Shared leadership and group interaction styles in problem-solving virtual teams

Abstract

Despite their prevailing growth, little systematic evidence exists regarding the effectiveness of computer-mediated "virtual" teams (VTs), especially in relation to their traditional counterpart, face-to-face teams (FtFTs). A partial least squares (PLS) analysis revealed that FtFTs were more likely to demonstrate higher levels of shared leadership and a constructive interaction style than were VTs. In turn, shared leadership and a constructive interaction style were shown to positively predict cohesion, whereas a defensive interaction style was shown to negatively predict cohesion. Shared leadership was also positively associated with a constructive interaction style and negatively associated with a defensive interaction style. Finally, task performance was shown to be a function of group cohesion.

Full text

Shared Leadership And Group Interaction Styles
In Problem-Solving Virtual Teams
Pierre Balthazard
1
, David Waldman
1
, Jane Howell
2
, Leanne Atwater
1
pb@asu.edu, waldman@asu.edu, jhowell@ivey.uwo.ca, leanne.atwater@asu.edu
1
Arizona State University West, School of Management, Phoenix, Arizona
2
University of Western Ontario, Richard Ivey School of Business, London, Ontario
Abstract
Despite their prevailing growth, little systematic
evidence exists regarding the effectiveness of computer-
mediated “virtual” teams (VTs), especially in relation to
their traditional counterpart, face-to-face teams (FtFTs).
A Partial Least Squares (PLS) analysis revealed that
FtFTs were more likely to demonstrate higher levels of
shared leadership and a constructive interaction style
than were VTs. In turn, shared leadership and a
constructive interaction style were shown to positively
predict cohesion, whereas a defensive interaction style
was shown to negatively predict cohesion. Shared
leadership was also positively associated with a
constructive interaction style and negatively associated
with a defensive interaction style. Finally, task
performance was shown to be a function of group
cohesion.
1. Introduction
The virtual organization, and its smaller version, the
virtual team (VT) represent new and growing organiza-
tion forms. VT members are geographically and often
temporally distributed, and oftentimes, the members
have different areas of expertise and may work in differ-
ent functional areas [1,2]. VTs can potentially give
organizations increased flexibility and responsiveness,
permitting geographically dispersed experts to rapidly
form a cohesive unit that can work on an urgent project.
When finished, the team can be disbanded and members
re-deployed to other projects; members may also serve
on multiple virtual teams simultaneously. Although the
VT has become an increasingly common work unit of
many organizations, it remains an evolving and rela-
tively unstudied organizational form.
New organizational forms can present many mana-
gerial challenges such as ineffective team development,
ambiguous roles for group members, lack of cohesion or
teamwork, and performance problems [3,4]. The overall
purpose of the present research is to define an assess-
ment protocol and test a model of the effectiveness of
problem-solving VTs compared to their more tradi-
tional, face-to-face equivalent (FtFTs). As shown in
Figure 1, the model follows a traditional input-
process-output conceptualization of group effective-
ness (e.g., [5]), highlighted by an initial focus on me-
dia type.
Figure 1. Model of the hypothesized relationships
between media, shared leadership, group interaction
styles, cohesion, and task performance
2. Shared Leadership and Group Inter-
action Styles
Much of the literature has questioned the
effectiveness of VTs, as compared to their FtF
counterparts (e.g., [6,7,8]). For example, Duarte and
Snyder [1] argued that VTs may not be appropriate
when issues are either highly emotional or ambiguous,
or when the team is newly formed or short lived. We
propose that to better understand the relative
effectiveness of VTs versus FtFTs, it is necessary to
understand potential mediating variables, including
leadership and group interaction styles.
One important characteristic underlying
theoretical efforts to identify the key functional roles
of team leaders is the assumption that the leader
interacts directly with team members in the processes
of team development and performance management
+
+
++
+
+
+
-
-
Media
Shared
Leadership
Defensive
Intera ctio ns
Constructive
Interactions
Cohesion
Task
Perform an ce
Team
Size
Proceedings of the 37th Hawaii International Conference on System Sciences - 2004
0-7695-2056-1/04 $17.00 (C) 2004 IEEE 1

(e.g., [9,10,11,12,13,14]). This underlying assumption
also characterizes the literature on virtual teams, which
often must provide these functions on their own in the
absence of a formal leader (e.g., [9]). However, very
few studies have examined how leader roles are
duplicated, substituted, or eliminated given that the team
may be widely dispersed in time and geography [9].
Studies have not measured the amount or type of
leadership present in the VT (e.g., [15,16]). In fact, a
review of studies of leadership in VTs found that of 12
studies addressing leadership, none measured leader
style [17]. Further, as Shamir [18] suggests, leadership
in VTs may be shared by team members as they interact
with one another. Thus, leadership may not be the
domain of just one assigned, elected, or emergent
individual. Indeed, Bell & Kozlowski [9] purport that
the challenge for VTs is to determine how leadership
functions such as coaching, mentoring, performance
management, and team development can be
accomplished by distributing the functions to the team
itself.
In the current study, we felt that it was important to
measure the degree to which leadership, specifically
defined in transformational terms, is displayed within a
team by its members. In a problem-solving context,
transformational leadership (TFL) can be characterized
by team members showing enthusiasm and confidence,
promoting understanding and appreciation for differing
views, and intellectually stimulating group members to
reexamine critical assumptions and look at problems in
new ways [19]. Research has demonstrated that within
FtFTs, transformational leaders are likely to increase
group performance because they are helpful at
overcoming social loafing among group members
[20,21], and because they infuse shared values and a
sense of unified purpose or common identity [22].
Despite the apparent importance of TFL in both FtF
and VT settings, little evidence exists suggesting which
of these media is most likely to foster its emergence.
