (1977) introduced it to
organizational scholars and psychologists. However, the concept is typically ill-defined by using
exemplars or by defining it by its outcomes (Antonakis et al 2011, van Knippenberg & Sitkin
2013), which cannot help research advance (MacKenzie 2003). Moreover, measures of charisma,
using typical behavioral questionnaires, might not be capturing the construct appropriately and
such measures of charisma might be outcomes of other factors themselves (Antonakis et al 2011,
2
Yukl 1999). Making matters even muddier is that entire research programs, which have
dominated the field for decades (Gardner et al 2010, Lowe & Gardner 2000), have been
developed around questionnaires, where the measures per se are used to define the theory. This
state of affairs is unfortunate because the theory has not been prospectively described and
operationalized (van Knippenberg & Sitkin 2013).
Adding to the confusion is that the concept of “charisma” is frequently equated to or
confounded with another hazy topic in organizational behavior, “transformational leadership”
(Yukl 1999). And, despite the best efforts of scholars to constructively critique these two
constructs so that leadership research can move on (e.g., van Knippenberg & Sitkin 2013),
equating two fundamentally different concepts like charisma and transformational leadership has
made the fog over the leadership landscape thicker still.
The goal of our review is to bring some clarity to the field by untethering charisma from
transformational leadership theory and measurement, pointing out the conceptual gaps,
definitional problems, and methodological shortcomings, and providing fresh ideas to researchers
so that a new genre of studies can help advance a concept that has immense importance for
society. Doing so is vital for theory development and the empirical research that it will trigger.
However, it is important for practice too because charismatic leaders wield enormous power and
can use this power to accomplish great good or evil; thus it is essential to understand what
charisma is, its antecedents, moderators, and consequences (Antonakis 2012). We think it is time
to take the construct of charisma to the next level.
In this review, we will take historical stock of charisma, tracing its origins and how it has
been conceptualized in the sociological (e.g., Shils 1965, Weber 1947) and organizational
psychology literatures (House 1977, Shamir et al 1993). In addition to identifying conceptual
denominators in definitions of charisma, we will compare and contrast various ways in which
3
charisma has been studied in terms of research approaches or designs (e.g., quantitative,
qualitative, theory), and how it has been measured, including content coding (Emrich et al 2001,
House et al 1991, Simonton 1988) and questionnaires measures (e.g., Bass & Avolio 1995). We
will also examine how it has been manipulated (e.g., Howell & Frost 1989), whether it can be
developed (Frese et al 2003), as well as its outcomes (Fanelli et al 2009). We will also highlight
how it should be studied in a causal manner (Antonakis et al 2010) and discuss implications for
practice and policy.
Note, typical of reviews of this nature, our positions, coverage, and critiques rely on our
professional judgments and how we have interpreted the development of the field over time;
however, given the nebulosity and heterogeneity of the charisma construct, and the fact that it has
been studied across various fields, to ensure that we were comparatively objective in our
coverage and positions we report too on some data regarding a systematic review of the literature
that we also undertook (for details, follow the Supplemental Material link in the online version of
this article or at http://www.annualreviews.org/). We also use co-citation analysis (c.f. White &
Griffith 1981), and in particular document co-citation analysis, to chronicle the core works in the
field (Tsai & Wu 2010). This analysis will provide us with a picture of the intellectual landscape
on which articles writing on charisma have built, and help us better understand whether these
studies have drawn from a single unified body of knowledge or from multiple unconnected
bodies of knowledge.
A brief history of research on charisma
The term “charisma” is usually credited to the writings of the Max Weber. He borrowed,
secularized, and expanded the term from literature on religion (Sohm 1892), and referred to
charisma as an extraordinary power, giving leaders salvationist qualities to deliver followers from
4
great upheaval (Weber 1947, Weber 1968). Variants of the term “charisma,” however, predate
Sohm as well as biblical sources, and have roots in classical Greek mythology—the graces,
Charites, and in particular the goddess Charis (Smith 1998). The term charis signifies much in
classical Greek; in addition to gratitude (as in eucharist), it can refer to charm, excitement,
beauty, pleasure as well as allurement (Maclachlan 1996).
Charis is a complex word that has been the basis of other words. Of course, we are
concerned with charisma in leaders and their influence over others and not of charisma as applied
in lay terms to individuals in general. It is not difficult to see why this word was chosen to
undergird and describe a type of leadership that can give leaders great power, that is, “the
discretion and the means to asymmetrically enforce one’s will” over entities like organizations or
individuals (Sturm & Antonakis 2015, p. 139). Also, although the term charisma was not
originally used by the ancient Greeks to describe a type of leadership per se, Aristotle (1954) did
describe how leaders should persuade followers by demonstrating character and defending
values, stoking follower emotions, and using strongly reasoned argumentation (the ethos, logos,
and pathos); in fact, Aristotle’s insights on the rhetoric of leadership, as well as the importance of
persuasive devices like metaphor has provided a basis on which modern ideas of charisma have
evolved (Den Hartog & Verburg 1997, Shamir et al 1994, Willner 1984).
The foundations: Sociology and political science
Weber’s ideas about charisma have been enormously influential, and were popularized by
sociologists, political scientists, psychologists, and management scholars (e.g., Bass 1985, Davies
1954, Downton 1973, Etzioni 1961, House 1977, Shils 1965). Weber, however, was not very
precise about the nature of charisma (Riesebrodt 1999, Shamir 1999). For instance, Weber (1968)
suggested that charismatic leaders held “specific gifts of the body and spirit [that are] not
5
accessible to everybody” (p. 19) and that charismatic authority resulted from “times of psychic,
physical, economic, ethical, religious, [and] political distress” (p. 18); for Weber, crisis and social
distress were thus important conditions from which charisma emerged. Weber (1947) suggested
that such leaders were mages having been “endowed with supernatural, superhuman, or at least
specifically exceptional powers or qualities” (p. 358). As a result, Weber (1968) argued that
charismatic leaders create devoted followers who help their leader on a mission that arises out of
“enthusiasm, or of despair and hope” (p. 49).
Important for Weber was the notion that the very goals of a charismatic leader are
different to those of institutional methods of influence. Weber viewed charisma as radical force
wherein “charismatic domination is the very opposite of bureaucratic domination,” working
against methodical, rational, and economic ideals (Weber 1968, p. 20). Charisma’s impact is felt
on an emotional level, is “revolutionary and transvalues everything; it makes a sovereign break
with all traditional or rational norms” (Weber 1968, p. 24); yet, with time, charisma’s effects
wane and bureaucracy takes over. The charisma-bureaucracy seesaw continues as relevant
contextual factors become propitious for one or the other.
Of course, Weber’s ideas and insights have been useful in terms of putting charisma on
the leadership nomological map (and indeed 74% of the articles we reviewed make reference to
Weber’s work; also, see Figures 1A and 1B). Yet, using Weber’s ideas makes it difficult to
understand the nature of charisma and to study it systematically, and raises several important
questions: What is the nature of the gift and how can it be measured? What kinds of supernatural
or superhuman qualities do these leaders have? Is charisma always born in the crucible of crisis?
We will address these questions as we review research on charisma, how it is defined, and the
nature of the charismatic effect.
[Figure 1A and 1B here]
6
Sociologists popularized Weber’s ideas, and were instrumental in laying the groundwork
from which psychological theories (e.g., House 1977) of charisma evolved. Sociologists were
grateful to Weber, and attempted to move the concept from one that was overly focused on
psychological attributes and “gifts” of charismatic leaders to one that looked at charisma in a
broader social context (Friedland 1964). Essentially, much of sociologist’s thinking concerned
charisma’s institutional (and structural) character, how individuals straddle organizational and
personal sources of influence, and how these forces can be reconciled to benefit both the
organization and the individual (Shils 1965).
Key to some of this thinking has to do with need to maintain an ordered and stable social
system in which individuals can develop, progress and find identity, and the role of charisma in
this process (Shils 1965). Sociologists were thus keen to look at the secular nature of charisma,
one in which the leader is seen as a “master of events,” one that is able to resolve “existential
chaos,” who “structures a cosmos,” and “provides guides for action and a promise for the future”
(Spencer 1973, p. 345). For Spencer charisma consisted of (a) skilled performance, engendering
awe, and (b) having representation (i.e., symbolic based on values), which creates enthusiasm;
these two components can depend on the personal characteristics of the leader but also on the
situation, and luck too. Thus, charisma is what Spencer termed the “historical product” (p. 352)
between the person and the situation; this position is reminiscent of ideas stemming from
interactional psychology (Endler & Magnusson 1976).
Although several other sociologists made important contributions, we focus on the
sociological point of view of Etzioni (1961, 1964), whose work on structuralism—based on Max
Weber and Karl Marx’s thinking—has important implications for organizational scholars. Etzioni
(1964) sought to describe how organizations exhibit both formal-rational and informal-human
needs that are at conflict with each other and which must be balanced so that the organization can
7
adapt. Key to Etzioni (1964) was how organizations handle differences in power and the goals
that are pursued in hierarchies (i.e., between management and workers). Thus, the leadership
style adopted by management was an important determinant of an organization’s success because
“Most organizations most of the time cannot rely on most of their participants to internalize their
obligations, to carry out their assignments voluntarily, without additional incentives” (Etzioni,
1964, p. 59). Given this principal-agent problem Etzioni (1964) differentiated three types of
power that leaders could use: (a) physical power entailing the use of threats or coercion, (b)
material power entailing the use of rewards, and (c) symbolic (i.e., charismatic) power, a type of
normative power that depends on the person (see also Etzioni, 1961). For Etzioni (1964) greater
commitment and less alienation will be displayed in workers when leaders exercise symbolic
power.
Political scientists also made important contributions to the concept of charisma. Davies
(1954) appears to be the first to use the term in this discipline. For him charisma was not “a
characteristic of leaders as such but a relationship between leader and followers” (p. 1083). As in
sociology, political scientists were eager to try and pin down what seemed to be a “vague”
(Friedrich 1961, p. 3) and “nebulous” construct (Tucker 1968, p. 732). For Tucker, charismatic
leaders were salvationist or messianic and arose from situational distress; thus, this type of
leadership is a process of social influence and not necessarily a position of authority. Tucker was
one of the first to identify concrete characteristics of charismatic leaders and suggested that such
leaders had good communication skills, were visionary, had a sense of mission, believed in the
righteousness of their values and ideals, and had confidence that the vision can be achieved.
Other important contributors to the charisma concept include Downton (1973), who wrote
a theory of inspirational and charismatic (as well as transactional) leadership in revolutionary
settings. Burns (1978) proposed a similar theory of transforming (and transactional) leadership
8
but did not directly refer to the work of Downton (i.e., interestingly, Burns footnoted Downton’s
work twice and this in a rather oblique way). Burns used the term “transforming leadership,”
which he described as “elevating, mobilizing, inspiring, exalting, uplifting, preaching, exhorting,
evangelizing” (p. 20) to refer to a type of leadership that seemed very reminiscent to what Weber
and Downton had called “charisma.” In his magnum opus, Leadership, Burns (1978) tiptoed
around the concept of charisma, which he thought was beyond analysis, and preferred instead to
use the term “heroic” and “ideological” leadership. Burns placed these terms under the umbrella
“transforming” leadership; this move was a rather unfortunate one because it appears to be the
source that muddied the conceptual waters, as we discuss in reviewing the work of Bass (1985)
below.
The edifice: Applied psychology and management
In the 1970 the state of leadership research was under severe threat (Greene 1977) with
calls even to abandon studying it (Miner 1975). It was House (1977) who set the stage to heave
leadership research out of the doldrums. Coincidently, charismatic leadership as a topic had a
messianic effect on leadership research at time when leadership as a scientific construct was not
taken seriously by many scholars working in applied psychology, organizational behavior, or
management (Antonakis 2012, see also Hunt 1999). It is encouraging to see how interest in
charisma has grown over time (see Figure 2).
[Figure 2 here]
Of course, scholars have a keen interest in understanding leadership, particularly its
psychological foundations, which has immense practical importance, particularly for
organizations. In doing so, however, such scholars were accused of having jettisoned the original
Weberian ideas—relating to crisis, revolution, and charisma as a very unusual characteristic of a
9
person—which were held so sacrosanct by sociologists, to study a more tame and ordinary form
of it (Beyer 1999). Such qualms were made early by sociologists too about the vulgarization of
the term “charisma” (Bensman & Givant 1975). However, studying a more “ordinary” version of
it provides important advantages because “organizational charisma” can be found and studied in
various settings (Shamir 1999), which are not necessarily crisis prone; indeed, crisis is not
necessary for charisma to emerge, as several commentators have suggested (Conger & Kanungo
1998, Etzioni 1961, House 1999, Jacquart & Antonakis 2015, Shamir & Howell 1999).
House (1977) presented the first psychological theory of charismatic leadership
suggesting that such leaders create intensive emotional interactions with their followers. He
suggested too that such leaders are usually, but perhaps not necessarily, born of crisis and are
seen as saviors as well as role models and objects of identification. House (1977) argued too that
these leaders challenge the status quo and “through their leadership major social changes are
accomplished” (p. 189). House (1977) presented an amalgam of characteristics and behaviors of
charismatic leaders, noting that charisma is essentially based on ideological goals; he noted too
that such leaders paid attention to image management, set high expectations, communicated
confidence in goal attainment, and role-modeled desired behaviors. House also carefully
discussed the psychological outcomes and states that charismatic leaders would have on
followers. For House, charisma was not really a “gift” or some magical ability, but rather a
complex interaction between the leader, the prevailing context, and follower needs. He articulated
several testable propositions about his theory and challenged researchers to measure charisma
and identify how it impacts followers.
The gauntlet that House (1977) threw down was taken up by Bass (1985); basing his work
on Burns (1978) and House (1977), Bass articulated a theory that bore some strong overlap to
that of Downton (1973)—but to whose work he did not refer. Bass’s (1985) ideas have had an
10
enormous impact on leadership search, as evidenced in Figures 1A and 1B (see also: Antonakis et
al 2014a, Gardner et al 2010, Lowe & Gardner 2000). We thus will focus on his theory
extensively, also because it is the reason why there is so much confusion about charisma and
transformational leadership; at the same time, we do also wish to credit other important
contributions that were made by other researchers to the study of charisma (Berlew 1974, e.g.,
Bryman 1992, Conger & Kanungo 1987, Conger & Kanungo 1988, Conger & Kanungo 1998,
Conger et al 2000, Gardner & Avolio 1998, House & Howell 1992, Howell 1988, Howell &
Frost 1989, Shamir 1995, Shamir et al 1993, Waldman & Yammarino 1999); some of these are
highlighted in Figure 1A.
