examining the differential longitudinal performance …leeds-faculty.colorado.edu/dahe7472/lorinkova...

EXAMINING THE DIFFERENTIAL LONGITUDINALPERFORMANCE OF DIRECTIVE VERSUS EMPOWERING

LEADERSHIP IN TEAMS

NATALIA M. LORINKOVAWayne State University

MATTHEW J. PEARSALLUniversity of North Carolina

HENRY P. SIMS JR.University of Maryland

This study integrates theories from the leadership and team development literatures toresolve ambiguity regarding the relative benefits of empowering and directive leader-ship in teams by focusing on their influence on team development processes over time.Empirical results based on longitudinal performance data from 60 teams suggest thatteams led by a directive leader initially outperform those led by an empowering leader.However, despite lower early performance, teams led by an empowering leader expe-rience higher performance improvement over time because of higher levels of teamlearning, coordination, empowerment, and mental model development. Implicationsfor current and future team leadership research are discussed.

In response to the increasing organizational reli-ance on teams to perform complex tasks, leadershipresearch has gradually shifted its focus toward ex-amining the important role of leaders in improvingteam performance and adaptation (e.g., Burke,Stagl, Klein, Goodwin, Salas, & Halpin, 2006;Klein, Ziegert, Knight, & Xiao, 2006). Team leader-ship research has concentrated on the leader behav-iors that promote, develop, and maintain teamperformance, and two distinct approaches—em-powering and directive leadership—have assumedspecial importance (Srivastava, Bartol, & Locke,2006; Yun, Faraj, & Sims, 2005).

Empowering leadership involves sharing powerwith subordinates and raising their level of auton-omy and responsibility, and it manifests throughspecific behaviors such as encouraging subordi-nates to express opinions and ideas, promoting col-laborative decision making, and supporting infor-mation sharing and teamwork (Arnold, Arad,Rhoades, & Drasgow, 2000; Chen, Sharma, Edinger,

Shapiro, & Farh, 2011; Pearce, Sims, Cox, Ball,Schnell, & Smith, 2003; Yun et al., 2005). Empow-ering leadership tends to create psychological own-ership of a task, heightened efficacy and commit-ment, and higher levels of coordination andcollective information processing (e.g., Cohen,Chang, & Ledford, 1997; Zaccaro, Rittman, &Marks, 2001).

Directive leadership, on the other hand, is asso-ciated with a leader’s positional power and is char-acterized by behaviors aimed at actively structuringsubordinates’ work through providing clear direc-tions and expectations regarding compliance withinstructions (e.g., House, 1971; Pearce et al., 2003;Somech, 2006; Yukl & Falbe, 1991). Directive lead-ers help followers resolve task and role ambiguityand provide external monitoring and feedback ontheir performance, reducing process loss and al-lowing team to execute decisions more quickly(House, 1996; Kahai, Sosik, & Avolio, 2004; Sa-gie, 1997).

Although researchers and the practitioner-ori-ented literature have advocated empowering overdirective leadership, the empirical evidence hasnot fully supported this view, and it is not clearthat empowering leadership is actually better forenhancing team performance (e.g., Ensley,Hmieleski, & Pearce, 2006; Kahai et al., 2004; Ma-

We would like to thank Ray Sparrowe and three anon-ymous reviewers for their insights and improvementsduring the review process. We are also grateful to SubraTangirala for his feedback and comments and to JohnHollenbeck, Matthias Spitzmuller, and Chris Barnes fortheir generous advice and support for our simulation.

573

� Academy of Management Journal2013, Vol. 56, No. 2, 573–596.http://dx.doi.org/10.5465/amj.2011.0132

Copyright of the Academy of Management, all rights reserved. Contents may not be copied, emailed, posted to a listserv, or otherwise transmitted without the copyright holder’s expresswritten permission. Users may print, download, or email articles for individual use only.

thieu, Gilson, & Ruddy, 2006; Yun et al., 2005). Theextant leadership research has thus far neglectedtwo important issues that may help shed light onthis question. First, the literature generally exhibitsa cross-sectional perspective on leader effective-ness, focusing on whether directive or empoweringleaders are beneficial overall but overlooking thequestion of when empowering leadership might bemost or least effective. Second, although researchacknowledges the importance of team memberskills and attitudes in the effectiveness of variousleadership approaches (e.g., Thompson & Vecchio,2009), it has yet to examine the direct influenceleaders have on the development process of teamsor why empowering leadership may provide itspromised benefits to teams over time (e.g., Klein etal., 2006; Yukl & Lepsinger, 2004). This is espe-cially true for newly formed teams, where timeplays a critical role in understanding the interplaybetween team development and leadership (e.g.,Kozlowski, Gully, Nason, & Smith, 1999; Kozlow-ski, Gully, Salas, & Cannon-Bowers, 1996; Yukl &Lepsinger, 2004).

In response, we address these key conceptualand empirical problems by (a) examining the rela-tive influence of empowering and directive leader-ship on the performance of teams over multiplephases of team interaction and development and(b) identifying critical behavioral, motivational,and cognitive mechanisms that explain the differ-ences between empowering and directive leader-ship in teams over time. Because of the influence ofleaders on internal team processes and emergentstates (e.g., Mathieu, Maynard, Rapp, & Gilson,2008), the choice of leadership approach in anyfocal phase of team development will have impli-cations for both current and future team interac-tions and performance.

Therefore, drawing on Kozlowski et al.’s (1999)model of team compilation and adaptation, we of-fer a more nuanced description of the costs andbenefits of empowering leadership in teams by firstarguing that the initial guidance and role defini-tions provided by directive leaders will help theirteams to outperform empowered teams in the shortterm. However, we argue that, owing to the sharedlearning and decision-making climate that empow-ering leaders create in the initial phases of teamdevelopment, and the resultant emergence of col-lective states and processes in teams, empoweredteam members will be able to attain higher levels ofperformance through the later phase of team devel-opment, while teams with directive leaders will be

less able to improve. We then identify behavioral,motivational, and cognitive processes and states inteams that explain the improved performance ofempowering versus directive leadership over time:team learning, behavioral coordination, psycholog-ical empowerment, and team mental models.

By doing so, we hope to extend the empoweringleadership literature in two important ways. First,by demonstrating that teams led by an empoweringleader achieve greater long-term performance im-provement, but at the cost of initial performancedelays, we help resolve ambiguity about the bene-fits of each leadership type through explicating thetrade-offs and boundaries of empowering leader-ship in teams. Second, by identifying key emergentstates and processes in teams that result from em-powering leadership and convey its influence intofuture performance, we shed light on the underly-ing question of why the initial performance costs ofempowering leadership may be worthwhile in thelong run.

Apart from contributing to the empowering lead-ership literature, this study also extends other ex-tant theories of leadership. In particular, the pre-scription of empowering or directive leadershipbased on follower readiness has been a central tenetof situational leadership theories (e.g., Hersey &Blanchard, 1969, 1982; Vroom & Jago, 1988). Forexample, Hersey and Blanchard (1982) highlightedfollower attributes such as emotional maturity andreadiness to be empowered that indicate the extentto which followers are able to share ideas and findmotivation in greater responsibility (e.g., Chen,Kirkman, Kanfer, Allen, & Rosen, 2007; Spreitzer,1995). However, situational leadership theorieshave found only limited empirical support (seeThompson & Vecchio, 2009), in part due to ambi-guity as to the nature of follower readiness—a par-ticularly complex consideration for leaders of col-lections of individuals in a team context. As wewill describe, this study also extends aspects ofsituational leadership theory by providing a morecomprehensive way of conceptualizing followerreadiness that is specific to the team level and bysuggesting limits to the notion that initially di-rected teams will eventually be ready to receiveand respond to empowering leadership.