However, work by Shamir and Ben-Ari [59] suggests
that VT technologies may produce more distant or
impersonal means of communication between leaders
and potential followers. That is, because VTs lack the
richness of social cues present in FtFTs (e.g., facial
expression, tone of voice), and because they can inhibit
communication due to the requirement to type responses,
we believe that TFL displayed in VTs will be lower
when compared to FtFTs. Individuals in a VT
environment also may not be as able to utilize the
impression management strategies so essential to the
formation of TFL [23]. The result could be a reduction
in the identification, trust-building, and emotional
processes that characterize TFL within VTs as compared
to FtF teams. Additionally, Bell & Kozlowski [9]
suggest that the spatial distance between team members
and the use of communications technology may impede
VT leadership in terms of mentoring, coaching, and
developmental functions. In short, we expect that:
H1: Face-to-face teams will be more likely to
demonstrate higher levels of shared TFL than virtual
teams.
As problem-solving groups perform their tasks,
member roles become highly interdependent, and the
need for well-orchestrated interactions, including
reciprocal communication and feedback is essential
[9,24]. One factor that has been shown to have a
substantial effect on a team’s ability to perform is its
interaction style [25,45]. Interaction style is best
understood in terms of the communication patterns in
which a group engages as it deals with the inherent
conflicts of task orientation and maintenance of
member relationships. Cooke and Szumal [45]
showed that group interaction styles are either
composed of constructive or defensive behaviors. The
constructive interaction style is characterized by a
balanced concern for personal and group outcomes,
cooperation, creativity, free exchange of information,
and respect for others’ perspectives. Conversely,
defensive styles are likely to include both passive and
aggressive behaviors. Passive behaviors place an
emphasis on limited information sharing, lack of
questioning, and lack of impartiality. Aggressive
behaviors place an emphasis on personal agendas and
ambitions being placed above concern for the group
outcome. Previous research has provided empirical
support for the distinction between constructive and
defensive styles (e.g., [25,26,27]).
An unanswered question is how does a
predominant interaction style materialize in a group?
For example, why does one group develop a
predominant constructive style, while another group
develops a predominant defensive style? Although a
variety of personal variables of individual participants
may be at play, we propose that the actual media type
(i.e., face-to-face versus virtual) may have an
important influence on the interaction style that
materializes within a group. For example, we expect
that a FtF context is more likely to engender the
constructive interaction style because it allows more
free exchange and expressions of nonverbal and
paralinguistic messages that will help members
develop respect for the perspectives of others in the
group. In contrast, VTs, because they lack the
richness of FtF communication, are less likely to foster
empathy and concern, and they are more likely to
produce information suppression, personal insults, or
other hostile communication [28,29]. In addition, the
VT medium is likely to reduce the thoroughness of
discussion and produce less critical thinking [30].
Proceedings of the 37th Hawaii International Conference on System Sciences - 2004
0-7695-2056-1/04 $17.00 (C) 2004 IEEE 2

It follows that VTs are more likely than FtFTs to
engage in a defensive style. Virtual interaction makes it
easier to either consciously “free ride”, or unconsciously
“socially loaf” [29]. This, coupled with a lack of
emotional cues, may promote a tendency for VT
members to sink into a passive mode. Moreover, the
lack of personal inhibition in VTs, together with
increased tendencies toward insults and profanity, could
cause more aggressive behaviors. In sum, these
arguments suggest that:
H2: Face-to-face teams will be more likely to
demonstrate a constructive interaction style than virtual
teams.
H3: Face-to-face teams will be less likely to demonstrate
defensive interaction styles than virtual teams.
Further, we expect that TFL will be associated with
group interaction style. Teams with members who
exhibit TFL behavior will develop an appropriate
climate or tone by, for instance, fostering a free flow of
ideas and show respect for differing opinions. They
demonstrate values stressing a balance between personal
concerns and achieving cooperative group outcomes.
Members of teams who exhibit TFL also have a
preference for action, rather than the passivity associated
with a defensive group style [31]. In sum, we expect
that:
H4: Shared TFL will be positively associated with a
constructive interaction style.
H5: Shared TFL will be negatively associated with a
defensive interaction style.
3. Leadership, Interaction Style, and
Cohesion
There is theoretical and empirical evidence that TFL
is associated with the degree of cohesion in a group
[32,33]. For example, as articulated by Waldman and
Yammarino [34], transformational leaders have a desire
to forge cohesion, and they do so by expressing
confidence in the ability of the group to pursue common
goals, thus enabling individuals within the group to
experience a heightened sense of self-efficacy.
Kozlowski et al. [12] stressed the developmental
function of team leadership in terms of enacting a
common orientation or coherence. Further, although not
examining cohesion specifically, Sosik, Avolio and
Kahai [35] examined the effects of TFL on group
potency (the group’s belief that it can be effective) in a
virtual group. They found that TFL was associated with
higher levels of potency than transactional leadership.
Based on this evidence, we propose the following:
H6: The level of shared TFL within teams will
positively predict their cohesion.
We also propose a link between group interaction
styles and cohesion. Logically, the cooperation, trust,
and free exchange of ideas associated with a
constructive interaction style should engender a desire
on the part of team members to stick together.
Conversely, the passivity or lack of interaction,
coupled with the aggressive pursuit of personal
agendas shown in defensive teams, should result in
less cohesion. Indeed, Warkentin, Sayeed, and
Hightower [36] provide evidence that, as compared to
VTs, FtFTs ultimately obtain a higher degree of
cohesion and have members who are more satisfied
with the decision process followed by their respective
teams because they suffer fewer communication
problems. VTs are unable to duplicate the normal
“give and take” of FtF discussions, and VT
discussions may appear to lack focus because of
multiple team members “talking” at once. The
unfortunate result may be a less cohesive group. In
sum:
H7: A constructive interaction style will be positively
associated with team cohesion.
