A Longitudinal Model of the Effects of Team Leadership and Group Potency on Group Performance

31
GROUP & ORGANIZATION MANAGEMENT Sivasubramaniam et al. / TEAM LEADERSHIP, GROUP POTENCY A Longitudinal Model of the Effects of Team Leadership and Group Potency on Group Performance NAGARAJ SIVASUBRAMANIAM WILLIAM D. MURRY Duquesne University BRUCE J. AVOLIO Binghamton University DONG I. JUNG San Diego State University In the current study, the authors examine how leadership within a team predicts levels of group potency and group performance over time. The authors predict that groups that rated themselves high on transformational team leadership behaviors soon after the groups were formed would see themselves as being more potent over time and also would achieve a higher level of group performance. These predictions were confirmed. The implications of the current study for fur- ther research on leadership within teams are discussed. It has been estimated that nearly two thirds of the Fortune 500 organizations utilize some form of teams in their organization. Motorola, for example, has approximately 4,000 teams operating in its facilities around the globe. Research evidence has documented the organization-wide benefits of imple- menting self-managed work teams (Cohen & Bailey, 1997). Nevertheless, the use of teams in both large and small organizations has raised a number of issues regarding the development and integration of such teams into existing organizational structures and cultures, and its impact on leadership practices (Avolio, Jung, Murry, & Sivasubramaniam, 1996; Kozlowski, Gully, Salas, & Cannon-Bowers, 1996). Leadership (or lack of it) has been identified as one of the leading causes of failures in implementing a team-based work system (Katzenbach, 1997; This was research was conducted while Nagaraj Sivasubramaniam and William Murry were affiliated with Binghamton University. Group & Organization Management, Vol. 27 No. 1, March 2002 66-96 © 2002 Sage Publications 66 by guest on September 27, 2016 gom.sagepub.com Downloaded from

Transcript of A Longitudinal Model of the Effects of Team Leadership and Group Potency on Group Performance

GROUP & ORGANIZATION MANAGEMENTSivasubramaniam et al. / TEAM LEADERSHIP, GROUP POTENCY

A Longitudinal Model of the Effectsof Team Leadership and GroupPotency on Group Performance

NAGARAJ SIVASUBRAMANIAM

WILLIAM D. MURRYDuquesne University

BRUCE J. AVOLIOBinghamton University

DONG I. JUNGSan Diego State University

In the current study, the authors examine how leadership within a team predicts levels of grouppotency and group performance over time. The authors predict that groups that rated themselveshigh on transformational team leadership behaviors soon after the groups were formed wouldsee themselves as being more potent over time and also would achieve a higher level of groupperformance. These predictions were confirmed. The implications of the current study for fur-ther research on leadership within teams are discussed.

It has been estimated that nearly two thirds of the Fortune 500 organizationsutilize some form of teams in their organization. Motorola, for example, hasapproximately 4,000 teams operating in its facilities around the globe.Research evidence has documented the organization-wide benefits of imple-menting self-managed work teams (Cohen & Bailey, 1997). Nevertheless,the use of teams in both large and small organizations has raised a number ofissues regarding the development and integration of such teams into existingorganizational structures and cultures, and its impact on leadership practices(Avolio, Jung, Murry, & Sivasubramaniam, 1996; Kozlowski, Gully, Salas,& Cannon-Bowers, 1996).

Leadership (or lack of it) has been identified as one of the leading causesof failures in implementing a team-based work system (Katzenbach, 1997;

This was research was conducted while Nagaraj Sivasubramaniam and William Murry wereaffiliated with Binghamton University.

Group & Organization Management, Vol. 27 No. 1, March 2002 66-96© 2002 Sage Publications

66

by guest on September 27, 2016gom.sagepub.comDownloaded from

Sinclair, 1992; Stewart & Manz, 1994). Yet, only a few models of team effec-tiveness (e.g., Gladstein, 1984; Kozlowski, Gully, Salas, et al., 1996) haveexplicitly considered leadership as one of the determinants of team effective-ness. Although many authors have focused on examining the leadership of asingle individual leading a team (Cohen, Chang, & Ledford, 1997; Manz &Sims, 1987), there has been no attempt to examine the impact of leadershipwithin or by the team. Dunphy and Bryant (1996), after reviewing the litera-ture on teams and identifying gaps, concluded that future research mustinclude leadership within teams when attempting to model team effectiveness.

In contrast to prior research, our focus was to expand earlier discussions ofteam leadership by examining the type of leadership that occurs withingroups or teams.1 The objectives of this article are twofold. The first is to dif-ferentiate the concept of team leadership from individual leadership and toexamine the influence of team leadership and group potency beliefs on groupperformance over time. The second is to explore the stability of differentteam leadership styles over a period of time. This study extends priorresearch in this area by considering the performance impacts of two con-structs—team leadership and potency—often identified as determinants ofteam effectiveness but seldom examined together.

THE TEAM LEADERSHIP CONSTRUCT

TEAM LEADERSHIP

We believe the team can influence each member just as the individualleader can influence his or her followers. Leadership, viewed as a socialinfluence process, goes beyond a single leader when observed at multiplelevels and can be vested in an individual or a group (Avolio & Bass, 1995).Other researchers have also characterized leadership as representing a col-lective influence process (Astin & Astin, 1996; Bowers & Seashore, 1966;House & Aditya, 1997). For instance, Bales (1954) suggested that it might bebeneficial for groups to allocate the task and relational leadership roles to dif-ferent individuals and called this “coleadership.” House and Aditya (1997)described collective leadership as specific leader behaviors distributed through-out the entire work unit, in which different members of the work unit enactthose behaviors contemporaneously. House and Aditya (1997) concluded,“The research by Bowers and Seashore (1966) clearly demonstrates that theexercise of leader behaviors can be shared by members of work units, as wellas conducted by formal work unit managers” (p. 459).

Sivasubramaniam et al. / TEAM LEADERSHIP, GROUP POTENCY 67

by guest on September 27, 2016gom.sagepub.comDownloaded from

Seers (1989) proposed the construct of Team-Member exchange (TMX)as a way to capture the mutuality in the relationships between a team memberand the group. High-quality TMX means that a focal member and his or hergroup have an excellent social as well as task relationship and individualsengage in behaviors not out of self-interest but to benefit coworkers. Seers(1989) argued that quality of lateral (within-group) relationships is moreimportant than vertical relationships (with supervisor) in predicting workgroup outcomes.

Building on this work, we define team leadership as the collective influ-ence of members in a team on each other. Manz and Sims (1987) describedthe differences between self-management leadership and traditional leader-ship behaviors of external leaders. External self-management leadershipresults in within-team behaviors such as self-reinforcement, self-evaluation,and self-expectation. In our view, these within-team behaviors that enableindividual members of the team to identify with and be motivated by the teamcan be termed team leadership. In many ways, team-level leadership is simi-lar to individual-level leadership in that the functional relationships hypothe-sized at the individual level are expected to be “isomorphic” with the nextlevel (House, Rousseau, & Thomas-Hunt, 1995).

Just as the supervisor is the focal person providing leadership in thesupervisor-follower dyadic relationship at the individual level, the teambecomes the referent in the team-member relationship at the group level ofleadership. Thus, when a team begins to assume a persona of its own, individ-ual members might be influenced by a collective model of “team think”(Martin, 1993). As the team identity becomes more salient to its members,being a part of that collective entity might inspire and challenge team mem-bers to superior performance levels in very much the same way as an individ-ual leader influences his or her followers to perform when they are inspired(Bass, 1998). This may help explain Bowers and Seashore’s (1966) findingthat peer leadership had a higher positive correlation with unit performancethan the leadership exercised by an individual manager.

In sum, we define team leadership in terms of how members of a groupevaluate the influence of the group as opposed to one individual within orexternal to the group. By focusing on the collective leadership of a team, weare not negating the importance of individual leadership but rather simplyfocusing on an additional form of leadership that can be observed in groups.Our theoretical meaning and operational definition of team leadership inte-grates the perspective taken by the team member in assessing leadership aswell as the level at which the phenomenon of leadership is examined. Ourapproach to measuring team leadership is called the “referent-shift consen-sus model” by Chan (1998) as there is a shift in the referent (from an

68 GROUP & ORGANIZATION MANAGEMENT

by guest on September 27, 2016gom.sagepub.comDownloaded from

individual to the team) prior to assessment. Our strategy for operationalizingleadership at the team level is analogous to a number of constructs that havebeen first operationalized at the individual level of analysis but were thenescalated to the group level. These constructs include team empowerment(Kirkman & Rosen, 1999), group affect (George, 1990), collective efficacy(Bandura, 1986), and group prosocial behavior (George & Bettenhausen,1991).

