Download as pdf or txt
Download as pdf or txt
You are on page 1of 15

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/51622789

Centrality and Charisma: Comparing How Leader Networks and Attributions


Affect Team Performance

Article  in  Journal of Applied Psychology · September 2011


DOI: 10.1037/a0024890 · Source: PubMed

CITATIONS READS
55 2,286

3 authors:

Prasad Balkundi Martin Kilduff


University at Buffalo, The State University of New York University College London
23 PUBLICATIONS   1,414 CITATIONS    62 PUBLICATIONS   6,242 CITATIONS   

SEE PROFILE SEE PROFILE

David A. Harrison
University of Texas at Austin
103 PUBLICATIONS   16,209 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

A meta-analysis of strength of weak ties View project

Knowledge Networks View project

All content following this page was uploaded by Martin Kilduff on 03 February 2014.

The user has requested enhancement of the downloaded file.


Journal of Applied Psychology © 2011 American Psychological Association
2011, Vol. 96, No. 6, 1209 –1222 0021-9010/11/$12.00 DOI: 10.1037/a0024890

Centrality and Charisma: Comparing How Leader Networks and


Attributions Affect Team Performance

Prasad Balkundi Martin Kilduff


University at Buffalo Cambridge University

David A. Harrison
University of Texas at Austin

When leaders interact in teams with their subordinates, they build social capital that can have positive
effects on team performance. Does this social capital affect team performance because subordinates come
to see the leader as charismatic? We answered this question by examining 2 models. First, we tested the
charisma-to-centrality model according to which the leader’s charisma facilitates the occupation of a
central position in the informal advice network. From this central position, the leader positively
influences team performance. Second, we examined the centrality-to-charisma model according to which
charisma is attributed to those leaders who are socially active in terms of giving and receiving advice.
Attributed charisma facilitates increased team performance. We tested these 2 models in 2 different
studies. In the first study, based on time-separated, multisource data emanating from members of 56 work
teams, we found support for the centrality-to-charisma model. Formal leaders who were central within
team advice networks were seen as charismatic by subordinates, and this charisma was associated with
high team performance. To clarify how leader network centrality affected the emergence of charismatic
leadership, we designed Study 2 in which, for 79 student teams, we measured leader networking activity
and leader charisma at 2 different points in time and related these variables to team performance
measured at a third point in time. On the basis of this temporally separated data set, we again found
support for the centrality-to-charisma model.

Keywords: social networks, leader, centrality, charisma, team performance

The benefits that accrue to an individual as a result of occupying network patterns rather than the cause of such patterns. From this
a central position in a work-related network include faster promo- perspective, an individual’s personality consists of “the accumu-
tions (Podolny & Baron, 1997) and greater compensation (Burt, lated emotional residue of managing concrete relationships with
2007). But is one of the benefits of centrality a more favorable specific individuals” (Burt, 1992, p. 263). This “social personality”
assessment of personality? In the social network literature, person- approach (Warner & Lunt, 1941, p. 26) has a long heritage, with
ality often tends to be seen in this way—as the outcome of social roots in the work of Simmel (1971) and Goffman (1959) concern-
ing the ways in which an individual’s participation in society
differentiates that individual from others.
This article was published Online First September 5, 2011. When one looks at relevant psychological theory and research,
Prasad Balkundi, School of Management, University at Buffalo; Martin the one area in which this social personality approach is, to some
Kilduff, Judge Business School, Cambridge University, Cambridge, Eng- extent, mirrored is with respect to charismatic leadership. Thus,
land; David A Harrison, McCombs School of Business, University of there are claims that “the personalities of leaders” are “imagined or
Texas at Austin. constructed by followers” (Meindl, 1995, p. 331) and discussion of
This research was supported by the Smeal College of Business at how followers help “the leader to construct a charismatic image”
Pennsylvania State University, Texas Tech University, and Dhruva College
(Gardner & Avolio, 1998, p. 32). In the present article, we look at
of Management, Hyderabad, India. We thank Swapna Balkundi, Brian
Becker, Fred Dansereau, Ajay Mehra, Narayan Rao, and Frankie Weinberg this social construction process in the context of work teams.
for helpful comments on previous drafts. Special thanks to Anand Nim- Research concerning work teams makes the assumption that “the
balkar for the information technology support. Prior versions of this article leader interacts directly with team members in the processes of
were presented at the Intra-Organizational Networks conference at Emory team development and performance management” (Kozlowski &
University, Atlanta, GA (October 2005); at the 2005 National Academy of Bell, 2003, p. 358). To the extent that leaders are somewhat
Management Conference in Honolulu, Hawaii (August 2005); and at the unaware of their own attributes (cf. Bem, 1972), we suggest that
Universities of Cyprus (Nicosia, Cyprus), Manchester (Manchester, Eng-
frequent interaction with subordinates can help develop socially
land), and Singapore.
Correspondence concerning this article should be addressed to Prasad
relevant aspects of personality such as charisma. Furthermore, we
Balkundi, School of Management, Department of Organization and Human suggest that this process of social interaction can provide subor-
Resources, University at Buffalo, 274 Jacobs Management Center, Buffalo, dinates with opportunities to attribute to leaders qualities such as
NY 14260. E-mail: balkundi@buffalo.edu charisma.

1209
1210 BALKUNDI, KILDUFF, AND HARRISON

We bring together two traditions that have often been kept evidence in organizational science about the combined effects of
separate—research on interpersonal interaction in small groups leader charisma and centrality. By jointly considering leader cha-
(e.g., Shaw, 1964) and research on team effectiveness (e.g., Koz- risma and leader network centrality, our analyses provide a new
lowski & Ilgen, 2006). In bringing these two traditions together, understanding of the structural and personal means through which
we address how charismatic leadership relates to the leader’s leaders promote better team performance. We go beyond the
informal interactions with team members, and we address the compelling prior evidence that a charismatic leader or a centrally
leader’s role in promoting team effectiveness. A primary contri- placed leader will have positive effects on team performance. We
bution of the current article is to investigate whether leaders who propose two models (summarized in Figure 1) that bring together
are central in terms of giving and receiving advice to many team charismatic and network explanations of leader influences.
members emerge as charismatic in the eyes of subordinates. A Charismatic leaders are typically depicted as extraordinary in-
second contribution is to investigate the consequences of the social dividuals capable of inspiring subordinates toward both strong
construction of charisma. We investigate how occupation of a commitment to team tasks and high satisfaction with membership
central position and attributions concerning charisma together pro- on the team (Shamir, House, & Arthur, 1993). Through the addi-
vide the leader with a mechanism to affect team performance. We tional effort sparked by such subordinate commitment and satis-
focus on the following questions: Does the charisma of the leader faction, charismatic leaders improve team performance (House,
affect team performance through facilitating the leader’s occupa- Spangler, & Woycke, 1991). Charismatic leaders articulate lofty
tion of a central position in the informal team advice network? Or goals and exhibit determination in accomplishing these goals.
does the network centrality of the leader affect team performance They thereby encourage their subordinates toward stronger moti-
by facilitating the attribution of charisma to the leader by team vation via a greater outlay of time and energy (Conger & Kanungo,
members? 1987). Hence, the higher the charisma of the leader, the more
Work teams are prevalent in modern organizations (Gerard, productive the team (Barling, Weber, & Kelloway, 1996; Bass,
1995), and meta-analyses have shown that charismatic leaders help Avolio, Jung, & Berson, 2003; Howell & Frost, 1989). From this
such teams perform well (DeGroot, Kiker, & Cross, 2000; Judge & perspective, a leader’s charisma directly affects team performance
Piccolo, 2004; Lowe, Kroeck, & Sivasubramaniam, 1996). How-
(Waldman, Ramirez, House, & Puranam, 2001). In the words of a
ever, little is known about how social relationships of leaders relate
recent study, “the leader inspires and activates subordinates to
to charisma. Even the little that is known is potentially contradic-
perform and achieve goals beyond normal expectations” (Keller,
tory. Recent research has shown that a leader’s charisma affects
2006, p. 202).
network structure (cf. Bono & Anderson, 2005), whereas other
The nature of charisma has been much discussed. Originally,
research has shown that social networks facilitate charismatic
charismatic leaders were thought to be in possession of extraordi-
attributions to the leader from the subordinates (Pastor, Meindl, &
nary attributes that attracted followers (Weber, 1947). From this
Mayo, 2002). We present two studies that help explain how leader
perspective, charismatic leaders possess transcendental powers
charisma and leader centrality combine to affect team perfor-
(Trice & Beyer, 1986). A conception of charisma as an inherent
mance. The first study looked at 56 work teams across four
quality of the individual continues in work that emphasizes the
different organizations. The follow-up study of 79 student teams
ways in which charismatic leaders influence followers through
addressed issues of temporal sequence raised in the first study.
their personality. For example, a meta-analysis of 14 samples of
leaders from over 200 organizations showed that extraversion and
How Leaders Affect Team Performance: Two Models
agreeableness positively predicted charismatic leadership (Judge
Research shows that, for team leaders, both charisma and & Bono, 2000).
within-the-team network centrality independently predict team per- A differing conception emphasizes the ways in which followers
formance (Balkundi & Harrison, 2006; Lowe et al., 1996). These attribute charisma to leaders through social construction processes
compelling meta-analytic findings illustrate the effects of leader (e.g., Howell & Shamir, 2005). From this point of view, charisma
charisma and leader centrality in isolation. Yet there is little is not primarily a personality characteristic but an attribution

