Content uploaded by Raina Brands
Author content
All content in this area was uploaded by Raina Brands on Jul 16, 2015
Content may be subject to copyright.
This article was downloaded by: [163.119.96.63] On: 30 March 2015, At: 09:11
Publisher: Institute for Operations Research and the Management Sciences (INFORMS)
INFORMS is located in Maryland, USA
Organization Science
Publication details, including instructions for authors and subscription information:
http://pubsonline.informs.org
The Leader-in-Social-Network Schema: Perceptions
of Network Structure Affect Gendered Attributions of
Charisma
Raina A. Brands, Jochen I. Menges, Martin Kilduff
To cite this article:
Raina A. Brands, Jochen I. Menges, Martin Kilduff (2015) The Leader-in-Social-Network Schema: Perceptions of Network
Structure Affect Gendered Attributions of Charisma. Organization Science
Published online in Articles in Advance 23 Mar 2015
. http://dx.doi.org/10.1287/orsc.2015.0965
Full terms and conditions of use: http://pubsonline.informs.org/page/terms-and-conditions
This article may be used only for the purposes of research, teaching, and/or private study. Commercial use
or systematic downloading (by robots or other automatic processes) is prohibited without explicit Publisher
approval, unless otherwise noted. For more information, contact permissions@informs.org.
The Publisher does not warrant or guarantee the article’s accuracy, completeness, merchantability, fitness
for a particular purpose, or non-infringement. Descriptions of, or references to, products or publications, or
inclusion of an advertisement in this article, neither constitutes nor implies a guarantee, endorsement, or
support of claims made of that product, publication, or service.
Copyright © 2015, INFORMS
Please scroll down for article—it is on subsequent pages
INFORMS is the largest professional society in the world for professionals in the fields of operations research, management
science, and analytics.
For more information on INFORMS, its publications, membership, or meetings visit http://www.informs.org
Organization Science
Articles in Advance, pp. 1–16
ISSN 1047-7039 (print) ISSN 1526-5455 (online) http://dx.doi.org/10.1287/orsc.2015.0965
© 2015 INFORMS
The Leader-in-Social-Network Schema: Perceptions of
Network Structure Affect Gendered Attributions of Charisma
Raina A. Brands
London Business School, London NW1 4SA, United Kingdom, rbrands@london.edu
Jochen I. Menges
WHU–Otto Beisheim School of Management, 40233 Düsseldorf, Germany, jochen.menges@whu.edu
Martin Kilduff
University College London, London WC1E 6BT, United Kingdom, mjkilduff@gmail.com
Charisma is crucially important for a range of leadership outcomes. Charisma is also in the eye of the beholder—an
attribute perceived by followers. Traditional leadership theory has tended to assume charismatic attributions flow to
men rather than women. We challenge this assumption of an inevitable charismatic bias toward men leaders. We propose
that gender-biased attributions about the charismatic leadership of men and women are facilitated by the operation of a
leader-in-social-network schema. Attributions of charismatic leadership depend on the match between the gender of the
leader and the perceived structure of the network. In three studies encompassing both experimental and survey data, we
show that when team advice networks are perceived to be centralized around one or a few individuals, women leaders
are seen as less charismatic than men leaders. However, when networks are perceived to be cohesive (many connections
among individuals), it is men who suffer a charismatic leadership disadvantage relative to women. Perceptions of leadership
depend not only on whether the leader is a man or a woman but also on the social network context in which the leader is
embedded.
Keywords : charismatic leadership; social networks; gender; schemas
History: Published online in Articles in Advance.
Introduction
To be an effective leader, it is crucial to be granted lead-
ership qualities by others. Subordinates confer leader-
ship qualities (such as charisma; see Weber 1947) on
some but not all formally appointed leaders. In this
perceptual process, current research suggests women,
relative to men, are disadvantaged both in how they are
perceived in formal leadership roles (Eagly et al. 1992)
and in how they are perceived in informal leadership
roles in social networks (Burt 1992). But this research
consensus may be overly focused on organizational con-
texts that display a traditional centralized structure. We
explore the possibility that men and women may be
differentially conferred with charismatic leadership qual-
ities depending on the structure of relationships (central-
ized or cohesive) in which these leaders are embedded.
We suggest a leader-in-social-network schema that
contains abstract information concerning leaders in
social network contexts (see Lord and Foti 1986).
This schema enables individuals to anticipate a match
between the gender of the leader and the structure of
the social network, thereby influencing attributions of
charismatic leadership. Schemas are organized represen-
tations of past behaviors and experiences that function as
theories about reality to guide the individual in constru-
ing new experiences (Baldwin 1992). Individuals tend
to simplify complex phenomena by filtering information
through these preexisting knowledge systems (Neisser
1976). Person-in-situation schemas, such as the leader-
in-social-network schema that we propose, tend to be
richer, more accessible, and more widely used by peo-
ple than simple schemas relating just to persons or script
schemas that describe event sequences. We draw from
cognitive psychology and schema research to offer a
new approach to the question of why certain people are
attributed with charismatic leadership.
Specifically, we suggest that people have expectations
of the types of leaders—men or women—appropriate
for different social network structures. When the indi-
vidual perceives that people in a team go to one or a few
others for advice (a network pattern indicative of status
or power centralization), then the individual is likely to
expect the leader to be a man. Centralization concerns
the extent to which there is a well-defined power or sta-
tus structure within a group, a context associated with
masculine gender stereotypes (Acker 1990) and with
actual male behavior in groups (Aries 1976). When the
individual perceives that people in a team go to many
others for advice (a network pattern indicative of cohe-
sion), then the individual is likely to expect the leader
to be a woman. Cohesion concerns the extent to which
group members are bonded together (Beal et al. 2003),
1
Downloaded from informs.org by [163.119.96.63] on 30 March 2015, at 09:11 . For personal use only, all rights reserved.
Brands, Menges, and Kilduff: The Leader-in-Social-Network Schema
2Organization Science, Articles in Advance, pp. 1–16, © 2015 INFORMS
a context associated with female gender stereotypes
(Eagly and Karau 2002) and with actual women’s behav-
ior in groups (Maccoby 1990). Our core proposition is
that people expect to see a male leader when the network
is perceived as hierarchical and a female leader when
the network is perceived as cohesive. Subjective evalua-
tions of leadership quality (as assessed by attributions of
charisma) will tend to be enhanced when expectations
are met relative to when expectations are violated.
We focus specifically on charismatic leadership in this
paper for two reasons. First, charisma is central to lead-
ership theory and research (Conger and Kanungo 1987).
Indeed, “charisma is the most important component in
the larger concept of transformational leadership” (Bass
1985, p. 34). Scholars have called for the examination
of the specific components of transformational leader-
ship, highlighting attributions of charisma as an impor-
tant field of study (van Knippenberg and Sitkin 2013).
Second, more than other aspects of leadership, charis-
matic leadership is in the eye of the beholder (Avolio
and Yammarino 1990). From its inception, charismatic
leadership has been defined as an attribute perceived
by followers: leaders cannot be charismatic unless fol-
lowers perceive them as such (Weber 1947). We follow
prior work in recognizing that individual judgments con-
cerning a leader’s charisma are more open to situational
influence than are attributions concerning more objec-
tive aspects of leadership such as transactional leader-
ship (Pastor et al. 2007).
Charismatic Leadership
Charismatic leadership theory has traditionally described
the special qualities and behaviors of leaders (Weber
1947), including behaviors such as challenging the sta-
tus quo and nurturing followers’ development (Conger
and Kanungo 1987). Charismatic leaders set grand, long-
term visions and inspire others to enact these visions
(Conger and Kanungo 1987). But recent theory has
shifted the focus toward a recognition of charisma as
a relationship between the leaders who exhibit certain
characteristics and the followers whose own characteris-
tics lead them to perceive leaders as charismatic (How-
ell and Shamir 2005). Leadership resides within the
social system that develops between leaders and follow-
ers (Balkundi and Kilduff 2005). To the extent that fol-
lowers endorse a leader as charismatic, they place more
trust in the leader, feel more satisfied and more moti-
vated, and experience higher levels of empowerment and
self-efficacy (Conger et al. 2000,Jung and Avolio 2000).
Attributions of charismatic leadership predict team per-
formance (Balkundi et al. 2011).
The process by which a follower attributes charisma
to a leader involves a cognitive matching between the
follower’s implicit leadership theory (comprising proto-
typical expectations of leader characteristics and behav-
iors) and the actual leader’s characteristics and behaviors
(as perceived by the follower) (Lord et al. 1984). If the
leader matches the prototype, the follower endows the
leader with a charismatic personality (Nye and Forsyth
1991), seeing the leader as a role model, someone with
a vision who sets high performance expectations that
the team collectively accepts. Given the availability of
implicit leadership theories, social perceivers may make
judgments based on little or no evidence (Lord and
Emrich 2000,Lord and Maher 1994). It is the charis-
matic leadership prototype that supplies missing infor-
mation and enables the charismatic attribution (Lord and
Foti 1986). Notably, the charismatic prototype is not
fixed: as follower or situational factors change, defini-
tions of charismatic leadership also change (Lord et al.
2001). For example, attributions of charismatic leader-
ship are susceptible to changing factors such as team
performance (Shamir 1992) and followers’ arousal (Pas-
tor et al. 2007).