THEORETICAL BACKGROUND

Given its focus on sharing power with employeesand emphasizing collaboration (Arnold et al., 2000;Manz & Sims, 1987), empowering leadership has

574 AprilAcademy of Management Journal

been touted as particularly suited to meet the needsof leading teams of people (e.g., Kirkman & Rosen,1999; Pearce et al., 2003). However, researchershave not always succeeded in finding support forthe proposed benefits of empowering leadership onteam intermediate and final outcomes (e.g., Ma-thieu et al., 2006), with a number of studies findingevidence that not empowering leadership, but itsobverse—directive leadership—leads to higherperformance in teams (e.g., Ensley et al., 2006; Ka-hai et al., 2004; Yun et al., 2005). Therefore, despitethe association of both leadership styles with pos-itive team outcomes (e.g., Sagie, Zaidman, Ami-chai-Hamburger, Te’eni, & Schwartz, 2002), the rel-ative benefits of empowering and directiveleadership in influencing team performance remainunclear (Somech, 2006; Sosik, Avolio, & Ka-hai, 1997).

Each style tends to enhance follower perfor-mance because both directive and empoweringleaders are actively attempting to improve teameffectiveness through thoughtful, planned behav-iors. Empowering leadership tends to benefit inter-dependent teams by establishing participative andcollaborative norms among members, encouragingthem to contribute ideas, decide on optimal coursesof action, and take responsibility for team perfor-mance. These types of behaviors tend to lead topositive individual and work group outcomesacross contexts (e.g., Pearce et al., 2003; Yukl,1998). For example, Zhang and Bartol (2010) re-cently provided evidence that empowering leader-ship enhances employee creativity, through its ef-fects on employee psychological empowerment,intrinsic motivation, and creative process engage-ment. Similarly, Ahearne, Mathieu, and Rapp(2005) found a positive relationship between leaderempowering behaviors and followers’ job perfor-mance attributable to increased levels of self-effi-cacy and adaptability. At the team level, the goal ofempowering leadership is to develop a team’s ca-pacity to perform autonomously (Manz & Sims,1987). Though similar to participatory leadership,which “involves the use of decision proceduresintended to allow other people some influence overthe leader’s decisions” (Yukl, 1998: 83), empower-ing leadership requires leaders to invest more trustin their followers by allowing high levels of discre-tion and decision-making authority to pass into thefollowers’ hands.

However, a directive approach, which is similarto the autocratic leadership style in Vroom andJago’s (1988) decision model and the “tough lead-

ership style” of McIntyre and Salas (1995), focuseson behaviors related to giving detailed directions,expecting subordinates to follow those instruc-tions, and making decisions with limited subordi-nate input. Research suggests that superiors’ direc-tiveness can make task accomplishment easier forfollowers by providing them with specific, role-relevant directions and helping them focus theirefforts toward their individual tasks (Fiedler, 1968;Kahai et al., 2004). Additionally, directive leader-ship helps each individual to be better aware ofhis/her own role and the availability of role re-sources (Yukl, 1998), reducing ambiguity aboutwhat each person does (Kahai et al., 2004; Pearce etal., 2003) and establishing clear rules for behavior(e.g., Katzenbach & Smith, 1993). For example, re-search has shown that directive leadership can leadto improved patient care through the assignment ofspecific actions for handling an emergency situa-tion (Yun et al., 2005). This is in contrast to em-powering leadership, in which the higher degree offreedom and discretion granted to team membersallows more potential paths toward attaining agoal, potentially decreasing task and role clarity(Evans, 1970; House, 1971).

Situational leadership theories—conceived as ameans of providing practical advice to managers asto when to best apply each different leadershipstyle—are based on the principle that the relativebenefits of each leadership type will depend on thecompetence, maturity, or readiness of followers(e.g., Goodson, McGee, & Cashman, 1989; Hersey &Blanchard, 1969, 1982; Thompson & Vecchio,2009). Subordinates with low levels of readiness,such as new employees with limited skills but suf-ficient motivation, are expected to benefit from atelling, or directive, leadership approach, whichentails giving specific instructions to subordinates,as well as monitoring their progress and providingfeedback (e.g., Muczyk & Reimann, 1987; Sims,Faraj, & Yun, 2009; Yun et al., 2005). On the otherhand, in the case of followers with higher levels ofreadiness and with sufficient competence, efficacy,and commitment, situational leadership theoriessuggest that participating and delegating stylesof leadership should be most effective (e.g.,Blanchard, 2007). In terms of situational leadershiptheory, empowering leadership falls most closelyin line with delegation, though as a practical matterit is difficult to distinguish between delegation andhigh levels of participatory leadership (see Thomp-son & Vecchio, 2009).

2013 575Lorinkova, Pearsall, and Sims

Follower readiness was initially defined as “theextent to which a follower has the ability and will-ingness to accomplish a specific task” (Hersey &Blanchard, 1988: 174). Addressing criticisms byscholars (see Northouse, 2007), Blanchard (2007)revised the definition with a clearer conceptualtypology that combines the follower attributes ofcompetence and commitment to create discrete lev-els for “developing” versus “developed” followers.However, though member competence may play animportant role in team effectiveness, the readinessof teams to respond effectively to empowering lead-ership is more complex, as it is based on the degreeto which a team has developed the cognitive, mo-tivational, and behavioral capabilities which arecritical for long-term team effectiveness (e.g.,Burke, Stagl, Salas, Pierce, & Kendall, 2006). Thusfar, situational leadership theory has failed to takeinto account the development of these attributeswhen conceptualizing team readiness.

Team Development

Teams do not simply manifest collective compe-tence and commitment when needed. Rather, theirreadiness is temporally bound by the focal stage oftheir members’ collective development (Kozlowskiet al., 1996; Zaccaro et al., 2001), and each leader-ship approach directly influences how effectivelyteams mature in critical phases of their evolution.According to Kozlowski et al.’s (1999) model of

team development, the process occurs over fourdistinct phases during the initial periods of teamcomposition and adaptation (see Figure 1, “Effectsof Leadership in Team Development Phases”). Dur-ing the first phase, team formation, a team’s mem-bers come together to begin the socialization pro-cess during which they develop an understandingof the team’s collective purpose and start to think ofthemselves as a team. During the second phase,task compilation, individual team members tend tofocus their attention on individual task masteryand proficiency and remain “primarily self-focusedon their individual task performance” (Kozlowskiet al., 1999: 262). During these initial phases, therole of the leader is primarily felt through socializ-ing the followers into the organization or group andensuring that they develop the necessary skills toperform their tasks.

In the third phase, role compilation, team mem-bers begin to learn to coordinate their performancewith other team members through a process of shar-ing and receiving role information and clarifyingrole expectations (Graen & Scandura, 1987; Katz &Kahn, 1978). In this phase, the influence of activeleadership begins to be felt most keenly, as leader-ship interventions during role compilation arelikely to influence the nature and the quality ofthese interactions (Kozlowski et al., 1999). Specif-ically, we argue that directive leadership and em-powering leadership will differentially affect teamdevelopment and performance because of their

FIGURE 1Effects of Leadership in Team Development Phasesa

• Team members come together.• Focus /training of individual skills and task.

Phases 1 and 2: TeamFormation and Task Compilation

Phase 3: Role

Phase 4: Team

• Directive leaders provide clear guidance and specific instructions to team members: Focus on immediate performance. Empowering leaders emphasize idea exchange and participative decision- making climate: Focus is on learning and positive team member interactions.