H8: A defensive interaction style will be negatively
associated with team cohesion.
4. Leadership and Performance
We further expect that shared TFL will have a
direct positive effect on task performance in both
FtFTs and VTs. TFL, which encourages individuals to
contribute ideas and work toward group effectiveness,
has indeed been shown to be advantageous to group
outcomes [37]. However, to date, research has not
examined the extent to which TFL is demonstrated in
VTs compared to FtFTs, nor has it examined the
degree to which the amount of such leadership shown
in teams is relevant to group outcomes in both media.
Nevertheless, based on the plethora of evidence
regarding the effectiveness of TFL (e.g., [37]), we
expect that:
H9: The level of shared TFL within teams will
positively predict their task performance.
The combined effects of prior predictions suggest
that cohesion may mediate the relationship between
TFL and task performance. That is, the relationship
between shared TFL and task performance might be
Proceedings of the 37th Hawaii International Conference on System Sciences - 2004
0-7695-2056-1/04 $17.00 (C) 2004 IEEE 3

largely dependent on the fact that TFL behavior
influences cohesion as a mediating variable. Stated
alternatively, shared TFL may ultimately affect task
performance through its promotion of cohesiveness.
Although we are not aware of prior research specifically
testing the mediational effect of cohesion on the
relationship between TFL and task performance, we
nevertheless suggest that:
H10: Team cohesion will mediate the relationship
between shared TFL and task performance.
Finally, based on the majority of prior research (cf.,
[5,38]), we expect to find a relationship between
cohesion and team performance. The performance-
enhancing effect of cohesion results largely from
members’ commitment to the goals of the group.
Further, Zaccaro, Gualtieri, and Minionis [11] obtained
results suggesting that cohesion can improve team
decision-making, especially when the team is under time
pressure, as was the case in the present research.
Accordingly, we posit that:
H11: Team cohesion will positively predict task per-
formance.
5. Method
5.1. Participants
Media type, shared TFL, constructive and defensive
interaction styles, cohesion, and task performance data
were collected from 336 members of 88 teams who had
completed the "Ethical Decision Challenge" [39], a
structured problem-solving exercise used for manage-
ment development and team building in classroom and
corporate settings. Participants were MBA and senior
undergraduate students in multiple sections of a Man-
agement Information Systems course that required a 12-
week long team project. Participants had a median age
of 29 years; 50.7% were male; and 72.2% were Cauca-
sian. They averaged 9.6 years of work experience, and
64.8% reported having held a supervisory position.
There were no significant demographic differences be-
tween subjects in the two experimental conditions. The
exercise, completed early in the team-building stage of
the project, was performed for course credit. The me-
dian number of participants per team was four, with 24
three-member teams, 56 four-member teams, and 8 five-
member teams. Given that 67% of the participants re-
ported knowing no members of their group well, groups
could be best described as zero-history teams.
5.2. Task
The "Ethical Decision Challenge" requires
participants to rank 10 biomedical and behavioral
research practices—all of which involve human
subjects— in terms of their relative permissibility and
acceptability [39]. It provides participants with an
opportunity to practice their skills in both ethical
analysis and group decision-making. Solutions to the
"Ethical Decision Challenge" are developed first on an
individual basis and then as a group. Individual and
team solutions are then compared to experts’ solution.
Comparisons between individual solutions and the
experts’ solution indicate how well participants have
solved the problem. Comparisons between
participants’ individual scores and their team’s score
indicate whether they were able to achieve synergy by
fully using and building on their collective knowledge
and skills (see [40]). In other words, the team's
performance should be better than any individual
performer if group synergy is achieved.
Although task complexity has not received much
attention with respect to VTs, it has critical
implications for the structure, processes, and
leadership of virtual teams. Based on Van de Ven,
Delbecq, and Koenig [41], as well as Thompson’s
topology [60], we would characterize the workflow
processes of the Ethical Decision Challenge as
intensive. That is, it represents an interdependent
arrangement where team members must diagnose,
problem solve, and/or collaborate simultaneously as a
team to accomplish their task. Such tasks are typically
quite challenging, with the need for greater levels of
synchronous collaboration and information sharing
among team members [42].
5.3. Technology
Participants in 42 teams completed a paper
version of the exercise with a FtF discussion, while
participants in 46 teams completed a Web version of
the exercise with a computer-mediated textual
discussion. The assignment of teams to the virtual or
FtF condition was random. In both scenarios,
participants completed the "Challenge" during a
regularly scheduled 90-minute class meeting. FtF
participants were given paper booklets of the exercise
and directed to the problem statement. Computer-
mediated participants were provided with the URL of
the home page for the Web version of the exercise and
directed to the same problem statement. Each
participant was then given 10 minutes to read the
situation and “Challenge”, followed by an additional
10 minutes to rank the items (e.g., permissibility and
acceptability of 10 behaviors). Those in FtFTs
indicated their ranking on a provided answer form,
whereas those in computer-mediated teams (i.e., VTs)
Proceedings of the 37th Hawaii International Conference on System Sciences - 2004
0-7695-2056-1/04 $17.00 (C) 2004 IEEE 4

submitted their personal solution via an interactive Web
form.
Based on earlier research and extensive pre-
testing, teams were given 35 minutes to discuss the
problem and provide the best possible consensus
ranking of the items —a ranking with which all team
members could "live with." This amount of time
allowed all FtFTs and VTs to complete the task
without excessive time pressure and without generating
participant fatigue or disinterest. The literature on VTs
identifies three key dimensions to characterize their
“virtualness:” relative permanence of the team, team
dispersion, and technological enablement. Although
synthetic and atypically short in duration, our text-
based virtual teams appear to be consistent with the
published theoretical boundaries of “virtualness” and
provide a consistent protocol for comparison with
FtFTs. More importantly, text-based communication
represents a clear baseline condition for VTs,
especially in contrast to others that might use a variety
of bandwidth-intensive technologies.