Leadership prototypes are implicitly held and function
largely outside of people’s awareness. People have gen-
erally little insight into how these prototypes affect—and
sometimes bias—their assessment of those in leadership
positions. Thus, it is not unusual to find women leaders
evaluated less favorably than men (Eagly et al. 1992)
despite both genders displaying similar behavior (Eagly
et al. 1995). Leadership prototypes are part of a wider
net of social-cognitive associations people hold in their
minds. Leadership prototypes influence, and are influ-
enced by, the nodes and connections within this broader
mental net, including associations with gender stereo-
types (Sczesny 2005). Gender bias occurs because the
prototype of an effective leader from which attributions
are drawn is typically considered to be male (Schein
1973). Women, by virtue of their gender, are seen as
less able to fulfill the requirements of leadership (role
congruity theory; Eagly and Karau 2002).
In our research we challenge the inevitability of biased
attributions concerning women’s leadership. By mov-
ing to the consideration of a leader-in-social-network
schema, we open up the possibility that schematic
attributions concerning leadership affect both genders,
depending on the social network context. We draw
from schema theory (Baldwin 1992,Lord and Maher
1994), cognitive social network research (e.g., Janicik
and Larrick 2005), and gender schema research (e.g.,
Eagly and Carli 2003) to forge a new approach to the
enduring question of how the expectations that people
bring to the workplace affect leadership evaluations.
The Leader-in-Social-Network Schema
Person-in-situation schemas contain information about
people and behavior typically found in specific social
situations. These schemas combine elements of person
schemas (that facilitate the categorization of people into
types; see Cantor and Mischel 1979) and script schemas
Downloaded from informs.org by [163.119.96.63] on 30 March 2015, at 09:11 . For personal use only, all rights reserved.
Brands, Menges, and Kilduff: The Leader-in-Social-Network Schema
Organization Science, Articles in Advance, pp. 1–16, © 2015 INFORMS 3
(that describe the appropriate sequence of events in
given situations; see Schank and Abelson 1977). These
schemas allow people to go beyond the information
available in any particular situation and fill in the gaps
in knowledge with details that would be expected on the
basis of past experience (Baldwin 1992).
We suggest that the process by which an observer
endorses an individual as a charismatic leader involves a
cognitive matching between the observer’s expectations
concerning the types of people suitable for leadership
and the extent to which the individual is perceived to
fulfill those expectations in particular social network
contexts. The leader-in-social-network schema develops
from repeated interactions individuals have and observe
within the social networks to which they belong (Janicik
and Larrick 2005). Individuals build up knowledge con-
cerning the extent to which patterns of advice giving and
receiving tend to be associated with men or women lead-
ers. We examine advice networks because advice giving
and taking relate to leader charisma (e.g., Balkundi et al.
2011) and leader performance (Balkundi et al. 2011).
We focus on social network perceptions because implicit
leadership theories draw on people’s internal subjective
representations rather than on external objective realities
(Lord 2005).
Perceptions of advice networks are likely to differ
along two structural dimensions—centralization, as indi-
cated by the perceived centralization of the network,
and cohesion, as indicated by the perceived density of
the network. Centralization concerns the extent to which
there is a well-defined power or status structure within
the team. Cohesion refers to the extent to which peo-
ple within a team have many rather than few connec-
tions with each other. The perception of a social network
structure as centralized cues expectations for a leader
who can control the levers of power and influence. By
contrast, the perception of a social network structured as
cohesive cues expectations for a leader who can relate
to others and manage complex interpersonal situations.
These leadership expectations, cued by the perceived
social network context, align with gender stereotypes,
leading to a bias toward male leaders in hierarchical
contexts and toward female leaders in cohesive con-
texts. Thus we suggest that the leader-in-social-network
schema offers two prototypical matches between leader
gender and social network structure. One prototypical
pattern involves a male leader in a centralized network
in which one or a few individuals dominate. The other
prototypical pattern involves a female leader in a cohe-
sive network in which many individuals interact.
Centralization: The Male Leader in a
Centralized Network
Centralization is the extent to which a network is dom-
inated by a single individual (Freeman 1978–1979). For
networks that depict advice interactions, centralization
can signal either status or power depending on whether
network relations involve either deference or resource
flows (see Magee and Galinsky 2008, p. 359). Advice
networks that are highly centralized feature clear peck-
ing orders and stratification of status and power.
When individuals perceive advice networks to be
highly centralized, they anticipate the familiar network
pattern of a star with ties radiating out to others from
the central actor. This image of a dominant actor at
the center of the network matches prototypes of men
leaders as controlling and dominant (Eagly and Karau
2002,Schein 1973), with high status and power (Eagly
and Steffen 1984). Because power and status differences
between individuals are particularly salient in centralized
networks, such contexts are likely to cue an expecta-
tion of a male leader. Indeed, networks in which infor-
mal power is perceived to be centralized in one or a
few individuals resemble the traditional, formal struc-
tures of command-and-control that are associated with
male stereotypes about leadership (Schein 1973). These
hierarchical structures are readily perceived (Zitek and
Tiedens 2012) and tend to reinforce male gender stereo-
types, thereby placing men at an advantage over women
(Acker 1990).
Networks in which informal status and power differ-
ences are perceived to be accentuated provide a percep-
tual frame that tends to diminish leadership attributions
to women and enhance leadership attributions to men by
cueing expectations about the type of individuals best
suited to be leaders. In centralized advice networks, indi-
viduals expect to encounter leaders who exhibit power,
dominance, courage, and boldness—characteristics that
are prescribed for men leaders but proscribed for women
leaders (Eagly and Karau 2002). As such, we expect
women leaders embedded in networks that are perceived
to be centralized to be attributed with less charisma than
are men. This bias arises not because of actual behav-
ioral differences between men and women as leaders
but because men, rather than women, fulfill the leader-
in-social-network expectations in centralized networks.
Attributions of charisma to female leaders (relative to
male leaders) in centralized contexts suffer, we suggest,
because women leaders represent a mismatch with the
charismatic leadership prototype, whereas men leaders
represent a match with the charismatic leadership proto-
type, that is triggered by the centralized context.
Hypothesis 1. To the extent that team advice net-
works are perceived to be centralized, women will be
seen as less charismatic leaders than men.
Cohesion: The Female Leader in a Densely
Connected Network
Density is a network-level property that refers to the
number of ties that exist within a network relative to
the number of ties that are possible (Wasserman and
Faust 1994). Advice networks characterized by many
Downloaded from informs.org by [163.119.96.63] on 30 March 2015, at 09:11 . For personal use only, all rights reserved.
Brands, Menges, and Kilduff: The Leader-in-Social-Network Schema
4Organization Science, Articles in Advance, pp. 1–16, © 2015 INFORMS
connections between group members (relative to the
number of connections possible) tend to be experienced
as cohesive (Sparrowe et al. 2001). Dense networks pro-
mote cohesion by reducing the likelihood of conflict
between team members (Labianca et al. 1998) and by
facilitating the development of shared understandings
and trust (Coleman 1988). Therefore, to the extent that
the individual perceives the advice network to be char-
acterized by many exchanges, the individual is likely to
anticipate a collaborative, communal setting character-
ized by mutuality and generalized reciprocity.
The anticipation of communality in cohesive net-
works relates closely to expectations concerning women
relative to men. Women are presumed to be “social
specialists” (Bales and Slater 1955,Meeker and
Weitzel-O’Neill 1977). They are expected to strive for
intimacy and solidarity in their interactions with oth-
ers (Moskowitz et al. 1994) and are associated with
enabling interaction styles that support and maintain
social exchanges (Maccoby 1990). Stereotypical women
leaders are expected to be supportive and nurturing
(Rosette and Tost 2010). Thus, networks in which
informal interactions are perceived to be cohesive pro-
vide a perceptual frame that matches women leaders.
By contrast, there is a mismatch between the pro-
totypical leader expected in a cohesive network and
the male leadership stereotype. Men are expected to
strive for status and dominance rather than communal-
ity (Diekman et al. 2004). This expectation for men
prescribes aggressive, forceful, and competitive behav-
ior to gain status and also proscribes yielding behavior
(Prentice and Carranza 2002).
The feminized-leadership context engendered by the
perception of cohesion, therefore, places a man leader
in a position where he is likely to be perceived in vio-
lation of core gender stereotypes and a woman leader in
a position where she is likely to be perceived as adher-
ing to core gender stereotypes. And, ironically, those
men who try to adapt to the prevailing communal norms
by, for example, sharing credit for their work, are likely
to find their modesty not only unappreciated but actu-
ally punished because modesty in a man violates gender
expectations (Moss-Racusin et al. 2010). Attributions of
charisma to male leaders (relative to female leaders) in
cohesive network contexts suffer, we suggest, because
male leaders represent a mismatch with the female lead-
ership prototype appropriate to the cohesive context.
Hypothesis 2. To the extent that team advice net-
works are perceived to be cohesive, men will be seen as
less charismatic leaders than women.
Our theoretical approach focuses on networks that
exhibit centralization or cohesion. To test the hypothe-
ses, we require perceptual data concerning networks that
are seen to be either centralized or cohesive.
Study 1
The purpose of Study 1 was to show, experimentally,
that the differing patterns anticipated by the leader-in-
social-network schema would affect the attributions of
charismatic leadership to men and women. Centraliza-
tion and cohesion are structural properties of networks,
but people’s perceptions of these network properties may
differ from the actual properties. Indeed, individuals
are unlikely to perceive network properties accurately
(see Kilduff et al. 2008). Rather, individuals tend to cog-
nitively impose systematic structure on the networks in
which they habitually participate (Freeman et al. 1987).