Bass reworked the entire notion of charisma with a focus on measuring it, and placed it
under a class of leader behavior he called “transformational leadership.” Similar to the title of his
book, Leadership and Performance Beyond Expectations, he defined transformational leadership
as a type of leadership that “motivates us to do more than we originally expected to do” (p. 20)
and “used to describe leaders who by the power of their person have profound and extraordinary
effects on their followers” (p. 35). Apart from using a very loaded term “transformational”
(Antonakis 2012), his definition is not helpful because transformational leadership is defined by
its outcomes (see section on “Defining Charisma” for further elaboration on this problem). Thus,
what is transformational leadership? Most would say that it is about leaders who are able to
transform; however, this is not a definition but a tautological statement.
Bass’s core idea was to articulate a theory of leadership going beyond the simple
provision of rewards or punishments contingent on performance, which he termed “transactional
leadership,” by proposing a style of leadership that explained extraordinary performance and
commitment in followers, which he termed “transformational leadership.” He identified this new
type of leadership by asking executives to describe exemplars. On the basis of this pilot study and
11
a review of the literature, he identified and also developed items that represented influence and
charisma (as well as transactional leadership). Then using judges, he identified what were seen as
the most prototypical items for the major classes of leader transformational and transactional
behavior and tested these items in various samples using factor analyses. Bass (1985) stated
explicitly that the results from these analyses “were the basis for structuring this book” (p. 207)
and hence the theory. Although he reviewed literature on the topic, primarily the works we cited
above (i.e., those of Weber, Burns, House, and others), the way in which he conceptualized
charisma was inductive, leading to conceptual ambiguities (van Knippenberg & Sitkin 2013,
Yukl 1999). Not providing a conceptual definition and using measures to develop a theory have
created a very unsatisfactory situation that has impeded advancement of this construct (van
Knippenberg & Sitkin 2013); still today, the nature of transformational leadership is not clear.
The theory is basically defined by its measurement instrument, the Multifactor Leadership
Questionnaire—the MLQ—which has undergone numerous revisions (Bass & Avolio 1995); it is
used to measure leadership via self or observer reports. For Bass (1985), the core of
transformational leadership was charisma (see also Bass 1990, p. 199)—charisma provided the
emotional component of leadership. He also suggested that charisma and transformational
leadership were asymmetrically related in that not all charismatic leaders were transformational.
In the current version of the theory (i.e., the MLQ), charisma is called “idealized influence”
having a (a) behavioral as well as an (b) attributional component to it; theoretically charisma is
also tapped by a factor named (c) inspirational motivation. These three dimensions, along with
the (d) rational component of transformational leadership called intellectual stimulation as well as
(e) a developmental oriented factor called individualized consideration, comprise what is known
as “transformational leadership” (Antonakis et al 2003).
12
Apart from issues concerning conceptual definitions, there are several problems with this
theory and questionnaire. By far the biggest problem has to do with equating transformational
and charismatic leadership: Although Bass argued that charisma is part of transformational
leadership most researchers confuse the two constructs or suggest they are one and the same.
These two concepts are distinct and one should not been seen as a subcomponent of the other—
that is, a leader could be transformational without being charismatic, or vice-versa (Yukl 1999).
Furthermore, the two constructs may even be even incompatible. Being charismatic may
constrain the extent to which the leader could be “transformational” (i.e., empowering and
developmental); additionally, being too developmental, attentive, and empowering may limit the
extent to which the leader can be attributed charisma because the leader will not be seen as
extraordinary enough (cf. Yukl 1999). Thus, it is unfortunate too that even well-meaning critics
(i.e., van Knippenberg & Sitkin 2013) have suggested that these two constructs are isomorphic
and this, on the basis of high correlations between scales of the constructs. That factors or items
from the two constructs—using what we think is an inappropriate questionnaire to measure
charisma as an independent variable—correlate highly is not proof that they are measuring the
same thing. This correlation can be explained by endogeneity bias (i.e., confounding variables) as
well as other sources of bias. We elaborate on this problem later (see the section titled The “MLQ
problem” and endogeneity).
In conclusion, and as will become clearer in the following sections too, charisma (a) is
construct that should be studied and modelled in its own right and must be unleashed from
transformational leadership, (b) requires a clear definition, (c) must be operationalized following
the definition, (d) cannot be measured as an independent variable using current behavioral
questionnaires and usual statistical methods (because of endogeneity), and (e) must be measured
13
in an objective way not prone to cognitive or endogeneity biases (see Appendix for a brief
introduction to the endogeneity problem).
Defining Charisma
Before we provide what we believe to be useful definition of charisma, we first discuss
how charisma should not be defined. We discuss this issue extensively because most research
programs on charisma, beginning with definitions, require a keelhauling. Key to understanding
charisma is first defining it correctly; however, there are many problems and issues with current
definitions of charisma, which have impeded our scientific progress.
Problems with current definitions
We begin with the simplest of the problems in prevailing definitions with two analogies:
Imagine if scholars (a) defined effective companies as having divine qualities or (b) measured
effective practices in companies to predict company effectiveness. The problems of undertaking
research in this manner are immediately evident. Scientific fields can only progress if the
phenomenon studied is defined precisely and that the nature of the underlying phenomenon is
unveiled without making for a tautological definition and estimation.
To better understand the problem, we begin by showcasing a typical definition of
charisma, which goes like this: “Charisma has been defined as a characteristic of individuals who
‘by force of their personal abilities are capable of having profound and extraordinary effects on
followers’” (Wowak et al 2014, citing House, 1977). What is the problem with such a definition?
Constructs should not be defined in terms of their outcomes, their antecedents, or using
exemplars (MacKenzie 2003). Useful definitions would require the construct, charisma, to be
independent of its effects and specify the nature of the phenomenon in a prospective way. Of
course, empirical validation of the effects of the construct on outcomes (e.g., worker
14
productivity) is required to demonstrate the construct’s utility; however, the construct cannot be
defined in terms of the outcomes it should produce. Yet, as has been recently noted (Antonakis et
al 2011, van Knippenberg & Sitkin 2013) charismatic leadership has usually been defined
precisely in this way. Worse, it is defined in undefined terms. For example, Davies (1954)
suggested that charisma is “miraculously-given power” (p. 1083). Such a definition is not clear as
to the nature of this power. Etzioni (1961, p. 203) stated charisma is a relational property and “the
ability of an actor to exercise diffuse and intense influence over the normative orientations of
other actors”; such a definition uses outcomes and it is not clear how or why these outcomes are
produced. We find the same problem with the definition of Spencer (1973, p. 352) for whom the
essence of charisma is evident “in an attitude of awe and enthusiasm.”
Yukl (1999) suggested that “The most useful definition seems to be in terms of
attributions of charisma to a leader by followers who identify strongly with the leader” (p. 294);
again, using outcomes in a definition cannot clarify the nature of the concept. Bass (1990, p. 220)
avoids defining charisma altogether, though the closest he gets to a definition is with the
following: “a person of strong convictions, determined, self-confident, and emotionally
expressive and his or her followers must want to identify with the leader as a person, whether or
not in a crisis.” Yet, this definition will not help science progress, as we explain below with
respect to the problem of defining charisma with antecedent traits (beyond the other problem of
including outcomes in the definition). Even House (1977) defined charisma by its outcomes.
After having articulated the effects that charismatic leaders have on followers, he noted that the
“term charismatic leadership will be used to refer to any leader who has the above “charismatic
effects” (p. 192). However, in his defense he was suggesting that studying leaders who create
these effects may help to identify their characteristics, which would help to distinguish them from
leaders who do not produce charismatic effects.
15
To determine where the state of the field is with respect to defining charisma, we coded
the definitions of charisma provided explicitly in our reviewed articles, to find the lowest level
conceptual commonalities of elements in definitions, showing which are useful or not for a
definition of charisma (see Table 1). It is unfortunate to note that most scholars have defined
charisma as some kind of unknown quality, ability, or gift of a leader. Many define charisma by
its outcomes, and most other elements in the definitions we found are simply not useful to
explicate the nature of the construct and to distinguish it from others. Defining charisma as an
attribution is also problematic because an attribution is an outcome—or has to be modelled as
mediatory process to predict another outcome—which depends on some sort of leader action.
Defining charisma as a social (relationship) process or leader behaviors and actions is not
very useful because all types of leadership are predicated on processes, behaviors, and actions.
Also, stating that charisma depends on follower characteristics again renders it is an endogenous
construct, and endogenous variables can only be used as outcomes or “instrumented” mediators,
and not as regressors; as such it would not vary randomly in individuals and be beyond
manipulation (and thus be very hard to study). Moreover, charisma does not require a crisis to
emerge (which in any case violates a good definition by specifying an antecedent). And, social
cognition, whether attribution or inference based is a foundation for the evaluation of all types of
social relationships (see Jacquart & Antonakis 2015). Furthermore, using antecedent traits or
traits per se to define charisma does not help elucidate the nature of the charisma construct
because such traits can be predictive of other forms of leadership too.
[Table 1 here]
16
A solution to the definitional problem
The elements of a definition of charisma that we consider to be useful include defining it
as a type of leadership whose nature is based on values (i.e., morals), beliefs and symbolism as
well as on emotion, which is expressive in its transmission of information. We omit using the
word “vision” in the definition, because vision is really quite a vague notion, which in any case
stems from using symbolic means of communication that are useful for triggering a mental image
and hence a vision (Antonakis et al 2011). Although we could use the term “ideology” in a
definition of charisma, we will omit it so as to not confuse charisma with “ideological
leadership,” a type of leadership that is predicated on defending some tradition (Mumford et al
2008). Of course, ideology is based on values; thus, ideologues can also be charismatic but not all
charismatic leaders are necessarily ideologues.
Also, to help better identify the nature of the charismatic effect, we draw from economics,
which has not made many contributions to leadership yet, but which has some sound thinking
with respect to the incomplete and asymmetrical nature of information between players in
markets (Spence 2002). The market we refer to is leader selection or emergence, which includes
(a) leaders (b) followers of leaders, who accord leaders status (informally) by virtue of following
them, and (c) leader selectors acting on behalf of principals, who formally appoint leaders to
positions of authority. In this market, leaders engage in signaling; signals can be thought of as
“things one does that are visible and that are in part designed to communicate” (Spence 2002, p.
434). Via signaling, leaders can win selection tournaments or be accorded status by followers,
whether they are formal or informal leaders. As regards followers, leaders need to signal them
some information about what kinds of actions they should engage in via leader role-modeling (cf.
Hermalin 1998) and from value systems leaders communicate. Also, leaders need to signal about
17
their own skills (Spence 2002). Thus, what choices leaders take or what actions they value can,
via signaling, help solve coordination problems in public goods situations by affecting follower
choices and actions as well as beliefs about what others may do.
Also, with respect to followers per se, given that the nature of leadership as an influencing
process is not one of authority but one of voluntary following (Hermalin 1998), we think that the
term “signaling” as a general mechanism of information communication should also appear in a
definition of charismatic leadership. Note that signaling can occur via verbal and nonverbal
communication modes (Awamleh & Gardner 1999, Frese et al 2003, Towler 2003, Willner
1984); and, interestingly, those who are high on use of verbal communication means are high on
nonverbal too (Antonakis et al 2011) presumably because rich verbal communication means (e.g.,
storytelling) require more nonverbal gesturing (Jacquart & Antonakis 2015, Towler 2003).
Important here too is to avoid using the term “influence” in a definition of charisma.
Influence connotes having an ability to impose oneself; however, what is the nature of this ability
and why does it occur? Moreover, influence in terms of being “influential” is an outcome. Hence,
“signaling” and “influence” are separate constructs and because of signaling the leader is able to
influence (the dependent variable) under certain conditions. Therefore, following the above, as
well as previous definitions of charisma (cf., Antonakis et al 2011, Tucker 1968) we provide the
following general definition: Charisma is a values-based, symbolic, and emotion-laden leader
signaling.
A charismatic leader is one who signals using the aforementioned mechanisms noted in
the definition. Following this definition, an individual can be charismatic without having any
influence whatsoever. For the charismatic effect to occur, and for the followers to willingly
succumb to the leader’s influence, the leader must be accepted by followers because he or she is
communicates values and a mission that appeals to the followers. Theoretically, the “connection”
18
(i.e., the charismatic effect) that a charismatic leader has with his or her followers, and why the
followers identify with the leader, stems from the leader (a) justifying the mission by appealing to
values that distinguish right from wrong, (b) communicating in symbolic ways to make the
message clear and vivid, and also symbolizing and embodying the moral unity of the collective
per se, and (c) demonstrating conviction and passion for the mission via emotional displays
(Antonakis et al 2011). To the extent that such signalling on the part of the leader resonates with
the values of the collective, the leader will appear prototypical to followers (Hogg 2001); indeed,
charismatic leaders can be loved but also much loathed by those who do not share in the leader’s
values (Tucker 1968). It follows too that the more prototypical the leader seems, the more
followers will attribute certain qualities to the leader like courage, wisdom, competence, and so
forth (for more precision, we should use the term "infer" instead of "attribute"; the former
concerns describing what one is like and the latter the cause of an outcome, see: Erickson & Krull
1999, Jacquart & Antonakis 2015).
Note, as far as we are concerned, and for the purpose of the above definition we are not
assuming that the leader’s signals necessarily carry accurate information about the competence or
moral righteousness of the leader’s ideals; however, observers will use these signals to attribute
certain qualities to the leader and to give him or her the benefit of the doubt in situations of
strategic uncertainty, whether this uncertainty comes from uncertainty about the previous
performance of the leader or environmental uncertainty (Jacquart & Antonakis 2015). These
elaborations should be considered in theories along with questions like: under what conditions
does charisma as a signal convey credible information to followers as well as other interested
parties (e.g., investors) about the skills of leader? That is, does charisma correlate with
unobserved leader abilities? Tellingly, Aristotle mentioned the following about the use of
symbolic communication (and the use of metaphor in particular, which is a core component of
19
symbolic communication, Antonakis et al 2011): “But the greatest thing by far is to have a
command of metaphor. . . . it is the mark of genius” (Aristotle & Butcher 2011). Also, can we
assume that the skills of charismatic leaders can be acquired at a lesser cost by more able leaders;
or, do they depend directly on the natural endowments of leaders, that is, their intelligence and
personality, which are heritable to a large degree (Bouchard & Loehlin 2001, Bouchard &
McGue 2003)? And, given the assumed difficulty for a low-ability leader to acquire such skills,
does the market for leaders naturally sort itself in a way such that only high-ability leaders signal
charisma?