Directed teams:Reduced capacity forlearning and integrationof expertise, reliance onindividual knowledge and leader coordination.Empowered teams:Focus is on performanceimprovement usingbehavioral routines and shared cognitionsdeveloped in phase 3.

•

•

•

Compilation

Compilation

a Adapted from Kozlowski et al. (1999).


differing influence on team role compilationprocesses.

During role compilation, team members shouldnavigate their role interactions, learning about theirown and their teammates’ responsibilities and ca-pabilities through a series of role identification andclarification behaviors. This navigation allowsthem to begin using their knowledge of one anotherto integrate their efforts, resulting in the develop-ment of routinized processes and the emergence ofshared cognitive structures (e.g., Pearsall, Ellis, &Bell, 2010). However, team members with directiveleaders largely bypass these processes, as the leadergives specific role guidance and provides detailedgoals and instructions for each team member, thusactively managing their interactions (e.g., Kors-gaard, Schweiger, & Sapienza, 1995; Sagie, 1996).For example, task execution directions given by aleader are likely to trigger individual member’s ef-forts to perform their role-specific task require-ments and to obviate the need to learn about theirteammates’ roles to ensure their efforts are inte-grated. Directive leaders, therefore, act as explicitcoordination mechanisms for their teams, remov-ing the need for the time-consuming developmentof learned routines and shared cognitions (Espi-nosa, Lerch, & Kraut, 2004). Thus, directive leadersare likely to quickly benefit initial team perfor-mance by focusing team members’ attention on ex-ecuting their specific tasks.

Empowering leaders, on the other hand, willtend to first work to establish norms for collabora-tive decision making during the role compilationphase. With behaviors emphasizing the open ex-change of ideas and suggestions within a team,empowering leaders will attempt to create an atmo-sphere of increased interaction among team mem-bers, which aids individuals’ understanding of“who they interact with to perform the task” (Koz-lowski et al., 1999: 266) and the development ofshared perceptions and understanding. Therefore,during the role compilation phase, teams led by anempowering leader are likely to focus on develop-ing knowledge of each other and experiencing pos-itive interpersonal interactions, rather than on per-forming (Kozlowski et al., 1999).

Directive leadership during the role compilationphase should therefore be most effective for perfor-mance in teams. Teams with a directive leader willbe able to sustain their focus on task performance,relying on the explicit guidance provided by theirleader. In contrast, because empowering leadersattend to team members’ involvement in decision

making and encourage experimentation, learning,and collaborative contributions (Arnold et al.,2000; Leana, 1987; Pearce et al., 2003; Srivastava etal., 2006), teams with an empowering leader willinitially take longer to effectively perform theirtasks. This should enable directively led teams tooutperform empowered teams during the role com-pilation phase of team development. Therefore, wehypothesize:

Hypothesis 1. Teams led by a directive leaderexhibit higher initial performance than teamsled by an empowering leader.

Although directive leaders offer teams an initialperformance advantage in the role compilationphase, team performance and adaptability in thefourth phase, team compilation, depend heavily onthe role interactions and implicit coordinationmechanisms that are expected to emerge during theprevious phase (Kozlowski et al., 1999). Throughemphasizing team members’ involvement in deci-sion making, open communication, and sharing ofideas and opinions, empowering leaders furtherenable learning and coordination by creating a cli-mate in which high levels of knowledge exchangeare encouraged (Burke et al., 2006; Manz & Sims,1987; Srivastava et al., 2006).

Through these processes, a team develops thecollective competence, confidence, and commit-ment that make empowering leadership most effec-tive in the team compilation phase. In directedteams, however, the initial performance gains dur-ing role compilation come at the price of a reducedcapacity for learning and integration of expertiseduring team compilation. Because directed teammembers have had less need to engage in role iden-tification behaviors and develop shared cognitionsabout their roles and team task, they are left lesscapable of adapting to their task environment anddealing with uncertainty and complexity. Theirteam is, therefore, less able to take advantage of thediverse skills and knowledge of its members and isdependent upon, and limited by, the guidance ofthe directive leader, who must continue to serve asthe sole explicit coordination mechanism forthe team.

Therefore, although directive leaders may bemore effective in maximizing the performance ofteams in the short term, their actions prevent theteams from developing the very processes andemergent states that would allow them to be readyfor the leader to follow the prescription of situa-tional leadership theory to switch to empowering


leadership as the teams mature. Empowering lead-ers, on the other hand, despite initial performancedelays due to their modeling and emphasizing fre-quent interpersonal knowledge and idea exchanges(instead of focusing specifically on task perfor-mance), should better enable the development ofinformation-sharing norms and cognitive struc-tures to prepare teams to make effective decisionsand skillfully perform their tasks over the long run.We hypothesize that:

Hypothesis 2. Teams led by an empoweringleader exhibit greater performance improve-ment in the later, team compilation, phase ofdevelopment than teams led by a directiveleader.

Although research on follower readiness inteams has focused on the individual attributes andperceptions of team members (Goodson et al., 1989;Mathieu, Ahearne, & Taylor, 2007; Vecchio, 1987),competence and commitment at the team level arefar more complex and hinge on the emergence ofshared teams states and processes that enable effec-tive task interaction and performance. For teamswith high levels of interdependence and differen-tiated expertise, long-term effective performanceand adaptation result from the development of cog-nitive, motivational, and behavioral states and pro-cesses that convey the effects of leadership ontodistal team performance (e.g., Ilgen, Hollenbeck,Johnson, & Jundt, 2005). Drawing on Kozlowski etal.’s (1999) model of team development and Burkeet al.’s (2006) model of team adaptation, we focuson four critical emergent states and processes thatserve both as markers of collective team readinessand mediators of the relative influence of empow-ering leadership on performance improvement:team learning, behavioral coordination, psycholog-ical empowerment, and team mental models. To-gether, we argue, these four states and processeswill explain the difference in the performance ofempowering leadership and directive leadershipover the later, team compilation, phase of teamdevelopment.

Though we extend the definition of team readi-ness to include these emergent states and pro-cesses, this theoretical approach fundamentallydiffers from situational leadership theory, whichsuggests that team readiness acts as an independentmoderator of leadership effectiveness, as the choiceof leadership style should be a response to thecurrent readiness of a team (e.g., Thompson & Vec-chio, 2009). However, by focusing on team devel-

opment as both an outcome of leadership and anenabler of future leadership, we differentiate teamreadiness as a more complex, shared attribute of anentire team.

Behavioral Processes:Team Learning and Coordination

During team compilation, team members turntheir focus to successful task completion and be-havioral adaptation. We suggest that the perfor-mance improvement in teams with empoweringleaders during team compilation will be due, inpart, to the increased ability of such teams to learnand to coordinate their knowledge and actions.

Team learning is “the process by which relativelypermanent changes occur in the behavioral poten-tial of the group as a result of group interactionactivities through which members acquire, share,and combine knowledge” (Burke et al., 2006: 1190;Edmondson, 1999) and is instrumental in allowingteams to understand and adapt to their environ-ment. This behavioral process is of particular im-portance for teams with distributed expertise,whose members have to synchronize and leveragediverse skills and abilities to perform complextasks. As noted previously, teams with empoweringleaders will engage in greater levels of role identi-fication and discussion during the role compilationphase, leading to a shared understanding of team-mates’ roles and capabilities. This knowledge actsas a cognitive foundation for information acquisi-tion and exchange during team compilation (Burkeet al., 2006), allowing for improved team perfor-mance over time.