Upon achieving a consensus solution, a team
representative either registered the ranking with the
facilitator (FtFTs) or submitted a Web form (VTs).
Lastly, each team member independently completed
three questionnaires: (1) the Group Styles Inventory™
questionnaire assessing the team’s interaction style
[43], (2) a TFL questionnaire where each team member
assessed the leadership exhibited by each of the other
members in his or her team, and (3) a quantitative
measurement of group cohesion. All questionnaires
were completed before receiving feedback on the
"experts' rank" or the quality of their own (and team’s)
solution.
5.4. Measures
As justified below, the level of analysis in the pre-
sent study is the group. We included team size as a
control variable in all analyses.
5.4.1. Shared transformational leadership.To
measure the level of shared TFL in the team,
participants assessed each team member by rating eight
behavioral statements taken from a short form of the
Multifactor Leadership Questionnaire (MLQ) [44].
Specifically, items were chosen that were judged by
the researchers to be potentially relevant to a 35-
minute team problem-solving task. These items
generally tapped charismatic, inspirational, and
intellectual stimulation leadership behaviors [44].
Items were answered on a five-point scale ranging
from not at all (1) to a very great extent (5). The
ratings were then summed and an average shared TFL
score was computed for each team.
5.4.2. Group interaction styles. Group interaction
styles were measured using 24 items drawn from the
Group Styles Inventory™ [45]. These measures
focus on the ways in which members of a group
interact with one another and approach their task
during a problem-solving session. Specifically, a
constructive interaction style was measured by four
subscales composed of three items each. Similarly, a
defensive interaction style was measured by four
subscales also comprised of three items each.
Participants indicated the extent to which each item
described the interaction style of their team using a
five-point response scale ranging from not at all (1)
to a very great extent (5). Responses to the items for
the constructive and defensive group interaction
subscales were averaged, and then an average score
was computed for each team on the two scales.
5.4.3. Group cohesion. Group cohesion was
measured by asking participants to rate five items
that dealt with group atmosphere and satisfaction
with the group [46,47]. Responses were rated on a
five-point scale ranging from 1 = strongly disagree
to 5 = strongly agree. The items for group cohesion
were summed and averaged for each team.
5.4.4. Task performance. We assessed task
performance by measuring “performance” and
“synergy.” Our performance measure has been
referred to as the “conventional scoring algorithm”
as it calculates the difference between group
consensus versus average individual performance on
problem-solving tasks like ours (see [40], p.321).
Our “synergy” measure calculates the difference
between the group consensus versus the score of the
best member in the team. Previous research has
shown that in many cases this synergistic “gain” is
negative, indicating a failure to fully use the
expertise of the group and a loss due to group
process [40].
5.5. Level of Analysis
The justification for the aggregation of the items
to the group level is provided by tests based on the
multiple-item estimator r
wg
for scales with moderately
skewed distributions [61]. A median r
wg
value of .7 is
considered sufficient agreement within the group on
any given measure [62]. The median r
wg(j)
estimate
across the 88 groups is .76 for the cohesion measure,
.86 for the shared leadership measure, .84 for the
constructive measure, and .72 for the defensive meas-
ure, indicating that there is sufficient agreement be-
tween members' reports, and that the amount of
within-team variance is moderately small relative to
the expected amount of variance in responses. Fur-
Proceedings of the 37th Hawaii International Conference on System Sciences - 2004
0-7695-2056-1/04 $17.00 (C) 2004 IEEE 5

thermore, results of a series of one-way analyses of
variance and Bartlett’s test for homogeneity of variance
within groups support the proportional consistency of
variance among the responses of members within the
same group, as compared to the responses of members
across groups [48]. Thus, team members’ individual
assessments of group phenomena were aggregated to the
group level of analysis.
5.6. Data Analyses
The hypotheses were tested using a structural
equation modeling procedure called Partial Least
Squares (PLS) [49]. PLS is recommended for predictive
research models where the focus is on theory
development and testing, and it is suitable for use with
smaller samples [50]. The path coefficients in a PLS
model are standardized regression coefficients. The
loadings of items on the constructs can be interpreted in
a similar way as factor loadings in more traditional
factor analysis. PLS avoids the problems of
inadmissible solutions and factor indeterminacy common
to covariance fitting approaches (such as LISREL or
EQS). However, its main disadvantage is that parameter
estimates are sub-optimal when the sample size is small,
or when the numbers of indicators per latent variable is
small. Our sample size of 88 teams exceeded the
minimum suggested requirements (see [50]).
In PLS, constructs that are measured with more than
one observed variable (indicator) represent either under-
lying factors of (i.e., reflective construct), or indices
produced by, the observed variables (i.e., formative
construct). In the present study, we modeled indicators
of TFL, constructive and defensive group interaction
styles, and cohesion reflectively because we expected
each of these constructs to be homogeneous. Our expec-
tation was supported by the high levels of variance
shared by the indicators of the constructs in question
with their respective constructs, and by the high values
of internal consistency, as discussed in the section be-
low.
6. Results
The results supported Hypotheses 1 and 2: the FtF
media type was positively related to TFL (for
performance and for synergy) and to a constructive
interaction style (for performance and for synergy).
However, Hypothesis 3, which stated that FtF teams
would be less likely to demonstrate defensive interaction
styles than VTs, was not supported.
Hypotheses 4 and 5 were supported: shared TFL
was positively related to a constructive interaction style
(for performance and for synergy) and negatively related
to a defensive interaction style (for performance and for
synergy). As postulated by Hypothesis 6, shared TFL
positively predicted cohesion (for performance and for
synergy). Consistent with Hypotheses 7 and 8, a
constructive interaction style positively predicted
cohesion (for performance and for synergy), and a
defensive interaction style negatively predicted
cohesion (for performance and for synergy).