Prior experimental work on cognitive social structures
shows that individuals bring schematic expectations to
bear on perceptions of social networks (Janicik and
Larrick 2005). We bring these lines of work together by
testing whether the same underlying network, presented
to appear either centralized or dense, triggers different
schematic expectations.
Method
Sample. We recruited 198 individuals (79 men,
81 women, and 38 who did not report their gender) from
an online sample. The participants were all located in
the United States, were mostly white (78%), and aver-
aged 32 years of age (SD =10002). They had an average
of 14 years’ work experience (SD =9098) and were all
currently employed full or part time.
Procedure. The experiment used a 2 (gender of
leader: man or woman) ×2 (advice network configu-
ration: status centralized or cohesive) between-subjects
vignette design. (The manipulations are available
as supplemental material at http://dx.doi.org/10.1287/
orsc.2015.0965.) Participants read about the leader of
Team Media in a professional services firm and about
the interactions within the team’s advice network. The
description of the team advice network was accompa-
nied by a network diagram. Participants were then asked
a number of questions about the leader.
Gender Manipulation. Participants in the woman
leader condition read about a leader named Michelle.
Participants in the man leader condition read about a
leader named Michael.
Network Structure Manipulation. On their screens,
participants saw a network of nodes (labeled with peo-
ple’s names) and lines (representing connections). We
set the centralization score of the network at 71% and
the density score at 33%. The leader was shown as con-
nected to every other team member (and so had a cen-
trality score of 1). The other members of the team were
given gender-ambiguous names to avoid the possibility
that individuals would make attributions about leaders
based on the depiction of their personal ties to men and
women in their team.
Downloaded from informs.org by [163.119.96.63] on 30 March 2015, at 09:11 . For personal use only, all rights reserved.
Brands, Menges, and Kilduff: The Leader-in-Social-Network Schema
Organization Science, Articles in Advance, pp. 1–16, © 2015 INFORMS 5
Although we presented all participants with the same
structural network, we arranged for different depictions
of this same network to be viewed depending on experi-
mental condition.1The first depiction, designed to repre-
sent centralization, resembled a star network. The second
depiction, designed to represent cohesion, resembled a
dense network. A pretest of this manipulation confirmed
that people tended to see the star network (M =1946,
SE =109) as significantly more status concentrated than
the cohesive network (M =1507, SE =105, F127=
841, p=001). Likewise, individuals tended to see the
cohesive network (M =1937, SE =088) as signifi-
cantly more cohesive than the star network (M =1669,
SE =094 F127=448, p=004).
Dependent Variables. Participants were asked to rate
the leader’s charismatic qualities on 15 items that
assessed typical aspects of charismatic leadership,
including articulating a vision, providing a role model,
setting high performance expectations, and fostering
acceptance of group goals (=093) (Podsakoff et al.
1990). As we report in detail in the results section, we
checked to see whether other (noncharismatic) aspects
of leadership were affected by the independent vari-
ables. Specifically, we asked participants to rate lead-
ers’ intellectual stimulation on three items (=088)
and leaders’ individualized consideration on four items
(=073). We also measured transactional leader-
ship using five items concerning the leader’s contingent
reward behavior (=060) (Podsakoff et al. 1990). All
responses were given on a Likert scale from 1 =strongly
disagree to 7 =strongly agree, and were averaged prior
to analysis.
Control Variable. We used gender as the control vari-
able in our experiments. Participants reported their gen-
der, which was coded as 0 =man and 1 =woman.
Attention and Manipulation Checks
Prior to analysis, we established that responses would
be excluded if (a) they came from the same IP address,
(b) response times were greater than two standard devia-
tions above the mean, or (c) the gender of the leader was
recalled incorrectly. Ten cases were excluded because
respondents violated condition (b).
Results
A 2 (gender of leader: man or woman) ×2 (network
configuration: cohesive or centralized) analysis of vari-
ance revealed a main effect of network condition on
attributions of leadership (F1197=1109, p=0001,
partial 2=005), suggesting that leaders in cohesive
networks (M =517, SE =011) were perceived to be
more charismatic than leaders in centralized networks
(M =463, SE =011). There was no main effect of
leader gender (F1197=033, p=056).
We predicted that charismatic attributions to leaders
would depend on the fit between leader gender and
network structure. And, indeed, there was a signifi-
cant interaction between the gender of the leader and
the configuration of the network (F1197=396, p=
0048, partial 2=002; see Figure 1). More specif-
ically, Hypothesis 1(in centralized networks, women
leaders would be seen as less charismatic than men lead-
ers) was supported. Even though participants saw the
same centralized network in both leader conditions, par-
ticipants who read about a leader named Michelle (M =
452, SE =0.12) saw her as less charismatic than par-
ticipants who read about a leader named Michael (M =
4.75, SE =0.19, t=−198, p=0.05). Hypothesis 2
(in cohesive networks, men leaders would be seen as less
charismatic than woman leaders) was also supported.
Even though participants saw the same cohesive network
in both leader conditions, participants who read about a
leader named Michael (M =4.69, SE =0.18) saw him
as less charismatic than participants who read about a
leader named Michelle (M =537, SE =0.13, t=265,
p=0009). The pattern of results was the same when
we used analysis of covariance with the gender of the
participant as a covariate on the reduced sample of the
160 people who reported their gender.
Additional Analyses. We tested whether our hypothe-
ses would be supported for other components of
transformational leadership (intellectual stimulation and
individualized consideration) and for transactional lead-
ership. However, as expected from our theoretical rea-
soning, the results were specific to charisma. We found
no effect of the interaction between network config-
uration and leader gender on intellectual stimulation
(F1159=2.34, p=013), individualized consideration
(F1159=0.27, p=060), or transactional leadership
(F1159=2.12, p=015).
Figure 1 Study 1: Attributions of Charismatic Leadership as
a Function of the Leader-in-Social-Network Schema
1
2
3
4
5
6
7
Cohesive
Centralized
Attributions of charismatic leadership
Michael
Michelle
p = 0.009
p= 0.05
Downloaded from informs.org by [163.119.96.63] on 30 March 2015, at 09:11 . For personal use only, all rights reserved.
Brands, Menges, and Kilduff: The Leader-in-Social-Network Schema
6Organization Science, Articles in Advance, pp. 1–16, © 2015 INFORMS
Discussion
Study 1 provides evidence in support of the purported
effects of the leader-in-social-network schema on attri-
butions of charisma. Women leaders were attributed
with less charisma than men leaders in the context
of centralized networks, whereas women leaders were
attributed more charisma than men leaders in the con-
text of cohesive networks. But does the leader-in-social-
network schema affect attributions about leaders in the
real world? Experiments allow us to establish causality
by providing a high degree of control, but they can exag-
gerate gender bias in leadership evaluations (Eagly et al.
1992). Thus testing our hypotheses in a field setting is
important for determining the validity of our theory.
Study 2
In our second study we tested the hypotheses in a sur-
vey of individuals working in teams in real organizations.
We also examined whether the perceptual-mismatch
effects on attributions of leadership tended to be exclu-
sively concerned with charisma or whether they extended
to other leadership outcomes including noncharismatic
aspects of transformational leadership, leadership effec-
tiveness (Giessner and van Knippenberg 2008), and
leader–follower relationship quality (i.e., leader–member
exchange; see Graen and Uhl-Bien 1995).
Method
Sample. We recruited and paid 149 (91 men and 58
women) U.S. resident full- or part-time employed indi-
viduals for a study on social networks. Respondents
were recruited from an online panel and represented a
wide variety of industries and occupations, were pre-
dominantly white (79.5%), and were on average 31 years
old (SD =8.87).
Procedure. To measure respondents’ perceived social
networks, we used the ego network method (Burt 1992).
Respondents were asked to list and describe all mem-
bers of their team in terms of age, gender, ethnicity, and
formal leadership responsibilities. Subsequently, respon-
dents were asked “who would you go to for advice
on work-related matters?” followed by a list of their
coworkers’ names. Respondents checked the names of
those coworkers from whom they sought advice. For
each team member in turn, respondents were then asked
to check the names of coworkers whom the team mem-
ber went to for advice. Thus each respondent provided a
complete network map concerning his or her perceptions
of who shared advice relations with whom in the team,
commonly referred to as a “slice” in cognitive social
structure research (Krackhardt 1987).
Because we were only interested in studying leaders,
we only included respondents’ observations of individu-
als in their team who had formal managerial status and
excluded respondents’ observations of those team mem-
bers who had no formal managerial responsibility. On
average, there were 2.6 people in each respondent’s team
with formal leadership roles. Thus the 149 respondents
to our survey provided 384 observations of the leaders
in their teams.
Measures
Leader Gender. Leader gender was coded as 0 =man
leader and 1 =woman leader.
Advice Network Centralization. Advice network cen-
tralization was measured as the average difference in
indegree centrality between the most central actor and
all others (Wasserman and Faust 1994).
Advice Network Cohesion. Advice network cohesion
was measured as network density, calculated by dividing
the total number of reported ties by the total number of
possible ties (Wasserman and Faust 1994).
Charismatic Leadership. As in Study 1, respondents
were asked to rate their leaders on 15 items that assessed
charismatic leadership (Podsakoff et al. 1990); =0094.