Similarly, behavioral coordination, which in-volves “information exchange and mutual adjust-ment of action in order to align the pace and se-quencing of team member contributions” (Marks,Mathieu, & Zaccaro, 2001: 368) is seen as a criticalbehavior for interdependent teams as it reflects thesmooth integration of skills and knowledge be-tween members. The shared understanding of eachother’s roles, duties, and capabilities in teams withempowering leaders will allow team members toexchange information and synchronize actions, im-proving task performance over time.

In contrast, teams with a directive leader, wholimits team members’ input in decision making anddoes not consult with subordinates, should be lesslikely to share the information and knowledge heldseparately by different team members. Because thedirective leader acts as the source for guidance and


direction, individual members will tend to focus ontheir own responsibilities rather than on commu-nicating and sharing with the team (Pearce et al.,2003; Sagie, 1996). Further, a directive leadershipstyle may hamper the development of psychologi-cal safety in the team, as the leader does not solicitand encourage input (Edmondson, 1999). Teammembers in directed teams may, therefore, be muchless likely to engage in the experimentation, reflec-tive communication, and knowledge codificationthat typify effective team learning and coordination(Gibson & Vermeulen, 2003). Without these relatedbehavioral processes, teams with directive leadersare less likely to capitalize on their strong start and,ultimately, are less able to improve their perfor-mance over time.

Therefore, we hypothesize:

Hypothesis 3. The greater performance im-provement associated with empowering com-pared to directive leadership during team com-pilation is partially explained by team learning.

Hypothesis 4. The greater performance im-provement associated with empowering com-pared to directive leadership during teamcompilation is partially explained by team be-havioral coordination.

Motivational State: Psychological Empowerment

The most direct influence of empowering leader-ship on team interaction manifests through a feel-ing of psychological empowerment in a team (Chenet al., 2011; Mathieu et al., 2006). Because empow-ered employees feel more competent and able toinfluence their team’s outcomes, empowerment hasconsistently been shown to lead to higher levels ofemployee commitment, innovation, citizenship be-haviors, and performance (e.g., Seibert, Wang, &Courtright, 2011; Spreitzer, 2008; Thomas & Velt-house, 1990; Yukl, 2010). These feelings of compe-tence and commitment that emerge from empower-ing leadership likewise make a team particularlyready for empowering leadership in the future. Em-powered team members will tend to be motivatedby collective ownership of their choices and work,to support and back each other up, and to worktoward adaptation and performance improvementduring team compilation (e.g., Chen et al., 2007;Kirkman & Rosen, 1999; Kirkman, Rosen, Tesluk, &Gibson, 2004).

Therefore we hypothesize that:

Hypothesis 5. The greater performance im-provement associated with empowering com-pared to directive leadership during teamcompilation is partially explained by teamempowerment.

Cognitive State: Team Mental Models

Mental models are collective knowledge struc-tures that allow team members to understand andform expectations concerning other team members’responsibilities, needs, and behaviors (Moham-med, Klimoski, & Rentsch, 2000). Because they en-able team members to interpret information in asimilar manner (Hinsz, Tindale, & Vollrath, 1997),team mental models have been positively related toteam processes such as communication (Marks,Zaccaro, & Mathieu, 2000), decision making, andsituational awareness (Mathieu, Heffner, Goodwin,Cannon-Bowers, & Salas, 2005), as well as perfor-mance in many situations (e.g., Edwards, Day, Ar-thur, & Bell, 2006; Marks et al., 2000).

During the role compilation phase, empoweredteam members will tend to engage in dyadic ex-changes with their teammates regarding the natureand scope of each team member’s roles and capa-bilities, helping them to understand the patterns ofbehavior and interaction that are emerging in theirteam. As team members explore the boundaries andextent of each others’ expertise and duties, theybegin to develop a clear picture of how their uniqueknowledge, skills, and abilities fit together and howbest they can integrate them (Katz & Kahn, 1978;Kozlowski et al., 1999).

Then, during team compilation, as teams shifttheir focus more toward task execution, their men-tal models allow them to revise and refine the flowof work and to continuously improve performance(Pearsall et al., 2010). Teams with directive leader-ship, however, have relied on the explicit directionfrom the leader rather than engaging in time-con-suming role identification interactions and willhave less developed mental models to help inte-grate their efforts.

Therefore, we hypothesize that:

Hypothesis 6. The greater performance im-provement associated with empowering com-pared to directive leadership during team com-pilation is partially explained by team mentalmodel development.


METHODS

Research Participants

Participants in this study included 300 seniorundergraduate students enrolled in an upper-levelmanagement course at a major mid-Atlantic univer-sity in the United States. Participants were arrayedin 60 five-person teams that engaged in a three-hour computer simulation task. In exchange fortheir participation, participants were given extracourse credit and were eligible for cash prizesbased on their performance ($250 per team).

Task

Participants were randomly assigned to teamsand, with the exception of each team leader, ran-domly assigned to team roles. They then workedcollectively to complete tasks within a networkedcomputer simulation—the Leadership Develop-ment Simulator (LDS)—originally developed forthe Squadron Officer School at Maxwell Air ForceBase and used to train and evaluate midcareer of-ficers (see Davison, Hollenbeck, Barnes, Sleesman,& Ilgen, 2012). The LDS is a complex task in whichteams of four staff members and one leader mustmanage a large number of assets (48 in each round)to discover and interact with targets on a sharedtask screen over ten rounds of decision making. Tobe successful, team members must collaborate,combining their efforts to achieve certain objectives(such as attacking a large target or verifying intelli-gence), which, in turn, produce a higher perfor-mance score.

Team members and their leader were seatedaround a table, each in front of a separate computerstation, and were able to freely talk with each otherthroughout the simulation. Each team member hada defined role with assigned responsibilities, ar-ranged in a functional hierarchical structure con-taining staff members (the four players) and a for-mal leader. Each staff member was in charge ofeither “operation” assets or “intelligence” assetsthat could be used to gather and employ informa-tion about enemy targets (described below).

Teams were in charge of assessing and integrat-ing multiple sources of information in the processof finding and engaging targets on the task screen,which include both threats and opportunities. Op-portunities manifested as enemy vehicles and basesthat they could destroy to gain offensive points.Threats manifested on the simulation screen as en-emy vehicles and launchers that were capable of

destroying team members’ assets, costing the teamdefensive points. Threats were commonly pairedwith opportunities to protect them from attack.Each team’s goal was to maximize its performanceby identifying and destroying enemy threats andopportunities while avoiding having their own as-sets destroyed by enemy threats.

Team member roles. Teams were functionallystructured with defined, specialized roles and ex-pertise, and each member was responsible for ful-filling the same role throughout the entire simula-tion. Except for the leader, each other member wasrandomly assigned to one of four roles in eitheroperations or intelligence positions. The two teammembers who primarily gathered informationabout the environment were labeled as intelligenceteam members, and the two members who engagedthreats and opportunities to score points and pro-tect bases were labeled as operations teammembers.

Members in the intelligence positions were as-signed to one of two intelligence roles—signals orhuman intelligence—and each managed two typesof discrete information assets (8 assets of each typefor a total of 16 per player and 32 per team) thatcould be placed anywhere on the game grid in eachround to identify threats and opportunities. Eachintelligence asset was effective in a different regionof the environment and could be deployed to asingle location on the screen to gather probabilisticinformation regarding that area of the environment.These assets could not be lost to enemy threats;however, they were only accurate (at 95 percent) inone area of the simulation environment, known asthe “sweet spot.” The sweet spot could be locatedin the upper half, the lower half, or the middle ofthe screen, and each of the four types of intelli-gence assets were only accurate in this specificportion of the task screen. In the rest of the screen,assets would either give no information or, occa-sionally, false information. Therefore, identifyingthe sweet spot for each type of asset was a criticalresponsibility of the entire team.