Figure 1. Competing models of leader centrality and leader charisma effects on team performance. A:
Charisma-to-centrality model. B: Centrality-to-charisma model.
SOCIAL NETWORKS AND LEADER CHARISMA 1211

negotiated between followers and leaders (House et al., 1991). Centrality-to-Charisma Model
Instead of being restricted to truly exceptional individuals of
historic importance, charisma can be understood as an attribute of Following Max Weber, charismatic leaders are often depicted as
leaders “at all levels of organizations” (Avolio & Yammarino, endowed with “supernatural, superhuman, or at least specifically
1990, p. 195). exceptional qualities” (Weber, 1947, pp. 358 –359). To the extent
If charisma is primarily a personality trait, then a model in that such qualities remove charismatic leaders from the realm of
which leaders use the magnetism of their charisma to move into ordinary humanity, it is hard to see how such leaders can function
central positions in social networks makes sense. If, however, in the intensive collaboration characteristic of small teams. The
charisma is mainly attributed to leaders by followers through centrality-to-charisma model proposes an alternative view in
social relationships, then it makes more sense to model charismatic which charisma itself is socially discovered and constructed
leadership as a process in which leaders build social network through interaction with subordinates. This perspective builds on
relationships with followers on the basis of which charismatic network attribution research that has shown how people central in
attributions are developed. We differentiate between these two networks tend to be seen by others as influential (Brass, 1984). If,
conceptions of charisma by formulating two competing theoretical in contrast to the Weberian view, charismatic leadership is rela-
models. tively difficult to detect (cf. Hermalin, 1998), the process of
leadership activity itself—the process by which the formal leader
engages with subordinates in providing advice and soliciting in-
Charisma-to-Centrality Model
put—may contribute to charismatic attributions.
Charisma in this model operates primarily as a conventional Thus, the centrality-to-charisma model posits that leader activity
personality variable. Charismatic individuals possess abilities— in work-related advice interactions with team members provides
such as high levels of self-confidence and a strong sense of the leader with opportunities to establish charisma that is advan-
righteousness (House, 1977)—that attract followers (Weber, tageous for promoting team effectiveness. Leaders who occupy
1947). Charismatic individuals communicate their vision and ex- central positions in informal advice networks have opportunities
pectations to and through these followers. To the extent that through social interaction to build trust, inspire confidence, gain
charismatic individuals attract informal ties from followers, they respect, and promote rapport with the group. Central leaders can
build central positions for themselves within team advice net- transmit to subordinates evidence of energy and charisma. Fur-
works. A central position in the informal network can complement thermore, the process of consulting with and giving advice to
formal power and provide a platform for enhancing team perfor- subordinates provides opportunities for subordinates to discover
mance (cf. Carmeli, Ben-Hador, Waldman, & Rupp, 2009). Cha- and construct leader charisma (Bass, 1985).
risma can enable the leader to be in the right place in terms of We emphasize the extent to which leaders provide advice and
combining a position of formal authority with a position of infor- also the extent to which leaders solicit advice. The extent to which
mal influence (Brass, 1984). people communicate with others in work groups correlates with
The charisma-to-centrality model suggests, therefore, that the the extent to which these central people are seen as influential
leader’s charisma facilitates movement toward and occupation of a (Brass, 1984). Note that we are not assuming that people in a team
central, advantageous position in the informal network. From this are able to assess accurately how central the leader is (cf. Casciaro,
advantageous position, the leader is able to positively influence 1998). Rather, we are suggesting that the more opportunities
team performance. The pattern of relationships between the leader afforded to individual team members to interact with the team
and team members explains variation in team performance leader, the more individuals are able to assess the qualities of the
(Balkundi & Kilduff, 2005). The leader’s influence derives not just leader.
from the authority that comes with the formal position (cf. Dobrev Within work groups, people evaluate charisma based on direct
& Barnett, 2005) but also, potentially, from the interpersonal contact rather than on media presentations or public speeches
regard that derives from occupation of a central position in the (House et al., 1991). Leaders who are active in soliciting advice
informal advice network concerning work-related issues (Brass, provide opportunities to subordinates to voice their issues. Such
1985; Sparrowe & Liden, 2005). A formal leader who is central in subordinate-focused behavior (i.e., individualized consideration—
the informal advice network has several direct ties with subordi- Antonakis & House, 2002) is an important aspect of leadership.
nates and can use these ties to transmit goals, anticipate problems, Openness to employee voice has been shown to predict member
solicit employee advice, and manage resource flows (Balkundi & satisfaction (Detert & Burris, 2007) and subordinates’ commitment
Harrison, 2006). When a subordinate buys into a network relation- to team goals and tasks (Vroom & Yetton, 1973). According to one
ship with a charismatic leader, this enables the leader to motivate meta-analysis, leaders who provide subordinates with individual-
the individual to go beyond self-interest to focus on team outcomes ized consideration tend to be seen by subordinates as charismatic
(cf. Keller, 2006, p. 203). (r ! .85; Lowe et al., 1996). Through soliciting advice, leaders win
Under the charisma-to-centrality model, the leader’s charisma the trust and respect of employees (MacLeod & Brady, 2007).
indirectly affects team performance by facilitating the leader’s In summary, individuals central in advice networks, in terms of
navigation to an advantageous berth within the informal network both giving and receiving advice, are likely to be intensively
structure of the team (cf. Bono & Anderson, 2005). Because involved in their work (Zagenczyk & Murrell, 2009), are likely to
noncharismatic leaders tend to remain on the periphery of their influence others concerning their perceptions of the team (Zagenc-
team networks, they lack influential relationships with subordi- zyk, Scott, Gibney, Murrell, & Thatcher, 2010), and are likely to
nates and are less likely to positively influence member motivation be seen by others as charismatic (Krackhardt, 1990). To the extent
and accomplishment, and therefore, team performance suffers. that centrally located team leaders are seen as charismatic, their
1212 BALKUNDI, KILDUFF, AND HARRISON

pronouncements are likely to be viewed as visionary, their encour- supervisors), 14 teams from Pharmaceutical Labs (each team rated
agement is likely to incite extra effort, and their teams are likely to by one of four supervisors), seven teams from Research Institute
perform well. (rated by one supervisor), and 15 teams from Applied Engineering
Projects (rated by one supervisor).
Study 1 Procedure. We surveyed team members at all four sites to
collect social network ties and demographic characteristics. Fur-
thermore, we asked subordinates questions about their leader’s
Method
charisma, and we later asked the upper level manager of each
Sample. We collected social network, charisma, and team leader to rate the performance of the (leader’s) team. That is, we
performance data at three sites in South India and one site in the obtained data from three sources: team members, leaders, and
United States. The sample size at each site ranged from 66 to 233 higher level supervisors of the leaders. In the case of Applied
people, with a mean of 170. The three Indian research sites Engineering Projects, the course instructor was the higher level
included Regional Hospital, a large general hospital that employed supervisor of the student teams and their leaders and therefore
233 people working in 25 teams concerned with patient care or provided team performance evaluations. At all four sites, we
administration. These teams provided diagnostics and medical distributed the same supervisor (upper level manager) and team
treatment to sick patients, whereas the work of the administrative member questionnaires. Participants at each site were promised
teams involved supporting the patient care teams and maintaining and given an overview of findings.
the hospital. Pharmaceutical Labs employed 212 people working At all four sites, we sent a letter to potential participants (in-
in 21 teams in South India to develop or produce pharmaceutical cluding team members, team leaders, and supervisors of the lead-
drugs. Research and development teams focused on the basic ers) describing the study, encouraging participation, and assuring
science of creating new or improved pharmaceuticals. The manu- complete confidentiality (Dillman, 2000). The letter also empha-
facturing teams translated the findings of the research and devel- sized the organizational and scientific importance of the study,
opment teams into large-scale production of the drugs. The without divulging specific variables or hypotheses. One week
government-run Research Institute employed 168 people working later, we sent each potential participant a package that included a
in eight teams to develop national policy guidelines concerning cover letter, the questionnaire itself, and two consent forms. After
nutrition and health. For example, one team comprised patholo- another week, we followed up nonrespondents with reminder
gists whose job it was to conduct research and develop health postcards. Then, after an additional week, we sent letters remind-
policies so as to reduce the occurrence of cancer. The U.S. research ing individuals of the study and enclosed replacement surveys.
site—Applied Engineering Projects— consisted of a class of 66 In a second wave of data collection, we contacted only the upper
advanced engineering students at a large state university working level supervisors of team leaders and requested that they evaluate
in 15 student teams on applied projects for a large aerospace team performance. The time lag of roughly two months separating
company. Throughout the semester the student teams made pre- the two waves of data collection allowed the teams to complete
sentations to the class and to the instructor. All the teams in India their current tasks. For example, the student engineering teams
and the United States were led by formally designated leaders. filled out the social network survey in the middle of their semester-
For each team, the members worked together through at least long projects, and the instructor rated team performance after the
two performance cycles before team performance was assessed teams submitted their project reports and made presentations to the
(Marks, Mathieu, & Zaccaro, 2001). In the case of the applied aerospace company (the sponsor of the project). The combination
engineering teams, the instructors rated the teams based on their of time-separated data and different respondents (team members
team projects and presentations made to external stakeholders. The vs. team supervisors) was designed to minimize problems with
supervisors at the hospital met the team leaders every Monday to common method variance.
keep abreast of performance developments and address any prob- Networks. Because we were interested in team-level network
lems that surfaced. In the remaining two sites, the supervisors measures, we determined membership in each team using converg-
interacted with the teams on a daily basis. Teams were character- ing evidence from different sources of information to permit
ized by interdependence that was reciprocal rather than pooled or cross-validation: interviews with human resources personnel, ar-
sequential (Thompson, 1967). chival records (including organizational charts), and team member
The total sample across all sites consisted of 472 people in 69 responses. To map the team advice network, at the beginning of
teams. The average response rate across teams was 82% (cf. the questionnaire each respondent was asked to scan an alphabet-
Sparrowe, Liden, Wayne, & Kraimer, 2001). We excluded 11 ical list of all team members and place checks next to the names of
teams from this total because the teams were dyads, consisting of those “people from whom you seek advice about work-related
only one member other than the leader. Furthermore, to avoid matters.”
statistical dependence in our sample, we also excluded one each of Leader centrality. Leaders can energize team members both
two pairs of teams that shared the same leader. In these cases, we by providing subordinates with advice and by seeking advice from
retained the team to which the leader devoted the most time. Our them. To capture advice seeking and giving by the team leader, we
final sample consisted of 413 people in 56 teams, with a mean calculated degree centrality (Borgatti & Everett, 2006): the total
team size of seven, with each team being rated in terms of its number of ties that each leader had in the team. Degree centrality
performance by one supervisor. The individuals who were ex- is the most theoretically appropriate centrality measure of the local
cluded from the analysis were not statistically different from those activity of the leader in the team. As a standard network textbook
retained in terms of age, gender, and education. Overall, there were explained, “Degree centrality stresses the local viewpoint and
20 teams from Regional Hospital (each team rated by one of four measures transaction activity” (Degenne & Forsé, 1999, p. 134).
SOCIAL NETWORKS AND LEADER CHARISMA 1213