Other Outcome Variables. To examine whether the
effects generalized to other aspects of leadership, we
included further outcome measures in this study. Lead-
ers’ individualized consideration (=0.66) and intel-
lectual stimulation (=0088) were assessed with the
same measure as in Study 1. In addition, we included a
six-item measure of leadership effectiveness (=0.85)
(Giessner and van Knippenberg 2008). We also exam-
ined our hypotheses in the context of perceived rela-
tionship quality between leaders and followers using a
seven-item measure of leader–member exchange (=
0087) (Graen and Uhl-Bien 1995). Respondents provided
ratings on 1–7 Likert scales.
Controls. Leaders’ demographic characteristics have
the potential to bias attributions of leadership qualities
(Rosette et al. 2008), so we controlled for leader age and
ethnicity. Individuals who occupy central roles within
networks tend to be seen as more charismatic (Balkundi
et al. 2011), so we controlled for the perceived central-
ity of the leader. We controlled for team size given that
network density is negatively correlated with the number
of people in the network and we controlled for demo-
graphic characteristics of the team in terms of the ratio
of men to women and how demographically diverse the
team was in respect to age and ethnicity. Finally, we
controlled for the gender of the respondent in the analy-
sis. We considered the inclusion of further control vari-
ables, including full- or part-time employment, industry
affiliation, and occupation, but because these variables
were not associated with any of the focal variables in
this study, they were dropped for the sake of model
parsimony.
Downloaded from informs.org by [163.119.96.63] on 30 March 2015, at 09:11 . For personal use only, all rights reserved.
Brands, Menges, and Kilduff: The Leader-in-Social-Network Schema
Organization Science, Articles in Advance, pp. 1–16, © 2015 INFORMS 7
Attention Checks
Prior to analysis, we established that responses would
be excluded if (a) they came from the same IP address
or (b) response times were greater than two standard
deviations above the mean. Five cases were excluded
because respondents violated condition (a).
Analysis
The data were nested in that each respondent potentially
provided observations concerning the multiple managers
present in the team. Observations from the same indi-
vidual were not independent, and therefore, a regression
analysis was inappropriate. To account for the depen-
dency in our data, we used random coefficient modeling
where target (level 1) and respondent (level 2) effects
were modeled. Our hypotheses predicted a cross-level
interaction between a respondent’s mean-centered per-
ceptions of network structure (level 2) and leader gen-
der (level 1) on attributions of charisma, modeled using
slopes-as-outcomes models.
Results
Table 1shows means, standard deviations, and correla-
tions. People attributed charisma to leaders perceived as
embedded in centralized networks (r=0.36, p=0.01),
and people in general saw women as less charismatic
than men (r= −0016, p=0.01). Table 2presents the
results of regression analyses that controlled for numer-
ous other variables. In line with the bivariate correla-
tion, Model 2 in Table 2shows that people attributed
more charisma to leaders to the extent they perceived
those leaders to be embedded in centralized advice net-
works (=0.02, p=0.001). But charismatic attribu-
tions were not influenced by leader gender (= −0015,
p=0015). Further, Model 4 in Table 2reveals no evi-
dence that people attributed charisma to their leaders on
the basis of how cohesive networks were perceived to be
(= −0028, p=0.55). Given these main effect results
(that differed from Study 1), the question remained as
to whether the hypotheses, which posited interaction
effects, received support.
Hypothesis Tests. Recall the prediction in Hypothe-
sis 1that in networks perceived as centralized, women,
relative to men, would be seen as less charismatic. As
Model 3 in Table 2shows, this hypothesis was sup-
ported: there was a significant cross-level interaction
between team member gender and perceived advice net-
work centralization on attributions of charisma (=
−0001, p=0.04). The simple slopes analysis (conducted
at 1.5 standard deviations from the mean2) confirmed
the expected pattern (see Figure 2). Individuals tended to
attribute less charisma to women than to men when they
perceived their team networks to be highly status con-
centrated (z= −2052, p=0.02). However, when central-
ization was perceived to be low, there was no significant
Table 1 Study 2: Means, Standard Deviations, and Correlations
Variable Mean SD 1 2 3 4 5 6 7 8 9 10 11 12 13
1Charisma 4042 1043
2Perceived centralization 17092 27099 0036∗∗
3Perceived cohesion 0041 0023 0003 −0018∗
4Leader gender a0049 0050 −0016∗∗ −0033∗∗ −0005
5Perceived centralization ×2001 10077 0008 0061∗∗ −0009 0031∗∗∗
Leader gender
6Perceived cohesion ×0011 0021 −0007 −0029∗∗ 0048∗∗ 0086∗∗∗ 0022∗∗
Leader gender
7Leader age 41029 12007 −0014∗∗ −0029∗∗ 0001 0024∗−0012∗0022∗∗∗
8Leader ethnicity b0035 0048 0016∗∗ 0045∗∗ −0005 −0007 0017∗∗ −0004 −0026∗∗
9Leader centrality 2005 2048 −0004 −0026∗∗ 005∗∗ −0009 −0011 0006 0006 −0007
10 Team size 7015 2035 0007 0025∗∗ −0034∗∗∗ −0007 0006 −0017 −0002 0012 0014
11 Team gender ratio 0035 0026 0006 0002 −0008 0023∗∗∗ 0033∗∗ 0033∗∗∗ −0014 0012 0001 0007
12 Team ethnicity variability 0054 0067 −0017 −0017∗−0003 0009 −0015 0016 −0002 0020∗0011 −0004 0013
13 Team age variability 7061 4082 −0019∗−0025∗∗ −0003 0033∗∗ −0015 0030∗∗ 0066∗∗∗ −0028∗∗ 0014 −0005 0011 0006
14 Respondent gender 0039 0048 −0022∗−0012 −0005 0047∗∗∗ 0013 0039∗∗ 0024∗∗ −0008 0010 −0008 0023∗∗ 0009 0021∗
a0=man, 1 =woman.
b0=white, 1 =not white.
∗p < 0005; ∗∗p < 0001 (two-tailed tests).
Downloaded from informs.org by [163.119.96.63] on 30 March 2015, at 09:11 . For personal use only, all rights reserved.
Brands, Menges, and Kilduff: The Leader-in-Social-Network Schema
8Organization Science, Articles in Advance, pp. 1–16, © 2015 INFORMS
Table 2 Study 2: Effects of Leader Gender and Perceived Social Network Structure on Attributions of Charismatic Leadership
Variable Model 1 Model 2 Model 3 Model 4 Model 5
Intercept 4070 4075 4080 4070 4070
400135 400125 400125 400135 400135
Individual variables
Leader ethnicity 0015 0007 0007 0015 0017
400125 400125 400115 400125 400125
Leader age −0001 −0001 −0001 −0001 −0001
400015 400015 400015 400015 400015
Leader’s perceived centrality 0013∗∗∗ 0014∗∗∗ 0014∗∗ 0014∗∗∗ 0015∗∗∗
400035 400045 400045 400035 400035
Respondent gender −0036 −0029 −0024 −0036 −0036
400245 400245 400255 400245 400245
Leader gender −0015 −0015 −0029∗−0015 −0015
400115 400125 400125 400115 400115
Team variables
Team size −0001 −0004 −0005 −0002 −0002
400055 400045 400045 400055 400055
Team gender ratio 0046 0025 0044 0042 0041
400495 400415 400435 400495 400495
Team ethnicity variability −0029†−0015 −0017 −0028 −0031†
400175 400155 400155 400175 400175
Team age variability −0004 −0002 −0003 −0004 −0004
400025 400035 400035 400025 400025
Perceived centralization 0002∗∗∗ 0002∗∗∗
400015 400015
Perceived cohesion −0029 −0056
400525 400545
Two-way interactions
Leader gender ×Perceived centralization −0001∗
400015
Leader gender ×Perceived cohesion 0085∗
400415
Level 1 R2a 0012 0017 0018 0012 0013
Level 2 R2a 0007 0010 0011 0008 0009
2b 8011∗∗∗ 9082∗∗ 0028 4063†
Notes.N=386 observations at level 1 and N=149 observations at level 2. Standard errors are in parentheses.
aCalculated as per Snijders and Bosker (1999).
bLikelihood ratio test of model fit.
†p < 0010; ∗p < 0005; ∗∗ p < 0001; ∗∗∗p < 00001 (two-tailed tests).
difference between the charisma attributed to men and
women (z=0099, p=0032).
Thus there is evidence of biased perceptions of
women’s leadership qualities. Does similar bias affect
perceptions of men’s leadership qualities? Recall that
Hypothesis 2predicted that in networks perceived as
cohesive, men, relative to women, would be seen as less
charismatic. As Model 5 in Table 2shows, this hypoth-
esis was supported: there was a significant interaction
between team member gender and perceived advice net-
work cohesion (=0085, p=0004). The simple slopes
analysis (conducted at 1.5 standard deviations from the
mean; see Preacher 2014) confirmed the expected pattern
(see Figure 3). People tended to attribute less charisma
to men than to women when they perceived their team
networks to be highly cohesive, an effect that was
marginally significant (z=1.68, p=0.09). However,
when cohesion was perceived to be low, there was no
significant difference between the charisma attributed to
men and women (z=0.44, p=0065).