Armed with a definition of charisma, researchers can focus on building a comprehensive
exposition of the nature of the charismatic effect including antecedents, mediatory mechanisms,
boundary (moderator) conditions, and outcomes, and this at different levels of analysis. Doing so
will allow for the development of a complete and general theory of charisma, ideally at the
interface of psychology and economics.
Finally, our definition is not concerned with the ultimate morality the leader per se (see
Howell 1988). Although we do, as researchers, care that leaders use their power to serve the
greater good, scientific definitions should not dabble in normative outcomes, which are more of a
philosophical concern. As far as our definition is concerned, that the signaling process is values
based suggests that the leader will be judged by the values and morals he or she communicates.
Also, the leader must appropriately communicate about actions in which the collective should
invest; leaders do so via beliefs and expectations and this using symbolic communication means
(e.g., metaphors) and displays of emotions in a correctly-calibrated and appropriate manner.
Thus, there are costs and benefits involved in signaling, not only in terms of having the ability to
produce the signal, which we assume will mostly be available to high-ability leaders, but also
with respect to what outcomes the signals can engender per se for the leader and the collective.
20
Leaders cannot say one thing and do another, or signal unrealizable actions because in the long
run they risk losing their credibility and hence the charismatic effect. Thus, leaders must
appropriately pitch their signal as a function of their ability. Hence, leaders who are charismatic
and whose values are accepted should in the long run have better performance outcomes than will
leaders who are charismatic and whose values are not accepted or, leaders who do are not
charismatic.
Studying charisma: Challenges and recommendations
Given the challenges we have identified with respect to how charisma is being defined, it
will not come as a surprise that studying this concept will prove to be difficult. In this section, we
briefly review and critique how charisma has been studied with respect to design,
operationalization, and the endogeneity problem.
Design of studies
Developing theory to explain phenomena including understanding causal relations
between variables as well as the conditions under which relations between the variables hold is an
important part of research (Bacharach 1989). Because of the enormous complexity of the
charisma construct and the diversity of researchers studying it, there has been a variety of ways to
go about studying charisma. What is the best way to study charisma? Are some ways more
impactful than others are? These questions are particularly interesting given the differences in
history and traditions in the two major disciplines that study charisma, sociology and psychology.
Researchers in the former are more open to qualitative (and “post-positivist” approaches)
whereas the latter are more quantitative and experimental.
Currently, the dominant way to study leadership is the quantitative design, as ascertained
recently in lifetime bibliometric study of articles published in the top field journal specializing in
21
leadership research, The Leadership Quarterly (Antonakis et al 2014a). Yet, the qualitative mode
of inquiry has been suggested as a potent complement but also an alternative to quantitative
methods; given the complex and contextually sensitive nature of the leadership phenomenon, the
qualitative method has been proposed as especially relevant and useful for studying charisma
(Conger 1998). However, in comparison to qualitative articles, quantitative—as do theory, review
and methodological—articles have a more significant impact on the field, as measured by the
citations received; furthermore, qualitative articles are typically under-represented in top-cited
papers (Antonakis et al 2014a).
As with previous research, we found that qualitative papers are undercited in comparison
to other types of articles (see online Supplemental Materials). Does the “citation market,” which
represents the collective wisdom of field as a whole have it right insofar as studying charisma is
concerned? We cannot answer this question. However, we believe that to advance a field we need
theories. Review articles, such as the current one, are helpful for taking stock and guiding future
research efforts. Then we need to test theories, which can only be done reliably in a quantitative
manner. Quantitative designs, if correctly done, offer many advantages for testing counterfactuals
and causal relations, and hence informing policy. This method of inquiry can still be used with
qualitative data because qualitative information can be coded in a straightforward way (in many
if not most cases).
With the increased availability of modern computing power and advanced statistical
methodology, contextual and multilevel information can now be readily coded and included in
statistical models. Given the current state of research for quantitative modeling across a whole
range of effects and conditions, and the capabilities of modern software, we think that researchers
should move more towards theory testing using experimental designs (Brown & Lord 1999). As
the old adage generally attributed to Kurt Lewin goes: “The best way to understand something is
22
to try to change it.” Also, appropriately-done field studies and “natural experiments” that can
avoid endogeneity threats are also very useful and probably more relevant too (Antonakis et al
2010). Going more towards the quantitative way will nonetheless require a radical overhaul of
how researchers go about measuring charisma, as we discuss next.
The “MLQ problem” and endogeneity
As mentioned recently by Day (2012, p. 862) “questionnaires remain a popular (if
misguided) approach to studying leadership. If you design and publish a brief, easy-to-administer
survey questionnaire, there is little doubt that researchers will use it. But, we should not lose sight
of the fact that a map is not the territory, and simply labeling a questionnaire as a measure of
‘leadership [or charisma] measure’ does not mean that it actually measures leadership [or
charisma].”
Are questionnaire measures really a problem? We think mostly yes and there are very
specific reasons why we take this position. Questionnaire measures of leadership, using ratings of
observers (typically subordinates, but also peers or supervisors) started becoming popular in the
1950s and 1960s (Fiedler 1967, Fleishman 1953, Katz et al 1951, Stogdill 1963). Currently, there
seems to be “create-a-questionnaire” bandwagon sweeping through our field to measure all sorts
of constructs; many of these are poorly defined and operationalized (Day & Antonakis 2013).
Researchers use questionnaire measures probably out of convenience or simply because they
have been trained to or because everyone else does it. It certainly is a “quick and dirty” way to
obtain data. Yet, such measures do not get to what charisma is—in terms of an independent
variable—no matter how big the statistical hammers are, which are used to “confirm” the factor
structure of the measures.
23
By and large the “new” leadership (Bryman 1992), which includes charismatic leadership,
dominates the leadership landscape (Antonakis et al 2014a) with the MLQ being the most-used
measurement instrument both of the “new” leadership (Antonakis & House 2014) as well as of
charisma, as the results of our review show (see online Supplemental Materials). The use of the
MLQ is growing, which is an unfortunate trend to observe because this instrument does not
measure charisma as an independent variable but mostly outcomes of charisma or charisma as an
endogenous variable (Antonakis 2012, Antonakis & House 2014, Shamir et al 1998). Constructs
that are endogenous cannot be used as regressors in a model to predict other outcomes unless
corrective action is taken to remove the endogeneity bias in the leadership construct (see
Appendix I. Also see Antonakis et al 2010, Antonakis et al 2014b, Bascle 2008, Duncan et al
2004, Larcker & Rusticus 2010).
To highlight the nature of the problem, we provide a brief example using an item from the
Idealized Influence Attributes scale of the MLQ (e.g., “Displays a sense of power and
confidence”). A leader could do more or less of the behavior or any other leader behavior for that
matter (Antonakis & House 2014, Bryman 1992, Hunt 1991), depending on various factors at the
leader level (e.g., how extraverted the leader is) or at the organizational level (e.g., selection,
training, or resources provision). These factors will correlate with outcomes of leadership too;
thus, omitting these factors from a predictive model cause endogeneity bias. Moreover, the
leader’s behavior may even depend on how performing a subordinate, workgroup, or
organization is (i.e., leaders become more confident with better performance, which does not
stem from the leader; or leaders may adjust their behavior and “crack the whip” more if followers
do not perform well). This bias is called simultaneity bias, a form of endogeneity that confounds
estimates. Indeed, many of the items of the MLQ measuring transformational leadership scale are
endogenous, as are items from most other leadership scales.
24
Apart from pure omitted variable bias, there are also biases affecting raters, stemming
from cognitive classification mechanisms. For instance, if a leader is effective due to other
factors that are not captured by the MLQ (e.g., instrumental/expert leadership, or charismatic
leadership in the true sense of the word), the leader will be classified as being effective, which
will trigger a cognitive “fill-in-the-blanks” mechanism (Cantor & Mischel 1977, Lord et al 1984);
in this way, any prototypically good factor of leadership will covary positively with the outcome
due to its correlation with the omitted cause (Antonakis & House 2014). There are also other
biases like affect for the leader, which may stem from many reasons other than those measured.
(see also Mount & Scullen 2001, Scullen et al 2000).
Finally, raters can also be biased directly because of “performance-cue” effects. That is,
raters who have knowledge of leader outcomes will be biased when rating behaviors that
theoretically can cause the outcomes (Jacquart & Antonakis 2015, Lord et al 1978, Meindl &
Ehrlich 1987). Again, via attribution mechanisms, raters are “filling-in-the-blanks” as described
above, but in this case the omitted variable is the outcome per se, which is also an endogenous
variable.
Questionnaire measures could certainly be used to study charisma; however, they must be
modeled as endogenous variables (i.e., outcome variables), predicted by what are known to be
exogenous variables such as manipulated variables or stable individual differences. Questionnaire
measures could be used as endogenous regressors of other endogenous variables if appropriate
statistical techniques are used to purge the scales from endogeneity bias (as discussed in the
Appendix); using these techniques can significantly change the validities of such measures
(Antonakis & House 2014). The use of such corrective procedures is unfortunately an exception
rather than the rule. Of course, issues of endogeneity bedevil all behavioral questionnaires, in
particular those that measure outcomes of leadership per se in an obvious way like quality of
25
leader-member exchange or LMX. Thus, much of the literature using questionnaire measures is
plagued by endogeneity issues and cannot inform theory or policy. To move forward we provide
suggestions about how charisma should be measured and a way that will avoid many of the issues
we identified above.
Operationalizing charisma: The way forward
As suggested in the previous section, using questionnaire measures of charisma is only
defensible if charisma is modeled as an endogenous variable. We think that the most promising
ways to measure charisma as an independent variable are either to (a) use unobtrusive and
objective measures that do not rely on perceptions of raters, like different types of archival data
(xxxxxxx in press) or (b) manipulate it directly in a laboratory or field experiment; perhaps one
day (c) better designed questionnaires will get to the gist of charisma.
As concerns unobtrusive and objective measures, Tucker (1968) was one of the first to
suggest objective measurement by recommending that markers of charisma be extracted from
various sources including biographical (i.e., historiometric) material, and this prior to a leader
achieving office so that charisma’s effects would not confounded with the outcomes of having
power. Given our definition of charisma (i.e., a values-based, symbolic, and emotion-laden leader
signaling), markers of leader signaling, both verbal and non-verbal, should be accessible in a
variety of artefacts, thus allowing for quantification and testing (Simonton 2003).
These artifacts could be speeches, video materials from interviews or other archival
sources or any other form of communication that are ideally detached from outcomes and reflect
an enduring pattern of leader signaling (with respect to values as well as symbolic and emotional
communication styles). Such an aggregate variable, if relatively stable over time and invariant
across situations (e.g., see Jacquart & Antonakis 2015, in rating speeches of U.S. presidential
26
contenders) will unlikely change as a function of what it predicts (i.e., outcomes) or of omitted
causes related to these outcomes.
Several researchers have used unobtrusive approaches by coding for markers of charisma.
For instance, one way to get a message across in a visual way (i.e., symbolically) is to use
metaphors, which can be reliably coded from speeches and which induces charisma (Mio et al
2005). Storytelling is also another technique, which is a bit more complex than metaphors are,
but which acts in a similar way (Towler 2003). Coding for value statements can be
straightforward (Frese et al 2003). In fact charismatic leaders use a variety of communication
techniques (Den Hartog & Verburg 1997, Shamir et al 1994) to communicate symbolically, and
these along with non-verbal delivery techniques (Awamleh & Gardner 1999, Frese et al 2003,
Towler 2003) can be reliably coded from text and video artefacts (refer to the "charismatic
leadership tactics" in Antonakis et al 2011, Antonakis et al 2012, Jacquart & Antonakis 2015).
Some simpler verbal techniques may also coded by computer (e.g., Davis & Gardner 2012).
To better understand charisma too, we think that more experimental studies should be
conducted, whether by exposing participants to different treatments (to intervene on the
participants’ leadership) or to expose them to someone trained to depict a particular leadership
style either in person or via video materials (Podsakoff et al 2013). To our knowledge, the first
study using a trained actor to manipulate charisma was undertaken by Howell and Frost (1989),
who manipulated charisma in terms of nonverbal delivery as well as verbal appeals. An example
of exposing participants to a video-taped manipulation includes Awamleh and Gardner (1999).
Depending on what is studied, even “paper people” experiments could be useful (e.g., Kosloff et
al 2010); however, these cannot fully capture the charismatic effect. Note, though, that most of
the experimental studies have been in the laboratory—such studies come with inherent
27
limitations insofar as external validity is concerned with respect to the ecological validity of the
task and the sample used.
We are hopeful too that researchers will develop questionnaire measures that better
capture charisma as an independent variable. However, getting to the point where charisma will
be adequately measured as an independent variable will be very challenging. Perhaps items could
better map on markers of charisma that researchers have manipulated (Antonakis et al 2011,
Frese et al 2003, Towler 2003), and which, because of their specificity, are not so prone to affect
or performance signal bias. For instance, perhaps researchers should consider asking raters to
evaluate to what extend the rated leader “uses metaphors when communicating,” “tell stories
often to convey a point,” “often poses rhetorical questions,” or “depicts choices in contrasts (i.e.,
black and white).” Raters, however, might find it hard to accurately recall such details, or such
details may even be imperceptible to them. That is, individuals may feel the consequence of
metaphor (or storytelling), but not explicitly discern and recall its use or consciously reflect on
such linguistic devices (cf. Lakoff & Johnson 1980). We thus think that extracting features from
video or archival material by trained coders or computer would be easier to use. Work sample
tests and role plays may also be useful in this regard.
State of the science and future research directions
Given its immense practical importance, we see a very rosy future for the charisma
construct. There is much to be done still in terms of discoveries. Of course, we need to first get
our theoretical house in order prior to using correct operationalizations of charisma and
appropriate causal tests. Researchers continue to be interested in this construct and at this rate we
think that some major advances will be made particularly if studying the construct from
multidisciplinary perspectives (e.g., psychology, economics, and biology).
28
In this section we briefly discuss where research using charisma should head, based on
our judgment, the coded articles, the gaps we have identified, and limitations we see in current
approaches. We supplement our suggestion, with brief reports on features we coded from the
articles (where relevant) so as to better understand and objectively demonstrate the kinds of
variables, contexts, samples and the like being studied by researchers (see Supplemental Material
link). We hope that this section will provide researchers with ideas about how to better study
charisma.