EXEMPLARY LEADERSHIP IN TEAMS

Highly effective teams have been characterized as having a clear centralfocus or vision and members willing to make sacrifices for the good of theteam’s mission or vision (Katzenbach & Smith, 1993). They actively buildeach member’s potential; they are cohesive and have members that fullyidentify with the team’s central purpose and values (Cohen, Ledford, &Spreitzer, 1996; Kozlowski, Gully, McHugh, Salas, & Cannon-Bowers,1996). Each of these characteristics is also important and typically discussedin the context of more traditional models of individual inspirational and/ortransformational leadership (Avolio et al., 1996; Bass & Avolio, 1994).Exemplary and/or transformational leaders develop in their followers ashared sense of purpose and/or vision and they work toward developing fol-lowers to their full potential. They are the “keeper,” if not champion, of theirgroup’s values, and often it is because of the leader’s efforts or sacrifices thatfollowers all come to identify with the vision or goals being pursued.

Current empirical research on teams indicates that exemplary leadershipcan have a positive impact on group motivation, efficacy, and performance(Cohen et al., 1997; Kumpfer, Turner, Hopkins, & Librett, 1993). Otherresearchers who qualitatively examined high-performing teams have alsocome to the same conclusion (Hackman, 1990; Katzenbach, 1997; Katzenbach& Smith, 1993; Manz & Sims, 1993). Effective if not exemplary leadershipby teams is required for achieving the highest levels of motivation and per-formance. Based on the current literature on teams, the type of social influ-ence process in teams that we predict makes a difference between a collectionof individuals being a group versus a high-performing team is leadership thathas been characterized as more transformational (Avolio et al., 1996). More-over, we are suggesting that the exemplary leadership shown at the team levelis virtually the same type of leadership that has been described betweenhighly effective leaders working with their followers.

At the individual level, transactional management-by-exception leader-ship has been shown to be associated with lower levels of motivation and sat-isfaction and poorer performance (Lowe, Kroeck, & Sivasubramaniam,

Sivasubramaniam et al. / TEAM LEADERSHIP, GROUP POTENCY 69

by guest on September 27, 2016gom.sagepub.comDownloaded from

1996). Management-by-exception leadership focuses members’ attention onmistakes and correction rather than developing each other’s potential. Teamspracticing such leadership will be seen tracking each other’s mistakes ratherthan successes, focusing on failures, and only monitoring individual mem-bers’ performance for errors. Groups exhibiting such leadership behaviorswould be unable to develop a sense of collective identity and are more likelyto remain a collection of inspectors rather than operating as a highly devel-oped team. Management-by-exception leadership behaviors, observed at thegroup level, are expected to have a negative impact on group motivation,potency, and performance.

A passive leader who avoids addressing problems or delays taking actionsuntil problems become serious is also unlikely to be effective. This laissez-faire style has been found to be negatively associated with follower satisfac-tion and performance (Bass & Avolio, 1994). We would expect groups thatdisplay similar patterns of laissez-faire behaviors to be dysfunctional andperform well below expectations. When leadership within the team is absent,members of the team are unlikely to identify with and be motivated by theteam’s goals, similar to how individual followers might behave in a situationin which the leader provides no support or direction when needed.

OVERVIEW OF THEORETICALMODEL AND HYPOTHESES

The many models of team effectiveness discussed in prior literature shareseveral common characteristics. All models of team effectiveness use thesystems perspective of inputs—process—outputs to conceptualize the pat-tern of relationships among the variables of interest. On the input side, priormodels have identified member-team characteristics (Campion, Medsker, &Higgs, 1993; Gladstein, 1984; Kozlowski, Gully, McHugh, et al., 1996), taskstructure (Cohen et al., 1996; Gladstein, 1984), team capacity (Klimoski &Mohammed, 1994), leadership (Avolio et al., 1996; Kozlowski, Gully, Salas,et al., 1996; Stewart & Manz, 1994), and organizational structure-culture(Lawler, 1992). Several models also include a set of (mediating) processvariables including group cohesiveness (Kozlowski, Gully, Salas, et al.,1996), norms (Bettenhausen & Murnighan, 1985; Cohen et al., 1996), collec-tive effort (Klimoski & Mohammed, 1994), and potency beliefs (Campionet al., 1993; Cohen et al., 1996; Shea & Guzzo, 1987). The outcomes havebeen operationalized in many different ways including job satisfaction(Campion, Papper, & Medsker, 1996; Gladstein, 1984), absenteeism (Cohenet al., 1996), cycle time (Silver & Bufanio, 1996), work unit productivity/

70 GROUP & ORGANIZATION MANAGEMENT

by guest on September 27, 2016gom.sagepub.comDownloaded from

sales (Campion et al., 1993; Gladstein, 1984), and ratings of work groupeffectiveness (Cohen et al., 1996; Gladstein, 1984; Manz & Sims, 1987).

The theoretical model presented in Figure 1 clarifies the relationshipsamong team leadership (antecedent), group potency2 (process), and groupperformance (outcomes). We use as a general framework for discussing theeffect of leadership in teams the team developmental model presented byKlimoski and Mohammed (1994). Klimoski and Mohammed (1994) attemptedto explain how a team formulates a shared understanding of its own behav-iors, values, and ideas. By developing a collective belief structure that isshared among members, teams begin to develop a sense of coherence(Smircich, 1983) and expectations of one another that can help facilitatelearning, allowing the team to determine its own direction and to ultimatelyto lead itself (Cannon-Bowers & Salas, 1990). Their model is particularly rel-evant to our investigation as they consider the effects of member perspectivesof leadership and potency on performance.

As team members build a set of expectations for each other that is linked tothe team’s mission, they are better able to identify their shared purpose,which helps each member to identify with what he or she is personally tryingto accomplish over time (Bettenhausen & Murnighan, 1985; Cannon-Bowers& Salas, 1990; Orasanu, 1990). Moreover, how a team thinks about itselfmay be a useful indicator of its stage of development with respect to leader-ship (Klimoski & Mohammed, 1994). For example, where members feel thateach individual on the team is out to fulfill his or her own self-interest, theirlevel of development may be characterized by a more transactional leader-ship orientation (e.g., “If you meet this expectation, then in return I will pro-vide this level of support”).

Several components of Klimoski and Mohammed’s (1994) model of teamperformance will be tested in the current study. In their model, they beginwith each individual’s potential to contribute, team composition, size, andresources available. Each of these factors contribute to what they call a“team’s capacity.” In the present study, these variables were controlled byrandomly assigning members to their respective groups and limiting groupsize to four or five members. Next in their model is leadership, which we haveoperationalized here as transformational, transactional management by excep-tion and laissez-faire leadership at the team level. Klimoski and Mohammed(1994) also identified several process variables including level of collectiveeffort and quality of interpersonal relations. We operationalize the “collec-tive effort” variable in their model as potency and measure performance interms of each group’s overall performance over a 4-month period. A signifi-cant component of the Klimoski and Mohammed model that was not directlyexplored in the current study was team members’ mental model of the team,

Sivasubramaniam et al. / TEAM LEADERSHIP, GROUP POTENCY 71

by guest on September 27, 2016gom.sagepub.comDownloaded from

except to the extent that we indirectly measure this based on members’ rat-ings of the team’s collective leadership profile.

IMPACT OF LEADERSHIP ON GROUP POTENCY

Shamir, House, and Arthur (1993) argued that transformational leader-ship can enhance the potency of groups or teams by making participation in agroup’s efforts more meaningful and tied to the collective identity of thegroup. Leadership by groups can highlight the importance of the task andhow the group has greater capability to take on difficult challenges. Burns(1978) described transformational leaders as people who work to increase thesalience and importance of the group, providing members with reasons tosacrifice their own self-interests for the greater good of the group, commu-nity, and society. Stated another way, people are more likely to sacrifice forsomething they believe in and can identify with versus something seen as atask one is obligated to perform. By collectively building faith in a betterfuture, transformational leadership among team members is expected toenhance the overall feelings of potency within a team. To the degree thattransformational leadership builds personal identification with a group (Shamiret al., 1993), along with a sense of confidence, a group’s level of potency andperformance is expected to be higher. Moreover, Guzzo, Yost, Campbell,and Shea (1993) argued that transformational leadership directly influencesgroup potency by boosting the confidence of team members and developingin them the belief that they will succeed.

72 GROUP & ORGANIZATION MANAGEMENT

Figure 1: Hypothesized Model of Effects of Team Leadership and Group Potency onGroup Performance

NOTE: Direct paths between leadership measures and group performance are shown in dashedlines to indicate hypothesized lack of effect.

by guest on September 27, 2016gom.sagepub.comDownloaded from

On the other hand, management-by-exception and laissez-faire leadershipbehaviors represent a style of leadership in groups in which it is likely thatmembers are more inclined to monitor each others’ mistakes or avoidaddressing problems that no one will take responsibility for in terms of struc-turing the group’s agenda, agreements, or expectations. In such situations,the group becomes more passive, less potent, and generally ineffective(Hackman, 1990; Silver & Bufanio, 1996). At best, leadership styles thatfocus on corrective actions or avoidant behaviors may have no direct effecton potency. However, at the individual level, each of these leadership styleshas been shown to lead to lower follower satisfaction and effectiveness, aswell as lower organizational performance (Lowe et al., 1996).