Additional Analyses. We examined whether the gen-
der biasing effects of centralization depended on
cohesion (and vice versa), but the three-way inter-
action between gender, centralization, and cohesion
was not significant (=0.01, p=0097). We also
tested whether other components of transformational
leadership—namely, intellectual stimulation and individ-
ualized consideration—would be affected by the leader-
in-network schema. With intellectual stimulation as the
dependent variable, there was no significant interaction
between perceived centralization and gender (=0.34,
p=0.74) or perceived advice network cohesion and
gender (=1.95, p=0.23). The results were similar
for individualized consideration: there was no signif-
icant interaction between perceived centralization and
Downloaded from informs.org by [163.119.96.63] on 30 March 2015, at 09:11 . For personal use only, all rights reserved.
Brands, Menges, and Kilduff: The Leader-in-Social-Network Schema
Organization Science, Articles in Advance, pp. 1–16, © 2015 INFORMS 9
Figure 2 Study 2: Attributions of Charisma to Men and
Women Leaders in the Context of Perceived
Centralization
1
2
3
4
5
6
7
Low centralization High centralization
Attributions of charismatic leadership
Men
Women
p = 0.02
p = 0.32
gender (=0.07, p=0.11) or perceived advice net-
work cohesion and gender (=0.77, p=0.64). Further-
more, with a measure of leadership effectiveness as the
dependent variable, there was no significant interaction
between perceived centralization and gender (=0.01,
p=0.59) or perceived advice network cohesion and
gender (=0.49, p=0.58). The results were similar
with leader–member exchange as the dependent variable:
there was no significant interaction between perceived
centralization and gender (=−001, p=0.97) or per-
ceived advice network cohesion and gender (=0.15,
p=0.66).
Discussion
The results of Study 2 show that men leaders were seen
as more charismatic than women leaders to the extent
that people perceived their team advice networks to be
highly centralized. This gender bias reversed when team
advice networks were perceived to be cohesive, however,
Figure 3 Study 2: Attributions of Charisma to Men and Women
Leaders in the Context of Perceived Cohesion
1
2
3
4
5
6
7
Low cohesion High cohesion
Attributions of charismatic leadership
Men
Women
p = 0.65
p = 0.09
so that women leaders were perceived to be more charis-
matic than men leaders. Further, these results (in support
of the hypotheses) were restricted to charismatic leader-
ship attributions.
Both studies presented so far, however, entail a limi-
tation in that they do not permit us to test the effects of
centralization and cohesion independently of one another.
Networks that feature dense connections tend to be
decentralized; thus the network properties of cohesion
and centralization tend to be negatively correlated. How-
ever, in the realm of perception, people are likely to dis-
tort both the extent of centralization and the extent of
connectedness beyond the empirically possible (Kilduff
et al. 2008). Study 3 allows us to capture indepen-
dent measures of perceived centralization and perceived
cohesion.
Study 3
In this study, we tested predictions concerning the
leader-in-social-network schema with a new approach to
assessing perceptions of centralization and cohesion in
advice networks (Mehra et al. 2014). We wanted to see
how respondents would attribute charisma to men and
women leaders when asked directly about the network
structures in which they participated. This approach
allowed us to evaluate attributions of charisma to men
and women leaders in cases when both centralization
and cohesion were perceived, thereby fully elaborating
our theoretical model.
Method
Sample. Respondents were 157 (79 men and 76 women)
U.S. resident full- or part-time employees representing
more than 20 different industries and a wide variety of
occupations. They were mostly white (77%) and on aver-
age 33 years old (SD =10.88) and had worked for an
average of 13 years (SD =9.99). We recruited and paid
these respondents through an online panel.
Procedure. After connecting to our website, respon-
dents were asked to complete two sets of questions,
the order of which was counterbalanced.3One set of
questions concerned the demographic characteristics and
charisma of respondents’ team leaders; the other set of
questions concerned respondents’ perceptions of their
team networks.
Measures
Leader Gender. As in Study 2, leader gender was
coded as 0 =man and 1 =woman.
Centralization and Cohesion. We presented respon-
dents with two visual scales adapted from Mehra et al.
(2014).4One scale represented perceived centralization
Downloaded from informs.org by [163.119.96.63] on 30 March 2015, at 09:11 . For personal use only, all rights reserved.
Brands, Menges, and Kilduff: The Leader-in-Social-Network Schema
10 Organization Science, Articles in Advance, pp. 1–16, © 2015 INFORMS
and ranged from completely centralized to completely
decentralized. The other scale represented perceived
cohesion and ranged from 0% to 100% density.
Charismatic Leadership. As in Studies 1 and 2,
respondents were asked to rate their leaders on 15 items
that assessed charismatic leadership (Podsakoff et al.
1990); =0096.
Other Outcome Variables. Leaders’ individualized
consideration and intellectual stimulation were assessed
with the same measures as in Studies 1 and 2. For indi-
vidualized consideration, =0.62; for intellectual stim-
ulation, =0.90.
Control Variables. As in Study 2, we controlled for
demographic characteristics of the leader, including age
and ethnicity, and the demographic composition of the
team in terms of age, ethnicity, and gender ratio. We also
again controlled for the gender of the respondents. Indi-
viduals who occupy central roles within networks tend
to be seen as more charismatic (Balkundi et al. 2011), so
we controlled for the perceived centrality of the leader.
We also controlled for individuals’ perceptions of their
own centrality in the network because people’s reactions
to the perceived degree of centralization in their team
advice networks may be affected by perceptions of their
own importance within those interactions. We measured
both leader and respondent centrality by describing cen-
trality accompanied by a visual depiction and then ask-
ing respondents to indicate their own and their leaders’
centrality on a five-point scale (1 =on the edge of the
network to 5 =in the center of the network). Finally,
because large networks tend to be less dense than small
networks, we controlled for team size.
Attention Checks
Prior to analysis, we established that responses would
be excluded if (a) they came from the same IP address
or (b) response times were greater than two standard
deviations above the mean. No cases met these rules
for exclusion; thus all respondents were included in the
analysis.
Analysis
We conducted a multiple linear regression to examine
how respondents’ perceptions of their team advice net-
works in terms of centralization and cohesion affected
the charisma they attributed to their leaders. Centraliza-
tion and cohesion were standardized prior to analysis.
In social networks, density and centralization are neg-
atively correlated. However, our survey design meant
that respondents were able to choose technically impos-
sible configurations (e.g., dense and highly centralized
networks). To account for this, we controlled for per-
ceptions of density in our analysis of centralization, and
vice versa.
Results
Table 3displays the descriptive statistics and correla-
tions among the variables. Interestingly, respondents in
this sample, who were allowed to assess the network
properties of centralization and cohesion independently,
saw leaders in cohesive networks as charismatic (r=
0.36, p=0.01) and leaders in centralized networks as
noncharismatic (r= −0017, p=0.05).
Table 4displays the regression results. Model 4 shows
the main effects of the three variables of interest, con-
trolling for the other variables. Paralleling the result of
the bivariate correlation, to the extent that respondents
saw their team advice networks as cohesive, they tended
to attribute more charisma to their leaders, regardless of
whether the leaders were men or women (=5.71, p=
0.0001). There were no significant effects for network
centralization or leader gender. Given these results, the
question remained: Were the hypotheses (that involved
interaction effects) supported?
Hypothesis Tests. Hypothesis 1predicted that to the
extent that team advice networks were perceived to be
centralized, women would be seen as less charismatic
leaders than men. As can be seen in Table 4, Model 5,
there was support for this hypothesis in the form of a
significant interaction between leader gender and advice
network centralization (= −5093, p=0.04; see Fig-
ure 4). Analyses of the interaction at plus and minus
1.5 standard deviations from the mean (see Preacher
2014) confirmed that when individuals perceived their
team advice networks to be highly centralized, women
leaders were attributed with less charisma than men
leaders (t= −1097, p=0.03). When centralization was
perceived to be low, however, there was no gender
bias in attributions of charismatic leadership (t=1.26,
p=0021).
Hypothesis 2predicted that to the extent that team
advice networks were perceived to be cohesive, men
would be seen as less charismatic leaders than women.
As Model 6 in Table 4shows, this hypothesis was not
supported in this study: the interaction between cohe-
sion and leader gender was not significant, suggesting
that attributions of charisma to men and women lead-
ers did not differ as a function of perceptions of team
advice network cohesion (=0.49, p=0085). Rather,
there continued to be a strongly significant main effect
of cohesion on attributions of leadership (=5.79, p=
0.001) and a nonsignificant effect of leader gender (=
−2004, n.s.). This pattern of results indicated that in
networks perceived to be cohesive, both men and women
were perceived as charismatic leaders.
Additional Analyses. These data allowed us to test
how respondents evaluated men and women leaders
when both centralization and cohesion were assessed
separately by respondents. We examined, therefore,
Downloaded from informs.org by [163.119.96.63] on 30 March 2015, at 09:11 . For personal use only, all rights reserved.
Brands, Menges, and Kilduff: The Leader-in-Social-Network Schema
Organization Science, Articles in Advance, pp. 1–16, © 2015 INFORMS 11
whether there was a three-way interaction between per-
ceived centralization, perceived cohesion, and leader
gender. The interaction term was not significant (=
−302, p=0019). Thus attributions of charisma to men
versus women leaders were not affected by whether or
not advice networks perceived as cohesive were also
seen as centralized.