With respect to the articles we coded for in our review, of the 280 articles we included,
most (i.e., 112 articles) were quantitative. Qualitative articles constituted the next largest category
(90), followed by theory (69) and review (9) articles. In term of journal targeting, most papers on
charisma appeared in field journals with The Leadership Quarterly publishing the most articles
(68 articles), followed by Leadership (10 articles), and Journal of Applied Social Psychology (10
articles); in the fourth place came a top general journal, Academy of Management Review (7
articles). The top-five most cited articles, in order, included a theory article by Shamir et al.
(1993), followed by another theory article (Conger & Kanungo 1987), a critique (Yukl 1999), an
empirical article (House et al 1991) and finally another theory article (Gardner & Avolio 1998).
As concerns quantitative papers, from which it was quite straightforward to extract
information, most articles studied charisma as a dependent variable, though about a third studied
it as an independent variable. Only a few studies modelled charisma as a mediator or moderator;
note, we identified 98 other quantitative articles measuring charisma but we excluded them
because they failed to model charisma correctly; that is, charisma was measured as an
endogenous variable but used as a regressor (note, from these 98 articles, 77 of these modelled
charisma as an independent variable and 13 as a mediator).
29
With respect to sample locations, half came from the U.S.A.; although most of the work
on charisma is being done there, it is encouraging to see that researchers from other countries are
interested in studying this concept too. Important to note here is that charisma (or descriptions
similar to the term) is universally endorsed as a prototypical characteristic of effective leaders
(Brodbeck et al 2000, Den Hartog et al 1999, Koopman et al 1999). Thus, we hope to see much
more research on the topic conducted outside of the U.S.A. Doing so will also help in
establishing charisma’s boundary conditions across different cultures.
In terms of samples used about half of the samples used undergraduate or graduate
students as participants (one reason being that we excluded many field studies using working
populations due to endogeneity concerns), followed by participants coming from private and
public firms; about a fifth were politicians, with the rest coming from other contexts. Moreover,
in terms of outcomes predicted by charisma, most were perceptual and only about a quarter were
objective. Although perceptual measures are interesting to study, we hope to see more studies
using charisma to predict objective outcomes, particularly at the organizational level. We still do
not know really if and how charisma influences organizational outcomes (Yukl 1999) as well as
other levels-of-analysis effects (Waldman & Yammarino 1999). Indeed, most conceptualizations
and theories have strong individual and dyadic focus (Antonakis & Atwater 2002) and more
research on group as well as on organizational-level outcomes is sorely needed (Beyer 1999,
Yukl 1999). Still it is reassuring to see that there is some evidence showing that charisma does
matter when using objective outcome measures (e.g., Flynn & Staw 2004).
Additionally, although we would expect most processes to “work” in the same way across
samples (and cultures too), we hope to see more research using realistic contexts and this in
design conditions that allow for strong causal inference. Yet again, we are disappointed by the
fact that we only found one field experiment done with working populations (Antonakis et al
30
2011). Still, it is encouraging to see that about a third of the studies at least used an experimental
protocol and about a quarter archival data (the rest being field data). With respect to using
experiments and in addition to seeing more field experiments we hope to see too more realistic
alternative treatments (i.e., incentivized); simply comparing a charismatic treatment to a placebo
treatment does not provide for a strong test. Likewise, tasks that are oftentimes used, particularly
in laboratory settings lack ecological validity and are low stakes—we hope to see more
consequential tasks having real world analogues (i.e., like work productivity). In terms of
manipulations, most studies manipulated the content of the speech or the communication style of
the leader, which we find as a good sign. Also about a dozen studies used trained actors—all of
which were experimental (as would be expected). We think that this type of design affords very
strong experimental control and should be used more often.
There are several other recommendations we have to move the field forward. First, we
need to know more about the mediators and moderators of the charismatic effect (van
Knippenberg & Sitkin 2013). At this time, we know more about moderators including leader
distance (Shamir 1995), crisis (Bligh et al 2004), leader prototypicality (Van Knippenberg & Van
Knippenberg 2005), leader confidence (De Cremer & Van Knippenberg 2004), and attributional
ambiguity (Jacquart & Antonakis 2015), among others. However, mediation mechanisms are still
in the realm of the theoretical (Shamir et al 1993) and this is so because, unfortunately most
researchers have not tested how the effects of charisma are channelled (mediated) appropriately
in a causal way (Antonakis et al 2010).
Second, we need more complete theories and more direct tests of the theories. For
instance, the highest cited paper we coded (Shamir et al 1993), which happens to be a theory
paper, still has not been appropriately tested—using a correct causal specification (see Antonakis
et al 2010)—with respect to mediatory mechanisms (i.e., identity states).
31
Third, methodological standards need to be drastically improved; publishing research that
finds “associations” and “correlations” does not help practice advance. Also, synthesizing
endogenous correlations in meta-analyses simply makes for more precise endogenous
correlations, which cannot shed more light on cause-effect relations. It is unfortunate to observe
that a meta-analysis by DeGroot et al (2000) has reported that the MLQ was used almost
exclusively by researchers to measure charisma as a predictor. Another meta-analysis only used
the MLQ charisma scale by design (Fuller et al 1996)! There is a sore need for a well-done meta-
analysis using correct measures of charisma and this in models that are properly and causally
specified.
Fourth, we hope to see more qualitative research, ideally quantified and tested
appropriately (Eagly & Antonakis 2014, Simonton 2003). Purists might argue that such studies
are not qualitative anymore; however, even researchers with strong qualitative credentials have
suggested that quantification of qualitative data can be very useful (Maxwell 2010); such studies,
we think, will be able to shed some unique light onto the charisma phenomenon. Most of the
qualitative research we surveyed used case studies; these can be very idiosyncratic that cannot
make for very generalized findings. Multiple cases and comparisons are much more useful
(Eisenhardt & Graebner 2007), particularly if quantified (Antonakis et al 2014a).
Fifth, what individual differences predict charisma? Although we have suggested
charisma, in its pure form is a marker of leader ability (and as such is trait-like), there should be
some stable variables with which it correlates; that is, other measureable individual differences
may share some common variance with charisma too. For instance, production of creative
metaphors—an important marker of charisma (Antonakis et al 2011)—is predicted by general
intelligence (Silvia & Beaty 2012). Some pointers are provided in Bono and Judge (2004) with
respect to personality. However, there are hardly any articles linking one of the most-studied
32
individual difference variable, intelligence, to charisma. Also, we do not know enough about how
male and female leaders are seen by others, and how effective they are, when using charismatic
sources of influence.
Finally, other interesting avenues to explore would be linking perceptual measures of
charisma to biological individual differences like facial appearance (Todorov et al 2005, Trichas
& Schyns 2012), height (Hamstra 2014) or other factors (e.g., hormones), as well as to looking at
neuroscientific correlates (Waldman et al 2011). As concerns the latter there is some very
interesting research showing how charismatic rhetorical strategies like storytelling affects neuro-
endocrinological functioning (Barraza et al 2015, Barraza & Zak 2009, Speer et al 2009). Such
research would also benefit from using twin studies so as to determine—and this using objective
measures of charisma—the extent to which charisma is heritable.
Managerial and cross-cultural implications
Given the immense attention that practitioners pay to leadership as well as the
development of leaders and leadership production systems (Day 2000, Day et al 2014), it is
important to consider the practical importance of charisma as well as its generalization across
contexts. First, we know that charisma is considered a prototypical characteristic of effective
leadership and this across a wide array of cultures (Den Hartog et al 1999). Thus, we think that
our recommendations regarding how charisma should be operationalized and studied can be
applicable to a wide range of situations and contexts; however, we still need more carefully
designed research to determine what bounds charisma’s relation with other variables as a function
of culture or other moderating factors.
Second, although there is still a dearth of studies demonstrating how charisma affects
macro-organizational performance (in terms of productivity measures, assets, etc.), we do know
33
that charisma matters whether using subjective or objective measures of leader success and this
on various levels of analysis (but mostly on the micro-level). Third, is charisma—like
intelligence, or a confident and extraverted personality—a gift? If it is, that charisma matters for
performance suggests it is critical to have appropriate leaders in place by using well-designed
selection systems. However, as suggested by Etzioni (1961) decades ago, charisma can be trained
and it is encouraging to see that there are scientific studies demonstrating this point (Antonakis et
al 2011, Frese et al 2003, Towler 2003). Such studies have shown that leaders can be made aware
of different signaling techniques, both verbal (e.g., use of metaphor, contrasts, stating the
sentiments of the collective) and nonverbal (e.g., gesturing, facial expressions), that can be used
to make the leaders more charismatic. Also, the gains that can be made are quite respectable, with
a d = .62 (e.g., see Antonakis et al 2011). Thus, charisma is a “gift” in the sense that leaders can
receive it via well designed interventions! However, these interventions cannot be done in a
cursory manner (Antonakis et al 2011).
As concerns leader development efforts, there is far too much fad-driven thinking in the
world of practice (Zaccaro & Horn 2003); why precisely this is the case is anybody’s guess but it
probably has to do with commercial interests of consulting companies, self-proclaimed gurus,
lack of scientific training among practitioners, but also from badly designed and insufficiently
tested leadership models emanating from the world of academia that works its way into practice.
As concerns the latter, immensely popular book authors (e.g., of “Good-to-great” or “Built-to-
last” type books), after having studied successful companies, suggested that charisma does not
matter for organizational performance. Such books, however, are based on very faulty logic and
statistics, including (a) what is commonly referred to as sampling on the dependent variable—
that is, assuming what successful cases have in common drives their success without having
compared these cases to a control group (Denrell 2003)—or (b) regression to the mean
34
(Kahneman 2011), a statistical phenomenon explaining why successful performance can depend
on luck or random causes and with time it will regress back to mediocrity. Thus, it is very
important for practitioners, but also academicians too, to critically examine whether the
interventions they are using are evidence-based (Dietz et al 2014).
To conclude, leadership and in particular its charismatic form is very important for teams,
organizations, even countries. Moreover, charismatic leaders can be formal or informal: charisma
can be exhibited across all levels of institutions, from supervisors to team players in leaderless
teams, all the way up to CEO or country presidents; it can also be exhibited in social movements
and can work from close or a distance. Thus, what we have described is not a phenomenon that is
uniquely linked to a position; it is a process of influence. As such, it is imperative that institutions
find ways to identify and develop this process, and then to harness it for the greater good.
Conclusions
In this review, we showed how thinking on charisma has developed over time.
Contrasting various literatures, we showed how charisma has gone from being conceived as a
rather unknown and apparently divine quantity whose nature was revolutionary to a more
domesticated version that can be observed in various settings and operationalized to predict
various outcomes. We showed that some of the biggest obstacles the charisma construct has faced
included the lack of proper definition of charisma, the confounding of charisma with
transformational leadership, and the very lax, and endogeneity-plagued standards that have been
used to measure and model charisma. We hope that our review has helped clear the way to
developing a unified theory of charisma so that it can be studied it in a more robust and causally-
defensible way.
35
Summary points
1. Charisma continues to occupy a prominent role in leadership research.
2. The early literature on charisma was very vague about the nature of construct.
3. The two major streams of charisma, which conceptualize charisma in rather different
ways include (a) sociology and political science and (b) applied psychology and
management.
4. Charisma is often confused with transformational leadership; these are separate constructs
and charisma should be studied in its own right.
5. Many definitions of charisma have been tautological or unclear; it is best to think of
charisma from a signaling theory point of view.
6. Given the limitations of questionnaire measures, biases of raters, and other endogeneity
biases, charisma must be measured and operationalized in an objective way when used as
an independent variable.
7. There is a need to study the impact of charisma on objective outcomes and using strong
causal designs.
8. Charisma can be developed.
Related resources
The following TEDx talk, delivered by the first author of the study, demonstrates the importance
of charisma. This resource could be a useful complement in education and training:
https://youtu.be/SEDvD1IICfE
36
References
Antonakis J. 2012. Transformational and Charismatic Leadership. In The nature of leadership,
ed. DV Day, J Antonakis, pp. 256-88. Thousand Oaks: Sage Publications
Antonakis J, Atwater LE. 2002. Leader distance: A review and a proposed theory. The
Leadership Quarterly 13: 673-704
Antonakis J, Avolio BJ, Sivasubramaniam N. 2003. Context and leadership: An examination of
the nine-factor full-range leadership theory using the Multifactor Leadership
Questionnaire. The Leadership Quarterly 14: 261-95
Antonakis J, Bastardoz N, Liu Y, Schriesheim CA. 2014a. What makes articles highly cited? The
Leadership Quarterly 25: 152-79
Antonakis J, Bendahan S, Jacquart P, Lalive R. 2010. On making causal claims: A review and
recommendations. The Leadership Quarterly 21: 1086-120
Antonakis J, Bendahan S, Jacquart P, Lalive R. 2014b. Causality and endogeneity: Problems and
solutions. In The Oxford Handbook of Leadership and Organizations, ed. DV Day, pp.
93-117. New York: Oxford University Press
Antonakis J, Fenley M, Liechti S. 2011. Can Charisma Be Taught? Tests of Two Interventions.
The Academy of Management Learning and Education 10: 374-96
Antonakis J, Fenley M, Liechti S. 2012. Learning charisma: Transform yourself into someone
people want to follow. Harvard Business Review June: 127-30
Antonakis J, House RJ. 2014. Instrumental leadership: Measurement and extension of
transformational–transactional leadership theory. The Leadership Quarterly 25: 746-71
Aristotle, Butcher SH. 2011. Poetics: CreateSpace
Aristotle, Roberts WR, Bywater I, Solmsen F. 1954. Rhetoric. New York,: Modern Library
Awamleh R, Gardner WL. 1999. Perceptions of leader charisma and effectiveness: The effects of
vision content, delivery, and organizational performance. The Leadership Quarterly 10:
345-73
Bacharach SB. 1989. Organizational theories: Some criteria for evaluation. Academy of
Management Review 14: 496-515
Barraza JA, Alexander V, Beavin LE, Terris ET, Zak PJ. 2015. The heart of the story: Peripheral
physiology during narrative exposure predicts charitable giving. Biological Psychology
105: 138-43
Barraza JA, Zak PJ. 2009. Empathy toward Strangers Triggers Oxytocin Release and Subsequent
Generosity. Annals of the New York Academy of Sciences 1167: 182-89
Bascle G. 2008. Controlling for endogeneity with instrumental variables in strategic management
research. Strategic Organization 6: 285-327
Bass BM. 1985. Leadership and performance beyond expectations. New York: The Free Press
Bass BM. 1990. Bass & Stogdill's handbook of leadership: Theory, research, and managerial
applications. New York: Free Press. xv, 1182 p. pp.