The data we collected consisted of directed ties: whether Person i linear model that examined team performance rating (Level 1)
reported going to Person j for advice. Because we wanted to differences across the different supervisors (Level 2). There was a
capture the total activity of each leader in both soliciting and significant difference between the raters (p " .05), and it was due
receiving advice, we symmetrized the data using the union rule—a to the one specific rater from the Research Institute. When we
tie was retained in the matrix if either Person i reported going to excluded this rater, there were no differences in performance
Person j for advice or if Person j reported going to Person i for ratings among the remaining raters (p $ .30).
advice. In either of these cases, interaction occurs between the two Perceptions of leader effectiveness. One of the key assump-
people. This interaction, irrespective of who initiates it, can lead to tions of both models proposed above is that charisma is an ante-
the exchange of information and provide leaders with influence cedent of performance. Cumulative evidence supports this formu-
opportunities (Krackhardt, 1990). We standardized centrality mea- lation (e.g., Judge & Piccolo, 2004). However, there is the
sures by team size to produce normalized degree centrality scores possibility that subordinates attribute charisma to a leader because
that varied between 0% and 100% (Wasserman & Faust, 1994). the leader is a proven high performer (Agle, Nagarajan, Sonnen-
Leader charisma. This was measured with seven items from feld, & Srinivasan, 2006; Keller, 1992; Meindl, 1995). To help
the Multifactor Leadership Questionnaire developed by Bass and control for this possibility, we assessed the subordinates’ percep-
his colleagues (Bass, 1985; Bass & Avolio, 1990; see Waldman et tions of the prior effectiveness of team leaders using two 5-point
al., 2001, for an example of the use of this scale). All team scale items modeled on two items recently (Foti, Knee, & Backert,
members were asked to rate their leader on items relating to 2008) added to the General Leadership Impression Scale (Lord,
communicating high performance expectations, generating re- Foti, & DeVader, 1984). Specifically, we asked subordinates in
spect, and so on. Each item was anchored on a 5-point scale each team to identify (a) how much the leader was instrumental to
ranging from not at all to frequently, if not always. We aggregated the overall effectiveness of the group and (b) how much the leader
leader charisma scores to the team level given that the mean rwg contributed to the performance of the group’s task. Scale endpoints
for team member reports of leader charisma was .85, with 87% of were labeled None and Extreme Amount. Coefficient alpha for this
teams having rwg values above .70 (James, Demaree, & Wolf, two-item measure was .87.
1984). Further tests justified this aggregation (ICC1 ! .30, p "
.01, ICC2 ! .72; Bliese, 2000). This suggests that team members Results and Discussion
had a common view of their leader’s charisma. Coefficient alpha
for the leader charisma scale was .92, suggesting high internal The descriptive statistics in Table 1 show that nine out of the 56
consistency. Other studies have reported similar levels of esti- leaders (16%) were female. On average, a leader was connected
mated reliability for these seven items (Waldman et al., 2001). via advice ties to 75% of the subordinates, as indicated in the mean
Team task performance. Each team was rated by an upper normalized degree centrality (M ! 75.03, SD ! 27.09). Subordi-
level manager (supervisor of team leaders) on a scale of 1 (very nates saw leaders as high performers who contributed to the
poor) to 7 (outstanding) for 11 items developed by Campion, effectiveness of their teams, as indicated by the mean score of 4.19
Papper, and Medsker (1996). Sample items included “The quality for perceptions of prior leader performance on a scale of 1 to 5.
of work done by this team was ____.” In their study of 42 teams, However, there was still substantial variance in those perceptions
Campion et al. found these items to have sound psychometric (SD ! 0.66), and responses covered the full range of the scale.
properties, including high variance across teams and good internal The zero-order correlations in Table 1 show that both leader
consistency. In the present study, the items were reliable (# ! .93). charisma (r ! .45, p " .01) and leader network centrality (r ! .35,
Network size. The network position of a person is partly p " .01) predicted team task performance. The correlations with
dependent on the size of the surrounding network. A member of a performance shown in Table 1 are quite comparable to other
large team can potentially forge more ties than a member of a studies, including meta-analytic estimates of effects for charisma
small team. To ensure standardized comparisons of leader central- (% ! .49; DeGroot et al., 2000) and leader centrality (% ! .29;
ity across teams of different sizes, we used normalized measures of Balkundi & Harrison, 2006). These psychometric and sociometric
degree centrality in our analyses (Borgatti, Everett, & Freeman, independent variables (leader charisma and leader centrality) also
2002). In addition, because team size itself has been predictive of correlated with each other (r ! .47, p " .01). The significant
team performance in prior studies (Guzzo & Dickson, 1996), we correlations further highlight the importance of the questions ad-
included this variable as a control. dressed in the models, which we examined through hierarchical
Leader gender. Research has indicated systematic effects of multiple regression analysis.
the gender of the leader on leadership style (Eagly & Carli, 2003; Charisma-to-centrality model. This model suggests that
Eagly & Johnson, 1990). Therefore, we incorporated a dummy charismatic leaders occupy central positions in the advice net-
variable for leader gender as a control. Males were coded as 1 and works of their teams and, through the occupation of these advan-
females as 2. tageous positions, facilitate how well teams accomplish their tasks.
Site. To control for the effects of the four different research From this perspective, the positive effect of leader charisma on
sites, we included three dummy variables in the regression analy- team performance (& ! .46, p " .01, as shown in Model 3, Table
sis. We also examined differences across substantive variables 2) should have been mediated by leader centrality. Was this
(leader centrality, leader charisma, team task performance, team mediation model supported? The answer is no. Although charis-
size, and leader effectiveness) across sites. Confidence intervals matic leaders tended to occupy central network positions (Model 4,
(CIs) overlapped for all variables with the exception of team task Table 3: & ! .42, p " .01) and teams with more central leaders
performance, where lower ratings were recorded for the Research tended to achieve higher performance (Model 2, Table 2: & ! .29,
Institute site. This difference was confirmed by a hierarchical p " .05), there is no indication in our results that the effect of
1214 BALKUNDI, KILDUFF, AND HARRISON

Table 1
Descriptive Statistics and Correlations Among Control, Independent, and Dependent Variables for Study 1

Variable M SD 1 2 3 4 5 6 7 8 9 10

Control variables
1. Institute 0.12 0.33 —
2. Engineering teams 0.27 0.48 (.23 —
3. Hospital 0.36 0.45 (.28 (.45!!! —
4. Pharmaceuticals 0.23 0.43 (.21 (.33!! (.41!! —
5. Team size 7.30 4.25 (.37!!! (.42!!! .29!! (.18 —
6. Leader gender 1.16 0.37 (.02 (.06 .08 (.12 (.07 —
7. Leader perceived effectiveness 4.19 0.66 .06 (.05 .23! (.25! .12 (.10 —
Independent and dependent variables
8. Leader centrality 75.03 27.09 (.29!! .31!! .05 (.18 (.30!! (.07 .12 —
9. Leader charisma 3.67 0.75 .09 (.10 .06 (.12 .00 (.08 .26!! .47!!! —
10. Team performance 41.29 7.06 (.35!!! .15 .07 .00 (.19 .13 (.04 .35!!! .45!!! —

Note. n ! 56.
!
p " .10. !! p " .05. !!!
p " .01.

leader charisma on team performance was due to leaders occupy- to-charisma model explained 23% of the variance in team perfor-
ing central positions in team advice networks. As Model 4, Table mance. We conducted several other analyses to rule out alternative
2, shows, after the inclusion of the mediator (leader centrality), explanations (see the Appendix for details).
leader charisma remained a significant predictor of team perfor-
mance (& ! .43, p " .01), and centrality had no detectable effect Summary
(& ! .09, ns). The Sobel (1982) test for mediation was nonsignif-
icant, t(47) ! 0.68, ns. The results show that teams led by charismatic leaders tend to be
Centrality-to-charisma model. This model suggests that high performing and that the extent to which leaders are seen as
charisma is attributed to those leaders who are socially active in charismatic depends on the extent to which leaders are central in
terms of giving and receiving advice and that this leader charisma team advice networks. These results support the centrality-to-
facilitates team performance. From this perspective, the positive charisma model over the charisma-to-centrality model. Yet cha-
effect of leader centrality on team performance (& ! .29, p " .05, risma and network centrality were measured at the same time, so
as shown in Model 2, Table 2) should have been mediated by we still did not know whether leaders who are central in team
leader charisma. Was this mediation model supported? The answer advice networks tend subsequently to receive attributions of cha-
is yes. Central leaders tended to be seen as charismatic (Model 2, risma from subordinates. Prior research could support the inter-
Table 3: & ! .48, p " .01), teams with charismatic leaders tended pretation that it is the charisma of the leader that directly affects
to have high performance (Model 3, Table 2: & ! .46, p " .01) both leader centrality (e.g., Bono & Anderson, 2005) and team
and, as Model 4, Table 2, shows, when the hypothesized media- performance (e.g., Keller, 2006). To more carefully assess the
tor—leader charisma—was introduced into the regression (& ! temporal ordering posited in the mediation models, we designed a
.43, p " .01), there was no significant effect of leader centrality on second, longitudinal study to test the relationships among the
team performance (& ! .09, ns). The Sobel (1982) test for medi- variables on a more homogeneous sample of teams with a more
ation was significant, t(47) ! 2.13, p " .05. Overall, the centrality- temporally sensitive research design.