Our second test of how respondents evaluated men
and women leaders when both centralization and cohe-
sion were perceived involved testing our hypotheses on
a subsample of respondents who indicated that they
perceived their team social networks to be both cen-
tralized and cohesive. We selected all respondents who
scored at or above the scale midpoint on perceptions of
both centralization and cohesion (N=93). Hypothesis 1
(men would be attributed more charisma than women in
networks perceived as centralized) was supported by a
significant two-way interaction between gender and cen-
tralization (= −0094, p=0003). Supplementary anal-
yses confirmed that to the extent respondents perceived
their networks to be highly centralized, men were evalu-
ated more favorably as leaders (t= −206, p=0001), but
when centralization was perceived to be lower, there was
no gender bias in attributions of leadership (t=1.29,
p=0020). Recall that this result was found in a sam-
ple in which all respondents perceived their networks
to be relatively cohesive. Thus, even when respondents
perceived their networks to be cohesive, a condition
expected to facilitate favorable evaluations of women’s
leadership, centralization still negatively affected percep-
tions of women’s charisma. Hypothesis 2(men would
be attributed less charisma than women in networks per-
ceived as cohesive) was not supported in this restricted
sample. The two-way interaction between leader gender
and density was not significant (=0.51, p=0014) in
the sample in which everyone perceived their networks
to be relatively centralized.
We also tested whether the leader-in-social-network
schema would affect the intellectual stimulation and
individualized consideration components of transforma-
tional leadership. It did not. We did not find a two-
way interaction between leader gender and perceptions
of centralization on either intellectual stimulation (=
−0028, p=0066) or individualized consideration (=
0055, p=0047). Likewise, we did not find a two-way
interaction between leader gender and perceptions of
cohesion on intellectual stimulation (=0022, p=0071)
or individualized consideration (= −0064, p=0038).
Discussion
Our leader-in-social-network theory suggests that when
individuals perceive mismatches between expectations
relating to the social network context and expectations
relating to the gender of the leader, individuals will
make gender-biased attributions of charismatic leader-
ship. For attributions concerning women leaders, we
Table 3 Study 3: Means, Standard Deviations, and Correlations
Variable Mean SD 1 2 3 4 5 6 7 8 9 10 11 12 13 14
1Charisma 4099 1017
2Perceived centralization 2093 1012 −0017∗
3Perceived cohesion 3080 1014 0036∗∗ −0021∗∗
4Leader gender a0041 0049 −0008 0002 −0005
5Perceived centralization ×0002 0062 −0015†0061∗∗∗ −0009 0002
Leader gender
6Perceived cohesion × −0002 0070 0024∗∗ −0008 0069∗∗∗ −0004 −0013
Leader gender
7Leader age 40084 12058 −0010 0012 −0019∗0006 −0011 0022∗∗
8Leader ethnicity b0017 0038 0006 0010 −0005 0010 0004 0005 −0030∗∗
9Leader centrality 2029 1058 0007 −0005 0004 −0001 −0003 −0005 0013 −0009
10 Team size 5059 5022 0002 −0003 −0004 −0011 0002 −0005 0001 −0005 0009
11 Team gender ratio 0068 1048 −0001 0003 0003 −0062∗∗∗ −0008 0007 −0005 0002 0016∗0002
12 Team ethnicity variability 0064 0031 −0010 −0001 0005 −0007 0001 0002 0010∗−0051 −0025∗∗ −0011 −0006
13 Team age variability 7071 4074 −0002 −0008 −0023∗∗ 0011 0003 −0023∗∗ 0053 −0007 0014†−0002 0004 0002
14 Respondent gender a0049 0050 0010 0003 −0010 0021∗0018∗−0010 0015†−0007 0001 −0004 −0061∗∗ 0001 −0003
15 Respondent centrality 3031 1023 0035∗∗∗ −0021∗∗ 0013 −0007 −0007 0003 −0009 0005 −0042∗∗∗ 0005 −0006 0007 −0013 0017∗
a0=man, 1 =woman.
b0=white, 1 =other.
†p < 0010; ∗p < 0005; ∗∗ p < 0001; ∗∗∗p < 00001 (two-tailed tests).
Downloaded from informs.org by [163.119.96.63] on 30 March 2015, at 09:11 . For personal use only, all rights reserved.
Brands, Menges, and Kilduff: The Leader-in-Social-Network Schema
12 Organization Science, Articles in Advance, pp. 1–16, © 2015 INFORMS
Table 4 Study 3: Effects of Leader Gender and Perceived Social Network Structure on Attributions of Charismatic Leadership
Variable Model 1 Model 2 Model 3 Model 4 Model 5 Model 6
Intercept 321∗∗∗ 331∗∗∗ 301∗∗∗ 327∗∗∗ 309∗∗∗ 325∗∗∗
076077072085084085
Control variables
Leader ethnicity −011 023 138 158 055 169
448449423427425433
Leader age −017 −018 −012 −013 −018 −012
014014013013013013
Leader centrality 292∗∗∗ 278∗∗ 232∗229∗225∗229∗
103103097099097099
Team size −003 −001 005 002 004 002
026026025026025026
Team gender ratio 381 358 521 279 392 275
523523493654678657
Team ethnicity variability −245 −213 −370 −379 −429 −372
54354351165451520
Team age variability 019 018 041 042 060†042
036036034034035034
Respondent gender 247 276 471 427 608†428
360361343355360356
Respondent’s perceived centrality 656∗∗∗ 619∗∗∗ 554∗∗∗ 536∗∗∗ 553∗∗∗ 537∗∗∗
124129119126125127
Predictor variables
Leader gender −202 −182 −204
362358363
Perceived centralization −146 −025 194 −027
143139172140
Perceived cohesion 579∗∗∗ 571∗∗∗ 593∗∗∗ 579∗∗∗
134138137183
Two-way interactions
Leader gender ×Perceived centralization −593∗
281
Leader gender ×Perceived cohesion 049
261
R2020∗∗∗ 015∗∗∗ 024∗∗∗ 024∗∗∗ 026∗∗∗ 023∗∗∗
Notes.N=157. Standard errors are in parentheses.
†p<010; ∗p<005; ∗∗p<001; ∗∗∗p<0001 (two-tailed tests).
Figure 4 Study 3: Attributions of Charisma to Men and
Women Leaders in the Context of Perceived
Centralization
1
2
3
4
5
6
7
8
Low centralization High centralization
Attributions of charismatic leadership
Men
Women
p = 0.21 p = 0.05
found support for this prediction: even when individuals
perceived their networks to be highly cohesive, women
were still negatively affected by centralization. For men
leaders, our findings from Study 3 are less clear given
the absence of support for the hypothesized negative
effect of cohesion on charismatic attributions for men.
More research is needed, but taken alone, the results
from Study 3 suggest that the prototypical pattern of
men leaders in centralized networks may be more potent
in its effects on attributed charisma than the prototypical
pattern of women leaders in cohesive networks.
General Discussion
In support of the leader-in-social-network schema per-
spective, results across three studies suggest that when
people’s expectations concerning team leaders are mis-
aligned, people attribute less charisma to leaders. Expec-
tations are cued both by the social network context of
Downloaded from informs.org by [163.119.96.63] on 30 March 2015, at 09:11 . For personal use only, all rights reserved.
Brands, Menges, and Kilduff: The Leader-in-Social-Network Schema
Organization Science, Articles in Advance, pp. 1–16, © 2015 INFORMS 13
work (centralized or cohesive team advice networks) and
by the gender of the leader. In support of Hypothesis 1,
people saw women as less charismatic than men when
centralized team advice networks cued people’s expec-
tations of male status and power. In support of Hypothe-
sis 2, people in two of the three studies saw men as less
charismatic than women when cohesive team advice net-
works cued people’s expectations of female caring and
communality.
Our hypotheses concerned interaction effects between
network structure and leader gender. The inconsistent
pattern of main effects across the studies suggests the
importance of taking an interaction approach to under-
standing why some leaders are considered more charis-
matic than others. In our research we found that the
effects of perceived network properties on attributions
of charismatic leadership were not straightforward but
required consideration of leader gender in the context of
network structure. It is interesting to note that cohesion
was positively related to charisma attributions in Stud-
ies 1 and 3 in which participants reacted to schematic
representations of team network structure, whereas cen-
tralization was positively to charisma in Study 2, in
which participants reported on the actual ties linking
team members to one another. One possible interpre-
tation of these findings is that when individuals are
allowed to react to “ideal types” of networks, they tend
to provide an espoused theory (see Argyris 1993) of
good leadership as a collective, shared exercise. But
when people are anchored to the actual ties they observe
in their teams, they reveal a more traditional theory-
in-use (cf. Argyris 1993), in which good leadership is
typified by command-and-control.
Contributions to Theory and Research
The research contributions derive from a focus on the
effects of people’s mismatched expectations. Network
cognition research has expanded our knowledge of how
people think about social networks (Brands 2013). In
particular, network cognition research has spent decades
examining bias in patterns of social network perceptions
(e.g., Kilduff et al. 2008). But this research has nei-
ther theorized nor researched how perceptions of net-
work structure affect perceivers’ attributions of qualities
(positive or otherwise) to others in the social network.
In putting forward theory and empirical research con-
cerning mismatched expectations, we move beyond the
almost exclusive focus on patterns of biased perceptions
in the minds of perceivers. We identify leaders within
work teams as potential victims or beneficiaries of per-
ceivers’ expectations that are cued by social network
context and targets’ attributes.
Thus we move forward, within the research program
of cognitive networks, the idea that social network struc-
tures, although facilitating outcomes for some categories
of people, may adversely affect outcomes for others.