In contrast, operations team members had fourdifferent types of assets: strike, escort, refuel, andinformation. Strike assets had the capability of cap-italizing on opportunities. Escort assets had thecapability of destroying threats. Refuel assets en-abled other assets to reach distant portions of theenvironment. Information assets gathered informa-tion from areas larger than the areas investigated byintelligence assets. In contrast to intelligence as-sets, the information that was gathered by informa-


tion assets was always completely accurate (at 100percent). These assets were controlled by the twooperation members—offensive and support. Theoffensive operation member controlled four strikeand four escort assets and was responsible for uti-lizing these assets in the entire task screen. Thesupport member controlled four refuel and fourinformation assets and was responsible for utilizingthese assets to support strike and escort vehicles(refuel assets) or coordinate grid searches with in-telligence members (information assets).

Each team’s assigned leader could approve orchange the decisions made by the offensive andintelligence team members.

Simulation environment. The environment inthe LDS consisted of a grid, 16 rows (1–16) by 16columns (A–P), totaling 256 squares. At the start ofthe simulation, teams were presented with a blankgrid. However, as shown in Figure 2, hidden

throughout the grid were threats and opportunities,which could be either small or large and eitherfixed or mobile. Threats attacked assets and bases,costing the teams points. Opportunities could bedestroyed by strike assets, gaining the team points.Teams needed a single asset to engage a small tar-get, but they needed two assets to engage largetargets. Fixed targets remained in the same squarethroughout the entire simulation, while mobile tar-gets moved around the grid.

Teams engaged in the simulation through a seriesof rounds, or decision-making periods. During thefirst three minutes of each round, team membersdeployed their assets on the simulation grid. Next,the team leader had one and a half minutes toreview the team decisions and make any desiredchanges to the asset allocation. Then, teams viewedthe results of their choices, seeing what threats andopportunities their assets had located and engaged,

FIGURE 2Leadership Development Simulator Screen


and had two minutes to analyze this feedback andmake decisions for the next round. In total, eachteam performed the task for ten rounds.

Procedure

Three to four weeks before the experimental ses-sion (upon signing up to participate in the study),participants completed an online questionnairethat assessed their natural leadership tendencies,described in the section on leadership manipula-tion below. After team members were randomlyassigned to teams, the leader of each team wasselected for training on the basis of his or her nat-ural inclination to behave in a directive versusempowering way, depending on experimental con-dition (as described below). Selected team leaderswere trained to exhibit desired leadership behav-iors in isolation from the rest of the team, immedi-ately before the start of the experiment.

Upon arriving at the laboratory, participants instaff positions were randomly assigned to opera-tions or intelligence roles, after which they com-pleted a pretask survey (capturing the control vari-ables) and were introduced to the other teammembers and the purpose of the team task. Next,the entire team was trained with a 30-minute pre-recorded presentation on how to operate the simu-lation. The team introduction and the trainingphases mirrored the team formation and task com-pilation phases of Kozlowski et al.’s (1999) frame-work. During these early team development phases,members were asked to focus on understanding theteam task (the team’s collective purpose), masteringtheir individual tasks, and developing the skillsnecessary for their task completion.

Following the whole team training, to reinforcethe manipulation each leader was asked to make astatement to his or her team and use prescriptedcomments consistent with the leadership style inwhich he or she had been trained. The Appendixlists these scripts. Then, participants performedfive rounds of the task, after which the simulationwas paused. This five-round period coincides withthe role compilation phase of team development,during which team members predominantly engagein role identification behaviors in an attempt tomaximize their role knowledge. At this point, lead-ers were again reminded about the prescriptedcomments available for their leadership style.Teams then performed the five additional rounds ofthe task, which mirror the team compilation phase,during which team members were expected to in-

tegrate their efforts in order to maximize perfor-mance. After the end of the simulation, participantscompleted a short survey measuring their percep-tions of the leader and psychological empower-ment, as well as the team mental model measure.Then participants were debriefed and thanked fortheir participation.

Measures and Manipulations

Leadership manipulation. In this study, we ma-nipulated leadership using a two-step approach ad-vocated by Durham, Knight, and Locke (1997) thatutilizes both selection and training to maximize theeffectiveness of leadership manipulation. Morespecifically, the leader was chosen on the basis ofhis/her personal predisposition to behave in anempowering or directive way and then additionallytrained to exhibit the desired leadership behavior.

As noted above, results from an online question-naire administered prior to participants reportingto the experimental venue was the basis for leaderselection. We used the ten-item Directive LeaderScale (Durham et al., 1997), which was adaptedfrom Cox and Sims (1996), to select directive lead-ers. Participants were asked to indicate on a scale(1 � “extremely uncomfortable,” and 7 � “ex-tremely comfortable”) the extent to which theywould feel comfortable performing directive leaderbehaviors while working in a group, such as “tak-ing charge of a group,” “giving instructions togroup members,” and “specifying others’ roles in agroup task.”

The individual scoring highest on the directiveleadership inclination was chosen for the role ofthe directive leader in teams randomly assigned tothe directive leadership condition. To maximizethe difference (and the influence of the manipula-tion), we used a similarly constructed survey toselect the empowering leaders in the empoweredteams: participants filled in a survey asking to whatextent they felt comfortable performing empower-ing behaviors. The items were adapted from theEmpowering Leadership Questionnaire (ELQ), de-veloped by Arnold and his colleagues (2000), andparticipants responded by indicating (1 � “ex-tremely uncomfortable” and 7 � “extremely com-fortable”) the extent to which they would feel com-fortable, while working in a group, performingempowering leader behaviors such as “encouragegroup members to assume responsibilities on theirown,” “advise group members to exchange infor-mation with one another,” and “listen to and con-


sider ideas and suggestions of group members.”The individual who scored highest on the ELQ in ateam randomly assigned to the empowering leader-ship condition was selected for the role of the em-powering leader. Team leaders were not informedof the reason for their selection.

The selected leaders were trained on site, in thesimulation lab, immediately before each experi-ment. Training has been shown to have an impacton leader actions (e.g., Towler, 2003) and to in-crease the use of desired leader behaviors (e.g.,Barling, Weber, & Kelloway, 1996; Towler, 2003),even when a manipulation is only 5 minutes long(e.g., Manz & Sims, 1986). In this study, team lead-ers received 30 minutes of training, the composi-tion of which depended on their experimental con-dition. Directive leaders were first exposed to a10-minute verbal presentation that explained anddetailed what kind of behaviors they were expectedto exhibit. Then they were shown a short movieclip (from Apollo 13) emphasizing the desired di-rective leader behaviors; the movie clip was fol-lowed by a 10-minute role-playing simulation thatagain emphasized the desired behaviors. For thelast 5 minutes of the training, directive leaderswere also trained and asked to develop a specificgame plan without input from team members andto ask team members to carry out the proposedgame plan (Yun et al., 2005).

The selected empowering leaders were exposedto a similar 30-minute training that included a 10-minute verbal presentation, which explained andemphasized the kind of behaviors the leader wasexpected to exhibit during the simulation; a shortmovie clip, also from Apollo 13, this time empha-sizing empowering leadership; and a 10-minuterole-playing exercise to strengthen the modeled be-haviors. For the last 5 minutes of the training, em-powering leaders were asked to develop a plan toinclude all the team members in setting team per-formance goals via exchanging ideas and informa-tion with one another.