Bass BM, Avolio BJ. 1995. MLQ Multifactor leadership questionnaire for research: Permission
set. Redwood City, CA: Mindgarden
Bensman J, Givant M. 1975. Charisma and modernity: The use and abuse of a concept. Social
research: 570-614
Berlew DE. 1974. Leadership and organizational excitement. California Management Review 17:
21-30
Beyer JM. 1999. Taming and promoting charisma to change organziations. The Leadership
Quarterly 10: 307-30
37
Bligh MC, Kohles JC, Meindl JR. 2004. Charisma under crisis: Presidential leadership, rhetoric,
and media responses before and after the september 11th terrorist attacks. The Leadership
Quarterly 2: 211-39
Bollen KA. 2012. Instrumental Variables in Sociology and the Social Sciences. Annual Review of
Sociology 38: 37-72
Bono JE, Judge TA. 2004. Personality and Transformational and Transactional Leadership: A
Meta-Analysis. Journal of Applied Psychology 89: 901-10
Bouchard TJ, Loehlin JC. 2001. Genes, evolution, and personality. Behavior Genetics 31: 243-73
Bouchard TJ, McGue M. 2003. Genetic and environmental influences on human psychological
differences. Journal of Neurobiology 54: 4-45
Brodbeck FC, Frese M, Akerblom S, Audia G, Bakacsi G, et al. 2000. Cultural ariation of
Leadership Prototypes Across 22 European Countries. Journal of Occupational and
Organizational Psychology 73: 1-73
Brown DJ, Lord RG. 1999. The utility of experimental research in the study of
transformational/charismatic leadership. The Leadership Quarterly 10: 531-39
Bryman A. 1992. Charisma and leadership in organizations. London: Sage Publications
Burns JM. 1978. Leadership. New York: Harper & Row
Cantor N, Mischel W. 1977. Traits as prototypes: Effects on recognition memory. Journal of
Personality and Social Psychology 35: 38-48
Conger JA. 1998. Qualitative research as the cornerstone methodology for understanding
leadership. The Leadership Quarterly 9: 107-21
Conger JA, Kanungo RN. 1987. Toward a behavioral theory of charismatic leadership in
organizational settings. Academy of Management Review 12: 637-47
Conger JA, Kanungo RN, eds. 1988. Charismatic leadership: The elusive factor in
organizational effectiveness. San Francisco: Jossey-Bass Publishers
Conger JA, Kanungo RN. 1998. Charismatic leadership in organizations. Thousand Oaks, CA:
Sage Publications.
Conger JA, Kanungo RN, Menon ST. 2000. Charismatic Leadership and follower effects.
Journal of Organizational Behavior 21: 747-67
Davies JC. 1954. Charisma in the 1952 Campaign. American Political Science Review 48: 1083-
102
Davis KM, Gardner WL. 2012. Charisma under crisis revisited: Presidential leadership, perceived
leader effectiveness, and contextual influences. The Leadership Quarterly 23: 918-33
Day DV. 2000. Leadership development: A review in context. The Leadership Quarterly 11:
581-613
Day DV. 2012. The future of leadership: Challenges and prospects. In The Oxford Handbook of
Leadership and Organizations, ed. DV Day, pp. 853-61. New York: Oxford
Day DV, Antonakis J. 2013. The future of leadership In The Wiley-Blackwell Handbook of the
Psychology of Leadership, Change and Organizational Development, ed. HS Leonard, R
Lewis, AM Freedman, J Passmore, pp. 221–35. Oxford: John Wiley & Sons
Day DV, Fleenor JW, Atwater LE, Sturm RE, McKee RA. 2014. Advances in leader and
leadership development: A review of 25 years of research and theory. The Leadership
Quarterly 25: 63-82
De Cremer D, Van Knippenberg D. 2004. Leader self-sacrifice and leadership effectiveness: The
moderating role of leader self-confidence. Organizational Behavior and Human Decision
Processes 95: 140-55
38
DeGroot T, Kiker DS, Cross TC. 2000. A Meta-Analysis to Review Organizational Outcomes
Related Charismatic Leadership. Canadian Journal of Administrative Sciences 17: 356-72
Den Hartog DN, House RJ, Hanges PJ, Ruiz-Quintanilla SA. 1999. Culture specific and cross-
culturally generalizable implicit leadership theories: Are attributes of
charismatic/transformational leadership universally endorsed? The Leadership Quarterly
10: 219-56
Den Hartog DN, Verburg RM. 1997. Charisma and rhetoric: Communicative techniques of
international business leaders. The Leadership Quarterly 8: 355-91
Denrell J. 2003. Vicarious learning, undersampling of failure, and the myths of management.
Organization Science 14: 227-43
Dietz J, Antonakis J, Hoffrage U, Krings F, Marewski J, Zehnder C. 2014. Teaching evidence-
based management with a focus on producing local evidence. Academy of Management
Learning & Education 13: 397-414
Downton JV. 1973. Rebel leadership: Commitment and charisma in the revolutionary process.
New York: The Free Press
Duncan GJ, Magnusson KA, Ludwig J. 2004. The Endogeneity Problem in Developmental
Studies. Research in Human Development 1: 59-80
Eagly AH, Antonakis J. 2014. Leadership. In APA Handbook of Personality and Social
Psychology: Attitudes and Social Cognition, ed. G Borgida, J Bargh, pp. 571-92.
Washington: APA Books
Eisenhardt KM, Graebner ME. 2007. Theory building from cases: Opportunities and challenges.
Academy of Management Journal 50: 25-32
Emrich CG, Brower HH, Feldman JM, Garland H. 2001. Images in words: Presidential rhetoric,
charisma, and greatness. Administrative Science Quarterly 46: 527–57
Endler NS, Magnusson D. 1976. Toward an interactional psychology of personality.
Psychological Bulletin 83: 956
Erickson DJ, Krull DS. 1999. Distinguishing Judgments About What From Judgments About
Why: Effects of Behavior Extremity on Correspondent Inferences and Causal
Attributions. Basic and Applied Social Psychology 21: 1-11
Etzioni A. 1961. A comparative analysis of complex organizations. New York: The Free Press
Etzioni A. 1964. Modern organizations. Englewood Cliffs, New Jersey: Prentice-Hall
Fanelli A, Misangyi VF, Tosi HL. 2009. In Charisma We Trust: The Effects of CEO Charismatic
Visions on Securities Analysts. Organization Science 20: 1011-33
Fiedler FE. 1967. A theory of leadership effectiveness. New York: McGraw-Hill
Fleishman EA. 1953. The Measurement of Leadership Attitudes in Industry. Journal of Applied
Psychology 37: 153-58
Flynn FJ, Staw BM. 2004. Lend me your wallets: The effect of charismatic leadership on external
support for an organization. Strategic Management Journal 25: 309-30
Frese M, Beimel S, Schoenborn S. 2003. Action training for charismatic leadership: Two
evaluations of studies of a commercial training module on inspirational communication of
a vision. Personnel Psychology 56: 671-97
Friedland WH. 1964. For a sociological concept of charisma. Social Forces 43: 18-26
Friedrich CJ. 1961. Political Leadership and the Problem of the Charismatic Power. Journal of
Politics 23: 3-24
Fuller JB, Patterson CEP, Hester K, Stringer DY. 1996. A quantitative review of research on
charismatic leadership. Psychological Reports 78: 271-87
39
Gardner WL, Avolio BJ. 1998. The charismatic relationship: A dramaturgical perspective.
Academy of Management Review 23: 32-58
Gardner WL, Lowe KB, Moss TW, Mahoney KT, Cogliser CC. 2010. Scholarly Leadership of
the Study of Leadership: A Review of The Leadership Quarterly's Second Decade, 2000-
2009. The Leadership Quarterly 12: 922-58
Gemmill G, Oakley J. 1992. Leadership: An Alienating Social Myth? Human Relations 45: 113-
29
Greene CN. 1977. Disenchantment with leadership research: Some causes, recommendations,
and alternative directions. In Leadership: The cutting edge. , ed. JG Hunt, LL Larson, pp.
57-67. Carbondale: Southern Illinois University Press
Hamstra MRW. 2014. ‘Big’ men: Male leaders’ height positively relates to followers’ perception
of charisma. Personality and Individual Differences 56: 190-92
Hermalin BE. 1998. Toward an Economic Theory of Leadership: Leading by Example. The
American Economic Review 88: 1188-206
Hogg MA. 2001. A social identity theory of leadership. Personality & Social Psychology Review
5: 184-200
House RJ. 1977. A 1976 Theory of Charismatic Leadership. In The Cutting Edge, ed. JG Hunt,
LL Larson, pp. 189-207. Carbondale: Southern Illinois: University Press
House RJ. 1999. Weber and the Neo-charismatic Leadership Paradigm: A Response to Beyer.
The Leadership Quarterly 10: 563-74
House RJ, Howell JM. 1992. Personality and charismatic leadership. The Leadership Quarterly
3: 81-108
House RJ, Spangler WD, Woycke J. 1991. Personality and charisma and the U.S. presidency: A
psychological theory of leader effectiveness. Administrative Science Quarterly 36: 364-96
Howell JM. 1988. Two faces of charisma: Socialized and personalized leadership in
organizations. In Charismatic leadership: The elusive factor in organizational
effectiveness, ed. JA Conger, RN Kanungo, pp. 213-36. San Francisco: Jossey-Bass
Publishers
Howell JM, Frost PJ. 1989. A laboratory study of charismatic leadership. Organizational
Behavior and Human Decision Processes 43: 243-69
Hunt JG. 1991. Leadership: A new synthesis. Newbury Park, CA: Sage Publications
Hunt JG. 1999. Tranformational/charismatic leadership's transformation of the field: An
historical essay. The Leadership Quarterly 10: 129-44
Jacquart P, Antonakis J. 2015. When does charisma matter for top-level leaders? Effect of
attributional ambiguity. Academy of Management Journal 58
Kahneman D. 2011. Thinking, fast and slow. New York: Farrar, Straus and Giroux. 499 p. pp.
Katz D, Maccoby N, Gurin G, Floor LG. 1951. Productivity, supervision and morale among
railroad workers. Ann Arbor, Michigan: Survey Research Center, Institute for Social
Research (University of Michigan)
Koopman PL, Den Hartog D, Konrad E, Akerblom S, Audia G, et al. 1999. National Culture and
Leadership Profiles in Europe: Some Results From the GLOBE Study. European Work
and Organizational Psychology 8: 503-20
Kosloff S, Greenberg J, Weise D, Solomon S. 2010. The effects of mortality salience on political
preferences: The roles of charisma and political orientation. Journal of Experimental
Social Psychology 46: 139-45
Lakoff G, Johnson M. 1980. Metaphors we live by. Chicago: University of Chicago Press. xiii,
242 p. pp.
40
Larcker DF, Rusticus TO. 2010. On the use of instrumental variables in accounting research.
Journal of Accounting and Economics 49: 186-205
Lord RG, Binning JF, Rush MC, Thomas JC. 1978. The effect of performance cues and leader
behavior on questionnaire ratings of leadership behavior. Organizational Behavior and
Human Performance 21: 27-39
Lord RG, Foti RJ, De Vader CL. 1984. A Test of Leadership Categorization Theory: Internal
Structure, Information Processing, and Leadership Perceptions. Organizational Behavior
and Human Performance 34: 343-78
Lowe KB, Gardner WL. 2000. Ten Years of the Leadership Quarterly: Contributions and
Challenges for the Future. The Leadership Quarterly 11: 459-514
MacKenzie SB. 2003. The dangers of poor construct conceptualization. Journal of the Academy
of Marketing Science 31: 323-26
Maclachlan B. 1996. The Age of Grace: Charis in Early Greek Poetry. Chichester: Princeton
Maxwell JA. 2010. Using Numbers in Qualitative Research. Qualitative Inquiry 16: 475–82
Meindl JR, Ehrlich SB. 1987. The romance of leadership and the evaluation of organizational
performance. Academy of Management Journal 30: 91-109
Miner JB. 1975. The uncertain future of the leadership concept. An overview. In Leadership
Frontiers, ed. JG Hunt, LL Larson, pp. 197-208. Kent, OH: Kent State University
Mio JS, Riggio RE, Levin S, Reese R. 2005. Presidential leadership and charisma: The effects of
metaphor. The Leadership Quarterly 16: 287-94
Mount MK, Scullen SE. 2001. Multisource feedback ratings: What do they really measure? In
How people evaluate others in organizations, ed. M London, pp. 155-76. Mahwah, NJ:
Lawrence Erlbaum
Mumford MD, Antes AL, Caughron JJ, Friedrich TL. 2008. Charismatic, ideological, and
pragmatic leadership: Multi-level influences on emergence and performance. The
Leadership Quarterly 19: 144-60
Podsakoff NP, Podsakoff PM, MacKenzie SB, Klinger RL. 2013. Are we really measuring what
we say we're measuring? Using video techniques to supplement traditional construct
validation procedures. Journal of Applied Psychology 98: 99-113
Riesebrodt M. 1999. Charisma in Max Weber's Sociology of Religion. Religion 29: 1-14
Scullen SE, Mount MK, Goff M. 2000. Understanding the latent structure of job performance
ratings. Journal of Applied Psychology 85: 956-70
Shamir B. 1995. Social Distance and Charisma - Theoretical Notes and an Exploratory-Study.
The Leadership Quarterly 6: 19-47
Shamir B. 1999. Taming charisma for better understanding and greater usefulness: A response to
beyer. The Leadership Quarterly 10: 555-62
Shamir B, Arthur MB, House RJ. 1994. The rhetoric of charismatic leadership: A theoretical
extension, a case study, and implications for research. The Leadership Quarterly 5: 25-42
Shamir B, House RJ, Arthur MB. 1993. The motivational effects of charismatic leadership: A
self-concept based theory. Organization Science 4: 577-94
Shamir B, Howell JM. 1999. Organizational and contextual influences on the emergence and
effectiveness of charismatic leadership. The Leadership Quarterly 10: 257-83
Shamir B, Zakay E, Breinin E, Popper M. 1998. Correlates of charismatic leader behavior in
military units: Subordinates' attitudes, unit characteristics, and superiors' appraisals of
leader performance. Academy of Management Journal 41: 387-409
Shils E. 1965. Charisma, Order, and Status. American Sociological Review 30: 199-213
41
Silvia PJ, Beaty RE. 2012. Making creative metaphors: The importance of fluid intelligence for
creative thought. Intelligence 40: 343-51
Simonton DK. 1988. Presidential style: Personality, biography, and performance. Journal of
Personality and Social Psychology 55: 928-36
Simonton DK. 2003. Qualitative and Quantitative Analyses of Historical Data. Annual Review of
Psychology 54: 617-40
Smith DN. 1998. Faith, reason, and charisma: Rudolf Sohm, Max Weber, and the theology of
grace. Sociological Inquiry 68: 32-60
Sohm R. 1892. Kirchenrecht: Verlag von Duncker & Humblot
Speer NK, Reynolds JR, Swallow KM, Zacks JM. 2009. Reading Stories Activates Neural
Representations of Visual and Motor Experiences. Psychological Science 20: 989-99
Spence M. 2002. Signaling in Retrospect and the Informational Structure of Markets. The
American Economic Review 92: 434-59
Spencer ME. 1973. What is charisma? British Journal of Sociology 24: 341-54
Stogdill RM. 1963. Manual for the Leader Behavior Description Questionnaire, Form XII.