Table 2
Summary of Regression Analyses Predicting Team Performance for Study 1

Model 1: controls Model 2: controls ' leader Model 3: controls Model 4: Model 3 ' leader
Variable only centrality ' charisma centrality

Institute (.31! (.29! (.27! (.27!


Engineering teams .06 (.03 (.01 (.03
Hospital (.01 (.05 (.01 (.02
Team size (.05 .01 (.07 (.05
Leader gender .12 .15 .15 .09
Leader perceived
effectiveness .01 (.03 (.11 (.12
Leader centrality .29!! .09
Leader charisma .46!!! .43!!!
F 1.37! 1.84! 3.49!!! 3.06!!!
Adjusted R2 .04 .10 .24 .23

Note. n ! 56; entries are standardized regression coefficients.


!
p " .10. !! p " .05. !!! p " .01.
SOCIAL NETWORKS AND LEADER CHARISMA 1215

Table 3 “from whom you seek advice about project-related matters.” We


Summary of Regression Analysis Predicting Leader Charisma constructed 79 advice matrices, one for each team, to capture these
and Leader Centrality for Study 1 data on social ties. Cell xij in the advice matrices represented
whether i went to j for advice (cell value ! 1) or not (cell value !
Leader charisma Leader centrality 0). Each matrix was then used to calculate the number of ties a
Variable Model 1 Model 2 Model 3 Model 4
leader had with subordinates. As in Study 1, when either a subor-
dinate sought advice from the leader or vice versa, it was counted
Institute (.08 (.04 (.10 (.06 as tie. To standardize for differences in team size, we used the
Engineering teams .13 (.01 .30! .24 normalized degree centrality measure that varies between 0% and
Hospital .04 (.06 .20 .18
100% (Wasserman & Faust, 1994). We followed the same proce-
Team size .04 .14 (.22 (.23
Leader gender (.07 (.02 (.11 (.08 dure for both waves of network data.
Leader perceived effectiveness .26! .20 .11 .01 Leader charisma. Twice (T1 and T2) during the semester,
Leader centrality .48!!! team members were asked to rate team leader behaviors on the
Leader charisma .42!!! Multifactor Leadership Questionnaire items used in Study 1 (Bass,
F 0.99 2.70!! 2.24! 4.12!!!
Adjusted R2 .01 .18 .12 .28 1985; Bass & Avolio, 1990). The scale was internally consistent
(# ! .93). The mean rwg across all teams was .96 at both T1 and
Note. n ! 56; entries are standardized regression coefficients. T2, with all teams having rwg values above .75 (James et al., 1984),
!
p " .10. !! p " .05. !!! p " .01. thereby providing support for aggregation. Other analyses also
supported the aggregation to the team level (T1 ICC1 ! .14, p "
Study 2 .01, ICC2 ! .39; T2 ICC1 ! .25, p " .01, ICC2 ! .58; Bliese,
2000).
Team task performance. Course instructors rated teams
Method
using the same 11-item scale used in Study 1 (Campion et al.,
Sample. To test the two competing models of the relationship 1996; # ! .94).
between charisma and network centrality in the context of team Rater effect. Because five different instructors rated the
performance, we collected multisource, three-wave survey data teams, we created four different dummy codes for rater.
from 356 persons in 79 four- or five-person teams. Median team Leader gender. Men were coded as 1 and women as 2.
size was four members. Teams worked on an organizational sim-
ulation project as part of a senior-level strategy course at a public Results and Discussion
southwestern U.S. university over a 16-week semester. The orga-
nizational simulation was capstone to the class and required Table 4 shows means, standard deviations, and zero-order cor-
knowledge of different functional inputs to the operation of a firm. relations from Study 2. In support of the centrality-to-charisma
In addition to the simulation, the teams submitted multiple case model, leaders central in team advice networks at T1 tended to be
studies that were independent of the content, but not the general seen as charismatic by members of their team at T2 (r ! .19, p "
principles, of the simulation. The majority of participants were .10) and were judged to lead more productive teams at T3 (r ! .37,
male (72.5%), and the average age of the respondents was 23 p " .01). In contrast, the correlational evidence did not support the
years. Each team was assigned a leader by the class instructor, and charisma-to-centrality model as strongly. Leaders rated as charis-
this leader served as the liaison between the team and the instruc- matic by subordinates at T1 were not found to move into central
tor. network positions at T2 (r ! .13, ns) nor were their teams judged
Procedure. The initial wave of data collection occurred dur- to be high performing at T3 (r ! .06, ns).
ing the first week of the simulation-based activities in the class These correlational trends were mirrored in the regression re-
(T1), approximately one month after the semester began. We sults shown in Table 5. In support of the centrality-to-charisma
measured the advice network ties within the team, as well as leader perspective, Model 2 in Table 5 shows that leaders who were
charisma. The second wave (T2) of data collection occurred 2 central at T1 were later seen as charismatic at T2 (& ! .23, p "
months later, when we measured the same two constructs again. .05). Moreover, the teams of T1-central leaders tended to perform
By then, team members had several weeks of working alongside well at T3 (& ! .33, p " .01). Sobel’s (1982) test confirmed the
one another and were working on the last stage of the simulation. partial mediation of charisma measured at T2 on the relationship
The third wave of data collection occurred at the end of the between leader centrality at T1 and team performance at T3,
semester (T3), when instructors were asked to rate the performance t(68) ! 2.71, p " .01. However, the charisma-to-centrality per-
of each project team. For the first two waves of data collection, of spective was not supported. As Table 5, Model 4, shows, leader
the 350 students in the course, 306 responded. Thus, the overall charisma at T1 did not significantly predict leader centrality at T2
response rate was 87% across the two waves. When these data (& ! .18, ns), nor was the charisma of leaders at T1 found to
were aggregated to the team level, we had missing data for eight contribute to the performance of teams at T3 (& ! .06, ns; Model
teams. These teams, which were subsequently dropped from the 5 in Table 5).
study, were not different from the remaining teams in terms of Table 4 shows that the two measures of leader centrality (at T1
background characteristics such as gender, age, and SAT scores of and T2) were highly correlated (r ! .49, p " .01) as were the two
team members. measures of leader charisma (r ! .63, p " .01). These across-time
Leader centrality. At both T1 and T2, we asked the respon- correlations suggest only modest change and raise the question,
dents to look through a list of team members and identify members Does either variable at T1 predict change in the other variable from
1216 BALKUNDI, KILDUFF, AND HARRISON

Table 4
Descriptive Statistics and Correlations Among Control, Independent, and Dependent Variables for Study 2

Variable M SD 1 2 3 4 5 6 7 8 9 10 11

1. Rater 1 0.19 0.39 —


2. Rater 2 0.20 0.40 (.24!! —
3. Rater 3 0.19 0.39 (.23!! (.24!! —
4. Rater 4 0.20 0.40 (.24!! (.25!! (.24!! —
5. Rater 5 0.21 0.41 (.25!! (.26!! (.25!! (.26!! —
6. Leader gender 1.36 0.48 .18 (.12 .07 .02 (.14 —
7. Leader centrality (T1) 85.82 23.65 .25!! (.18!! .06 (.16 (.04 .06 —
8. Leader charisma (T1) 3.72 0.60 (.05 (.07 .08 .20!! (.15 .15 .18 —
9. Leader centrality (T2) 81.05 26.89 .22!! (.18! .05 (.15 .05 (.06 .49!!! .13 —
10. Leader charisma (T2) 3.79 0.67 (.13 (.19 (.06 .17 .21! (.03 .19! .63!!! .23!! —
11. Team performance (T3) 3.51 0.63 (.05 .14 .14 (.41!!! .17 .08 .37!!! .06 .37!!! .22!! —

Note. n ! 79. T1 ! first wave of data collection; T2 ! second wave of data collection; T3 ! third wave of data collection.
!
p " .10. !! p " .05. !!! p " .01.