To aid leaders in acting in a directive or empow-ering manner, we gave them a “cheat sheet” with alist of key verbal prompts that were suggested foruse during interaction with the team throughoutthe simulation. The phrases listed on this cheatsheet (shown in the Appendix), reflect generalrules of the simulation rather than specific, exper-tise-based comments and served simply to rein-force the manipulation. Finally, the leaders weregiven a statement to read before the beginning ofthe experiment; the text of the statement empha-

sized either directive or empowering leaderbehaviors.

Team performance. Team performance was cal-culated as the sum of a team’s offensive (pointsgained) and defensive points (points lost) duringeach round of the simulation and reflected theteam’s primary objective, maximizing its score.Teams received points each time they engaged anopportunity or neutralized a threat and lost pointseach time one of their assets was destroyed by athreat or a threat reached the team base. Specifi-cally, teams scored 4 points for capitalizing on asmall opportunity and 16 points for capitalizing ona large opportunity; they scored 2 points for de-stroying a small threat and 4 points for destroying alarge threat; and they lost 8 points for each assetdestroyed by a threat and each time a mobile threatreached a base.

Team learning. Team learning was assessed bythe number of intelligence assets sent to the correctsweet spots (described above) during each round ofthe simulation. This measure was based on thedegree to which a team identified the areas inwhich its intelligence assets were accurate, whichrequired team members’ joint experimentation andreflective communication about the results of theexperimentation (Gibson & Vermeulen, 2003). Atthe beginning of the simulation, team memberswere unaware in which part of the grid each type ofintelligence asset would accurately identify threatsand opportunities. Without accurate information,team members would be unable to engage enemyopportunities and threats, making sweet spot iden-tification critical for team success. To learn thesweet spot areas for each type of asset, team mem-bers in all roles had to exchange information re-ceived from their own assets, experiment and sharetheir mistakes and successes, integrate the place-ment of their assets to “double source” a gridsquare and verify which assets were effective there,and consolidate the information received fromtheir assets with the rest of the team. Once teamslearned where intelligence assets were accurate,they were able to send them to the grids on the taskscreen where they would be most effective—thesweet spots—to identify targets and opportunitiesfor scoring. Therefore, the number of intelligenceassets sent to the sweet spots indicated the effectivetransfer and application of team learning ineach round.

Behavioral coordination. Our measure of behav-ioral coordination focuses on the alignment of teammember actions toward their interdependent task,


as described by Marks et al. (2001). Specifically, wemeasured the number of times during each roundthat strike and escort assets received refueling sup-port from the operations support role to investigateand destroy enemy opportunities and threats iden-tified by intelligence. All missions sent to thenorth, or top half of the game grid, required refuel-ing support, without which the tasked assets wouldrun out of fuel and be destroyed. For a strike orescort mission to survive a trip to the north, multi-ple team members must agree on how best to allo-cate their respective resources and then simultane-ously commit them to the same game grid square.

Team empowerment. Team empowerment de-scribes a team’s collective sense of having authorityand responsibility to control their work and wasmeasured with a six-item measurement instrumentadapted from Mathieu and colleagues (2006) (� �.87). Each team member indicated the extent towhich they agreed or disagreed (1 � “strongly dis-agree” and 5 � “strongly agree”) with the followingexemplary statements: “In our team members areresponsible for deciding how to achieve our goals,”“My team is empowered to change our work pro-cesses in order to improve our performance,” and“In our team members have a great deal of freedomin deciding how we will do our work.” We usedJames, Demaree, and Wolf’s (1984) rwg agreementindex to justify aggregating individual members’response to the team level (mean rwg � .91). Inaddition, we calculated and report here the intra-class correlations (ICCs), which represent whethermeasures are sufficiently reliable to model effectsat the team level. ICC1, indicating the reliability ofa single rating of the team construct, equaled .49,and ICC2, which represents the reliability of theaverage (mean) team members’ responses, equaled.79. Thus, reliability statistics, coupled with ourtheoretical reasoning, provided support for reliableteam-level effects, and we proceeded with aggregat-ing the individual members’ scores to arrive to asingle score for each team (Bliese, 2000).

Team mental models. To assess team mentalmodel development, we adapted Marks et al.’s(2000) concept-mapping technique, which was spe-cifically designed for tasks such as this (e.g., Ellis,2006). Following the experimental task, team mem-bers were given a task scenario accompanied byeight blank spaces (two per team member) thatneeded to be filled with one of eight concepts thatrepresented different aspects of team member rolesand capabilities. Team members completed themaps by placing concepts that best represented the

actions of each team member on the diagram. Thesemental maps reflect the models held by team mem-bers throughout the performance session and wereconstructed specifically for this study and based onteam members’ roles during the experimental task.For example, for the operations support role, teammembers would need to accurately identify thatthey possessed refueling and information assetsand when they would deploy them.

To arrive at the team-level mental models score,we used each team member’s individual conceptmap and compared it to each of the other teammembers’ concept maps, dividing the number ofmatching responses by the total number of possibledyadic matches to obtain a total team mental modelsimilarity score ranging between 0 and 6.

Control variables. In this study, we controlledfor two team ability and experience factors thatmay play a role in team performance and learning:team cognitive ability, in terms of average GPA perteam (e.g., LePine, 2003), and team members’ de-gree of computer game experience, which was mea-sured with a three-item measure (� � .80) askingparticipants to indicate their level of computergame experience (e.g., “How skilled are you inplaying computer games?” [1 � “unskilled” and 5� “very skilled”]). The individual scores were av-eraged to arrive at a team-level computer gameexperience score.

RESULTS

Manipulation Checks

To ensure that leader behaviors were consistentwith their experimental condition, we measuredthe extent to which team members perceived thattheir leader behaved in an empowering or a direc-tive manner with two five-item instrumentsadapted from Pearce et al. (2003). We administeredthe instruments to participants after the end of thesimulation to avoid biasing team members’ percep-tions of their leader during the simulation execu-tion (Durham et al., 1997). One of the instrumentsmeasured perceived directive leadership, and theother measured perceived empowering leadership.A sample item for empowering leadership read,“The team leader gives the team autonomy andfreedom of action,” and a sample item for directiveleadership read, “The team leader defines tasks andresponsibilities for group members.” Both mea-sures used a scale ranging from 1, “strongly dis-agree,” to 5, “strongly agree.” The mean rwg for


perceived directive leadership was .77, and eachwork group rwg value exceeded .70, which is com-monly used as a cutoff to justify aggregation ofindividual-level measures to the group level (Klein& Kozlowski, 2000). Similarly, the mean rwg forperceived empowering leadership was .84. Intra-class aggregation statistics also justified aggregation(Bliese, 2000; James, 1982; Kirkman et al., 2004;Mathieu et al., 2006) to the team level (perceiveddirective leadership: ICC1 � .22, ICC2 � .63; per-ceived empowering leadership: ICC1 � .28, ICC2� .68).

Results indicated that the participants in the di-rective condition perceived their leader to be sig-nificantly more directive (mean � 3.77, s.d. � 0.49)than those in the empowering condition (mean �3.12, s.d. � .45; t[58] � 5.35, p � .01). Similarly,teams in the empowering condition viewed theirleaders as more empowering (mean � 3.96; s.d. �0.37) than teams in the directive condition (mean �3.05; s.d. � 0.52; t[58] � 7.81, p � .01]). Takentogether, these results provide support for the ef-fectiveness of the two leadership manipulations.