Columbus, OH: Fisher College of Business: The Ohio State University.
Sturm RE, Antonakis J. 2015. Interpersonal Power: A Review, Critique, and Research Agenda.
Journal of Management 41: 136-63
Todorov A, Mandisodza AN, Goren A, Hall CC. 2005. Inferences of competence from faces
predict election outcomes. Science 308: 1623-26
Towler AJ. 2003. Effects of charismatic influence training on attitudes, behavior, and
performance. Personnel psychology 56: 363 - 81
Trichas S, Schyns B. 2012. The face of leadership: Perceiving leaders from facial expression. The
Leadership Quarterly 23: 545-66
Tsai W, Wu C-H. 2010. Knowledge combination: A cocitation analysis. Academy of
Management Journal 53: 441-50
Tucker RC. 1968. The theory of charismatic leadership. Daedalus 97: 731-56
Van Knippenberg B, Van Knippenberg D. 2005. Leader self-sacrifice and leadership
effectiveness: the moderating role of leader prototypicality. Journal of Applied
Psychology 90: 25
van Knippenberg D, Sitkin SB. 2013. A Critical Assessment of Charismatic—Transformational
Leadership Research: Back to the Drawing Board? The Academy of Management Annals
7: 1-60
Waldman DA, Balthazard PA, Peterson SJ. 2011. Leadership and Neuroscience: Can We
Revolutionize the Way That Inspirational Leaders Are Identified and Developed?
Academy of Management Perspectives 25: 60-74
Waldman DA, Yammarino FJ. 1999. CEO charismatic leadership: Levels-of-management and
levels-of-analysis effects. Academy of Management Review 24: 266-85
Weber M. 1947. The theory of social and economic organization. New York: The Free Press
Weber M. 1968. Max Weber on charisma and institutional building (Ed. S. N. Eisenstadt).
Chicago: The University of Chicago Press
White HD, Griffith BC. 1981. Author cocitation: A literature measure of intellectual structure.
Journal of the American Society for information Science 32: 163-71
Willner AR. 1984. The spellbinders: Charismatic political leadership. New Haven: CT.: Yale
University Press
Wowak AJ, Mannor MJ, Arrfelt M, McNamara G. 2014. Earthquake or glacier? How CEO
Charisma Manifests In Firm Strategy Over Time. Strategic Management Journal
42
xxxxxxx. in press. Archival Data in Micro-Organizational Research: A Toolkit for Moving to a
Broader Set of Topics". Journal of Management
Yukl GA. 1999. An evaluation of conceptual weaknesses in transformational and charismatic
leadership theories. The Leadership Quarterly 10: 285-305
Zaccaro SJ, Horn ZNJ. 2003. Leadership theory and practice: Fostering an effective symbiosis.
The Leadership Quarterly 14: 769-806
43
Figure 1A: The core of the intellectual landscape of the socio-scientific study of charisma
Note: This document co-citation network map shows up to .10% most co-cited works within the
overall co-citation network derived from the 280 articles analyzed in the review. Each node
represents one document (only first authors are named) and edges represent the co-citations of
two documents. Darker and thicker edges indicate more co-citations. Node size indicates the
number of other documents along which this node is co-cited (i.e., its degree), and darker color
indicates greater strength (where strength is the sum of the weights of all edges to which a node is
connected). An interactive version of this network can be accessed online at the following
address: http://www.hec.unil.ch/jantonakis/annualreviews/index.html
44
Figure 1B: An author co-citation map of the socio-scientific study of charisma
Note: This co-citation map is based on first authors only. Difference in node size indicates relative difference in the frequency of occurrence.
Sources that are frequently co-cited are positioned close to each other. Only the 200 strongest edges are represented here. Node color indicates
membership to a common densely connected cluster, as identified by an algorithm. This figure also distinguishes discipline clusters (a) sociology
and political science (green nodes), (b) applied psychology and management (red nodes), and (c) social psychology (blue nodes).
45
Figure 2: Publishing Trends: Number of articles using the term “charisma” or variants
Note: We searched for articles using the term charisma and all combinations thereof in the Web of Science, published in 2014 or
before. The search returned 2,438 articles of which we included 280 (see Figure 2). We excluded articles not discussing charisma per
se or only tackling it cursorily. We excluded articles not written in English, book reviews, focusing on non-human charisma, scale-
development and testing articles, commentaries, editorials, book chapters, and conference abstracts. We excluded quantitative articles
and meta-analyses using an endogenous measure of charisma as an independent variable, moderator or mediator, and failing to
“instrument” charisma correctly (Antonakis et al 2010). For details on the coding refer to the online Supplementary Material.
0
25
50
75
100
125 150 175
200
Number of Articles
1954 1964 1974 1984 1994 2004 2014
Year
Included Excluded Total
Inclusion decision for coding
46
Table 1: Definitions of Charisma
Element of definition
Percentage of
definitions
using attribute
(weighted)
Percentage of
definitions
using attribute
(unweighted)
Useful
for a
definition?
1 Quality, ability, gift of the leader
84.95
71.43
No
2 Exceptional, extraordinary, exemplary individual
62.37
28.57
No
3 Defined in terms of an outcome
22.58
42.86
No
4 Vision, ideology, values, morals, beliefs, mission, symbols of leader
15.05
28.57
Yes
5 Followers' attribution (including group prototypicality of leader)
13.98
18.37
No
6 Social process (interaction, relationship)
11.83
22.45
No
7 Divine related
8.60
16.33
No
8 Emotion-based
8.60
16.33
Yes
9 Leader's behaviors and actions
8.60
16.33
No
10 Followers' characteristics (need, motives, background, self-esteem)
7.53
14.29
No
11 Expressive communication
6.45
12.24
Yes
12 Any contextual circumstances (crisis, social situation)
6.45
12.24
No
13 Leader trait (e.g., self-confidence, persistence, passion, optimism, honest, reliable)
4.30
8.16
No
Note: The weighted data uses 93 articles as the unit of analysis, some of which used the same definition. That is (a) 41 articles used the
definition “a certain quality of an individual personality by virtue of which he is set apart from ordinary men and treated as endowed
with supernatural, superhuman, or at least specifically exceptional powers or qualities” (coded under categories 1 and 2), and (b) 5
articles used the definition “An extraordinary quality of a person, regardless of whether this quality is actual, alleged, or presumed”
(coded under categories 1, 2, and 5). The unweighted data uses the definition as the unit of analysis.
47
Appendix
Endogeneity is an undesirable property of an estimator that renders estimates inconsistent, and
such estimates do not capture the causal relation between variables and hence cannot inform
policy (for a detailed introduction to the problem, see: Antonakis et al 2010, Antonakis et al
2014b). Inconsistent estimates do not converge to the true population values, asymptotically (i.e.,
as the sample increases). An endogenous variable is one that depends on other variables.
Exogenous variables—for example, a manipulated variables in an experiment, variables that vary
naturally in nature (e.g., temperature), are fixed by some process (e.g., latitude), are heritable to a
large extent or fixed in adulthood (e.g., intelligence and personality), are cyclical (e.g., election
cycles) and so forth—cannot vary as a function of other variables in or omitted from a model.
However, an endogenous variable is caused by other variables (i.e., q); if those variables are
omitted from the model but correlate with the outcome (i.e., y) that the endogenous variable (i.e.,
x) is supposed to predict, the coefficient of x on y, that is, the ordinary least squares (OLS) or
maximum likelihood (ML) estimate,
𝐶𝑜𝑣(𝑦,𝑥)
𝑉𝑎𝑟(𝑥)
, cannot be interpreted and will be confounded; this
confounding depends on the strength and direction of the relation of q to y and q to x (Antonakis
et al 2010). Endogenous variables can be used as regressors only if they have been
“instrumented”—that is, an exogenous source of variance (z) is used to purge the estimate from
endogeneity bias in a model of the form z x y. Note, this mediation model cannot be
estimated using the “usual” mediation methods in the organization sciences, whether
bootstrapped or not; an instrumental variable estimator must be used to estimate this model.
Because z is exogenous, it will not vary as a function of omitted causes of y or x; thus the
instrumental variable estimate,
𝐶𝑜𝑣(𝑦,𝑧)
𝐶𝑜𝑣(𝑥,𝑧)
, whether estimated by two-stage least squares (2SLS) or
maximum likelihood (where cross equation disturbances of x and y are correlated, see Antonakis
et al 2010) will be consistent and capture the true effect of x on y (Bollen 2012). To the extent
that x is a true cause of y—and as the instrumental variable formula shows—z must correlate both
with y and with x.
On-line supplementary materials
Charisma: An ill-defined and ill-measured gift
John Antonakis
Department of Organizational Behavior
University of Lausanne
john.antonakis@unil.ch
Nicolas Bastardoz
Department of Organizational Behavior
University of Lausanne
nicolas.bastardoz@unil.ch
Philippe Jacquart
EM-LYON Business School
jacquart@em-lyon.com
Boas Shamir
Department of Sociology and Anthropology
The Hebrew University of Jerusalem
Contents
Introduction ...................................................................................................................................... 1
Coding protocol ................................................................................................................................ 1
Which types of articles have had the biggest impact? ..................................................................... 3
Coding trends ................................................................................................................................... 4
For the correlation matrix of the key variables in coded papers, refer to Table 13. ........................ 7
Co-citation analyses ......................................................................................................................... 7
Table 1: Predicting Citations by Article Type ............................................................................... 10
Table 2: Trend for conceptualization of charisma.......................................................................... 11
Table 3: Trend for type of article ................................................................................................... 12
Table 4: Trend for sample occupation ............................................................................................ 13
Table 5: Trend for type of design ................................................................................................... 14
Table 6: Trend for type of variable ................................................................................................ 15
Table 7: Trend for type of leader ................................................................................................... 16
Table 8: Trend for measure of charisma ........................................................................................ 17
Table 9: Trend for use of MLQ and other measures ...................................................................... 18
Table 10: Trend for location of the sample .................................................................................... 19
Table 11: Trend for design of qualitative studies ........................................................................... 20
Table 12: Qualitative articles location of the sample ..................................................................... 21
Table 13: Correlation matrix of key variables in coded papers ..................................................... 22
Figure 1A: The intellectual landscape of the socio-scientific study of charisma ........................... 23
Figure 1B: The intellectual landscape of the socio-scientific study of charisma ........................... 24
Figure 2: Source co-citation analysis of the socio-scientific study of charisma ............................ 25
References ...................................................................................................................................... 26
1
Introduction
In this document, we present details on the coding protocol we used, coder reliability, as
well as results and trends we uncovered from the features we coded from articles. We also
discuss the co-citation analysis more extensively and present additional results.
Coding protocol
We developed a coding manual based on discussions between the authors about relevant
features to be extracted from the articles. The coding categories were refined after numerous
rounds of discussion following the coding of a sample of papers. We trained two graduate
students to undertake the coding. From the 280 articles that had to be coded
1
, we first used 54
articles as training materials and each of these articles was independently coded by the graduate
students. We also used this initial phase to refine the coding variables and categories. The first
and second author of the article resolved inconsistencies between the coders. In the second phase,
each coder was assigned 113 articles to code. Once completed, we gave each coder a set of 20
randomly determined articles previously coded by the other coder to compute the kappa statistic.
These 40 articles generated 422 coding events. The coders agreed on 87.44% of the coding
events. The agreement that could have been expected due to chance would have been 24.11%.
Thus, the agreement statistic (Landis & Koch 1977), κ = .83, SE = .02, z = 34.31, p < .001,
indicated that agreement was above chance.
1
Important to note is that there were substantially more quantitative articles in the initial pool (210 articles
altogether); however, we discarded 98 articles because they did not take appropriate corrective actions with respect
to using (a) the MLQ or similar measures as independent variables or (b) used MLQ-type measures as endogenous
regressors (i.e., mediators). Also, although we included in this review only quantitative articles that reported
unconfounded estimates when charisma was used as a regressor, and which thus may be of better quality than the
other categories of articles, we did not rate qualitative articles for quality given that there are no standards for doing
so. Note though that in Antonakis et al (2014) all quantitative and qualitative articles were included in the review
regardless of quality, and we found the same results as they did in the present articles (i.e., that qualitative articles are
undercited). There are several possibilities as to why qualitative articles are less appreciated by researchers in
general, and we do not discuss these here extensively (for a discussion see Antonakis et al 2014).
2
For each coded article, we extracted the following information:
(a) Age of article: we subtracted the year of publication from 2014;
(b) Citations: downloaded from WoS on 6 April 2015;
(c) 5-year WoS impact factor of the journal (which is highly stable: e.g., as a
demonstration, and using all journals in Applied Psychology, Business, Management, and Social
Psychology, over the periods 2001-2011 gives an ICC1 > .90) or the average impact factor across
the years available (coded missing for unranked journals);
(d) Information on the journal category and the impact factor of each category obtained
from WoS;
(e) Author affiliations: we utilized information from a well-known ranking source (QS
World University Ranking). Because rankings remain relatively stable over time (see Antonakis
et al 2014), we used the average rankings over the years 2010-2014. Each affiliation got a
ranking and rankings were reverse coded and averaged across coauthors for each article to obtain
a collective affiliation score. Unranked universities received the lowest ranking + 1;
(f) Number of cited references: Extracted from WoS;
(g) Article type: Categories included quantitative, qualitative, theory, and review. Note, if
an article coded qualitative data quantitatively, we coded the article as quantitative; if an article
made an empirical contribution, either quantitatively or qualitatively, after it had developed a
theory or some propositions, it was coded as quantitative or qualitative;
(h) Definition of charisma: coded if the authors defined charisma, either in their own
words or citing someone else definition of charisma.