T1 to T2? The answer is no. If we rerun the analysis in Table 5, charisma is a go-between from T1 network centrality to subse-
Model 2, controlling for charisma at T1 in predicting charisma at quent (T2) charisma. However, when examining the other path-
T2, then leader centrality is no longer significant (& ! .09, ns). If way, there is no evidence for T1 centrality coming between T1
we rerun the analysis in Table 5, Model 4, controlling for centrality charisma and T2 centrality.
at T1 in predicting centrality at T2, then charisma remains non- The primary focus of this article is on the team performance
significant (& ! .09, ns). implications of leader centrality and leader charisma. Thus, we
Is there then evidence of mediation? We used a bootstrap also used bootstrap mediation to check which paths to team per-
mediation testing procedure (Preacher & Hayes, 2008) appropriate formance were significant. The results confirmed the regression
for the small sample size of Study 2. Furthermore, given the small tests reported above. The charisma-to-centrality perspective was
size, we wanted to increase the statistical power and sensitivity to not supported. The pathway from leader charisma (T1) to leader
detect effects and thereby avoid incorrectly accepting the null. centrality (T2) to team performance (T3) was not significant
Thus, we used p " .10 (two-tailed) to test these paths, including because neither the total effect of charisma (T1) on team perfor-
whether either path—from centrality to charisma or from charisma mance, t(68) ! 1.39, ns, nor the direct effect of charisma (T1) on
to centrality—was significant (cf. Lim & Ployhart, 2004). We team performance, t(68) ! (0.99, ns, was significant.
found evidence to support the idea that team members made up The centrality-to-charisma model was supported, as shown
their minds at T1 concerning the leader and that these attributions by our examination of the pathway from centrality (T1) to
then carried over to T2: Leader charisma at T1 mediated the effect charisma (T2) to performance (T3). The total effect of centrality
of leader centrality at T1 on charisma at T2, t(68) ! 1.73, p " .10, (T1) on performance was positive and significant, t(68) ! 3.40,
but charisma at T1 did not predict centrality at T2 via prior p " .01, as was the direct effect of centrality (T1) on perfor-
centrality, t(68) ! 0.56, ns. These mediation tests suggest that T1 mance, t(68) ! 2.08, p " .05. The indirect effect was signifi-

Table 5
Summary of Regression Analyses in Study 2 Predicting Charisma (T2), Leader Centrality (T2), and Team Performance (T3)

Charisma (T2) Leader centrality (T2) Team performance (T3)

Model 4: Model 3 Model 5: controls ' leader Model 6: Model 5


Model 1: only Model 2: Model 1 ' Model 3: only ' leader charisma centrality (T1) ' leader ' leader charisma
Variable controls leader centrality (T1) controls (T1) charisma (T1) (T2)

Rater 1 (.26! (.30!! .17 .16 (.26!! (.16


Rater 2 (.31!! (.27! (.18 (.19 .03 .14
Rater 3 (.19 (.20 .02 (.01 (.05 .06
Rater 4 (.02 (.02 (.14 (.20 (.43!!! (.37!!!
Leader gender .00 (.01 (.11 (.14 .11 (.14
Leader centrality (T1) .23!! .36!!! .33!!!
Leader charisma (T1) .18 .06 (.16
Leader charisma (T2) .33!!
F 2.03 2.12! 1.54 1.74 4.76!!! 5.17!!!
Adjusted R2 .05 .08 .03 .06 .26 .31

Note. n ! 79; entries are standardized regression coefficients. T1 ! first wave of data collection; T2 ! second wave of data collection; T3 ! third wave
of data collection.
!
p " .10. !! p " .05. !!! p " .01.
SOCIAL NETWORKS AND LEADER CHARISMA 1217

cant as indicated by the CIs created by a bias-corrected boot- enthusiastically communicate about team goals. The very charisma
strap procedure: 90% CI unstandardized coefficient [0.0002, of individual leaders (something that might appear to be uniquely
0.0045] (Preacher & Hayes, 2008). Overall, the evidence points representative of their authentic selves) appears to be derived in
to leader charisma (T2) as mediating the relationship between part from the level of leader network activity.
leader centrality (T1) and team performance.
Contribution to Leadership Research
Summary This article complements recent research concerning how sub-
ordinates gain influence in organizations (Sparrowe & Liden,
The results suggest two conclusions. First, the bulk of the
2005). Our research has looked at the other side of the picture,
evidence is consistent with the idea that leader networking pre-
from the perspective of the leader concerned principally with
cedes and perhaps drives judgments of leader charisma. Second,
improving team performance, whereas prior research emphasized
enhanced charisma is likely to prove useful to leaders in their
the importance of subordinates gaining the sponsorship of super-
efforts to improve team performance. Study 2 provides further
visors. Our research emphasizes that charismatic power derives
evidence that leaders who establish themselves at the center of
from network interactions (cf. Krackhardt, 1990). We add a further
team advice networks tend to be seen as charismatic by team
link in addressing the issue of team performance as an outcome of
members and that this charisma facilitates how well the team
leader network activity and leader charisma.
achieves task goals.
Why did we not find support for the charisma-to-centrality
The evidence from the bootstrap mediation tests suggests that
model (cf. Burkhardt & Brass, 1990)? One possibility is suggested
leader centrality may affect charisma (T2) via prior charisma (T1)
by the finding that charismatic leaders are regarded by followers
and that this is a fully mediated relationship. These results suggest
with awe (Shamir, Arthur, & House, 1994; Shils, 1965). In the
that charismatic attributions may be based on the initial network
presence of leaders high in charisma (and therefore high in power),
ties of the leader and that these initial attributions may remain
followers tend to feel emotionally inhibited and deferential (Kelt-
stable over time.
ner, Gruenfeld, & Anderson, 2003). To the extent that leader
We should recognize the limitations of Study 2. Given the small
charisma inhibits followers from approaching or responding ex-
size of the teams, relatively modest amounts of variance were
pressively to leaders, charisma is unlikely to prove useful in
explained with respect to the charisma and centrality of leaders.
establishing leader centrality.
The conclusion that leader networking predicts leader charisma,
Recent research has suggested that it is the local network that
which, in turn, facilitates team performance, must remain tentative
matters for individual outcomes in organizations rather than the
pending studies that experimentally manipulate the variables and
larger interpersonal sphere represented by the whole organization
establish causality. Although our results diminish the likelihood of
of contacts (Burt, 2007). Our research has focused on this local
dual pathways of effects of leader centrality and charisma on team
network of interaction within the boundary of a work team where
performance, we have not explored other possible models. For
network dynamics are likely to be proximate and intense. Despite
example, it is possible that central (but noncharismatic) or charis-
the starring role of teams in the developing history of social
matic (but noncentral) leaders might run effective teams (cf. Brass
network research (e.g., the experimental studies of small groups—
& Burkhardt, 1993).
Mullen, Johnson, & Salas, 1991; see Balkundi & Harrison, 2006,
for a review), the study of the social networks of leaders in teams
General Discussion has been surprisingly scant in applied psychology, as recognized
by the editors of a recent special issue on social networks and
The cumulative results from the two studies provide overall
organizations (Brass, Galaskiewicz, Greve, & Tsai, 2004, p. 800).
support for the centrality-to-charisma model. Team leaders who
In our focus on how team leaders negotiate the pathways to
are central in advice networks (in terms of a bustle of advice
successful team performance, we add to the nascent organizational
interactions with many members) tend to be seen as charismatic by
literature concerned with the networks of team leaders (cf. Mehra,
subordinates. This charisma inspires and facilitates team perfor-
Dixon, Brass, & Robertson, 2006).
mance. Our research provides suggestive evidence that a
The nature of charisma has long been debated. Most researchers
leadership-relevant aspect of personality— charisma—may derive
argue that charisma has an inherent personality facet and may be
from occupation of a structural position in the network, compatible
trait based (Bacha, 2010; Waldman & Javidan, 2009). Others argue
with a social personality approach to leadership emergence.
that charisma is attribution based and is in the eye of the beholders
From a social network perspective, a central position in a team
(e.g., Pastor et al., 2002). One resolution to this debate is the
network offers the incumbent both motive and opportunity to
possibility that individuals differ with respect to their level of
become a player. In this sense, “a person’s social environment
inherent charismatic potential but that this potential requires social
elicits a specific personality” (Burt, 1992, p. 262). This view is not
situations in which it can be discovered and developed (cf. Howell
as radical as it may sound from the perspective of personality
& Frost, 1989; Pastor, Mayo, & Shamir, 2007). Further research
psychology, which also recognizes that personality traits require
can explore the extent to which charismatic attributions can be
appropriate situations to be exhibited and channeled (Winter, John,
fostered through network interactions.
Stewart, Klohnen, & Duncan, 1998). In the current research, we
have been looking at an individual trait that requires considerable
Future Directions
social construction in the minds of others. For this personality trait
to be recognized by subordinates, interaction opportunities have to If charisma is an attribution cued by the extent to which leaders
be available for the formal leader to demonstrate confidence and are active in advice networks, this raises the question of whether or
1218 BALKUNDI, KILDUFF, AND HARRISON