Measurement Timing

To test whether the timing of our measures coin-cided with the theorized phases of team develop-ment, we focused on the degree to which teamsengaged in risky, experimental behaviors, whichare more common in earlier phases—specifically,we counted the number of unescorted missions. InLDS, when an operations asset is sent on a mission,it may be destroyed by enemy threats unless ac-companied by escort assets. However, using escortslimits the number of possible missions in eachround, making the number of unescorted missionsa proxy for risk taking by a team. Results show thatthe number of unescorted missions in the role com-pilation phase (rounds 1–5; mean � 2.40, s.d. �1.60) were significantly higher than in the teamcompilation phase (rounds 6–10; mean � 1.07, s.d.� 1.32; t[294] � 8.24, p � .01), suggesting thatteams in the team compilation phase had shiftedtheir focus away from experimentation toward theroutines and performance improvement associatedwith team compilation.

Data Analyses

In keeping with the nature of our hypotheses, weused three different statistical analyses. First, weused a two-sample t-test to test our first hypothesis

(Hypothesis 1), examining the difference in theoverall level of team performance in the initial (rolecompilation) phase of team interaction between thetwo conditions, directed teams and empow-ered teams.

To test the nature of the leadership-time interac-tion and performance improvement hypotheses(Hypotheses 2–4), we estimated random coefficientgrowth models (RCM), following the six-step modelestimation procedure of Bliese and Ployhart (2002).Leadership was entered as a dichotomous level 2variable (“empowering leadership” � 1 and “direc-tive leadership” � 0), and the team performancescores at each round were entered as a level 1outcome. We conducted all analyses using the non-linear and linear mixed effects package (“nlme”) inthe statistical software R, version 2.14.0. First, wespecified the level 1 model (steps 1 through 4; seeBliese and Ployhart [2002] for a detailed descrip-tion of the procedure) in which teams’ performancescores were examined for growth over time. Havingcorrectly specified the level 1 model, we examinedthe level 2 model (steps 5 and 6), in which leader-ship is entered as a time-invariant team-level vari-able and the cross-level interaction between timeand leadership is used to estimate the extent towhich leadership accounts for between-team differ-ences in slope parameters for the time-performancerelationship (Hypothesis 2). Repeating the stepsoutlined above, we again used RCM to examine themediating effects of team learning (Hypothesis 3)and team behavioral coordination (Hypothesis 4)on performance improvement during the teamcompilation phase of team development (rounds 6through 10), following the recommendations ofMacKinnon, Fairchild, and Fritz (2007).

Finally, to test our fifth and sixth hypothesis weused generalized least square (GLS) regressionanalysis to regress the team performance score foreach performance round of the team compilationphase on the relevant variable (independent vari-able and mediators) and control variables. The useof GLS allowed us to use the one-time measures ofteam empowerment and team mental models(which made it impossible to utilize RCM for Hy-potheses 5 and 6) at the same time accounting forthe data heteroskedasticity and correlations amongthe dependent variable multitime longitudinal ob-servations. Having established the relevant regres-sion coefficients, we then constructed bootstrappedconfidence intervals to estimate the strength of theindirect path (Preacher, Rucker, & Hayes, 2007).


Hypothesis Testing

The summary of means, standard deviations, andcorrelations among the study variables of interest isprovided in Table 1. None of the control variableswere significantly correlated with team perfor-mance; however, to increase the robustness of ourresults, we controlled for them in the subsequentanalyses.

Hypothesis 1. Our first hypothesis proposes thatteams led by a directive leader exhibit higher initialperformance than teams led by an empoweringleader. Results of a two-sample t-test comparing theperformance results of directive versus empower-ing teams during the role compilation phase ofteam development (rounds 1–5) indicated that di-rected teams (mean � 7.33, s.d. � 14.30) signifi-cantly outperformed empowered teams (mean ��3.52, s.d. � 19.01); t[58] � 2.50, p � .05). There-fore, our first hypothesis was supported.

Hypothesis 2. Our second hypothesis proposesthat teams led by an empowering leader exhibitgreater performance improvement in the team com-pilation phase than teams led by a directive leader.To test this hypothesis, we built and tested a ran-dom coefficient growth model following the six-step procedure outlined in our data analyses sec-tion. At steps 1 through 3, we specified our level 1model and found that (as expected), there was apositive and significant linear trend for the influ-ence of time on team performance (� � 1.62, p �.01). Further, the significant slope variation amongteams (as suggested by a log-likelihood comparisonof the model with fixed versus random slope ef-fects: –2 log-likelihood � 211.61, p � .01) indicated

that the model with the random slopes fitted thedata significantly better. Modeling heteroskedastic-ity as an exponent of the (time) covariate at step 4additionally improved the fit of our level 1 model(�2 log-likelihood � 57.79, p � .01).

In steps 5 and 6 of the RCM procedure, we spec-ified the level 2 model and, as shown in Table 2, inthe final level 2 model, the interaction term involv-ing leadership and time was positive and signifi-cant (� � 0.34, p � .05), providing support forHypothesis 2. To better understand the cross-levellongitudinal influence of leadership (specifically,empowering leadership, coded 1), we plotted theinteraction in Figure 3.

As shown in Figure 3, teams led by an empow-ering leader exhibited lower initial performanceand greater performance improvement than teamsled by a directive leader, supporting both Hypoth-eses 1 and 2. We then tested the mediating role offour behavioral, cognitive, and motivational pro-cesses and states in explaining the difference inperformance between teams with empowering anddirective leaders during team compilation.

Hypotheses 3 and 4. Hypotheses 3 and 4 proposethat team learning and behavioral coordination, re-spectively, partially explain the difference in theperformance of empowering and directive leader-ship during the team compilation phase of teamdevelopment. Given their nature, these hypothesesshould be treated as mediated longitudinal moder-ation (Muller, Judd, & Yzerbyt, 2005). To supportthese hypotheses, we had to demonstrate that thetwo mediators—team learning and behavioral coor-dination—were significantly related to team perfor-

TABLE 1Descriptive Statistics of the Study Variablesa

Variables Mean s.d. 1 2 3 4 5 6 7 8

1. Leadership condition 0.50 0.502. Team cognitive ability 3.49 0.18 .153. Computer game experience 3.61 0.56 �.04 �.074. Team learning (rounds 6–10) 76.78 25.04 .26* .02 .105. Behavioral coordination (rounds 6–10) 25.08 7.2 .12 �.04 �.11 .24*6. Team empowerment 3.98 0.38 .45** �.12 .13 .12 .077. Team mental models 4.85 1.14 .25* �.08 .09 .21 .02 .198. Midpoint team performance 1.91 17.57 �.24* .01 �.18 .02 �.08 �.09 �.119. Final team performance 10.97 51.58 .03 .11 .16 .24* .17 .12 .23 .29*

a n � 60. Leadership condition is a dichotomous variable comparing empowering leadership (coded 1) with directive leadership (coded0). Team learning is the total number of team learning behaviors during the team compilation phase (rounds 6–10). Behavioral coordinationis the total number of coordinated missions during the team compilation phase (rounds 6–10).

* p � .05** p � .01


mance improvement while simultaneously leadingto a drop in the significance of the interaction term,involving time and leadership. Such a result wouldindicate that the differential influence of empow-ering and directive leadership over time becameless significant when team learning and team be-havioral coordination were included in the equa-

tion (see MacKinnon et al.[2007] for a detailed ex-planation of the procedure).