Although several authors have argued for a first-order relationship betweenteam leadership and efficacy and/or potency (Kozlowski, Gully, Salas, et al.,1996; Shamir et al., 1993), we could find only one empirical study that exam-ined the relationship between team leadership and potency. Kumpfer et al.(1993) examined the mediating effects of leadership in groups on team mem-ber satisfaction, efficacy, and effectiveness. They hypothesized that anempowering style of team leadership will increase team efficacy, which inturn will positively impact team effectiveness. The 65 participants in theirstudy were members of community coalitions for the prevention of alcoholand drug abuse. Kumpfer et al. (1993) reported positive relationships betweena group’s level of empowering leadership and levels of team efficacy andbetween efficacy and group effectiveness.

The first hypothesis for the current study builds directly on the model pre-sented by Klimoski and Mohammed (1994) as well as the conceptual workby Shamir et al. (1993), who have argued that transformational leadershipenhances the collective identity of a group, which in turn positively contrib-utes to a group’s potency level. Thus,

Hyptothesis 1: Transformational team leadership will be positively related to per-ceptions of group potency over time, whereas management-by-exception andlaissez-faire leadership will be negatively related to group potency.

EFFECT OF GROUP POTENCY ONLEADERSHIP AND PERFORMANCE

Although several researchers have tested the temporal relationship betweenpotency and performance (e.g., Pearce & Gallagher, 1999), we know of nostudy that has examined the reciprocal relationship between team leadershipand potency. We argued earlier that transformational team leadership wouldresult in higher levels of team potency. Teams that have higher levels ofpotency are expected to believe that what they are doing within their team is

Sivasubramaniam et al. / TEAM LEADERSHIP, GROUP POTENCY 73

by guest on September 27, 2016gom.sagepub.comDownloaded from

the “right” approach or direction. Hence, they can be expected to continue todisplay higher levels of transformational team leadership and lower levels ofmanagement-by-exception and laissez-faire leadership. Thus, our secondhypothesis is:

Hypothesis 2: Potency beliefs will be positively related to subsequent time periodtransformational leadership behaviors and negatively related to subsequentperiod management-by-exception and laissez-faire leadership behaviors.

Our second hypothesis is based on the idea that groups that have devel-oped strong potency beliefs will be encouraged to continue progroup behav-iors as exemplified by team transformational leadership. Groups with lowpotency beliefs might spiral downward, continuing to exhibit the ineffectivemanagement-by-exception and laissez-faire leadership behaviors.

Guzzo et al. (1993) identified group potency as an important cognitiveinfluence on team performance. Guzzo and his colleagues have found a posi-tive relationship between group potency and team performance (Guzzo et al.,1993; Shea & Guzzo, 1987). Several authors have also reported general sup-port for the potency-performance linkage (Cohen & Denison, 1990; Spink,1990). Campion and his colleagues (Campion et al., 1993, 1996) found thatpotency beliefs of team members significantly predicted employee satisfac-tion, managerial judgments of team effectiveness, as well as productivity. Infact, the measure of potency was the only variable that significantly predictedall of the criterion measures in both studies.

Silver and Bufanio (1996) examined the relationships among grouppotency, group goals, and task performance among 25 student groups. Theirresults showed that group potency was significantly and positively related tosubsequent task performance. They also made the case for the performance-efficacy spirals proposed by Hackman (1990) and further developed byLindsley, Brass, and Thomas (1995). Groups that started off performing wellgot even better as time passed, whereas groups that got into difficulty early onfound their problems compounded over time.

Drawing parallels from the above works, it can be argued that groups thatquickly develop shared beliefs of openness, participation, empowerment,trust, and challenging each other’s beliefs (characteristics of transformationalteam leadership) develop a strong can-do attitude (potency beliefs) that trans-lates to superior performance. Groups that start off exhibiting the transactionalmanagement-by-exception or the avoidant laissez-faire leadership behaviorsmay not develop the cohesion and trust among team members and spend con-siderable time monitoring others’ performance for errors rather than findingways to challenge the status quo.

74 GROUP & ORGANIZATION MANAGEMENT

by guest on September 27, 2016gom.sagepub.comDownloaded from

Hypothesis 3: Group potency beliefs will be positively related to subsequentgroup performance.

DIRECT AND INDIRECT EFFECTS OFLEADERSHIP ON GROUP PERFORMANCE

Hackman (1990) concluded that more effective groups developed sharedbeliefs concerning the empowerment of each other, expected levels of partic-ipation, openness to challenging each other’s perspective, a stronger sense ofmission and focus, and a higher level of trust more quickly than less effectivegroups. Vinokur-Kaplan (1995) examined the effectiveness of 15 psychiatrictreatment teams using Hackman’s (1990) model of team effectiveness, con-cluding overall effectiveness was best predicted when they included in theiranalysis feelings about how each team member influenced or led others. Howteam members influence and/or lead each other has been linked to levels ofsubsequent team potency and performance (Kumpfer et al., 1993; Salas,Dickinson, Converse, & Tannenbaum, 1992). However, evidence of thedirect impact of team leadership on group performance is quite limited. Theonly evidence for direct impact of team leadership on performance can befound in the works of Manz and Sims (1987) and Cohen et al. (1997). Theyfound that different dimensions of self-management leadership had a signifi-cant influence on work group performance. Although they conceptualizedself-management leadership as provided by an external leader, the behaviorsdescribed by them are similar to our conceptualization of team transformationaland constructive transactional leadership, albeit at a different level of analysis.

Because the development of teams represents a time-based process, weexamined how the group’s leadership and level of potency were manifestedover time. Specifically, groups that initially see themselves as more trans-formational would also exhibit higher initial levels of potency. These initiallevels of potency are expected to positively predict the group’s subsequenttransformational leadership behaviors, potency beliefs, and overall perfor-mance (Guzzo et al., 1993). Alternatively, groups characterized by management-by-exception and laissez-faire styles of leadership would fail to establish aclear set of positive shared expectations, determining each other’s contribu-tions to the group based on post hoc corrective action. Such groups are notexpected to establish a collective or shared effort and would not exhibit highlevels of potency over time. This then leads to our fourth hypothesis. Guzzoet al. argued that the impact of leadership on effectiveness could be viewedas mediated by the impact of leadership on group potency. In line withtheir argument, we expected that group potency beliefs will mediate therelationship between team leadership behaviors and group performance and

Sivasubramaniam et al. / TEAM LEADERSHIP, GROUP POTENCY 75

by guest on September 27, 2016gom.sagepub.comDownloaded from

that the direct effects of team leadership on performance will be less than theindirect effects through group potency. We present this hypothesized mediat-ing relationship in Figure 1 using a dashed line, which signifies less directimpact of team leadership on performance than the indirect effects.

Hypothesis 4: Transformational, management-by-exception, and laissez-faireleadership behaviors will not have a direct effect on group performance; rather,their effects will be mediated through level of perceived group potency.

To summarize, there have been relatively few attempts to capture whatconstitutes team leadership, except in the form of examining a specific teamleader’s influence on the team’s potency, development, and performance.What should be clear at this point in our discussion is that leadership within ateam should not imply there is no leadership of the team. Bowers and Sea-shore (1966) began their discussion of leadership within groups by suggest-ing that effective leaders set the pattern or standard for mutual leadership,which followers then supply each other. Bradford and Cohen (1984) used theconcept of balance in their discussion of shared leadership. Balance is whenmembers complement each other’s competencies, adding to the overallpotency of the group or team. We have attempted to extend these earlier argu-ments by operationalizing team leadership using leadership constructs thathave received a substantial amount of attention at the individual level of anal-ysis. Moreover, we intend to pick up where Avolio et al. (1996) left off, sug-gesting the next logical step in assessing team leadership was “to examine thereciprocal relationship between measures of team leadership, potency andperformance” (p. 201).

METHOD

THE SETTING

The study was conducted at a large public university in the NortheasternUnited States. Subjects were undergraduate students majoring in businessand enrolled in two sections of an introductory organizational behaviorcourse. At the beginning of the semester, students were randomly assigned tofour- or five-person groups to form 42 groups of mixed gender and ethnicityto complete a series of group assignments throughout the semester.

A total of 182 students participated in phase I of the current study. Of thetotal number of students, there were 155 who completed both phases.Because this was a longitudinal study of the effects of team leadership on

76 GROUP & ORGANIZATION MANAGEMENT

by guest on September 27, 2016gom.sagepub.comDownloaded from

team performance, we used the data from those 155 students who completedboth phases of the study. We excluded one more respondent from furtherconsideration as the person was the lone respondent from that group. Thefinal sample had 154 students representing 41 groups. The mean age of thesample was 21, and 48% were females.

RESEARCH PROCEDURES

Participants completed all survey measures individually during class timeroughly 3 weeks into the semester (Time 1) and 10 weeks after the initialassessment (Time 2). Groups interacted on a weekly basis to complete caseassignments and experiential exercises during the first 3 weeks of the semes-ter. They were also required to turn in the first of several assignments beforethe first phase of the survey assessment (Time 1) was completed. None of thegroups received any formal feedback on their first assignment before com-pleting the survey instruments. The second phase of data collection was com-pleted 2 weeks before the end of the semester (Time 2) and a week before thegroups’ final project was due. Students were not given their grades for groupprojects until after the last day of classes. Although not monitored systemati-cally, all of the groups met regularly outside class hours to discuss and com-plete the assigned projects.