Table 1. PLS analysis of the hypothesized
relationships
Hy po t he s e s
Pat h
coefficient
(
df = 87
)
t
Path
coefficient
(
df = 87
)
t
H1: Medi a
→
leadership
0.4
4.16
***
0.4 4.24***
H2: Medi a → constructive interaction
0.19
2.16
**
0.19 2.27**
H3: Medi a → defensive behaviors
0.07 -0.78 0.07 -0.73
H4: Leader shi p → constructive interaction
0.57
7.18
***
0.57 8.07***
H5: Leader shi p → defensive interaction
-0.43 -4.56*** -0.43 -4.74***
H6: Leader shi p → cohesion
0.13 1.75* 0.12 1.78*
H7: Constructive interaction → cohesion
0.42 4.71*** 0.42 5.12***
H8: Defensive interaction → cohesion
-0.46 -4.71*** -0.46 -4.69***
H9: Leader shi p → task performance
0.01 0.32 -0.12 -0.52
H11: Cohesion → task performance
0.21 2.02** 0.26 1.88*
Cont rol : Team size
→
task performance
0.08 0.30 -0.04 -0.01
*p<. 05; **p<.01; ***p<.001
Note: The
t
-val ues ar e estimated based on a jackknif e procedure f or which 88 sampl es were generated.
Perf ormance Synergy
Contrary to Hypothesis 9, the results indicated
that the relationship between shared TFL and task
performance was not significant. However,
Hypothesis 11, which posited a positive relationship
between cohesion and task performance, was
supported (for performance and for synergy). We
further attempted to evaluate Hypothesis 10, which
posited that cohesion would mediate the relationship
between shared TFL and task performance. Since the
relationship between shared TFL and task performance
was not significant, mediation effects involving
leadership were not substantiated according to the
work of Baron and Kenny [cf. [52]).
Finally, the literature suggested that team size
might have an effect on task performance, so a control
path was included between these two variables. The
results indicated that team size was not related to task
performance.
7. Discussion
The objective of this study was to explore, via
structural equation modeling, how the choice of a
media type can ultimately influence task performance
in problem-solving teams. Shared TFL and group
interaction styles appear to be somewhat dependent on
media type. As compared to VTs, FtFTs have a
Proceedings of the 37th Hawaii International Conference on System Sciences - 2004
0-7695-2056-1/04 $17.00 (C) 2004 IEEE 6

greater tendency to develop shared TFL and a
constructive interaction style. However, teams using
both media types were found to be equally likely to
produce defensive behaviors.
Overall, these results are consistent with the
findings and suggestions of others. For example, Straus
and McGrath [7] found FtFTs to be more cohesive than
VTs. Our findings also support the proposals of Bell &
Kozlowski [9] who suggested that a more informed
explanation for the relationship between media type and
cohesion would include the analysis of leadership and
group interaction styles since they contribute to (or
inhibit) the level of cohesiveness. Specifically, design
and workflow characteristics of VTs will influence the
leadership functions that are critical for the team’s
coherence, seamlessness, integration, and effectiveness
[12]. Clearly, in our study, constructive teams produced
greater cohesiveness while defensive behaviors inhibited
team cohesiveness.
We further attempted to evaluate the mediational
role of cohesion in the relationship between shared TFL
and task performance. We originally expected that
shared TFL would have a direct effect on task
performance, a finding that was not substantiated.
However, our findings are in line with those of Ross and
Offermann [51] who also showed no relationship
between TFL and objective task performance. Since the
relationship between shared TFL and team performance
was not significant, mediation effects involving
leadership were not clearly substantiated. It is
interesting to note how newer conceptualizations of
mediation (e.g., [52]) do not necessitate a relationship
between the independent variable (i.e., transformational
leadership) and the dependent variable (performance) in
order to demonstrate that mediation has indeed occurred.
In line with prior research, we obtained a direct
relationship between cohesion and task performance.
Yet, the nature of the relationship may not be altogether
clear. Specifically, does greater cohesion lead to better
performance, or does performance cause groups to
become more cohesive? In our study, it is likely that
greater cohesion led to better task performance since
participants had no objective criteria to assess the quality
of their teams’ solutions prior to assessing cohesiveness.
Although the cohesion-performance relationship was
significant in our study, it may actually be weaker for
our short-lived groups than for enduring groups where
past performance can further affect cohesion in an
ongoing manner.
7.1. Practical Implications
In total, our findings suggest that VTs may
ultimately face cohesion and performance difficulties,
largely because of problems engendered in terms of
leadership and interaction styles. For such groups, a key
issue may be how to facilitate the development and
subsequent maintenance of group cohesion to improve
the effectiveness of these emerging groups. Findings
such as these should help organizations recognize the
critical role of team leadership in terms of the
development and shaping of team processes,
generating shared goal commitment, and shaping a
constructive climate (e.g., [12,13,45]). Unfortunately,
the ability of leaders seems severely restricted by the
lack of face-to-face contact within VTs.
Daft and Lengel [53] argued that such problems
can be addressed by the use of richer interaction media
to enhance social presence. Others have shown that
synchronous communication is superior to
asynchronous communication for intensive tasks that
require considerable information sharing and complex
collaborative work [9,24]. Our study suggests that
synchronous, but text-based, systems may not provide
a rich enough interaction environment in comparison
to, say, videoconferencing. Certainly, virtual groups
generally report a preference for using video when
such technology is available [54]. Other experimental
studies have provided corroboration, but suggest that
performance differences attributable to video are
significant only in short-term, or time-limited groups
[55]. It seems that virtual groups are able to generate
productivity-enhancing levels of cohesion given
sufficient time, but that this may be accelerated by the
use of higher-bandwidth technologies. Unfortunately,
the costs and invasiveness of establishing high-
bandwidth interactions in VTs make them oftentimes
impractical for common use.