We build on prior research showing that network struc-
tures affect people differently depending on people’s
characteristics. Thus, women (compared with men) are
less likely to benefit from networks featuring structural
holes (Burt 1992) and may be disadvantaged by network
homophily (Ibarra 1992). The dynamic interplay of cog-
nitive expectations and individual characteristics is on
the frontier of network research (Burt et al. 2013).
A further contribution of our research is to the area
of leadership. First, we advance the cognitive social
network approach to leadership (Balkundi and Kilduff
2005) by articulating the ways in which perceptions
of network context and leader gender relate to attribu-
tions of charisma to the leader. Prior work has exam-
ined how the configuration of ties around leaders affects
others’ estimations of leader charisma (Balkundi et al.
2011). But this prior work assumed that network roles
benefited leaders regardless of leaders’ individual char-
acteristics. We draw on social psychological research
concerning the ubiquity and potency of gender schemas
in evaluating leaders’ behaviors (Eagly and Karau 2002,
Eagly et al. 1992) to show that the social network con-
text of the perceiver differentially affects attributions of
charisma to men and women leaders.
In doing so, we raise a question for theory concern-
ing gender and leadership. There is a well-documented
association between gender stereotypes and agentic-
communal orientation (Eagly and Steffen 1984). When
applied to leadership, this means that when women enact
the agentic behaviors necessary for leadership, they are
penalized for being insufficiently communal, resulting in
negative evaluations of their leadership abilities (Eagly
and Karau 2002). This existing research on gender and
leadership has tended to focus on leaders’ behaviors in
accounting for gender bias in leadership. However, our
research examines how social interactions that are per-
ceived to occur around leaders, regardless of leaders’
participation in these exchanges, shape perceptions of
leadership qualities. To the extent that these interactions
are perceived by followers as arising unprompted by the
leader, one implication of our findings is that perceptions
of agency and communality arise not only from men’s
and women’s behavior but also from the social context
in which those behaviors occur. However, it may be that
leaders are perceived by the followers to be responsible
for cultivating the social structure in their teams along
centralized and cohesive dimensions. If this is the case,
then our research highlights an important boundary con-
dition to the agency penalty normally applied to women
leaders in that it suggests that women can be agentic in
networks as long as they cultivate cohesive and therefore
gender appropriate networks. Future research can clarify
these differing interpretations of our findings.
Another contribution of this research is to redress the
prevailing consensus that women relative to men are
disadvantaged as leaders. Although much research has
Downloaded from informs.org by [163.119.96.63] on 30 March 2015, at 09:11 . For personal use only, all rights reserved.
Brands, Menges, and Kilduff: The Leader-in-Social-Network Schema
14 Organization Science, Articles in Advance, pp. 1–16, © 2015 INFORMS
sought to identify the behaviors of charismatic leaders,
little attention has been paid to demographic character-
istics such as gender, as noted in a recent review (Walter
and Bruch 2009). This oversight is particularly notable
given that bias against female leaders is a key theme
in leadership research (Eagly et al. 1992). In our work
we move away from the prevailing consensus to suggest
that gender biases are context dependent and are cued,
in part, by the social network structural characteristics
of centralization and cohesion.
Further, in terms of a contribution to leadership
research, this study underpins the need to disaggre-
gate multidimensional leadership concepts and study
specific leadership dimensions. Our research shows
effects specifically for charisma, which is one dimen-
sion of the broader concept of transformational leader-
ship (Bass 1999). Because the leader-in-social network
schema engages the charismatic leadership prototype,
its effects are bound to attributions of charisma and
leave unaffected other (noncharismatic) dimensions of
transformational leadership (i.e., individualized consid-
eration, intellectual stimulation) as well as transactional
leadership, general evaluations of leadership, and ratings
of the relationship quality between the leader and the fol-
lower. Even though the customary pattern is to downplay
differences between charismatic leadership and transfor-
mational leadership (van Knippenberg and Sitkin 2013),
our research shows that charismatic leadership, because
of its unique roots in attributions of charisma, should
not be equated with transformational leadership but be
studied as a specific construct in its own right.
Limitations and Future Research
Across three studies we mandated that respondents
(recruited exclusively from online panels) had work
experience with full- or part-time jobs in the United
States. Data collected through online panels are of simi-
lar or better quality than data collected from respondents
in specific contexts such as students at a university or
employees of a single company (Buhrmester et al. 2011).
Data from multiple sources improve external validity.
But because our respondents worked in a wide range
of jobs, industries, and organizations, we were unable
to estimate the effects of organizational culture on their
attributions of leadership. It is likely that some orga-
nizations (relative to others) feature cultures conducive
to women leaders. Future research could examine how
organizational culture itself represents a rich repository
of cognitive expectations affecting how men and women
leaders are perceived.
Cognitive social network research typically examines
the relationship between perceived networks and actual
networks (e.g., Kilduff et al. 2008). In our Studies 2
and 3, we were unable to confirm the veracity of reports
of network structure. However, perceptions of social
networks represent phenomena of interest in their own
right (Krackhardt 1987), given that if people perceive
situations as real, then these situations have real con-
sequences (Thomas and Thomas 1928). Our research
focused on outcomes of the mismatch between expec-
tations cued by perceived network structure and expec-
tations cued by leader gender. Future research could
examine whether the effects we describe are rooted in
the actual press of network structure or whether these
effects are mainly the result of biased social network
perceptions.
Supplemental Material
Supplemental material to this paper is available at http://dx.doi
.org/10.1287/orsc.2015.0965.
Endnotes
1The manipulations are available as supplemental material to
this paper.
2Our hypotheses concern networks that show distinct patterns
of perceived centralization and cohesion. Accordingly, we plot-
ted interactions at 1.5 standard deviations from the mean to
reflect networks that are distinctly centralized or cohesive (in
line with pertinent recommendations; see Preacher 2014).
3Robustness checks indicated results were unaffected by the
order in which the materials were presented.
4The measures are available as supplemental material to this
paper.
References
Acker J (1990) Hierarchies, jobs, bodies: A theory of gendered orga-
nizations. Gender Soc. 4(2):139–158.
Argyris C (1993) Knowledge for Action: A Guide to Overcoming Bar-
riers to Organizational Change (Jossey-Bass, San Francisco).
Aries E (1976) Interaction patterns and themes of male, female, and
mixed groups. Small Group Res. 7(1):7–18.
Avolio BJ, Yammarino FJ (1990) Operationalizing charismatic lead-
ership using a levels-of-analysis framework. Leadership Quart.
1(3):193–208.
Baldwin MW (1992) Relational schemas and the processing of social
information. Psych. Bull. 112(3):461–484.
Bales RF, Slater PE (1955) Role differentiation in small decision-
making groups. Parsons T, Bales RF, eds. Family, Socialization,
and Interaction Process (Free Press, Glencoe, IL), 259–306.
Balkundi P, Kilduff M (2005) The ties that lead: A social network
approach to leadership. Leadership Quart. 16(6):941–961.
Balkundi P, Kilduff M, Hanison DA (2011) Centrality and charisma:
Comparing how leader networks and attributions affect team per-
formance. J. Appl. Psych. 96(6):1209–1222.
Bass BM (1985) Leadership: Good, better, best. Organ. Dynam.
13(3):26–40.
Bass BM (1999) Two decades of research and development in trans-
formational leadership. Eur. J. Work Organ. Psych. 8(1):9–32.
Beal DJ, Cohen RR, Burke MJ, McLendon CL (2003) Cohesion and
performance in groups: A meta-analytic clarification of construct
relations. J. Appl. Psych. 88(6):989–1004.
Brands RA (2013) Cognitive social structures in social network
research: A review. J. Organ. Behav. 34(S1):S82–S103.
Buhrmester M, Kwang T, Gosling SD (2011) Amazon’s Mechanical
Turk. Perspect. Psych. Sci. 6(1):3–5.
Downloaded from informs.org by [163.119.96.63] on 30 March 2015, at 09:11 . For personal use only, all rights reserved.
Brands, Menges, and Kilduff: The Leader-in-Social-Network Schema
Organization Science, Articles in Advance, pp. 1–16, © 2015 INFORMS 15
Burt RS (1992) Structural Holes: The Social Structure of Competition
(Harvard University Press, Cambridge, MA).
Burt RS, Kilduff M, Tasselli S (2013) Social network analysis:
Foundations and frontiers on advantage. Annual Rev. Psych.
64(1):527–547.
Cantor N, Mischel W (1979) Prototypes in person perception.
Berkowitz L, ed. Advances in Experimental Social Psychology,
Vol. 12 (Academic Press, New York), 3–52.
Coleman JS (1988) Social capital in the creation of human capital.
Amer. J. Sociol. 94(Supplement):S95–S120.
Conger JA, Kanungo RN (1987) Toward a behavioral theory of charis-
matic leadership in organizational settings. Acad. Management
Rev. 12(4):637–647.
Conger JA, Kanungo RN, Menon ST (2000) Charismatic leadership
and follower effects. J. Organ. Behav. 21(7):747–767.
Diekman AB, Goodfriend W, Goodwin S (2004) Dynamic stereotypes
of power: Perceived change and stability in gender hierarchies.
Sex Roles 50(3–4):201–215.
Eagly AH, Carli LL (2003) The female leadership advantage: An
evaluation of the evidence. Leadership Quart. 14(6):807–834.
Eagly AH, Karau SJ (2002) Role congruity theory of prejudice toward
female leaders. Psych. Rev. 109(3):573–598.
Eagly AH, Steffen VJ (1984) Gender stereotypes stem from the dis-
tribution of women and men into social roles. J. Personality Soc.