not such charisma is active for the leader only within the team. For team members (Pastor et al., 2002). By being centrally active in the
example, when the team leader leaves the room after addressing network of advice ties, a leader has opportunities to communicate
the team and enters a different social context, is the charisma of the directly to team members a clear vision for how they can work
leader left behind? That is, if the leader checks charisma at the toward team goals. The route to effective leadership involving
door of the team room and there is no spillover effect of charisma charisma may be less daunting than has been thought (cf. Conger,
on other aspects of the leader’s work life, charisma would be 1999). A leader who lacks the “vision thing” in public speeches
ephemeral, emerging and vanishing as the leader crosses the may still garner attributions of charisma as the leader of a small
team’s threshold. Charisma would be a local phenomenon trig- group, although this localized charisma may not emerge in the
gered by the specific social interactions present or absent in public arena (cf. Galvin, Balkundi, & Waldman, 2010). Visionary
specific groups. rhetoric may be necessary to communicate charisma to large
An alternative possibility is that the attribution of charisma in numbers of distant people, but the informal network of advice and
one social context changes the individual’s self-concept and be- interaction may accomplish the same effect within small teams.
havior such that the individual attracts charismatic attributions in
other social settings. Leaders who have been seen as charismatic in Limitations
their own team may begin to assert their importance and distinc-
tiveness as they interact with people from around the firm. Thus, Our research was limited to the specific context of small-to-
in an interesting reversal of the self-fulfilling prophecy effect medium sized teams in which interactions with leaders were in-
(Merton, 1968), followers could create the conditions under which tense. Although we found that leader centrality predicted charisma,
leaders’ self-efficacy is promoted. In the spirit of Meindl’s (1990) leader centrality within entire organizations might relate nega-
approach to leadership, it would be the subordinates who created tively to attributions of charisma (cf. Brass, 1984). The relation-
and maintained leadership efficacy in terms of the romance of ship between centrality and attributions of charisma may well
leadership. Further research exploring the possible extent of spill- depend on how network boundaries are specified (cf. Kilduff &
over effects of attributed charisma (cf. Avolio, 2007) is needed. Brass, 2010). Furthermore, the teams in our studies comprised
Also, the question arises as to the cross-situational stability of knowledge workers and students engaged in tasks with high re-
individuals’ centralities. Evidence suggests that leadership emer- ciprocal interdependence. Standard leadership theory would antic-
gence in leaderless groups is related to such personality factors as ipate that in dealing with subordinates who possess substantial
social dominance (e.g., Anderson & Kilduff, 2009). Other research domain-specific expertise, leaders should be participative rather
indicates that the extent to which individuals receive friendship than autocratic (Vroom & Jago, 2007). The leaders in the teams we
nominations from others is, in part, genetically determined studied garnered attributions of charisma through participative
(Fowler, Dawes, & Christakis, 2009), but there is no work estab- management—soliciting advice from and providing advice to team
lishing such stability for advice ties. Considering the individual’s members. This management style may not be appropriate for teams
network centrality as a stable trait opens up numerous research in other kinds of situations. Further research should examine
possibilities that have been hitherto neglected. whether these centrality-to-charisma effects can be generalized to
Our results highlight the importance of centrality within teams, other contexts and other kinds of teams such as those involving
but what about network connections to people outside of teams? manufacturing or physical performance (Sparrowe & Liden,
Prior work suggested that frequency of communication with non– 2005).
team members relates negatively to team performance (Ancona & Although we have added to the picture of how individual-
Caldwell, 1992) and that diversity of external ties is unrelated to difference variables and social network positions together contrib-
team effectiveness (Oh, Chung, & Labianca, 2004). However, ute to understanding performance outcomes in organizations (cf.
team leader centrality in the external friendship network of group Mehra, Kilduff, & Brass, 2001), we recognize that there are many
leaders has been shown to affect team performance (Mehra et al., other individual-difference and personality variables that could be
2006). Given this inconclusive pattern of results, we looked again examined. Our choice of charisma was guided by its importance in
at Study 1 for which we did have data concerning external ties of the leadership literature and its potential to relate positively to the
team leaders. We found no significant effects on team performance network positions of leaders (cf. Bono & Anderson, 2005). Future
or leader charisma of team leaders’ external ties measured as (a) research can look more deeply at the question of how other
leaders’ ties to their supervisors and as (b) leaders’ betweenness socially attributed aspects of personality can be developed through
centrality assessed organization wide. This is clearly an area for social network dynamics.
future research. An alternative explanation for the relationship between leader
charisma and team performance might be that team performance
Practical Implications drives charisma rather than the other way around—performance
could be the independent variable. Subordinates might take the
A leader who is central in a team advice network relative to one performance reputation of leaders into account in estimating their
who is isolated is likely to find multiple opportunities to dispense charisma, as suggested by the romance of leadership approach
and gather advice, communicate work-related matters with team (Meindl, Ehrlich, & Dukerich, 1985). Indeed, in Study 1, we found
members, and, thereby, construct a charismatic social personality that subordinates’ ratings of leaders’ prior performance signifi-
that is a valuable resource in motivating and directing team per- cantly predicted subordinates’ estimates of leaders’ charisma. We
formance. Thus, the current research helps peel away the mystique therefore made sure that the dependent variable in Study 1—team
surrounding charismatic leadership to reveal the workaday process task performance—was measured well after subordinates rated
by which some leaders emerge as charismatic in the eyes of their leaders’ charisma (cf. Marks et al., 2001), and we ensured that
SOCIAL NETWORKS AND LEADER CHARISMA 1219

performance ratings were provided by upper level managers rather Avolio, B. J., & Yammarino, F. J. (1990). Operationalizing charismatic
than by subordinates. We also controlled for prior perceived per- leadership using a levels-of-analysis framework. Leadership Quarterly,
formance in our regressions predicting subsequent performance. In 1, 193–208.
Study 2, we did not face the problem of attribution based on prior Bacha, E. (2010). The relationships among organizational performance,
performance because the teams were newly created in the context environmental uncertainty, and employees’ perceptions of CEO cha-
risma. Journal of Management Development, 29, 28 –37.
of research.
Balkundi, P., & Harrison, D. (2006). Ties, leaders, and time in teams:
In this article, we have presented two studies using identically Strong inference about network structure’s effects on team viability and
measured variables, thus facilitating a comparison of internal and performance. Academy of Management Journal, 49, 49 – 68.
external validity. Study 1 might be seen as offering higher external Balkundi, P., & Kilduff, M. (2005). The ties that lead: A social network
validity in that it combined data from two countries and across approach to leadership. Leadership Quarterly, 6, 941–961.
different firms. Yet the simultaneity of network and charisma Barling, J., Weber, T., & Kelloway, E. K. (1996). Effects of transforma-
measures undermined claims concerning the temporal ordering of tional leadership training on attitudinal and financial outcomes. Journal
the constructs. In contrast, Study 2 might be seen as possessing of Applied Psychology, 81, 827– 832.
higher internal validity given that measurements across the time Bass, B. M. (1985). Leadership and performance beyond expectations.
periods facilitated a test of the temporal sequence by which central New York, NY: Free Press.
leaders emerged as charismatic. However, given that Study 2 was Bass, B. M., & Avolio, B. J. (1990). The Multifactor Leadership Ques-
tionnaire. Palo Alto, CA: Consulting Psychologists Press.
based on a student sample, critics might question its external
Bass, B. M., Avolio, B. J., Jung, D. I., & Berson, Y. (2003). Predicting unit
validity. In general, then, the two studies offer complementary
performance by assessing transformational and transactional leadership.
strengths and weaknesses. Journal of Applied Psychology, 88, 207–218.
Although we used time-ordered research designs with multiple Bem, D. J. (1972). Self-perception theory. In L. Berkowitz (Ed.), Advances
sources of data, uncommon in contemporary studies in organiza- in experimental social psychology (pp. 1– 62). New York, NY: Aca-
tions (Scandura & Williams, 2000), the internal validity of future demic Press.
research can be further strengthened. Prior work has shown that Bliese, P. D. (2000). Within-group agreement, non-independence, and
individuals placed in central network positions tend to be seen as reliability: Implications for data aggregation and analysis. In K. J. Klein
leaders (Freeman, 1979; Mullen et al., 1991). An experimental & S. W. Kozlowski (Eds.), Multi-level theory, research, and methods in
design that manipulated network structure in this way and then organizations: Foundations, extensions, and new directions (pp. 349 –
measured charisma would increase confidence in the veracity of 381). San Francisco, CA: Jossey-Bass.
the centrality-to-charisma model. The analyses in the present ar- Bono, J. E., & Anderson, M. H. (2005). The advice and influence networks
of transformational leaders. Journal of Applied Psychology, 90, 1306 –
ticle can be considered conservative tests given that the estimated
1314.
effect of charisma on outcomes is almost cut in half when re-
Borgatti, S. P., & Everett, M. G. (2006). A graph-theoretic perspective on
searchers employ either longitudinal research designs or different centrality. Social Networks, 28, 466 – 484.
data sources (Judge & Piccolo 2004). The research designs in both Borgatti, S. P., Everett, M. G., & Freeman, L. C. (2002). Ucinet for
Study 1 and Study 2 incorporate both of these restrictive research Windows: Software for social network analysis. Lexington, KY: Ana-
design facets. lytic Technologies.
In our examination of leadership in teams, we have provided Brass, D. J. (1984). Being in the right place: A structural analysis of
suggestive evidence that leaders who occupy central roles are individual influence in an organization. Administrative Science Quar-
likely to be seen as charismatic by team members. If our research terly, 29, 518 –539.
has an overall message, it is that team leaders active in providing Brass, D. J. (1985). Men’s and women’s networks: A study of interaction
and eliciting advice may achieve for themselves a type of charis- patterns and influence in organizations. Academy of Management Jour-
matic social capital that promotes team performance. nal, 28, 327–343.
Brass, D. J., & Burkhardt, M. E. (1993). Potential power and power use:
An investigation of structure and behavior. Academy of Management
References Journal, 36, 441– 470.
Agle, B. R., Nagarajan, N. J., Sonnenfeld, J. A., & Srinivasan, D. (2006). Brass, D. J., Galaskiewicz, J., Greve, H. R., & Tsai, W. (2004). Taking
Does CEO charisma matter? An empirical analysis of the relationships stock of networks and organizations: A multilevel perspective. Academy
among organizational performance, environmental, and top management of Management Journal, 47, 795– 817.
team perceptions of CEO charisma. Academy of Management Journal, Burkhardt, M. E., & Brass, D. J. (1990). Changing patterns or patterns of
49, 161–174. change: The effects of a change in technology on social structure and
Ancona, D. G., & Caldwell, D. F. (1992). Bridging the boundary: External power. Administrative Science Quarterly, 35, 104 –127.
activity and performance in organizational teams. Administrative Sci- Burt, R. (1992). Structural holes: The social structure of competition.
ence Quarterly, 37, 634 – 665. Cambridge, MA: Harvard University Press.
Anderson, C., & Kilduff, G. J. (2009). Why do dominant personalities Burt, R. (2007). Secondhand brokerage: Evidence on the importance of
attain influence in face-to-face groups? The competence-signaling ef- local structure for managers, bankers, and analysts. Academy of Man-
fects of trait dominance. Journal of Personality and Social Psychology, agement Journal, 50, 119 –148.
96, 491–503. Campion, M. A., Medsker, G. J., & Higgs, A. C. (1993). Relations between
Antonakis, J., & House, R. J. (2002). An analysis of the full-range lead- work group characteristics and effectiveness: Implications for designing
ership theory: The way forward. In B. J. Avolio & F. J. Yammarino effective work groups. Personnel Psychology, 46, 823– 850.
(Eds.), Transformational and charismatic leadership: The road ahead Campion, M. A., Papper, E. M., & Medsker, G. J. (1996). Relations
(pp. 3–33). Amsterdam, the Netherlands: JAI Press. between work team characteristics and effectiveness: A replication and
Avolio, B. J. (2007). Promoting more integrative strategies for leadership extension. Personnel Psychology, 49, 429 – 452.
theory-building. American Psychologist, 62, 25–33. Carmeli, A., Ben-Hador, B., Waldman, D. A., & Rupp, D. E. (2009). How
1220 BALKUNDI, KILDUFF, AND HARRISON