As an alternative approach, Weisband [56]
suggests that teams meet face-to-face prior to working
virtually. Again, this solution might not be feasible
for teams that choose the virtual environment to
overcome spatial or geographic distance.
In response to the quandary, Manz & Sims [57]
implied that typical VT leadership functions must be
accomplished by substitutes and by distributing to the
team itself. Others suggest that VT leaders will need
to implement a system in which team members will be
able to regulate their own performance patterns as a
team [9,12,58]. Although much work remains, our
findings suggest that teams can and should assess
interaction styles in VTs, beginning at formative
stages, and then proactively manage any interaction
challenges that could threaten team performance.
More importantly, we used a set of instruments that
could provide team members with indicators of shared
leadership, interaction style, and cohesion. These
instruments could enable team members to ultimately
maximize their performance.
Proceedings of the 37th Hawaii International Conference on System Sciences - 2004
0-7695-2056-1/04 $17.00 (C) 2004 IEEE 7

7.2. Limitations and Future Research
We should note that our study involved a short-
duration consensus-type task that has its pros and cons.
On the positive side, our comparative research protocol
was designed in a way that a large number of FtFTs and
VTs performed systematically the identical task in the
same time frame and provided feedback by answering
validated instruments. In contrast, much extant research
in VTs has been either anecdotal and/or limited in the
number of teams examined (e.g., [64]). On the other
hand, as suggested above, VTs may fare better when
members have a longer time period to get acquainted.
Some have argued that VTs are inappropriate when the
team is newly formed or short-lived [1,55]. However, it
is also important to remember that many real-world VTs
are indeed ad hoc. Congruent with our protocol, they are
composed temporarily to solve urgent problems and then
disbanded. Ultimately, in using our accelerated protocol,
we do not purport to create surrogates for the
longitudinal activities and performance of ongoing
virtual teams. Instead, our protocol more realistically
defines a discrete and objective assessment procedure
capable of assisting virtual teams with team development
and performance management functions.
A related limitation of our study was that our teams
were not actual work groups in organizations, but rather
student groups assembled for a semester. The risks of
poor performance were not as great as might have been
in an actual work setting, and the benefits of excellent
performance were equally limited.
Organizations can be naïve about the advantages,
problems, and disadvantages of emerging organizational
forms, such as VTs. They may perceive the benefits of
VTs largely in terms of reduced staff travel costs and
freedom to work from anywhere. However, they are
often unaware of their potential disadvantages in terms
of group interaction, leadership, and cohesion problems.
This combination of misconceptions and rapid changes
in technologies can make informed decision-making
regarding the use of VTs quite difficult.
The present study has addressed some of the
potential limitations of VTs. At the same time, we
recognize that additional research is warranted
comparing behaviors of FtFTs to those of geographically
and/or temporally dispersed VTs and examining the
effects of leadership and group interaction styles.
Further research is needed to establish whether our
results hold true in a field design with computer-
supported groups working on actual (as opposed to
simulation-based) problem-solving tasks. This would
help capture the motivational element that may have
been somewhat missing from the present study. It may
also yield insights into the relative strength of the effects
of leadership and interaction styles on real-world VTs.
8. References
[1] Duarte, D. L., & Snyder, N. T. Mastering virtual teams.
San Francisco: Jossey-Bass. 1999.
[2] Townsend, A., DeMarie, S., & Hendrickson, A. “Virtual
teams: Technology and the workplace of the future.” Acad-
emy of Management Executive, 12, 1998. pp.17-29.
[3] Kozlowski, S. W .J. “Training and developing adaptive
teams: Theory, principles, and research.” In J.A. Cannon-
Bowers & E. Salas (Eds.), Decision making under stress:
Implications for training and simulation Washington, DC:
APA Books. 1998. pp. 115-153.
[4] DeSanctis, G., & Poole, M. S. “Transitions in teamwork
in new organizational forms.” In B. Markovsky, M. J.
Lovaglia, L. Troyer, & E. J. Lawler (Eds.), Advances in
Group Processes. Greenwich, CT: JAI Press. vol. 14, 1997.
pp. 157-176.
[5] Guzzo, R. A., & Dickson, M. W. “ Teams in organiza-
tions: Recent research on performance and effectiveness.”
Annual Review of Psychology, 47, 1996. pp. 307-338.
[6] Lipnack, J. & Stamps, J. Virtual Teams: Reaching across
space, time, and organizations with technology. New York:
John Wiley. 1997.
[7] Straus, S. G., & McGrath, J. E. “Does the medium really
matter? The interaction of task type and technology on
group performance and member reactions.” Journal of
Applied Psychology, 79, 1994. pp. 87-97.
[8] Tan, B. C. Y., Wei, K. K., Huang, W. W., & Ng, G. N.
2000. “A dialogue technique to enhance electronic commu-
nication in virtual teams.” IEEE Transactions on Profes-
sional Communication, 43: 153-165.
[9] Bell, B. S. & Kozlowski, S. W. J. “A typology of virtual
teams: Implications for effective leadership.” Group &
Organization Management, 27. 2002. pp. 14-49.
[10] Fleishman, E. A., Mumford, M. D., Zaccaro, S. J.,
Levin, K. Y., Korotkin, A. L., & Hein, M. B. “Taxonomic
efforts in the description of leader behavior: A synthesis and
functional interpretation.” Leadership Quarterly, 2. 1991.
pp. 245-287.
[11] Zaccaro, S. J., Gualtieri, J., & Minionis, D. “Task
cohesion as a facilitator of team decision making under
temporal urgency.” Military Psychology, 7. 1995. pp. 77-
93.