MEASURES

Ratings of team leadership were collected using the Team Multi-FactorLeadership Questionnaire (TMLQ) developed by Bass and Avolio (1994).Avolio, Sivasubramaniam, Murry, and Jung (1999) assessed the psychometricproperties and dimensionality of the TMLQ in three independent samplesusing traditional and confirmatory factor analytic procedures. Their resultsindicated that five factors were reliably measured with respect to assessingteam leadership behavior. The first three scales, inspirational motivation,intellectual stimulation, and individual consideration, represented team trans-formational leadership. Also, the individualized consideration scale includeditems that represented team members providing recognition and rewards to eachother for effective performance. The fourth factor represented management-by-exception leadership, and the fifth dimension was avoidant/laissez-faireleadership. Avolio et al. also found evidence to conceptualize the five-factormodel as a hierarchical model, with the three transformational scales rep-resented by a single higher order factor. Due to our sample size limitationsand the high intercorrelations between the three transformational leader-ship scales, we aggregated the three scales to form a general scale labeled

Sivasubramaniam et al. / TEAM LEADERSHIP, GROUP POTENCY 77

by guest on September 27, 2016gom.sagepub.comDownloaded from

“transformational leadership.” Our strategy of representing the threetransformational factors by a single factor is consistent with the approachused by other researchers (Wofford, Goodwin, & Whittington, 1998; Yammarino,Dubinsky, & Spangler, 1998).

In all, there were 14 items in the transformational team leadership scale.The transformational scale reliability was .91 in Time 1 and .89 in Time 2.Management-by-exception team leadership was assessed by 4 items. Thescale reliability was .61 in both time periods. Laissez-faire leadership wasassessed by 5 items and had acceptable scale reliabilities in both time periods.In Time 1 it was .75, and in Time 2 it was .87. Group potency was assessed byusing the 8-item scale developed by Guzzo and his colleagues (Guzzo et al.,1993). This instrument has been used widely and has been found to haveacceptable measurement properties (Kirkman & Rosen, 1999). The scalereliability was .91 for Time 1 and .92 for Time 2. Mean scores for these fourscales were used as the indicators of the respective constructs. To minimizeany confounding effects due to practice or order, item order was scrambledfor the two administrations of the survey. The time between the two surveyadministrations was approximately 10 weeks.

Table 1 shows sample items for each of the survey scales. All team leader-ship items were measured using 5-point responses scales: 0 (not at all); 1(once in awhile); 2 (sometimes); 3 (fairly often); 4 (frequently, if not always).Group potency items were measured using a 5-point rating scale: 0 (stronglydisagree) through 4 (strongly agree).

Measurement of team performance has always been a problem in priorresearch. Use of perceptual measures as indicators of performance obviouslyincreases the possibility of monomethod bias when assessing relationshipsamong survey measures of leadership, potency, and performance. Usingsuch perceptual measures makes it difficult to separate the effects of leader-ship on performance from the confounding effects of attribution bias thatmay be associated with using a paper-and-pencil measure of performance(e.g., “I rate the leadership of my team as being outstanding, and so my rat-ings of performance should also be up to the standards of leadership for myteam”). To avoid such problems, we used each group’s grade assigned by therespective instructors in the two participating classes as a measure of perfor-mance. The final group grades were determined at the end of the semester,well after the second phase of data collection was completed. To ensure com-parability across the two sections, we standardized the grades within eachsection. We then used z scores as the standardized measure of performancefor all subsequent analyses. The instructors were unaware of the results of thesurvey data prior to allocating grades for the group projects.

78 GROUP & ORGANIZATION MANAGEMENT

by guest on September 27, 2016gom.sagepub.comDownloaded from

ANALYSES

Before aggregating the survey data collected from the two class sections,we first tested whether the variance-covariance matrices for all survey mea-sures were equivalent across the two class sections using Box’s M test forhomogeneity of variance. This is necessary to ensure the relationships amongthe variables were not moderated by the differences in the two classes or theinstructors. The χ2 statistic was insignificant (χ2 = 65.16, df = 55, p < .17),indicating the pattern of interrelationships among the survey measures for thetwo sections were equivalent. Based on these results, we combined the twodata sets and performed all subsequent analyses without considering the sec-tion participants were drawn from to participate in this study.

All of the constructs examined in this study were conceptualized, assessed,and analyzed at the group level. Chan (1998) and Mathieu (1991) providedjustification for empirically escalating individual perceptions to a higherlevel of analysis once conceptual arguments have been formulated. Chan(1998) developed a compositional model to guide empirical tests for aggre-gation. He argued that an individual’s perceptions will form a functional rela-tionship between constructs at different levels of analysis and that processesassociated with this functional relationship define this higher level of analy-sis. Mathieu (1991) concurred that higher level constructs exert an indirectinfluence on individual perceptions.

Sivasubramaniam et al. / TEAM LEADERSHIP, GROUP POTENCY 79

TABLE 1

Sample Items in Team Multi-Factor LeadershipQuestionnaire and Group Potency

Scale Item

Transformationalleadership

Members of my team talk about how trusting each other can help over-come their difficulties

Members of my team envision exciting new possibilitiesMembers of my team emphasize the importance of being committed to

our beliefsMembers of my team work out agreements about what’s expected

from each otherManagement by

exceptionMembers of my team closely monitor each other’s performance for

errorsMembers of my team direct attention toward failure to meet standards

Laissez-faire Members of my team avoid addressing problemsMembers of my team wait until things have gone wrong before taking

actionGroup potency Our group believes it can be very productive

by guest on September 27, 2016gom.sagepub.comDownloaded from

To empirically test the aggregation of our constructs to the group level, weused James, Demaree, and Wolf’s (1984) procedure for estimating within-group interrater agreement. The James et al. (1984) rwg conforms to therequirements of Chan’s third compositional model, the “referent-shift con-sensus model.” Within-group consensus is used to justify aggregation ofindividuals’ perceptions of collective leadership to represent the value of thehigher level construct, namely, team leadership. The mean rwg estimates forall survey measures across the two time periods were well above the accept-able levels of .7 suggested by James et al. (1984) to support our conceptual-ization of these constructs. The mean rwg for leadership scales rangedbetween .75 and .87, and rwg for potency was .87 in Time 1 and .85 in Time 2.

The hypothesized longitudinal model was tested using AMOS 3.61(Arbuckle, 1997). Because we were using scale means instead of individualitems as indicators of leadership and potency, we used the proceduresdescribed by Jöreskog and Sörbom (1989) to specify the latent variablecausal model. Specifically, factor loadings were set equal to the square rootof the reliability of each scale score, and corresponding error variance valueswere fixed to be equal to one minus the scale reliability times the variance ofthe scale score. Scale reliabilities for transformational leadership, manage-ment by exception, laissez-faire, and potency for both time periods werecomputed and used to specify the structural equation models. Reliability ofgroup performance (instructor-assigned grade standardized by section) wasarbitrarily set at .95.

Fit indices suggested by Arbuckle (1997), Medsker, Williams, and Holohan(1994), and Jöreskog and Sörbom (1989) were used to evaluate the adequacyof the model fit. They were the χ2 test, goodness-of-fit index, ComparativeFit Index (CFI), and χ2 divided by degrees of freedom. In addition, we alsoexamined the root mean square error of approximation (RMSEA) suggestedby Browne and Cudeck (1993), as well as modelR2 for practical significance.

Because our sample size may have been too small to derive robust param-eter estimates using the maximum likelihood estimation procedures, we useda two-pronged strategy to ensure parameter estimates were robust. First, wecarefully examined the sample distributions for any departures from multivariatenormality. We used the Kolmogrov-Smirnov (K-S) test to verify our assump-tion that all measures were normally distributed. The K-S tests were insignif-icant, indicating the distribution patterns for our measures did not exhibit anynoticeable departure from normality assumptions. Examination of outliers inthe sample using the Mahalanobis distance measure from the centroidrevealed that one group was significantly away from the mean response pat-terns. Exclusion of this group from the sample did not change our estimates,

80 GROUP & ORGANIZATION MANAGEMENT

by guest on September 27, 2016gom.sagepub.comDownloaded from

minimizing our concerns about the effects of sample size on our results. Wealso performed a jackknife procedure in which we estimated the modelparameters after excluding one case from the sample. Mean estimates of theparameters from the jackknife procedure were compared with the originalestimates for the full sample to reveal any biases in our estimates.

CONSIDERING TEMPORAL STABILITY

Because the study is based on panel data, time data assumptions and prop-erties need to be incorporated in both the conceptual model as well as the ana-lytical procedures (Bergh, 1993). To consider the time effects specificallywithin the structural equation model, we hypothesized significant effects foreach of the four variables measured over the two time periods. Specifically,we modeled the relationship between Time 1 and Time 2 measures of trans-formational leadership, management by exception, laissez-faire, and grouppotency. Significant parameter estimates would indicate that each measurehas some stable characteristics over time. Furthermore, the inclusion of thesepaths in the structural equation model means the effects of one variable, suchas potency, on a target variable such as leadership, is actually related to thechanges in the target variable over time. For instance, if we ignore the stabil-ity of transformational leadership behaviors over time, then interpreting themediating effects of potency on group performance becomes problematicdue to violations of assumptions for longitudinal designs. Specifically, ourpredictions of performance may be affected by misspecification of the longi-tudinal model. By incorporating time-based estimates in our model, we canalso draw substantive conclusions based on the significance of these parame-ter estimates. For instance, if the stability coefficients were close to 1, thenwe can safely say that certain behaviors are evident at a very early stage in thegroup’s development, and delayed interventions may have difficulty achiev-ing the desired impact over time.