leaders cultivate social capital and nurture employee vigor: Implications leadership. Organizational Behavior and Human Decision Processes,
for job performance. Journal of Applied Psychology, 94, 1553–1561. 43, 243–269.
Casciaro, T. (1998). Seeing things clearly: Social structure, personality, Howell, J. M., & Shamir, B. (2005). The role of followers in the charis-
and accuracy in social network perception. Social Networks, 20, 331– matic leadership process: Relationships and their consequences. Acad-
351. emy of Management Review, 30, 96 –112.
Conger, J. A. (1999). Charismatic and transformational leadership in James, R. L., Demaree, R. G., & Wolf, G. (1984). Estimating within-group
organizations: An insider’s perspective on these developing streams of interrater reliability with and without response bias. Journal of Applied
research. Leadership Quarterly, 10, 145–179. Psychology, 69, 85–98.
Conger, J. A., & Kanungo, R. N. (1987). Toward a behavioral theory of Judge, T. A., & Bono, J. E. (2000). Five-factor model of personality and
charismatic leadership in organizational settings. Academy of Manage- transformational leadership. Journal of Applied Psychology, 85, 751–
ment Review, 12, 637– 647. 765.
Cortina, J. M. (2003). Organizational research methods: Apples and or- Judge, T. A., & Piccolo, R. (2004). Transformational and transactional
anges (and pears, oh my!)—The search for moderators in meta-analysis. leadership: A meta-analytic test of their relative validity. Journal of
Organizational Research Methods, 6, 415– 439. Applied Psychology, 89, 755–768.
Degenne, A., & Forsé, M. (1999). Introducing social networks. London, Keller, R. T. (1992). Transformational leadership and the performance of
England: Sage. research and development project groups. Journal of Management, 18,
DeGroot, T., Kiker, D. S., & Cross, T. C. (2000). A meta-analysis to review 489 –501.
organizational outcomes related to charismatic leadership. Canadian Keller, R. T. (2006). Transformational leadership, initiating structure, and
Journal of Administrative Sciences, 17, 356 –371. substitutes for leadership: A longitudinal study of research and devel-
Detert, J. R., & Burris, E. R. (2007). Leadership behavior and employee opment project team performance. Journal of Applied Psychology, 91,
voice: Is the door really open? Academy of Management Journal, 50, 202–210.
869 – 884. Keltner, D., Gruenfeld, D. H., & Anderson, C. (2003). Power, approach,
Dillman, D. A. (2000). Mail and Internet surveys: The tailored design and inhibition. Psychological Review, 110, 265–284.
method (2nd ed.). New York, NY: Wiley. Kilduff, M., & Brass, D. J. (2010). Organizational social network research:
Dobrev, S. D., & Barnett, W. P. (2005). Organizational roles and transi- Core ideas and key debates. Academy of Management Annals, 4, 317–
tions to entrepreneurship. Academy of Management Journal, 48, 433– 357.
449. Kozlowski, S. W. J., & Bell, B. S. (2003). Work groups and teams in
Eagly, A. H., & Carli, L. L. (2003). The female leadership advantage: An organizations. In W. C. Borman, D. R. Ilgen, & R. Klimoski (Eds.),
evaluation of the evidence. Leadership Quarterly, 14, 807– 834. Handbook of psychology: Industrial and organizational psychology (pp.
Eagly, A. H., & Johnson, B. T. (1990). Gender and leadership style: A 333–375). New York, NY: Wiley.
meta-analysis. Psychological Bulletin, 108, 233–256. Kozlowski, S. W. J., & Ilgen, D. R. (2006). Enhancing the effectiveness of
Foti, R. J., Knee, R. E., Jr., & Backert, R. S. G. (2008). Multi-level work groups and teams. Psychological Science in the Public Interest, 7,
implications of framing leadership perceptions as a dynamic process. 77–124.
Leadership Quarterly, 19, 178 –194. Krackhardt, D. (1990). Assessing the political landscape: Structure, cog-
Fowler, J., Dawes, C., & Christakis, N. (2009). Model of genetic variation nition, and power in organizations. Administrative Science Quarterly,
in human social networks. PNAS: Proceedings of the National Academy 35, 342–369.
of Sciences of the United States of America, 106, 1720 –1724. Lim, B. C., & Ployhart, R. E. (2004). Transformational leadership: Rela-
Freeman, L. C. (1979). Centrality in social networks: I. Conceptual clari- tions to the five-factor model and team performance in typical and
fication. Social Networks, 1, 215–239. maximum contexts. Journal of Applied Psychology, 89, 610 – 621.
Galvin, B. M., Balkundi, P., & Waldman, D. A. (2010). Spreading the Lord, R. G., Foti, R. J., & DeVader, C. L. (1984). A test of leadership
word: The role of surrogates in charismatic leadership processes. Acad- categorization theory: Internal structure, information processing, and
emy of Management Review, 35, 477– 494. leadership perceptions. Organizational Behavior and Human Perfor-
Gardner, W. L., & Avolio, B. J. (1998). The charismatic relationship: A mance, 34, 343–378.
dramaturgical perspective. Academy of Management Review, 23, 32–58. Lowe, K. B., Kroeck, K. G., & Sivasubramaniam, N. (1996). Effectiveness
Gerard, R. J. (1995). Teaming up: Making the transition to self-directed, correlates of transformational and transactional leadership: A meta-
team-based organizations. Academy of Management Executive, 9, 91– analytic review of the MLQ literature. Leadership Quarterly, 7, 385–
93. 425.
Goffman, E. (1959). The presentation of self in everyday life. Garden City, MacLeod, D., & Brady, C. (2007). The extra mile: How to engage your
NY: Doubleday. people to win. Harlow, England: Financial Times.
Guzzo, R. A., & Dickson, M. W. (1996). Teams in organizations: Recent Marks, M. A., Mathieu, J. E., & Zaccaro, S. J. (2001). A temporally based
research on performance and effectiveness. Annual Review of Psychol- framework and taxonomy of team processes. Academy of Management
ogy, 47, 307–338. Review, 26, 356 –376.
Harrison, D. A., Price, K., & Bell, M. (1998). Beyond relational demog- Mehra, A., Dixon, A. L., Brass, D. J., & Robertson, B. (2006). The social
raphy: Time and the effects of surface- and deep-level diversity on work network ties of group leaders: Implications for group performance and
group cohesion. Academy of Management Journal, 41, 96 –107. leader reputation. Organization Science, 17, 64 –79.
Hermalin, B. (1998). Toward an economic theory of leadership: Leading Mehra, A., Kilduff, M., & Brass, D. J. (2001). The social networks of high
by example. American Economic Review, 88, 1188 –1206. and low self-monitors: Implications for workplace performance. Admin-
House, R. J. (1977). A 1976 theory of charismatic leadership. In J. G. Hunt istrative Science Quarterly, 46, 121–146.
& L. L. Larson (Eds.), Leadership: The cutting edge (pp. 189 –273). Meindl, J. R. (1990). On leadership: An alternative to the conventional
Carbondale, IL: Southern Illinois University Press. wisdom. In B. M. Staw & L. L. Cummings (Eds.), Research in organi-
House, R. J., Spangler, D. W., & Woycke, J. (1991). Personality and zational behavior (Vol. 12, pp. 159 –203). Greenwich, CT: JAI Press.
charisma in the U.S. presidency: A psychological theory of leader Meindl, J. R. (1995). The romance of leadership as a follower-centric
effectiveness. Administrative Science Quarterly, 36, 364 –396. theory: A social constructionist approach. Leadership Quarterly, 6,
Howell, J. M., & Frost, P. J. (1989). A laboratory study of charismatic 329 –341.
SOCIAL NETWORKS AND LEADER CHARISMA 1221