Psych. 46(4):735–754.
Eagly AH, Karau SJ, Makhijani MG (1995) Gender and the effec-
tiveness of leaders. Psych. Bull. 117(1):125–145.
Eagly AH, Makhijani MG, Klonsky BG (1992) Gender and the eval-
uation of leaders: A meta-analysis. Psych. Bull. 111(1):3–22.
Freeman LC (1978–1979) Centrality in social networks conceptual
clarification. Soc. Networks 1(3):215–239.
Freeman LC, Romney AK, Freeman SC (1987) Cognitive struc-
ture and informant accuracy. Amer. Anthropologist 89(2):
310–325.
Giessner SR, van Knippenberg D (2008) License to fail: Goal defi-
nition, leader group protoypicality, and perceptions of leadership
effectiveness after leader failure. Organ. Behav. Human Decision
Processes 105(1):14–35.
Graen GB, Uhl-Bien M (1995) Relationship-based approach to lead-
ership: Development of leader-member exchange (LMX) theory
of leadership over 25 years: Applying a multi-level multi-domain
perspective. Leadership Quart. 6(2):219–247.
Howell JM, Shamir B (2005) The role of followers in the charismatic
leadership process: Relationships and their consequences. Acad.
Management Rev. 30(1):96–112.
Ibarra H (1992) Homophily and differential returns: Sex differences
in network structure and access in an advertising firm. Admin.
Sci. Quart. 37(3):422–447.
Janicik GA, Larrick RP (2005) Social network schemas and the
learning of incomplete networks. J. Personality Soc. Psych.
88(2):348–364.
Jung DI, Avolio BJ (2000) Opening the black box: An experimental
investigation of the mediating effects of trust and value congru-
ence on transformational and transactional leadership. J. Organ.
Behav. 21(8):949–964.
Kilduff M, Crossland C, Tsai W, Krackhardt D (2008) Organi-
zational network perceptions versus reality: A small world
after all? Organ. Behav. Human Decision Processes 107(1):
15–28.
Krackhardt D (1987) Cognitive social structures. Soc. Networks
9(2):109–134.
Labianca G, Brass DJ, Gray B (1998) Social networks and perceptions
of intergroup conflict: The role of negative relationships and third
parties. Acad. Management J. 41(1):55–67.
Lord RG (2005) Implicit leadership theory. Schyns B, Meindl JR,
eds. Implicit Leadership Theories (Information Age Publishing,
Charlotte, NC), ix–xiv.
Lord RG, Emrich CG (2000) Thinking outside the box by looking
inside the box: Extending the cognitive revolution in leadership
research. Leadership Quart. 11(4):551–579.
Lord RG, Foti RJ (1986) Schema theories, information processing,
and organizational behavior. Sims HP Jr, Gioia DA, eds. The
Thinking Organization (Jossey-Bass, San Francisco), 20–48.
Lord RG, Maher KJ (1994) Leadership and Information Process-
ing: Linking Perceptions and Organizational Performance (Rout-
ledge, New York).
Lord RG, Foti RJ, De Vader CL (1984) A test of leadership cat-
egorization theory: Internal structure, information processing,
and leadership perceptions. Organ. Behav. Human Performance
34(3):343–378.
Lord RG, Brown DJ, Harvey JL, Hall RJ (2001) Contextual con-
straints on prototype generation and their multilevel conse-
quences for leadership perceptions. Leadership Quart. 12(3):
311–338.
Maccoby EE (1990) Gender and relationships: A developmental
account. Amer. Psychologist 45(4):513–520.
Magee JC, Galinsky AD (2008) Social hierarchy: The self-reinforcing
nature of power and status. Acad. Management Ann. 2:351–398.
Meeker BF, Weitzel-O’Neill PA (1977) Sex roles and interper-
sonal behavior in task-oriented groups. Amer. Sociol. Rev. 42(1):
91–105.
Mehra A, Borgatti SP, Soltis S, Floyd T, Ofem B, Halgin DS, Lopez-
Kidwell V (2014) Imaginary worlds: Using visual network scales
to capture perceptions of social networks. Brass DJ, Labianca
G, Mehra A, Halgin DS, Borgatti SP, eds. Contemporary Per-
spectives on Organizational Social Networks (Emerald Group
Publishing, Bingley, UK), 315–336.
Moskowitz DS, Suh EJ, Desaulniers J (1994) Situational influences
on gender differences in agency and communion. J. Personality
Soc. Psych. 66(4):753–761.
Moss-Racusin CA, Phelan JE, Rudman LA (2010) When men break
the gender rules: Status incongruity and backlash against modest
men. Psych. Men Masculinity 11(2):140–151.
Neisser U (1976) Cognition and Reality: Principles and Implica-
tions of Cognitive Psychology (W. H. Freeman and Company,
New York).
Nye JL, Forsyth DR (1991) The effects of prototype-based biases on
leadership appraisals: A test of leadership categorization theory.
Small Group Res. 22(3):360–379.
Pastor JC, Mayo M, Shamir B (2007) Adding fuel to fire: The impact
of followers’ arousal on ratings of charisma. J. Appl. Psych.
92(6):1584–1596.
Podsakoff PM, MacKenzie SB, Moorman RH, Fetter R (1990) Trans-
formational leader behaviors and their effects on followers’ trust
in leader, satisfaction, and organizational citizenship behaviors.
Leadership Quart. 1(2):107–142.
Preacher K (2014) Interaction effects in MLR, LCA, and MLM.
Accessed October 20, 2014, http://www.quantpsy.org/interact/
interactions.htm.
Prentice DA, Carranza E (2002) What women and men should be,
shouldn’t be, are allowed to be, and don’t have to be: The
contents of prescriptive gender stereotypes. Psych. Women Quart.
26(4):269–281.
Downloaded from informs.org by [163.119.96.63] on 30 March 2015, at 09:11 . For personal use only, all rights reserved.
Brands, Menges, and Kilduff: The Leader-in-Social-Network Schema
16 Organization Science, Articles in Advance, pp. 1–16, © 2015 INFORMS
Rosette AS, Tost LP (2010) Agentic women and communal lead-
ership: How role prescriptions confer advantage to top women
leaders. J. Appl. Psych. 95(2):221–235.
Rosette AS, Leonardelli GJ, Phillips KW (2008) The white stan-
dard: Racial bias in leader categorization. J. Appl. Psych. 93(4):
758–777.
Schank RC, Abelson RP (1977) Scripts, Plans, Goals, and
Understanding: An Inquiry into Human Knowledge Structures
(Lawrence Erlbaum Associates, Hillsdale, NJ).
Schein VE (1973) The relationship between sex role stereotypes
and requisite management characteristics. J. Appl. Psych. 57(2):
95–100.
Sczesny S (2005) Gender stereotypes and implicit leadership theories.
Schyns B, Meindl JR, eds. Implicit Leadership Theories (Infor-
mation Age Publishing, Charlotte, NC), 159–172.
Shamir B (1992) Attribution of influence and charisma to the
leader: The romance of leadership revisited. J. Appl. Soc. Psych.
22(5):386–407.
Snijders TAB, Bosker RJ (1999) Multilevel Analyses: An Introduction
to Basic and Advanced Multilevel Modeling (Sage, London).
Sparrowe RT, Liden RC, Wayne SJ, Kraimer ML (2001) Social net-
works and the performance of individuals and groups. Acad.
Management J. 44(2):316–325.
Thomas WI, Thomas DS (1928) The Child in America: Behavior
Problems and Programs (Alfred A. Knopf, New York).
van Knippenberg D, Sitkin SB (2013) A critical assessment of
charismatic-transformational leadership research: Back to the
drawing board? Acad. Management Ann. 7(1):1–60.
Walter F, Bruch H (2009) An affective events model of charis-
matic leadership behavior: A review, theoretical integration, and
research agenda. J. Management 35(6):1428–1452.
Wasserman S, Faust K (1994) Social Network Analysis: Methods and
Applications (Cambridge University Press, New York).
Weber M (1947) The Theory of Social and Economic Organization
(Oxford University Press, New York).
Zitek EM, Tiedens LZ (2012) The fluency of social hierarchy: The
ease with which hierarchical relationships are seen, remembered,
learned, and liked. J. Personality Soc. Psych. 102(1):98–115.
Raina A. Brands is an assistant professor in organizational
behavior at London Business School. She received her Ph.D.
from the University of Cambridge. Her research focuses on
social networks and cognitions and their consequences for the
careers of women.
Jochen I. Menges is a professor of leadership and human
resource management at WHU–Otto Beisheim School of Man-
agement. He obtained his doctoral degree from the University
of St. Gallen. His research focuses on social dynamics between
leaders and followers and on emotions in organizations.
Martin Kilduff is a professor of organizational behavior in
the Department of Management Science and Innovation, Uni-
versity College London. He received his Ph.D. from Cornell
University. His research focuses on the microfoundations and
consequences of individuals’ social networks.
This work is licensed under a Creative Commons Attribution 4.0 International
License. You are free to copy, distribute, transmit and adapt this work, but you
must attribute this work as “Organization Science. Copyright 2015 INFORMS.
http://dx.doi.org/10.1287/orsc.2015.0965, used under a Creative Commons
Attribution License: http://creativecommons.org/licenses/by/4.0/.”
Downloaded from informs.org by [163.119.96.63] on 30 March 2015, at 09:11 . For personal use only, all rights reserved.