RESULTS

Descriptive statistics and a correlation matrix for all variables included inthis study are presented in Table 2. The pattern of correlations was similar inboth Time 1 and Time 2 and between these measures and group performance.In general, team leadership behaviors were stable as indicated by the meanscores for each leadership scale. Significant bivariate relationships betweenteam leadership, potency, and performance indicated that transformationalleadership and group potency, assessed as early as 3 weeks into the semester,

Sivasubramaniam et al. / TEAM LEADERSHIP, GROUP POTENCY 81

by guest on September 27, 2016gom.sagepub.comDownloaded from

TABLE 2

Descriptive Statistics and Correlations of Study (N = 41)

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

Time 11. Transformational leadership 2.54 0.41 .912. Management by exception 1.58 0.41 .20 .613. Laissez-faire 0.77 0.36 –.47 .30 .754. Group potency 2.77 0.43 .80 –.07 –.61 .91

Time 25. Transformational leadership 2.47 0.42 .66 .14 –.47 .44 .896. Management by exception 1.54 0.39 –.18 .33 .38 –.23 –.05 .617. Laissez-faire 0.97 0.57 –.44 .04 .60 –.44 –.61 .16 .878. Group potency 2.78 0.49 .68 .16 –.52 .61 .80 –.03 –.65 .92

9. Group performancea 0 1.00 .34 .17 –.30 .42 .45 .07 –.46 .54 .95

NOTE: Absolute values of correlations greater than .30 are significant at p < .05; values greater than .37 are significant at p < .01. Italicized items are scalereliabilities.a. Group performance is standardized within each section.

82

by guest on Septem

ber 27, 2016gom

.sagepub.comD

ownloaded from

were each significant predictors of group performance at the end of thesemester. Group size, which varied between three to five, was not signifi-cantly related to any of the variables. Results of the AMOS analysis reportedbelow confirm these general observations.

TEST RESULTS OF MODELS

Table 3 summarizes the results of our analyses comparing all three modelsspecified earlier. The chi-square statistic testing the full model was signifi-cant (χ2 = 30.71, df = 19, p < .022), indicating the model did not fit the dataadequately. However, other fit indices were generally more supportive of thefull model. The CFI was well above .9, and the χ2/df ratio was below the 2 to 5range suggested by Marsh and Hocevar (1985) as indicative of adequate fit.RMSEA was .124 and above the .08 cutoff recommended by Browne andCudeck (1993). However, the estimated RMSEA was not statistically greaterthan .05 (p < .084). These results indicate the full model fits the data reason-ably well. The jackknife estimates were robust, and the bias in jackknife esti-mates was close to zero and insignificant. We computed the critical ratio foreach parameter using the jackknife estimates to assess the significance ofeach parameter.

TESTS OF HYPOTHESES

Effects of leadership on potency. Parameter estimates for the full modelare presented in Figure 2. Only transformational leadership assessed in Time1 had a significant positive effect on potency (γ = 1.009, p < .01). Although asimilar pattern was observed for the relationship between transformationalleadership and potency in Time 2, laissez-faire had a significant negativeimpact on group potency only in Time 2. The absolute effect sizes differedacross the two time periods. Transformational leadership in Time 2 had amuch lower impact on potency as compared to Time 1 results. These resultsprovide support for our hypothesis that transformational leadership would bepositively associated with a contemporaneous measure of potency. However,we did not find support for our hypothesis that management-by-exceptionteam leadership would be negatively associated with same-period potency.

Effects of potency on leadership. Potency beliefs assessed in Time 1 hadno effect on management-by-exception and laissez-faire leadership assessedin Time 2 beyond any variance explained by temporal stability. Contrary toexpectations, potency beliefs developed in Time 1 had a significant and

Sivasubramaniam et al. / TEAM LEADERSHIP, GROUP POTENCY 83

by guest on September 27, 2016gom.sagepub.comDownloaded from

TABLE 3

Estimates of Path Coefficients of Three Study Models

Maximum Likelihood Estimatesof the Model Parametersa Jackknife Estimatesb

Parameter Standardized Unstandardized Unstandardized SE Biasc Critical Ratio

TFL1 → Potency1 1.009 1.078 1.083 .340 .005 3.19**MBEA1 → Potency1 –0.312 –0.329 –0.331 .281 –.002 –1.18LF1 → Potency1 0.096 0.116 0.122 .421 .006 0.29TFL1 → TFL2 1.320 1.381 1.386 .390 .005 3.56***MBEA1 → MBEA2 0.502 0.479 0.483 .225 .004 2.14*LF1 → LF2 0.782 1.242 1.248 .440 .006 2.84**Potency1 → TFL2 –0.660 –0.647 –0.651 .362 –.004 –1.80*Potency1 → MBEA2 –0.304 –0.275 –0.275 .182 .000 –1.51Potency1 → LF2 0.078 0.102 0.106 .337 .004 0.31Potency1 → Potency2 0.313 0.339 0.340 .145 .001 2.35*TFL2 → Potency2 0.589 0.652 0.653 .132 .001 4.96***MBEA2 → Potency2 0.182 0.218 0.221 .158 .003 1.40LF2 → Potency2 –0.286 –0.236 –0.237 .094 –.001 –2.53**Potency2 → Performance 0.574 1.180 1.181 .297 .001 3.98***χ2 30.708 30.730 .022

NOTE: TFL1 = transformational leadership, Time 1; TFL2 = transformational leadership, Time 2; MBEA1 = management by exception, Time 1; MBEA2 = man-agement by exception, Time 2; LF1 = laissez-faire, Time 1; LF2 = laissez-faire, Time 2; Potency1 = Time 1; Potency2 = Time 2.a. N = 41; model degrees of freedom = 19.b. Means of the estimates from 41 samples using the jackknife procedure; sample size for each jackknife sample = 40.c. The difference between original maximum likelihood estimates and jackknife estimates.*p < .05. **p < .01. ***p < .001.

84

by guest on Septem

ber 27, 2016gom

.sagepub.comD

ownloaded from

negative impact on transformational leadership in Time 2 (β = –.66, p < .05).Hence, our second hypothesis was not supported.

Effects of potency on performance. Group potency in Time 2 significantlyand positively predicted standardized group performance (β = .574, p <.001). The full model explained nearly 33% of the variance in group perfor-mance, providing support for our third hypothesis.

Direct versus indirect effects of leadership. To test whether the mediationalor direct effects model was a better representation of the data, we fitted amodified full model with the direct paths from leadership to performanceestimated. We estimated the direct effects of each leadership factor sepa-rately. This resulted in dropping an additional degree of freedom as com-pared to the original full model. Model statistics did not improve over themore parsimonious original model in any of the three direct effects models(transformational: χ2 = 30.06; management by exception: χ2 = 30.46;laissez-faire: χ2 = 30.29). As expected, the difference in model fit was notsignificant, indicating the more parsimonious original model provides asgood a fit as the revised direct effects model. Moreover, the parameter esti-mate for the direct paths from team leadership variables to group performancewas not significant, whereas the path from group potency to performancewas significant. We then computed estimates of the indirect effects of leader-ship on performance. Indirect effects of transformational leadership and

Sivasubramaniam et al. / TEAM LEADERSHIP, GROUP POTENCY 85

Figure 2: LongitudinalModel of Effects of TeamLeadership andGroup Potency in Time1 and Time 2 on Group Performance

NOTE: CFI = Comparative Fit Index; RMSEA = root mean square of approximation.

by guest on September 27, 2016gom.sagepub.comDownloaded from

laissez-faire assessed in both time periods were significant, whereas manage-ment by exception did not explain any additional variance in group potencyor performance beyond that explained by other leadership variables. Theindirect effects of Time 1 transformational leadership and laissez-faire ongroup performance were .859 (p < .003) and –.351 (p < .05), respectively. InTime 2, the indirect effects of transformational leadership and laissez-fairewere .769 (p < .002) and –.275 (p < .05), respectively. These results providestrong support for our fourth hypothesis.

Temporal stability. The stability coefficients across time were all positiveand significant even though the magnitude of stability varied across the fourmeasures. Transformational leadership ratings were more than twice as sta-ble as management-by-exception and laissez-faire leadership ratings. Itappears that early ratings of team transformational, management-by-exceptionand laissez-faire leadership remained quite stable over the 3-month period oftime (γ = 1.320, p < .001; γ = .502, p < .05; γ = .782, p < .01, respectively).