We also looked at trends regarding some main characteristics of quantitative and, to a
lesser extent, qualitative articles. The next section provides more information on the trends and
the variables coded.
3
Which types of articles have had the biggest impact?
We examined whether article types were associated with different citation rates, which are
one of the best gauges of article impact (Bergh et al 2006, Lokker et al 2008). In order to see how
the “citation market” for charisma appreciates these different modes of inquiry we modelled the
citations as a function of various control variables that have been established to predict citations
(see Table 3 in Antonakis et al 2014). Given that we are predicting citations across a large time-
span from articles stemming from various journals and disciplines, we included the following
additional controls (a) a quadratic term for article age given the “decay” of citations over time
(Aksnes 2003), (b) the impact factor of the journal, given that papers published in higher impact
journals typically collect more citations (Larivière & Gingras 2010) and (c) the mean impact
factor of all journals across all fields in which the journal is listed because fields have substantial
different citations rates due to varying research practices (Antonakis & Lalive 2008).
Results from a zero-inflated negative binomial model (Blevins et al 2015) showed that we
could predict a hefty amount of the variance in citations, whether using lifetime citations or one
that models current influence (i.e., citations per year). Looking at lifetime citations indicates that,
holding constant the rest of the controls at the means, quantitative (38.13 citations), review (55.85
citations), and theory (37.87) papers all received significantly more citations than did qualitative
(27.53 citations) papers. We found similar results when predicting citations per year (see Table
1).
[Table 1 here]
To supplement the above analysis on predicting citations, we also looked at the
composition of the top-forty cited articles. Most used a quantitative design (i.e., 18), followed by
theory papers (15), qualitative papers (4) and reviews (3); this distribution was significantly
4
different (likelihood ratio
2
(3) = 17.77, p < .001), as would be expected given the different
proportion of coded articles. Thus, to make for a fairer comparison, in terms of proportion of
articles in the top forty as a proportion of articles coded, two-tailed binomial probability tests
showed that quantitative articles are overrepresented in the top-40 (p = .09), as were theory
papers (p = .07); review articles were appropriately represented (p > .10), and qualitative papers
were underrepresented (p < .01). These findings, coupled with the previous findings on predicting
citations suggest that researchers will get more “cites for their buck” by publishing quantitative,
theory or review papers on the topic of charisma.
Coding trends
In order to determine whether a trend was present in features of the coded data, we
estimated robust regression models (Huber 1964), which discount outliers (using Stata 13). At
first, we included the year and the year squared as regressors (we included the latter to capture
possible quadratic effects). If the coefficient for the year squared was significant (at p<.05) and
positive, we reported a quadratic positive trend (“++” sign in column trend). If the coefficient
was not significant, we then ran a simple linear model with only year as regressor; if the latter
model was significant, we reported a linear trend (“+” sign in column trend). If the regression did
not converge (e.g., when there were too many zero observations), we replicated the same
procedure just described above but using an OLS estimator with robust standard errors.
We will now turn to the coded variables of interest, presenting how they were coded as
well as descriptions and trends.
1. The conceptualization of charisma: This category examined the perspective from which
the authors conceptualized charisma, including conceptualized as (a) personality traits or
combination of traits; (b) a set of leader behaviors; (c) a perception or an attribution of the leader;
5
(d) attitudes, emotions or affect towards the leader; and (e) a relationship between the leader and
others. Each article was classified into a single category only based on the predominant view of
the author (see Table 2).
[Table 2 here]
2. Article type: Reflecting whether the article was quantitative, qualitative, theory or
review (see Table 3).
[Table 3 here]
When an article was included as a quantitative piece, we coded for the following variables (note:
all analyses were done at the study level):
3. Sample occupation: This variable concerned the targets of interest in the study, or put in
another way, the focal entities studied in the statistical model. Categories comprised (a) managers
working in private or public companies; (b) undergraduate students; (c) graduate and EMBA
students; (d) politicians; and (e) others and non-specified (see Table 4).
[Table 4 here]
4. Design of the study: Here we coded the method to collect data on charisma. Relevant
categories are (a) experiment, whether in laboratory, field or quasi-experiments; (b) field study;
(c) archival data; and (d) coded interviews (see Table 5).
[Table 5 here]
5. How charisma was used as a variable: This variable captures how the charisma variable
was included in the statistical model. Charisma was coded as being (a) an independent variable
(IV); (b) a moderator; (c) a mediator; and (d) a dependent variable (DV). If a study included
charisma both as an IV and as a DV, we coded for both. Note: as explained in our article, we
excluded coded articles that used an endogenous measure of charisma as an IV, moderator or
6
mediator, and failed to properly instrument charisma. However, when charisma was a mediator,
we still included articles that would have been rejected otherwise when a hypothesis was linking
the IV with charisma in a substantive manner. Only first-stage regressions were coded for these
types of article (i.e., considering charisma as a DV). Articles testing charisma as a dependent
variable were always included, even if the model predicting charisma was to be incorrectly tested
(see Table 6).
[Table 6 here]
6. The target leader: Categories included (a) real leader I (e.g., leader with whom the rater
has a direct relationship, leader under study and rated by their followers, or leader rated by the
researcher from archival data); (b) real leader II (e.g., target in a video used as a stimulus material
or leader with whom the follower has no direct relationship); (c) hypothetical leader (i.e., when a
subject has to think of a leader, not someone in particular); and (d) actor (see Table 7).
[Table 7 here]
7. Measure of charisma: Categories included (a) leader scales, including for example, the
MLQ (Avolio et al 1995), the CKS (Conger & Kanungo 1998), TLI (Podsakoff et al 1990), or the
CLIO (De Hoogh et al 2004), coded independently, (b) measures coded from characteristics of
the leader, (c) measures coded from his verbal rhetoric and non-verbal behaviors, (d) direct
measures (e.g., is he/she charismatic?) and (e) other measures. Note, we excluded coding features
from articles that manipulated charisma in an experiment, even if they performed a manipulation
check using a measure of charisma (see Table 8).
[Table 8 here]
7
8. Endogenous charisma used as regressor: Here we report on the type of measures (MLQ
vs. others) used in studies included in our review versus the studies rejected because charisma
was endogenous (see Table 9).
[Table 9 here]
9. Location: This category refers to the region in which the sample was located. Because
no distinctive trends would be visible if we presented the data at the country level, we regrouped
the sample locations by regions, including (a) North-America; (b) Europe; (c) Asia; (d) others; (e)
cross-National; and (f) not specified (see Table 10).
[Table 10 here]
10. Qualitative articles: When an article was coded as a qualitative piece, we coded for its
design and the location of the leader under study. Only percentages are reported for the latter (see
Tables 11 & 12).
[Tables 11 & 12 here]
For the correlation matrix of the key variables in coded papers, refer to Table 13.
[Table 13 here]
Co-citation analyses
We conducted co-citation analyses at the author, document, and source units of analysis.
Document co-citation analysis: We examined how many times any given pair of
documents (i.e., articles, chapters, or books) have been cited by the 280 articles included in this
review. This analysis can help us understand the intellectual base from which these 280 studies
build on. As will be clear from below, we found that these studies have drawn from a single
unified body of knowledge rather than from multiple unconnected bodies of knowledge.
8
Using the Sci2 (2009) tool, we extracted a total of 11,364 unique references (nodes) from
the 280 articles included in the review
2
. The largest connected component is 11,311 nodes large
indicating that the intellectual landscape of charisma we see from these documents is a highly
cohesive one. From this initial citation network, we further extracted a network of 661,660 co-
citations (edges). For visualization purposes, we kept edges only greater than one (i.e., we
excluded pairs of documents that were only cited together once in our initial set of 280 articles)
and we deleted isolates. The resulting network consists of 1,615 documents and 28,619 co-
citations (Figure 1A). An interactive version of this network can be accessed online at the
following address: http://www.hec.unil.ch/jantonakis/annualreviews/index.html. Figure 1B shows
the same network keeping only the top 0.5% largest edges (i.e., edges greater than 17).
[Figures 1A and 1B here]
Author co-citation analysis: Using VOSViewer (Van Eck & Waltman 2010), we mapped
the author co-citation network of our sample of empirical studies of charisma. Note that this
network is based only on first authors. From 7,193 first authors, we retained the top 50 authors
(i.e., authors who have been cited at least 28 times). The resulting network can be seen in the
main paper (Figure 1B).
Source co-citation analysis: Additionally, we thought to understand the network of
sources (i.e., journals, books, etc.) cited by empirical studies of charisma. Again, using
VOSViewer, we mapped the source co-citation network of our full sample of empirical studies of
charisma. From the resulting 5,856 sources, we retained all sources which had been cited at least
2
We preprocessed the data by running the built-it detect duplicate algorithm. Note, typographical errors aside, a
given work may still referenced in multiple ways. This issue is particularly important with regard to Weber’s work,
which is at times referenced to in the original German version or in an English translation, as part of collected works,
or as other works about Weber. As a result of this issue, and because this analysis is at the document level, the
influence of Weber’s work is somewhat diluted (e.g., in Figure 1). In other words, what we graphically show is a
lower bound estimate of the influence of Weber’s work. In fact, among the 280 papers we analyzed we found that
Weber’s work was the most frequently cited (i.e., 73.94% cited Weber’s work; however, 55.71% cited Bass, 54.64%
cited House, 50.71% cited Conger, and 45.71% cited Shamir).
9
28 times (n = 62). The resulting map can be seen in Figure 2. This figure nicely shows the
contribution of (a) sociology and political science (blue nodes), (b) applied psychology (green
nodes), and (c) management (red nodes) to the study of charisma.
[Figure 2 here]
10
Table 1: Predicting Citations by Article Type
(Model)
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
(9)
(10)
Article age
.23**
.20**
.20**
.20**
.22**
.15**
.13**
.11**
.10**
.10**
(5.67)
(5.17)
(5.45)
(5.78)
(7.52)
(4.42)
(4.10)
(3.46)
(4.65)
(4.65)
Article age
2
-.00**
-.00**
-.00**
-.00**
-.00**
-.00**
-.00**
-.00**
-.00**
-.00**
(4.10)
(3.79)
(4.09)
(4.52)
(5.89)
(3.52)
(3.40)
(2.78)
(3.82)
(3.82)
Cited Refs
.02**
.01**
.01**
.01**
.01**
.01**
.01**
.01**
.01**
.01**
(7.87)
(10.87)
(6.28)
(4.16)
(4.50)
(7.13)
(9.59)
(8.51)
(7.43)
(7.43)
# of Authors
.44*
.30
.24
.18*
.19**
.35*
.27*
.21
.16+
.16+
(2.34)
(1.95)
(1.67)
(2.34)
(2.60)
(2.29)
(1.96)
(1.61)
(1.96)
(1.96)
Quantitative
.93**
.60**
.36*
.33*
.78**
.54**
.32**
.32**
(6.48)
(3.31)
(2.09)
(1.97)
(6.50)
(4.54)
(3.20)
(3.20)
Review
.99**
.56
.71**
.71**
1.00**
.70**
.77**
.77**
(3.32)
(1.75)
(2.66)
(2.61)
(3.85)
(2.89)
(3.64)
(3.64)
Theory
1.00**
.70**
.34*
.32*
1.03**
.77**
.38**
.38**
(4.45)
(3.27)
(2.17)
(2.09)
(4.25)
(3.42)
(3.60)
(3.60)
Agg. impact factor field
1.31**
.67*
.64*
1.16**
.65*
.65*
(3.09)
(2.38)
(2.25)
(2.79)
(2.30)
(2.30)
Author ave. affilliation rank
a
.00**
.00*
.00*
.00**
.00**
.00**
(4.54)
(2.48)
(2.29)
(4.23)
(2.63)
(2.63)
Impact factor journal
.30**
.31**
.24**
.24**
(4.25)
(4.08)
(4.69)
(4.69)
Constant
-.98*
-1.17**
-2.84**
-2.07**
-2.31**
-2.04**
-2.27**
-3.74**
-3.13**
-3.13**
(2.24)
(3.48)
(6.49)
(4.73)
(5.63)
(3.72)
(4.86)
(8.64)
(8.69)
(8.68)
Pseudo-R square
b
.49**
.54**
.59**
.69**
.67**
.37**
.43**
.50**
.59**
.59**
Omitted category for article type is “Qualitative” Model 1-4: predicting times cited using zero-inflated negative binomial regression; Model 5
predicting times cited using negative binomial regression; Model 6-9: predicting times cited per year using zero-inflated negative binomial
regression; Model 10 predicting times cited per year using negative binomial regression; n = 280 (models 1-3, 6-8); n = 276 (models 4, 5, 9, and
10)—the reduced n-size is because some journals do not have an impact factor.
a
Reverse coded thus a higher number indicates higher rank;
b
Wald
test for change in r-square from additional regressors from previous model (for Models 1, 5, 6, and 10 it is the Wald test for the r-square for the full
equation); pseudo R-square is based on the Cox-Snell (1989) method. The inflation coefficient is significant for final models (i.e., 4 and 8) as is the
dispersion parameter (i.e., ln(α) for models 4, 5, 9, and 10). 12.50% of articles were not cited. Cluster robust z-statistics in parentheses (using
unique journal categories and category combinations as per the Web of Science journal impact factor listings); ** p<.01, * p<.05, +p=.05.