Meindl, J. R., Ehrlich, S. B., & Dukerich, J. M. (1985). The romance of equations models. In S. Leinhart (Ed.), Sociological methodology (pp.
leadership. Administrative Science Quarterly, 30, 78 –102. 290 –312). San Francisco, CA: Jossey-Bass.
Merton, R. K. (1968). Social theory and social structure. New York, NY: Sparrowe, R. T., & Liden, R. C. (2005). Two routes to influence: Integrat-
Free Press. ing leader-member exchange and network perspectives. Administrative
Mullen, B., Johnson, C., & Salas, E. (1991). Effects of communication Science Quarterly, 50, 505–535.
network structure: Components of positional centrality. Social Networks, Sparrowe, R. T., Liden, R. C., Wayne, S. J., & Kraimer, M. L. (2001).
13, 169 –185. Social networks and the performance of individuals and groups. Acad-
Oh, H., Chung, M. H., & Labianca, G. (2004). Group social capital and emy of Management Journal, 44, 316 –325.
group effectiveness: The role of informal socializing ties. Academy of Thompson, J. D. (1967). Organizations in action. New York, NY:
Management Journal, 47, 860 – 875. McGraw-Hill.
Pastor, J. C., Mayo, M., & Shamir, B. (2007). Adding fuel to fire: The Trice, H. M., & Beyer, J. M. (1986). Charisma and its routinization in two
impact of followers’ arousal on ratings of charisma. Journal of Applied social movement organizations. In B. M. Staw & L. L. Cummings
Psychology, 92, 1584 –1596. (Eds.), Research in organizational behavior (Vol. 8, pp. 113–164).
Pastor, J. C., Meindl, J. R., & Mayo, M. C. (2002). A network effects Greenwich, CT: JAI Press.
model of charisma attributions. Academy of Management Journal, 45, Vroom, V. H., & Jago, A. G. (2007). The role of situation in leadership.
410 – 420. American Psychologist, 62, 17–24.
Podolny, J. M., & Baron., J. N. (1997). Resources and relationships: Social Vroom, V. H., & Yetton, P. (1973). Leadership and decision making.
Pittsburgh, PA: University of Pittsburgh Press.
networks and mobility in the workplace. American Sociological Review,
Waldman, D. A., & Javidan, M. (2009). Alternative forms of charismatic
62, 673– 693.
leadership in the integration of mergers and acquisitions. Leadership
Preacher, K. J., & Hayes, A. F. (2008). Asymptotic and resampling
Quarterly, 20, 130 –142.
strategies for assessing and comparing indirect effects in multiple me-
Waldman, D. A., Ramirez, G. G., House, R. J., & Puranam, P. (2001). Does
diator models. Behavior Research Methods, 40, 879 – 891.
leadership matter? CEO leadership attributes and profitability under
Reagans, R. E., Zuckerman, E. W., & McEvily, B. (2004). How to make
conditions of perceived environmental uncertainty. Academy of Man-
the team: Social networks vs. demography as criteria for designing
agement Journal, 44, 134 –143.
effective projects in a contract R&D firm. Administrative Science Quar-
Warner, W. L., & Lunt, P. S. (1941). The social life of a modern commu-
terly, 49, 101–133.
nity. New Haven, CT: Yale University Press.
Scandura, T. A., & Williams, E. A. (2000). Research methodology in Wasserman, S., & Faust, K. (1994). Social network analysis: Methods and
organizational studies: Current practices and implications for future applications. New York, NY: Cambridge University Press.
research. Academy of Management Journal, 43, 1248 –1264. Weber, M. (1947). The theory of social and economic organization (A. M.
Shamir, B., Arthur, M. B., & House, R. J. (1994). The rhetoric of charis- Henderson & T. Parsons, Trans.). Glencoe, IL: Free Press.
matic leadership: A theoretical extension, a case study, and implications Winter, D. G., John, O. P., Stewart, A. J., Klohnen, E. C., & Duncan, L. E.
for research. Leadership Quarterly, 4, 25– 42. (1998). Traits and motives: Toward an integration of two traditions in
Shamir, B., House, R. J., & Arthur, M. B. (1993). The motivational effects personality research. Psychological Review, 105, 230 –250.
of charismatic leadership: A self-concept based theory. Organization Zagenczyk, T., & Murrell, A. J. (2009). It is better to receive than to give:
Science, 4, 577–594. Advice network effects on job and work-unit attachment. Journal of
Shaw, M. E. (1964). Communication networks. In L. Berkowitz (Ed.), Business and Psychology, 24, 139 –152.
Advances in experimental social psychology (pp. 111–147). New York, Zagenczyk, T., Scott, K. D., Gibney, R., Murrell, A. J., & Thatcher, J. B.
NY: Academic Press. (2010). Social influence and perceived organizational support: A social
Shils, E. (1965). Charisma, order, and status. American Sociological Re- networks analysis. Organizational Behavior and Human Decision Pro-
view, 30, 199 –213. cesses, 111, 127–138.
Simmel, G. (1971). On individuality and social norms. Chicago, IL: Zohar, D., & Tenne-Gazit, O. (2008). Transformational leadership and
University of Chicago Press. group interaction as climate antecedents: A social network analysis.
Sobel, M. E. (1982). Asymptotic intervals for indirect effects in structural Journal of Applied Psychology, 93, 744 –757.

(Appendix follows)
1222 BALKUNDI, KILDUFF, AND HARRISON

Appendix

Additional Tests

Study 1 Gazit, 2008), and team network centralization (i.e., the extent to
which each team network was concentrated around one or a few
Other Approaches to Centrality central actors). We also examined the leader’s own (betweenness)
The measure of leader centrality that we used in the tests centrality in the entire organization (cf. Reagans, Zuckerman, &
mentioned in the main text counted team members to whom advice McEvily, 2004) given that we had these data. As with the other
was given plus team members from whom advice was received. controls, these variables did not have statistically significant or
However, individuals who gave and received advice from the unique relationships with team performance, and their inclusion
leader were counted only once. Both receiving advice and giving did not change the direction or the substantive effects of our main
advice provide opportunities for leader influence and subordinate antecedents of team performance. To simplify presentation, there-
sensemaking. In the current sample, leaders gave advice to a fore, and retain the most power for model comparisons (i.e., allow
significantly higher percentage of subordinates than they received for unmediated effects of both charisma and leader centrality), we
advice from (mean indegree ! 63.83%, mean outdegree ! excluded these nonsignificant network variables from reported
37.55%), t(55) ! 4.19, p " .01. The indegree centrality of leaders analyses.
did not in itself predict team performance when we repeated the
regression analysis reported in Table 2 (& ! .13, ns), but indegree Other Control Variables
did predict leader charisma when we repeated the regression
Previous research has suggested that team outcomes can be
analysis reported in Table 3 (& ! .37, p " .01). The effects of
affected by member diversity in age, gender, and task interdepen-
advice seeking (leader outdegree) on leader charisma (& ! .26,
dence (Campion, Medsker, & Higgs, 1993; Harrison, Price, &
p " .01) and team performance (& ! .23, p " .15) were similar to
Bell, 1998). In our sample, none of these control variables pre-
those reported in the tables, but the effects were weaker. If, as our
dicted team performance or leader charisma.
centrality-to-charisma model suggests, charismatic attributions are
forged in processes of work-related advice exchange, then the
incorporation of both advice giving and advice receiving measures Moderation?
of leader centrality provide more robust estimates of opportunities
Both our models propose a series of direct effects. However,
for interaction than either indegree or outdegree alone.
another possibility is that leader centrality and charisma moderated
There are other network measures that capture different aspects
one another in their prediction of team performance. That is,
of centrality that are often considered relevant in organizational
charisma might facilitate performance only when the leader has
research (e.g., Brass, 1984) but that appear to have only limited
developed dyadic advice ties to a large number of subordinates (or
relevance to the measurement of centrality in relatively small work
vice versa). We did not find evidence to support this contingent
teams. Betweenness centrality is the extent to which a person
combination (cf. failure of a similar interactive model in Mehra et
serves as a broker by occupying an intermediary position on the
al., 2001). More specifically, the interaction term was nonsignifi-
shortest paths connecting disconnected others (Freeman, 1979).
cant for Degree Centrality ) Charisma (& ! .21, ns). We repeated
We reanalyzed the data, replacing degree with betweenness.
this analysis with a curvilinear model (Cortina, 2003), and the
Leader betweenness centrality predicted neither charisma nor team
results remained nonsignificant (& ! .29, ns).
performance in either model. We repeated this exercise using
eigenvector centrality, a measure that takes into account the cen-
trality of members to whom the leader is connected. Leader eigen- Study 2
vector centrality did predict both leader charisma and team per- Using a subset of the reported results, we replicated and con-
formance. This raised the question as to whether it offered firmed the findings of Study 1 using T2 leader centrality and leader
explanatory power beyond that provided by degree. On repeating charisma measures. Leader centrality (T2) did predict charisma
the analysis in Table 2 with both the degree and eigenvector (T2; &! .26, p " .05) and subsequent team performance (&! .37,
measures of centrality included in the regression, only the degree p " .01). Charisma predicted team performance even after con-
measure was found to be a significant predictor of leader charisma trolling for leader centrality (&! .23, p " .05). A Sobel (1982) test
(& ! .43, p " .05). found that the indirect effect of leader network centrality on
performance through leader charisma was significant, t(68) !
Other Network Structures 2.93, p " .05.
We also examined the effects of other possible control variables
derived from network structures, including team network density Received February 15, 2010
(i.e., the number of ties within the team network divided by the Revision received April 18, 2011
maximum number of ties that were possible; Zohar & Tenne- Accepted April 20, 2011 !

View publication stats

You might also like