To assess effects of team leadership behaviors at a very early stage in agroup’s development, we estimated the indirect effects of leadership vari-ables measured in Time 1 on group performance. As indicated above, theindirect effects of transformational and laissez-faire on final performancewere significant. These results eliminate possible concerns that awareness ofhow well the group performed in the assigned projects may explain the sig-nificant findings of this study. At the time of the first survey administration,the groups had no performance information as they had not even begun work-ing on their final projects. Our results suggest that perceptions of team leader-ship early on has a significant indirect impact on final outcomes throughmediating variables such as group potency and continued leadership behaviors.

SUMMARY AND DISCUSSION

The current study represents one of the first attempts to examine how teamleadership predicts levels of group potency and performance over time. Asnoted earlier, although some attention has been given to leadership of teams,there has been relatively little attention given to leadership within or byteams. Our goal was to begin moving work that has been completed on theindividual leadership of teams to the group or team level.

The basis for our hypotheses came from two sources. First, there has beenan extensive amount of work on transformational leadership conducted withthe individual leader as the unit of analysis. This body of research has

86 GROUP & ORGANIZATION MANAGEMENT

by guest on September 27, 2016gom.sagepub.comDownloaded from

generally shown that leaders who are rated more transformational have fol-lowers and units that perform at higher levels as compared to leaders whoexhibit management-by-exception and/or avoidant/laissez-faire behaviors(Bass, 1997; Lowe et al., 1996). Second, we utilized the model of team effec-tiveness offered by Klimoski and Mohammed (1994) as a basis for formulat-ing our hypotheses and to set up the causal model tested in the current study.We also supplemented our theoretical framework with work completed byShamir et al. (1993) and Avolio et al. (1996) to further substantiate our pre-diction that transformationally led teams would see themselves as morepotent and ultimately would achieve a higher level of group performanceover time.

SUMMARY OF OUR FINDINGS

Results of the current study largely confirmed our hypotheses. The fullmodel presented in Figure 1 accounted for nearly 33% of the variance ingroup performance. Our first hypothesis, that team leadership will signifi-cantly influence group potency beliefs, was supported. Team transformationalleadership positively related to potency in both time periods. Contrary to ourHypothesis 2 prediction, that potency would positively affect team trans-formational leadership, we found that potency beliefs in Time 1 were signifi-cantly and negatively related to team transformational leadership in Time 2.Our third hypothesis, that group potency would predict group performance,was strongly supported. We also found strong support for our fourth hypoth-esis, which predicted the relationship between team leadership and perfor-mance would be mediated by the group’s potency beliefs. The direct effectsof team leadership style on performance were not significant, whereas theindirect effects were significant. Moreover, the direct effects model had aless optimal fit than the more parsimonious, indirect effects model.

One interesting pattern in the results concerned the stability of leadershipand potency ratings observed over time. The pattern that emerged indicatedthat groups that saw themselves as more transformational at Time 1 also sawthemselves as more transformational and potent at Time 2. These early rat-ings of team transformational leadership and potency predicted higher levelsof group performance. Conversely, groups that started out as more laissez-faire in the first few weeks of operation appeared to remain so over time,showing a negative impact on subsequent levels of both potency and groupperformance. Also, the more effective team transformational leadershipbehaviors were more stable than potency beliefs. Initial perceptions of teamtransformational leadership explained nearly 44% of the variance in subsequent

Sivasubramaniam et al. / TEAM LEADERSHIP, GROUP POTENCY 87

by guest on September 27, 2016gom.sagepub.comDownloaded from

ratings of transformational leadership even though the two evaluations weregathered 10 weeks apart. These findings seem to indicate that the “good”teams got even better, and “poor” teams became even worse, over time.

Our results parallel findings that have been reported for studies focusingon individual leadership styles. Our findings indicated that transformationalleadership can occur in teams and it can impact group potency and perfor-mance in ways that parallel individual leader and follower interactions.Indeed, it appears, as has been suggested by Kozlowski, Gully, McHugh, etal. (1996), Stewart and Manz (1994), Manz and Sims (1987), and Shamiret al. (1993), that team or collective leadership is at least one factor in predict-ing the subsequent effectiveness of teams. Our results also confirm somerecent findings pertaining to the effects of leadership in teams reported byCohen et al. (1997), Vinokur-Kaplan (1995), Kumpfer et al. (1993), andSalas et al. (1992). How team members evaluated the collective influenceearly on in the team’s development did affect their subsequent perceptions ofpotency and the team’s performance on the group project.

EFFECT OF POTENCY BELIEFS ON RATINGS OFTRANSFORMATIONAL LEADERSHIP

We had hypothesized that group potency at Time 1 would have a positiveimpact on team transformational leadership at Time 2 but reported the oppo-site effect. Our results suggest that teams that developed a can-do attitude atTime 1 assessments may have suffered a letdown at Time 2. Although theseresults were at first surprising, we now believe they are supported by anec-dotal reports on other teams in the literature. For example, early success forsports teams that have not developed mature systems of norms and controlsfor group behaviors can result in an initial overestimation of their true capa-bilities and/or potency. Members of such teams might develop feelings ofinvincibility during a honeymoon period and discontinue engaging in moreeffective leadership behaviors in favor of proselytizing behaviors. Over time,this could result in a reappraisal of team behaviors and a loss in confidence,perhaps resulting in a realization that they were not as potent or transformationalas they might have first believed. We believe this provides a possible expla-nation for our results, yet future research needs to replicate these findings andexamine this pattern more systematically. For instance, researchers can col-lect qualitative data on how team members describe their team early on ver-sus at later stages of development. In addition, the effects observed herecould be examined experimentally when teams are given bogus feedbackabout how well or poorly they have performed initially. Using these approachesmay help explain the unexpected result obtained in this study.

88 GROUP & ORGANIZATION MANAGEMENT

by guest on September 27, 2016gom.sagepub.comDownloaded from

NEXT STEPS IN RESEARCH ON TEAM LEADERSHIP

There are a number of interesting avenues that can be pursued in subse-quent research on team leadership. First, the results here need to be replicatedon other samples that have a more diverse membership with respect to age,ethnicity, and work experience. Second, the measures of team leadershipused here must be further refined, with specific emphasis on measuring themore transactional components of team leadership. Although the correctiveaspects of transactional leadership were measured as a separate scale in thisstudy, the contingent reward and recognition aspect of transactional leader-ship that defines roles and expectations within a team was not measured as aseparate scale. Further work on scale development by Avolio et al. (1999)identified a new dimension of team leadership behavior that combined con-sideration shown to members with the structure set up for member interac-tions. However, they did not identify contingent reward and recognition as aseparate and distinct factor.

Third, evidence provided here indicates that some groups will “naturally”start out and evolve as more transformational, whereas others will be moretransactional from the start and become more so over time. This is potentiallya very significant issue for the timing of training interventions for team lead-ership development. Specifically, there may be some optimal band of time inwhich one can intervene to provide the necessary direction and structure for agroup to get it on track toward developing its full team leadership potential.Here, groups were asked to function on their own without any external super-vision or designated leader. A number of authors have suggested that it maybe appropriate for some groups to begin with more external supervision,allowing the team to take more responsibility for its own actions and deci-sions over time (Cohen et al., 1997; Kozlowski, Gully, Salas, et al., 1996;Manz & Sims, 1987). Results of the current study indicate this may be a par-ticularly useful strategy for groups that are becoming increasingly too correc-tive or avoidant over time.

One key practical question is, When may it be too late to intervene torecover a group that is accelerating toward ineffective team leadership? Inthe present study, by the third week, there was a very strong sense of therebeing a laissez-faire or management-by-exception style of leadership ingroups, which predicted the same style 10 weeks later. So, if only based onthe current study, we would recommend a very early intervention, particu-larly with groups that have members who are more likely to be transactional(“What’s in it for me?”). As far as we know, there has not been any systematicresearch undertaken to determine the optimal time span for intervention withrespect to developing teams and team leadership. With the rapid shift to

Sivasubramaniam et al. / TEAM LEADERSHIP, GROUP POTENCY 89

by guest on September 27, 2016gom.sagepub.comDownloaded from

team-based systems, this would appear to be a very practical area in which toconduct future research.

Related to the optimal time span for intervention, we must also to deter-mine how feedback can be used to improve team leadership development andperformance. Prussia and Kinicki (1996) reported that a group’s perceptionof itself, such as in terms of collective efficacy, mediated the impact of per-formance feedback on group effectiveness. It would be interesting for futureresearch to explore how the content and timing of feedback could impact theacceleration of team development and effectiveness.3 Again, early interven-tions with respect to feedback coupled with the appropriate content may helpto accelerate a team’s development in the desired direction.

Another area for potential future research is to tease out the actual patternsof behavior that result in groups becoming teams and teams becoming moretransformational in the collective sense described here. For example, to buildtrust, the group may first establish expectations via contracts and agreementswith its members. Over time, as members fulfill their obligations to agree-ments, they may begin to exhibit higher levels of trust in each other. Thespeed at which this occurs will likely depend on the nature of the group and itscharacteristics. More heterogeneous groups with little history of beingtogether may take longer to work through agreements and to build trust witheach other. These groups may also require more external leadership initiallyto launch the group but, over time, may establish enough trust to be self-sufficient with leadership within the group.