[12] Kozlowski, S. W. J., Gully, S. M., McHugh, P. P.,
Salas, E., & Cannon-Bowers, J. A. “A dynamic theory of
leadership and team effectiveness: Developmental and task
contingent leader roles.” In G. R. Ferris (Ed.), Research in
personnel and human resource management, 14. Greenwich,
CT: JAI Press. 1996. pp. 253-305.
Proceedings of the 37th Hawaii International Conference on System Sciences - 2004
0-7695-2056-1/04 $17.00 (C) 2004 IEEE 8

[13] McGrath, J. E. Leadership behavior: Some requirements
for leadership training. Washington DC: U.S. Civil Service
Commission. 1962.
[14] Roby, T.B. “The executive function in small groups.” In
L. Petrullo & B.M. Bass (Eds.), Leadership and interpersonal
behavior. New York: Holt, Reinhart & Winston. 1961.
[15] George, J., Easton, G., Nunmaker, J., & Northcraft, G. “A
study of collaborative group work with and without computer-
based support.” Information Systems Research, 1. 1990. pp.
394-415.
[16] Hiltz, S. R., Johnson, K., & Turoff, M. “Experiments in
group decision making: Communication process and outcome
in face-to-face versus computerized conferences.” Human
Communication Research, 13. 1986. pp. 225-252.
[17] Avolio, B., Kahai, S., & Dodge, G. “E-leadership: Impli-
cations for theory, research and practice.” Leadership Quar-
terly, 11. 2000. pp. 615-668.
[18] Shamir, B. Leadership in boundaryless organizations:
Disposable or indisposable? Unpublished manuscript, Hebrew
University. 1997.
[19] Bass, B. M., & Avolio, B. A. The multifactor leadership
questionnaire. Palo Alto, CA: Consulting Psychologists Press.
1990.
[20] Kahai, S., Sosik, J., & Avolio, B. Effects of leadership
style, anonymity, and rewards in an electronic meeting system
environment. Working paper, Center for Leadership Studies,
Binghamton University. 2000.
[21] Sheppard, J. “Productivity loss in performance groups: A
motivation analysis.” Psychological Bulletin, 113. 1993. pp.
67-81.
[22] Shamir, B., & Howell, J. M. “Organizational and contex-
tual influences on the emergence and effectiveness of charis-
matic leadership.” Leadership Quarterly, 10. 1999. pp. 257-
283.
[23] Gardner, W. L., & Avolio, B. J. “The charismatic relation-
ship: A dramaturgical perspective.” Academy of Management
Review, 23. 1998. pp. 32-58.
[24] Hollingshead, A. B., McGrath, J. E., & O’Connor, K. M.
“Group task performance and communication technology: A
longitudinal study of computer-mediated versus face-to-face
work groups.” Small Group Research, 24. 1993. pp. 307-333.
[25] Cooke, R. A., & Szumal, J. L. “Measuring normative
beliefs and shared behavioral expectations in organizations: the
reliability and validity of the organizational culture inventory.”
Psychological Reports, 72. 1993. pp. 1299-1330.
[26] Cooke, R. A., & Rousseau, D. M. “Behavioral norms and
expectations.” Group and Organization Studies, 13. 1988. pp.
245-273.
[27] Xenikou, A., & Furnham, A. “A correlational and factor
analytic study of four questionnaire measures of organiza-
tional culture.” Human Relations, 49. 1996. pp. 349-371.
[28] Hollingshead, A. B. “Information suppression and
status persistence in group decision making: the effects of
communication media.” Human Communication Research,
23. 1996. pp. 193-219.
[29] Weisband, S. “Group discussion and first advocacy
effects in computer-mediated and face-to-face decision-
making groups.” Organizational Behavior and Human
Decision Processes, 53. 1992. pp. 352-380.
[30] Sambamurthy, V., Poole, M. S., & Kelly, J. “The ef-
fects of variations in GDSS capabilities on decision-making
processes in groups.” Small Group Research, 24. 1993. pp.
523-546.
[31] Waldman, D. A., Bass, B. M., & Einstein, W. O.
“Leadership and outcomes of performance appraisal proc-
esses.” Journal of Occupational Psychology, 60. 1987. pp.
177-186.
[32] Conger, J. A., & Kanungo, R. N. Charismatic leader-
ship in organizations. Thousand Oaks, CA: Sage. 1998.
[33] Curphy, G. J. “An empirical investigation of the effects
of transformational and transactional leadership on organiza-
tional climate, attrition, and performance.” In K. E. Clark,
M. B. Clark, & D. P. Campbell (Eds.), Impact of leadership:
Greensboro, NC: Center for Creative Leadership. 1992. pp.
177-187.
[34] Waldman, D. A., & Yammarino, F. J. “CEO charis-
matic leadership: Levels-of-management and levels-of-
analysis effects.” Academy of Management Review, 24.
1999. pp. 266-285.
[35] Sosik, J., Avolio, B., & Kahai, S. “Effects of leadership
style and anonymity and group potency and effectiveness in
a group decision support system environment.” Journal of
Applied Psychology, 82. 1997. pp. 89-103.
[36] Warketin, M. E., Sayeed, L, & Hightower, R. “Virtual
teams versus face-to-face teams: An exploratory study of a
web based conference system.” Decision Sciences, 14.
1997. pp. 29-64.
[37] Lowe, K. B., Kroeck, K. G., & Sivasubramaniam, N.
“Effectiveness correlates of transformational and transac-
tional leadership: A meta-analytic review of the MLQ litera-
ture.” Leadership Quarterly, 7. 1996. pp. 385-425.
[38] Evans, C. R., & Dion, K. L. “Group cohesion and
performance: A meta-analysis.” Small Group Research, 22.
1991. pp. 175-186.