11
Table 2: Trend for conceptualization of charisma
Year
1954
1961
1964
1965
1966
1967
1968
1969
1971
1972
1973
1974
1975
1976
1978
1979
1980
1982
1983
1985
1986
1987
1988
1989
1990
Perception / Attribution 0 0 2 3 0 0 1 0 1 0 0 1 2 0 1 1 1 0 0 0 0 2 0 0 0
Personality traits 0 1 0 0 0 1 3 0 0 0 0 2 1 0 1 0 3 0 1 1 0 0 1 0 0
Behavior 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 1
Relationship 1 0 0 0 1 1 0 1 0 2 1 0 1 1 3 0 0 2 0 1 1 0 1 3 0
Attitude 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
Total 1 1 2 3 1 2 4 1 1 2 1 3 4 1 5 1 4 2 1 2 1 3 2 3 1
1991
1992
1993
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Total
Trend
Perception / Attribution 2 2 0 2 2 3 4 7 1 0 2 2 6 4 3 4 7 6 7 4 6 3 6 98 ++
Personality traits 0 1 1 0 0 3 3 1 4 3 1 1 2 3 2 5 6 4 4 1 2 4 4 70 ++
Behavior 0 1 1 0 1 0 2 4 2 5 1 2 1 2 3 1 3 3 2 6 3 6 7 58 ++
Relationship 2 0 0 1 0 2 0 3 2 1 2 0 1 1 2 1 1 3 2 2 3 1 0 50 +
Attitude 0 1 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 1 0 0 1 4
Total 4 5 2 3 3 8 9 15 9 10 6 5 10 10 10 11 17 16 15 14 14 14 18 280
12
Table 3: Trend for type of article
Year
1954
1961
1964
1965
1966
1967
1968
1969
1971
1972
1973
1974
1975
1976
1978
1979
1980
1982
1983
1985
1986
1987
1988
1989
1990
Quantitative 1 0 0 0 0 0 0 0 0 0 0 1 0 0 1 0 0 0 0 0 0 0 0 1 0
Qualitative 0 0 1 1 1 2 2 0 1 1 0 0 1 1 1 1 3 2 0 1 0 1 1 1 0
Theory 0 1 1 2 0 0 2 1 0 1 1 2 3 0 3 0 1 0 1 1 1 2 1 1 1
Review 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
Total 1 1 2 3 1 2 4 1 1 2 1 3 4 1 5 1 4 2 1 2 1 3 2 3 1
1991
1992
1993
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Total
Trend
Quantitative 1 2 0 1 3 5 3 8 1 5 2 3 9 6 4 4 8 6 6 11 5 6 9 112 ++
Qualitative 1 1 0 0 0 1 1 3 5 4 3 1 0 3 2 6 5 7 5 3 4 6 7 90 ++
Theory 2 2 2 2 0 2 5 3 3 0 1 1 1 1 2 1 4 1 4 0 3 2 1 69
Review 0 0 0 0 0 0 0 1 0 1 0 0 0 0 2 0 0 2 0 0 2 0 1 9 ++
Total 4 5 2 3 3 8 9 15 9 10 6 5 10 10 10 11 17 16 15 14 14 14 18 280
13
Table 4: Trend for sample occupation
Year
1954
1974
1978
1989
1991
1992
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Total
Trend
Private & Public companies 0 0 0 0 0 0 0 1 0 1 1 1 0 0 2 2 3 1 2 3 2 1 6 1 1 3 31 +
Undergraduate students 0 0 0 1 0 1 1 2 3 0 3 0 3 1 3 7 3 2 3 7 3 3 1 2 3 4 56 +
Graduate & EMBA students 0 0 0 0 0 1 0 0 0 0 2 0 0 0 1 2 1 1 0 0 1 0 1 2 0 1 13
Politicians 0 1 0 0 1 0 0 0 2 1 1 0 4 0 0 4 2 1 0 1 1 0 1 2 2 1 25
Others 1 0 1 0 0 1 0 0 0 1 1 0 0 2 0 0 0 0 0 1 0 3 4 0 1 1 17
Total 1 1 1 1 1 3 1 3 5 3 8 1 7 3 6 15 9 5 5 12 7 7 13 7 7 10 142
14
Table 5: Trend for type of design
Year
1954
1974
1978
1989
1991
1992
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Total
Trend
Experiments 0 0 0 1 0 1 0 2 2 0 6 0 3 0 5 5 1 2 2 6 2 3 3 2 2 3 51 +
Field Study 1 0 1 0 0 1 0 1 1 2 1 1 1 2 0 3 5 3 3 3 2 3 6 2 3 4 49 ++
Archival Data 0 1 0 0 1 0 0 0 2 1 1 0 3 1 0 5 2 0 0 2 2 1 4 2 2 1 31
Interview 0 1 0 0 0 0 1 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 2 5
Total 1 2 1 1 1 2 1 3 5 3 8 1 7 3 5 13 9 5 5 11 6 7 13 6 7 10 136
15
Table 6: Trend for type of variable
Year
1954
1974
1978
1989
1991
1992
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Total
Trend
Dependent Variable 0 1 1 0 1 1 0 2 4 3 3 1 3 3 5 8 8 4 5 5 4 5 9 5 4 6 91 ++
Independent Variable 1 0 0 1 0 1 1 1 1 0 5 0 4 0 0 5 0 1 1 6 2 2 5 0 3 3 43
Mediator 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 1
Moderator 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 2
Total 1 1 1 1 1 3 1 3 5 3 8 1 7 3 5 13 8 5 6 11 6 7 15 6 7 9 137
16
Table 7: Trend for type of leader
Year
1954
1974
1978
1989
1991
1992
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Total
Trend
Real Leader I 0 0 1 0 0 1 1 2 2 2 1 1 3 2 4 3 4 3 3 4 4 2 8 1 2 4 58 +
Real Leader II 1 1 0 0 1 0 0 0 3 1 2 0 4 1 0 7 3 0 2 2 2 1 4 3 4 3 45 ++
Hypothetical Leader 0 0 0 0 0 1 0 0 0 0 1 0 0 0 1 3 1 2 0 4 0 4 0 0 0 1 18
Actor 0 0 0 1 0 0 0 1 0 0 4 0 0 0 0 0 0 0 0 1 0 0 1 2 1 2 13
Total 1 1 1 1 1 2 1 3 5 3 8 1 7 3 5 13 8 5 5 11 6 7 13 6 7 10 134
17
Table 8: Trend for measure of charisma
Year
1954
1974
1978
1991
1992
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Total
Trend
MLQ 0 0 0 0 0 0 2 2 1 1 0 0 2 1 2 1 3 4 1 2 2 4 1 3 4 36 ++
Others 0 1 1 0 2 0 0 1 0 4 0 1 0 0 1 3 1 1 3 1 1 0 2 3 2 28 ++
Coded from verbal & non-verbal 0 0 0 1 0 0 0 0 0 0 0 3 1 3 4 2 0 0 2 2 1 2 2 0 0 23
Coded from characteristics of leader 0 0 0 1 0 0 0 2 1 1 0 3 0 0 2 0 0 0 1 1 0 2 1 2 1 18
Direct measure 1 0 0 0 0 1 0 0 1 0 0 0 0 0 1 1 1 2 0 0 0 2 0 0 1 11
CKS 0 0 0 0 0 0 0 0 0 0 1 0 0 1 3 1 0 0 0 0 2 0 0 0 1 9
TLI 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 2 0 0 0 3
CLIO 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 1 0 0 0 0 1 0 0 3
Total 1 1 1 2 2 1 2 5 3 6 1 7 3 5 13 9 5 8 8 6 6 12 7 8 9 131
18
Table 9: Trend for use of MLQ and other measures
1954
1974
1978
1990
1991
1992
1993
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Total
Trend
MLQ Excluded 0 0 0 1 0 1 1 1 0 0 1 1 0 2 4 2 3 2 3 6 3 5 5 5 4 4 5 59 ++
MLQ Included
0 0 0 0 0 0 0 0 2 2 1 1 0 0 2 1 2 1 3 4 1 2 2 4 1 3 4 36 +
Total MLQ
0 0 0 1 0 1 1 1 2 2 2 2 0 2 6 3 5 3 6 10 4 7 7 9 5 7 9 95 ++
Other Measures Excluded 0 0 0 0 0 1 0 1 0 1 2 0 1 0 1 0 0 2 3 0 2 3 2 6 3 9 4 41 ++
Other Measures Included 1 1 1 0 2 2 0 1 0 3 2 5 1 7 1 4 11 8 2 4 7 4 4 8 6 5 5 95 ++
Total Other Measures 1 1 1 0 2 3 0 2 0 4 4 5 2 7 2 4 11 10 5 4 9 7 6 14 9 14 9 136 ++
Total 1 1 1 1 2 4 1 3 2 6 6 7 2 9 8 7 16 13 11 14 13 14 13 23 14 21 18 231
19
Table 10: Trend for location of the sample
Year
1954
1974
1978
1989
1991
1992
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Total
Trend
North America 1 0 1 1 1 1 0 2 4 2 6 1 6 1 2 6 3 1 3 4 5 3 6 3 4 3 70 +
Europe 0 0 0 0 0 0 0 0 0 0 0 0 0 0 2 2 4 0 1 4 0 2 5 2 1 2 25 ++
Not Specified 0 0 0 0 0 0 0 1 0 0 1 0 0 1 0 3 1 0 0 2 0 1 1 1 1 2 15 +
Cross-National 0 1 0 0 0 0 0 0 0 0 1 0 0 1 1 1 0 1 1 1 1 0 0 0 1 2 12
Asia 0 0 0 0 0 1 1 0 0 1 0 0 1 0 0 1 0 1 0 0 0 1 0 0 0 1 8
Others 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 2 0 0 0 0 1 0 0 0 4
Total 1 1 1 1 1 2 1 3 5 3 8 1 7 3 5 13 8 5 5 11 6 7 13 6 7 10 134
20
Table 11: Trend for design of qualitative studies
Year
1964
1965
1966
1967
1968
1971
1972
1975
1976
1978
1979
1980
1982
1985
1987
1988
1989
1991
1992
1997
1998
1999
2000
2001
2002
2003
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Total
Trend
Case Study 1 1 1 2 2 1 1 1 1 1 1 2 2 1 1 1 1 1 0 1 1 2 5 3 2 1 3 0 6 3 5 4 2 4 2 7 73 ++
Others 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 1 0 1 1 0 0 2 0 0 1 1 1 0 3 0 12 +
Interview 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 2 1 0 0 0 1 0 5
Total 1 1 1 2 2 1 1 1 1 1 1 3 2 1 1 1 1 1 1 1 1 3 5 4 3 1 3 2 6 5 7 5 3 4 6 7 90
21
Table 12: Qualitative articles location of the sample
Region
Studies
Percentages
North America
29
32.22%
Europe
22
24.44%
Asia
10
11.11%
Cross-national
10
11.11%
Africa
8
8.89%
South America
7
7.78%
Not relevant
3
3.33%
Oceania
1
1.11%
22
Table 13: Correlation matrix of key variables in coded papers
Mean
S.D.
1
2
3
4
5
6
7
8
9
10
11
1. Time cited
31.97
69.84
2. Cites/per year
2.37
3.89
.90
3. Years
14.62
13.22
.13
-.05
4. Cited References
62.95
36.24
.15
.35
-.40
5. Number of author
1.90
1.12
.14
.26
-.32
.27
6. Affiliation rank
198.90
179.52
.09
.06
.07
.01
-.05
7. Agg. Impact factor field
1.41
.40
.20
.30
-.35
.31
.40
-.08
8. Impact factor journal
2.60
2.00
.49
.60
-.16
.39
.34
.05
.45
9. Quant. Article
.41
.49
.01
.06
-.27
.02
.48
-.02
.39
.24
10. Qual. Article
.32
.47
-.22
-.28
.05
-.14
-.37
.01
-.46
-.38
-.57
11. Theory Article
.24
.43
.23
.18
.30
.03
-.19
-.01
.02
.12
-.46
-.39
12. Review Article
.03
.18
.02
.13
-.10
.24
.11
.06
.06
.04
-.15
-.13
-.10
Note: times cited as of 6 April 2015.
23
Figure 1A: The intellectual landscape of the socio-scientific study of charisma
Note: Full document co-citation network. Figure 1B shows the top half-percent most co-cited works
within this network.
24
Figure 1B: The intellectual landscape of the socio-scientific study of charisma
Note: Document co-citation network showing only the 0.5% most co-cited works. Each node represents
one document (only first authors are named) and edges represent the co-citations of two documents.
Darker and thicker edges indicate more co-citations. Node size indicates the number of other documents
with which this node is co-cited (degree), and its color indicates the sum of the weights of all edges to
which this node is connected to (strength)—darker color indicates greater strength.
25
Figure 2: Source co-citation analysis of the socio-scientific study of charisma
Note: Difference in node size indicates relative difference in the frequency of occurrence. Sources that are frequently co-cited are positioned close
to each other. The 200 strongest edges are represented. Node color indicates membership to a common densely connected cluster, as identified by
an algorithm. The blue cluster contains sociology and political sources, the red cluster contains management sources, and the green cluster contains
psychology sources.
26
References:
Aksnes DW. 2003. Characteristics of highly cited papers. Research Evaluation 12: 159-70
Antonakis J, Bastardoz N, Liu Y, Schriesheim CA. 2014. What makes articles highly cited? The
Leadership Quarterly 25: 152-79
Antonakis J, Lalive R. 2008. Quantifying scholarly impact: IQp versus the Hirsch h. Journal of
the American Society for Information Science and Technology 59: 956-69
Avolio BJ, Bass BM, Jung DI. 1995. MLQ Multifactor leadership questionnaire: Technical
Report. Redwood City, CA: Mindgarden
Bergh DD, Perry J, Hanke R. 2006. Some predictors of SMJ article impact. Strategic
Management Journal 27: 81-100
Blevins DP, Tsang EWK, Spain SM. 2015. Count-Based Research in Management: Suggestions
for Improvement. Organizational Research Methods 18: 47-69
Conger JA, Kanungo RN. 1998. Charismatic leadership in organizations. Thousand Oaks, CA:
Sage Publications.
Cox DR, Snell EJ. 1989. Analysis of binary data. London ; New York: Chapman and Hall. vii,
236 p. pp.
De Hoogh A, Den Hartog D, Koopman P. 2004. De ontwikkeling van de CLIO: een vragenlijst
voor charismatisch leiderschap in organisaties. Gedrag en Organisatie 17: 354-81
Huber PJ. 1964. Robust estimation of a location parameter. Annals of Mathematical Statistics 35:
73-101
Landis JR, Koch GG. 1977. The measurement of observer agreement for categorical data.
Biometrics 33: 159-74
Larivière V, Gingras Y. 2010. The impact factor's Matthew Effect: A natural experiment in
bibliometrics. Journal of the American Society for Information Science and Technology
61: 424-27
Lokker C, McKibbon KA, McKinlay RJ, Wilczynski NL, Haynes RB. 2008. Prediction of
citation counts for clinical articles at two years using data available within three weeks of
publication: retrospective cohort study. BMJ 336: 655-57
Podsakoff PM, MacKenzie SB, Moorman RH, Fetter R. 1990. Transformational leader behaviors
and their effects on follower's trust in leader, satisfaction, and organizational citizenship
behaviors. The Leadership Quarterly 1: 107-42
Sci2 Team. 2009. Science of Science (Sci2) Tool. Indiana University and SciTech Strategies:
https://sci2.cns.iu.edu
Van Eck NJ, Waltman L. 2010. Software survey: VOSviewer, a computer program for
bibliometric mapping. Scientometrics 84: 523-38