At present, we do not know whether the members of groups that weremore transformational had higher levels of identification and commitment totheir work as Shamir et al. (1993) would predict. Future research needs toexamine how groups that described themselves as more transformationalbuilt trust, identification, and commitment into their tasks. For instance, didone member make a difference in getting the group to identify with its work?Was it a collection of members that each made the difference in terms of iden-tification? Or, in different groups, did both occur and have a similar impacton the group’s level of development with respect to transformational leader-ship, identification, and higher levels of performance? It seems safe to saythat, at the present time, we know very little about these dynamics in teamformation.

One final area to explore, which was only indirectly examined here, wasthe team’s mental or shared model. For example, do teams that see them-selves as more potent and transformational have models of themselves thatare more coherent and shared? Conversely, do groups that see themselves astransactional have less coherent mental models of themselves? We believe

90 GROUP & ORGANIZATION MANAGEMENT

by guest on September 27, 2016gom.sagepub.comDownloaded from

that by identifying the models shared by group members, we will also gainmore insight into the identification processes in teams and their effects onsubsequent performance.

LIMITATIONS

There are a number of limitations of the current study that must beaddressed in future research on team leadership. First, the sample character-istics are not necessarily similar to the typical work group in terms of age,work experience, and prior exposure to teamwork. Second, subjects werepooled from a convenience sample and therefore there may be characteristicsof the sample that may have biased the obtained effects upward or downward.Third, only perceptions of team leadership were collected from team mem-bers. It would be useful to have an outside view of the team’s leadership toconfirm the evaluations made by members on the team. Fourth, the collectionof potency and leadership ratings within the same time period may haveaffected the relationship observed between these two measures, likely inflat-ing it due to common source effects. Fifth, although the groups did notreceive any formal feedback on how they performed on the final projectbefore all survey ratings were obtained, they did have performance feedbackon the miniprojects that groups worked on during the course of the semester.It is possible their ratings of leadership and potency in Time 2 may have beenbiased to conform to the performance cues they had received from instruc-tors. However, evaluations of team leadership at Time 1, when the studentgroups had not received any feedback, predicted group performance viagroup potency. This provides some support for the model of team effective-ness tested in the current study. Sixth, the measures of team leadership werelimited to a set of constructs contained in the multifactor model of leadership.Extending measures of team leadership to other constructs seems well worthpursuing in future research.

Considering these limitations, this is one of the first studies to examine theimpact of leadership within or by teams on independently collected measuresof performance and to do so over time. One advantage of beginning suchresearch with students with relatively little work experience is that we mini-mize the potential problem of some members assigned to the team havingsubstantially more experience working in teams then others. In addition, stu-dent teams here were self-managed, as the instructors’ primary role was toassign them to groups. Thus, the current study can provide a preliminarybaseline from which we can examine how more heterogeneous groups func-tion over time.

Sivasubramaniam et al. / TEAM LEADERSHIP, GROUP POTENCY 91

by guest on September 27, 2016gom.sagepub.comDownloaded from

PRACTICAL IMPLICATIONS AND CONCLUSIONS

The current study reaffirms the importance of examining the effects ofteam leadership on subsequent group performance. Moreover, the findingsindicated that if some groups were simply left alone, they would become lesseffective over time and that “good” leadership does not necessarily spontane-ously emerge in those groups. Based on our results, there is some reason tobelieve that appropriate and early leadership interventions could have a sig-nificant and positive impact on subsequent group/team performance.

An issue that we have not addressed in this study is the selection of mem-bers into the group. The characteristics of members selected for teamworkcan obviously affect the type of development strategy one selects to workwith a specific team and type of leadership that may be more effective at theonset of the team’s formation.

What appears to be clear from this study’s results is that the characteristicsdiscussed with respect to high-performing teams are the same characteristicsthat have often been attributed to exemplary leaders in organizations whoinspire followers to achieve the highest levels of commitment and perfor-mance. It is these transformational characteristics that Burns (1978) firstidentified for individual leaders that may apply equally well to describing thecollective leadership of high-performing teams.

NOTES

1. The authors have used the terms team and group interchangeably even though weacknowledge that teams are much more than groups in terms of their stage of development, iden-tity, trust, interdependence, and empowerment.

2. Potency, as conceptualized in this study, is in line with the description provided by Guzzo,Yost, Campbell, and Shea (1993) and is different from a related construct, collective efficacy.Although potency refers to a generalized belief about team performance, collective efficacyrefers to individual beliefs about the group’s ability to perform a specific task. Because teamsstudied here were asked to perform several assignments and not just one assignment, we felt thatthe more generalized potency beliefs was a more appropriate representation of team memberconfidence than the more task-specific collective efficacy.

3. To examine the effects of performance feedback on group performance, we usedself-reported measure of group effectiveness as an intervening variable between Time 1 andTime 2. We found this interim performance measure was not significantly related to final perfor-mance and it did not mediate the effects of Time 1 potency on Time 2 leadership measures. Webelieve that our potency measure may have captured any available performance informationand, hence, chose to exclude the self-reported measure of effectiveness from our model.

92 GROUP & ORGANIZATION MANAGEMENT

by guest on September 27, 2016gom.sagepub.comDownloaded from

REFERENCES

Arbuckle, J. L. (1997). AMOS users’ guide: Version 3.6. Chicago: Small Waters.Astin, H. S., & Astin, A. W. (1996). A social change model of leadership development guide-

book, version 3. Los Angeles: University of California, Los Angeles.Avolio, B. J., & Bass, B. M. (1995). Individual consideration viewed at multiple levels of analy-

sis: A multi-level framework for examining the diffusion of transformational leadership.Leadership Quarterly, 6, 199-218.

Avolio, B. J., Jung, D. I., Murry, W. D., & Sivasubramaniam, N. (1996). Building highly devel-oped teams: Focusing on shared leadership processes, efficacy, trust, and performance. InM. M. Beyerlein, D. A. Johnson, & S. T. Beyerlein (Eds.), Advances in interdisciplinarystudies of work teams (pp. 173-209). Greenwich, CT: JAI.

Avolio, B. J., Sivasubramaniam, N., Murry, W. D., & Jung, D. I. (1999). A preliminary valida-tion of the Team Multifactor Leadership Questionnaire. Manuscript in preparation, BinghamtonUniversity.

Bales, R. F. (1954). In conference. Harvard Business Review, 32, 44-50.Bandura, A. (1986). Social foundations of thought and action: A social cognitive theory.

Englewood Cliffs, NJ: Prentice Hall.Bass, B. M. (1997). Does the transactional-transformational leadership paradigm transcend

organizational and national boundaries? American Psychologist, 52, 130-139.Bass, B. M. (1998). Transformational leadership: Industrial, military and educational impact.

Hillsdale, NJ: Lawrence Erlbaum.Bass, B. M., & Avolio, B. J. (1994). Improving organizational effectiveness through transformational

leadership. Thousand Oaks, CA: Sage.Bergh, D. D. (1993). Watch the time carefully: The use and misuse of time effects in manage-

ment research. Journal of Management, 19, 683-705.Bettenhausen, K., & Murnighan, J. K. (1985). The emergence of norms in competitive decision

making groups. Administrative Science Quarterly, 30, 350-372.Bowers, D. G., & Seashore, S. E. (1966). Predicting organizational effectiveness with a four-

factor theory of leadership. Administrative Science Quarterly, 11, 238-263.Bradford, D. L., & Cohen, J. (1984).Managing for excellence: The guide to developing high per-

formance organizations. New York: John Wiley.Browne, M. W., & Cudeck, R. (1993). Alternative ways of assessing model fit. In K. A. Long &

K. A. Bollen (Eds.), Testing structural equation models. Newbury Park, CA: Sage.Burns, J. M. (1978). Leadership. New York: Free Press.Campion, M. A., Medsker, G. J., & Higgs, A. C. (1993). Relations between work group charac-

teristics and effectiveness: Implications for designing effective work groups.Personnel Psy-chology, 46, 823-847.

Campion, M. A., Papper, E. M., & Medsker, G. J. (1996). Relations between work team charac-teristics and effectiveness: A replication and extension.Personnel Psychology,49, 429-452.

Cannon-Bowers, J. A., & Salas, E. (1990, May). Cognitive psychology and team training:Shared mental models in complex systems. Paper presented at the annual conference of theSociety for Industrial and Organizational Psychologists, Miami, FL.

Chan, D. (1998). Functional relations among constructs in the same content domain at differentlevels of analysis: A typology of composition models. Journal of Applied Psychology, 83,234-226.

Sivasubramaniam et al. / TEAM LEADERSHIP, GROUP POTENCY 93

by guest on September 27, 2016gom.sagepub.comDownloaded from

Cohen, S. G., & Bailey, D. E. (1997). What makes teams work: Group effectiveness researchfrom the shop floor to the executive suite. Journal of Management, 23, 239-290.

Cohen, S. G., Chang, L., & Ledford, G. E. (1997). A hierarchical construct of self managementleadership and its relationship to quality of work life and perceived group effectiveness.Per-sonnel Psychology, 50, 275-308.