[39] Balthazard, P. A. Ethical Decision Challenge — Inter-
net edition (adapted from R. A. Cooke). Arlington Heights
IL: Human Synergistics/Center for Applied Research. 2000.
Proceedings of the 37th Hawaii International Conference on System Sciences - 2004
0-7695-2056-1/04 $17.00 (C) 2004 IEEE 9

[40] Cooke, R. A., & Kernaghan, J. A. “Estimating the differ-
ence between group versus individual performance on prob-
lem-solving task.” Group and Organization Studies, 12. 1987.
pp. 319-342.
[41] Van de Ven, A. H., Delbecq, A. L., Koenig, R. “Determi-
nants of coordination modes within organizations.” American
Sociological Review, 41. 1976. pp. 322-328.
[42] Kozlowski, S. W. J., Gully, S. M., Nason, E. R., & Smith,
E. M. “Developing adaptive teams: A theory of compilation
and performance across levels and time.” In D. R. Ilgen & E.
D. Pulakos (Eds.), The changing nature of work performance:
Implications for staffing, personnel actions, and development.
San Francisco: Jossey-Bass. 1999. pp. 240-292.
[43] Balthazard, P. A. Group Styles Inventory — Internet
edition (adapted from R. A. Cooke and J. C. Lafferty). Arling-
ton Heights IL: Human Synergistics/Center for Applied Re-
search. 1999.
[44] Bass, B., M., & Avolio, B. A. Improving organizational
effectiveness through transformational leadership. Thousands
Oaks, CA: Sage. 1994.
[45] Cooke, R. A., & Szumal, J. L. “The impact of group
interaction styles on problem-solving effectiveness.” Journal
of Applied Behavioral Science, 30. 1994. pp. 415-437.
[46] Cook, J. D. The experience of work: A compendium and
review of 249 measures and their use. New York: Academic
Press. 1981.
[47] O’Reilly, C., Caldwell, D., & Barnett, W. “Work group
demography, social integration, and turnover.” Administrative
Sciences Quarterly, 34. 1989. pp. 21-37.
[48] Kozlowski, S. W. J., & Hattrup, K. “A disagreement about
within-group agreement: Disentangling issues of consistency
versus consensus.” Journal of Applied Psychology, 77. 1992.
pp. 161-167.
[49] Fornell, C., & Bookstein, F. L. “Two structural equation
models: LISREL and PLS applied to consumer exit-voice
theory.” Journal of Marketing Research, 19. 1982. pp. 440-
452.
[50] Barclay, D., Higgins, C., & Thompson, R. “The partial
least squares (PLS) approach to causal modeling: Personal
computer adoption and use as an illustration.” Technology
Studies, 2. 1995. pp. 285-309.
[51] Ross, S. M., & Offermann, L. R. “Transformational lead-
ers: Measurement of personality attributes and work group
performance.” Personality and Social Psychology Bulletin, 23.
1997. pp. 1078-1086.
[52] Kenny, D. A., Kashy, D. A., & Bolger, N. “Data analysis
in social psychology.” The handbook of social psychology (4
th
ed.). 1998. pp. 233-265.
[53] Daft, R., & Lengel, R. H. “Organizational information
requirements, media richness and structural design.” Man-
agement Science, 32. 1986. pp. 554-571.
[54] Olson, J., Olson, G., & Meader, D. “What mix of video
and audio is useful for small groups doing remote real-time
design work.” In Proceedings of the ACM Conference on
Human Factors in Computing Systems: Denver, CO, May.
1995. pp. 362-368.
[55] Walther, J. B. “Computer-mediated communication:
Impersonal, interpersonal, and hyperpersonal interaction.”
Communication Research, 23. 1996. pp. 1-43.
[56] Weisband, S. Assessing divergent and convergent
processes in face-to-face and computer-mediated groups.
Working paper, University of Arizona. 2000.
[57] Manz, C. C., & Sims, H. P. “Leading workers to lead
themselves: The external leadership of self-managing work
teams.” Administrative Science Quarterly. 32. 1987. pp. 106-
128.
[58] Zaccaro, S. J., & Burke, C. S. “Team versus crew lead-
ership: Differences and similarities.” In R. J. Klimoski
(Chair), When is a work team a crew – and does it matter?
Symposium conducted at the 13
th
annual Conference of the
Society for Industrial and Organizational Psychology, Dal-
las, TX. 1998.
[59] Shamir, B., & Ben-Ari, E. “Leadership in an open
army: Civilian connections, interorgaizational frameworks,
and changes in military leadership.”
In J. G. Hunt, G. E. Dodge, & L. Wong (Eds.), Out-of-the-
box leadership: Transforming the 21
st
century army and
other top-performing organizations Greenwich, CT: JAI
Press. 1999. pp. 15-40.
[60] Thompson, J. D. Organizations in action. New York:
McGraw-Hill. 1967.
[61] James, L. R., Demaree, R. G., & Wolf, G. “R(
wg
) - An
assessment of within-group inter-rater agreement.” Journal
of Applied Psychology, 78. 1993. pp. 306-309.
[62] George, J. M. “Personality, affect, and behavior in
groups.” Journal of Applied Psychology, 75. 1990. pp. 107-
116.
[63] Baron, R., & Kenny, D. “The moderator-mediator
variable distinction in social psychological research: Con-
ceptual, strategic, and statistical considerations.” Journal of
Personality and Social Psychology, 51. 1986. pp. 1173-
1182.
[64] Maznevski, M .L., & Chudoba, K. M. “Bridging space
over time: global virtual team dynamics and effectiveness.”
Organization Science, 11. 2000. pp. 473-492.
Proceedings of the 37th Hawaii International Conference on System Sciences - 2004
0-7695-2056-1/04 $17.00 (C) 2004 IEEE 10