Cohen, S. G., & Denison, D. R. (1990). Flight attendant teams. In J. R. Hackman (Ed.), Groupsthat work (and those that don’t): Creating conditions for effective teamwork (pp. 382-397).San Francisco: Jossey-Bass.

Cohen, S. G., Ledford, G. E., & Spreitzer, G. M. (1996). A predictive model of self-managingwork team effectiveness. Human Relations, 49, 643-676.

Dunphy, D., & Bryant, B. (1996). Teams: Panaceas or prescriptions for improved performance?Human Relations, 49, 677-699.

George, J. (1990). Personality, affect, and behavior in groups. Journal of Applied Psychology,75, 107-116.

George, J., & Bettenhausen, K. (1991). Understanding prosocial behavior, sales performance,and turnover: A group-level analysis in a service context. Journal of AppliedPsychology, 75,698-709.

Gladstein, D. (1984). Groups in context: A model of task group effectiveness. AdministrativeScience Quarterly, 29, 499-517.

Guzzo, R. A., Yost, P. R., Campbell, R. J., & Shea, G. P. (1993). Potency in groups: Articulatinga construct. British Journal of Social Psychology, 3, 87-106.

Hackman, J. R. (1990). Groups that work (and those that don’t): Creating conditions for effec-tive teamwork. San Francisco: Jossey-Bass.

House, R. J., & Aditya, R. N. (1997). The social scientific study of leadership: Quo vadis? Jour-nal of Management, 23, 409-473.

House, R., Rousseau, D. M., & Thomas-Hunt, M. (1995). The meso paradigm: A framework forintegrating micro and macro organizational behavior.Research inOrganizational Behavior,17, 71-114.

James, L. R., Demaree, R. G., & Wolf, G. (1984). Estimating within-group interrater reliabilitywith and without response bias. Journal of Applied Psychology, 69, 85-98.

Jöreskog, K. G., & Sörbom, D. (1989). LISREL 7: User’s reference guide. Mooresville, IN: Sci-entific Software International.

Katzenbach, J. R. (1997). Teams at the top: Unleashing the potential of both teams and individ-ual leaders. Boston: Harvard Business School Press.

Katzenbach, J. R., & Smith, D. K. (1993). The wisdom of teams. Boston: Harvard BusinessSchool Press.

Kirkman, B., & Rosen, B. (1999). Beyond self-management: Antecedents and consequences ofteam empowerment. Academy of Management Journal, 42, 58-74.

Klimoski, R., & Mohammed, S. (1994). Team mental model: Construct or metaphor? Journal ofManagement, 20, 403-437.

Kozlowski, S.W.J., Gully, S. M., McHugh, P. P., Salas, E., & Cannon-Bowers, J. A. (1996). Adynamic theory of leadership and team effectiveness: Developmental and task contingentleader roles. In G. R. Ferris (Ed.), Research in personnel and human resource management(Vol. 14, pp. 253-205). Greenwich, CT: JAI.

Kozlowski, S.W.J., Gully, S. M., Salas, E., & Cannon-Bowers, J. A. (1996). Team leadershipand development: Theories, principles, and guidelines for training leaders and teams. InM. M. Beyerlein, D. A. Johnson, & S. T. Beyerlein (Eds.), Advances in interdisciplinarystudies of work teams (pp. 253-291). Greenwich, CT: JAI.

94 GROUP & ORGANIZATION MANAGEMENT

by guest on September 27, 2016gom.sagepub.comDownloaded from

Kumpfer, K. L., Turner, C., Hopkins, R., & Librett, J. (1993). Leadership and team effectivenessin community coalitions for the prevention of alcohol and other drug abuse. Health Educa-tion Research, 8, 359-374.

Lawler, E. E., III. (1992). The ultimate advantage: Creating the high involvement organization.San Francisco: Jossey-Bass.

Lindsley, D. H., Brass, D. J., & Thomas, J. B. (1995). Efficacy-performance spirals: A multi-level perspective. Academy of Management Review, 20, 645-678.

Lowe, K. B., Kroeck, K. G., & Sivasubramaniam, N. (1996). Effectiveness correlates oftransformational and transactional leadership: A meta-analytic review of MLQ literature.Leadership Quarterly, 7, 385-425.

Manz, C. C., & Sims, H. P., Jr. (1987). Leading workers to lead themselves: The external leader-ship of self-managing work teams. Administrative Science Quarterly, 32, 106-128.

Manz, C. C., & Sims, H. P., Jr. (1993). Business without bosses: How self-managing teams arebuilding high performance companies. New York: John Wiley.

Marsh, H. W., & Hocevar, D. (1985). Application of confirmatory factor analysis to the study ofself-concept: First- or higher-order factor models and their invariance across groups. Psy-chological Bulletin, 97, 562-582.

Martin, D. (1993). Teamthink: Using the sport connection to develop, motivate, and manage awinning business team. New York: Penguin.

Mathieu, J. E. (1991). A cross-level nonrecursive model of the antecedents of organizationalcommitment and satisfaction. Journal of Applied Psychology, 76, 607-618.

Medsker, G. J., Williams, L. J., & Holohan, P. J. (1994). A review of current practices for evalu-ating causal models in organizational behavior and human resource management research.Journal of Management, 20, 439-464.

Orasanu, J. (1990, September). Sharedmentalmodels and crewperformance. Paper presented atthe 34th annual meeting of the Human Factors Society, Orlando, FL.

Pearce, C. L., & Gallagher, C. A. (1999, August). Confidence at the group level of analysis: Alongitudinal investigation of the development and impact of potency in teams. Paper pre-sented at the annual meetings of the Academy of Management, Chicago, IL.

Prussia, G. E., & Kinicki, A. J. (1996). A motivational investigation of group effectiveness usingsocial-cognitive theory. Journal of Applied Psychology, 81, 187-198.

Salas, E., Dickinson, T. L., Converse, S. A., & Tannenbaum, S. I. (1992). Toward an understand-ing of team performance and training. In R. W. Sweezey & E. Salas (Eds.), Teams: Theirtraining and performance (pp. 3-29). Norwood, NJ: Ablex.

Seers, A. (1989). Team-member exchange quality: A new construct for role-making research.Organizational Behavior and Human Decision Processes, 43, 118-135.

Shamir, B., House, R. J., & Arthur, M. B. (1993). The motivational effects of charismatic leader-ship: A self-concept based theory. Organizational Science, 4, 577-594.

Shea, G. P., & Guzzo, R. A. (1987). Group effectiveness: What really matters? Sloan Manage-ment Review, 28, 25-31.

Silver, W. S., & Bufanio, K. M. (1996). The impact of group efficacy and group goals on grouptask performance. Small Group Research, 27, 347-359.

Sinclair, A. (1992). The tyranny of a team ideology. Organization Studies, 13, 611-626.Smircich, L. (1983). Concepts of culture and organizational analysis. Administrative Science

Quarterly, 28, 339-358.Spink, K. S. (1990). Group cohesion and collective efficacy of volleyball teams. Journal of

Sport and Exercise Psychology, 12, 301-311.

Sivasubramaniam et al. / TEAM LEADERSHIP, GROUP POTENCY 95

by guest on September 27, 2016gom.sagepub.comDownloaded from

Stewart, G., & Manz, C. C. (1994, May). Leadership for self-managing work teams: A theoreti-cal integration. Paper presented at the annual conference of the Society for Industrial andOrganizational Psychologists, Nashville, TN.

Vinokur-Kaplan, D. (1995). Treatment teams that work (and those that don’t): An application ofHackman’s group effectiveness model to interdisciplinary teams in psychiatric hospitals.Journal of Applied Behavioral Science, 31, 303-327.

Wofford, J. C., Goodwin, V. L., & Whittington, J. L. (1998). A field study of a cognitiveapproach to understanding transformational and transactional leadership. Leadership Quar-terly, 9, 55-84.

Yammarino, F. J., Dubinsky, A. J., & Spangler, W. D. (1998). Transformational and contingentreward leadership: Individual, dyad, and group level of analysis. Leadership Quarterly, 9,27-54.

Nagaraj Sivasubramaniam, Ph.D. (Florida International University), is an assistantprofessor of management at Duquesne University. His current research interestsinclude examining the organizational impacts of individual and team leadership, strate-gic human resource management, and Internet technologies.

William D. Murry, Ph.D. (Virginia Tech), is an assistant professor of management atDuquesneUniversity.His current research interests focus on leadership, levels of analy-sis, and sexual harassment in the workplace.

Bruce J. Avolio, Ph.D. (University of Akron), is currently a professor of management atBinghamtonUniversity and the codirector of the Center for Leadership Studies. His cur-rent research interests include examining how computer mediation affects the impact ofleadership on team processes and performance. He will be moving to the University ofNebraska in Fall, 2002.

Dong I. Jung, Ph.D. (Binghamton University), is an associate professor of managementat San Diego State University. His research interests include transformational leader-ship, group development, and cross-cultural comparative management.

96 GROUP & ORGANIZATION MANAGEMENT

by guest on September 27, 2016gom.sagepub.comDownloaded from