In testing the first step of the mediated modera-tion hypothesis, we found that leadership (� �9.72, p � .05) as well as the interaction betweentime and leadership (� � 1.19, p � .05) were posi-tively and significantly related to team learningduring the team compilation phase (Hypothesis 3).Similarly, leadership (� � 5.69, p � .05) and theinteraction between time and leadership (� � 0.77,p � .05) were significantly related to team behav-ioral coordination during team compilation (Hy-pothesis 4). Then, in the next step of our hypothesistesting, we included all of the independent vari-ables (including the two mediator variables) in theRCM equation for team performance. As shown inTable 3, which presents the results of this step, thecoefficient for the interaction term involving timeand leadership failed to reach significance (� �1.24, n.s.), whereas both team learning behaviors (�� 0.39, p � .05) and behavioral coordination (� �1.26, p � .01) were positively and significantlyrelated to team performance over time.

To estimate the strength of the indirect effect, wefollowed the recommendation of Bauer, Preacher,and Gil (2006) and calculated the 95% confidence

TABLE 2Results of Random Coefficient Models for Team

Performance over Timea

Model and ParameterParameterEstimate s.e. t

Final level 2 modelIntercept �28.18 26.31 �1.07Time 2.13* 1.01 2.11Team cognitive ability 2.60 6.63 0.39Computer game experience 4.40 4.20 1.05Leadership �3.42 3.00 �1.14Time � leadership 0.34* 0.16 2.12

a For all level 1 parameter estimates, df � 538; for cross-levelinteraction parameters in level 2 analyses, df � 538; for param-eters preceding intercept variation in level 2 analyses, df � 56.Leadership is a dichotomous variable with 1 � “empoweringleadership” and 0 � “directive leadership.”

* p � .05

FIGURE 3Interactive Influence of Time and Leadership on Team Performance


interval of the indirect effect by employing theMonte Carlo method for assessing mediation(MCMAM). The calculated confidence interval es-timating the strength of the team learning indirecteffect (CI � 0.03; 1.07] did not include zero, indi-cating a significant indirect effect through teamlearning (Hypothesis 3). The confidence intervalaround the indirect effect of team behavioral coor-dination (CI � 0.05; 2.24) also excluded zero, indi-cating that behavioral coordination accounted forsignificant variance in team performance improve-ment (Hypothesis 4). Cumulatively these findingssupported both hypotheses.

Hypotheses 5 and 6. Hypothesis 5 proposes thatteam empowerment partially explains the greater

performance improvement for empowered teams(compared to directed teams) in the team compila-tion phase of team development. Hypothesis 6 pro-poses that team mental models also serve as a si-multaneous mediator to explain the greaterperformance improvement of empowered teams.As noted in the Methods section, we used GLSregression to account for the mediating role of teamempowerment (Hypothesis 5) and mental models(Hypothesis 6) in the team compilation phase ofdevelopment (rounds 6–10). Our results supportedboth Hypothesis 5 and Hypothesis 6. Specifically,in line with Shrout and Bolger’s (2002) recommen-dations, the leadership condition was positivelyand significantly related to each of the mediators (�� .12, p � .05: Table 4, model 1; and � � .53, p �.01, Table 4, model 2), providing support for ourreasoning that leadership explains variability inteam motivational (team empowerment) and cogni-tive (team mental models) states. Then, in the finalstep of the mediation testing, we found that teamempowerment (� � 26.14, p � .01, Table 4, model3) was significantly and positively related to teamperformance, and the interaction term involvingtime and leadership failed to reach significance.The 95% bootstrapped confidence interval estimat-ing the strength of the indirect effect excluded zero(CI � 0.95; 5.95). Together, these findings providedsupport for Hypothesis 5.

Team mental model development was also posi-tively and significantly related to team perfor-mance when included in the estimation of the fullmediation model (together with team empower-ment, leadership condition, and the interaction be-tween time and leadership condition: � � 2.26, p �.05). The interaction of time and leadership failed

TABLE 3Results of Random Coefficient Models for

Mediators Testinga

Model and ParameterParameterEstimate s.e. t

Final level 2 modelIntercept �37.32 82.26 �0.45Time 3.58 1.92 1.86Team cognitive ability 21.68 20.45 1.06Computer game experience �1.23 12.86 �0.10Leadership 12.37 17.69 0.70Time � leadership 1.24 2.59 0.50Team learning 0.39* 0.17 2.17Behavioral coordination 1.26** 0.47 2.69

a For all level 1 parameter estimates, df � 234; for cross-levelinteraction parameters in level 2 analyses, df � 234; for param-eters preceding intercept variation in level 2 analyses, df � 55.Leadership is a dichotomous variable with 1 � “empoweringleadership” and 0 � “directive leadership.”

* p � .05** p � .01

TABLE 4Results of GLS for Testing Team Empowerment and Team Mental Models as Mediators

VariableModel 1:

Empowerment as OutcomeModel 2:

Team Mental Models as OutcomeModel 3:

Performance as Outcome

Intercept 1.96** 1.61 �15.47**Team cognitive ability 0.51* 0.07 3.31Computer game experience 0.07 0.75* 0.08Leadership 0.12* 0.53** 15.58Team empowerment 26.14**Team mental models 2.26*Time 3.87Time � leadership 2.60df 295 295 285

* p � .05** p � .01


to reach significance in the full mediation model,and the confidence interval estimating the strengthof the indirect effect excluded zero (CI � 0.17;2.51), supporting Hypothesis 6.

DISCUSSION

Recent theoretical developments and organiza-tional trends have highlighted the benefits of em-powering leadership for individuals and teams(Chen et al., 2011; Srivastava et al., 2006; Zhang &Bartol, 2010). However, research examining the ef-fectiveness of empowering leadership on team per-formance has been far from conclusive, with anumber of studies finding evidence that directiveleadership may actually lead to higher perfor-mance, especially in the cases of action and projectteams (Ensley et al., 2006; Yun et al., 2005).

To help resolve this debate and to extend existingtheory to clarify some of the theoretical ambiguityregarding their relative benefits, we investigatedthe influence of empowering and directive leader-ship on the performance of newly formed teamsover time. Our results indicated that althoughteams with directive leaders started performingwell more quickly, their performance plateaued,whereas the emergent cognitions and improvedlearning and coordination capabilities of empow-ered teams allowed them to improve over time.These findings offer a number of theoretical andpractical implications.

Theoretical Implications

Departing from previous research, which has pre-dominantly considered leadership types in isola-tion (e.g., Mathieu et al., 2006; Srivastava et al.,2006), we directly compare two distinct leadershiptypes to resolve conflicting evidence in the litera-ture about the effectiveness of these leadership ap-proaches for team performance. Doing so providesinsight into both when and why empowering anddirective leadership approaches are most effectivein teams and contributes to debate as to the limitsand benefits of empowerment (e.g., Cotton, Voll-rath, Lengnick-Hall, & Froggatt, 1990; Leana, Locke,& Schweiger, 1990; Spreitzer, 1995; Wagner, 1994).

Conceptually, our findings confirm the existingnotion about the positive influence of empoweringleadership found in the literature for individualperformance (e.g., Ahearne et al., 2005; Zhang &Bartol, 2010) and for long-standing top manage-ment teams (Srivastava et al., 2006) and extend

them to action and project teams that undergo teamdevelopment before reaching their full potential(Kozlowski et al., 1999). However, by demonstrat-ing that empowering leadership comes at an initialperformance cost, we highlight an importantboundary condition to empowering styles that mayhelp explain some of the inconsistent or weakereffects on performance described in the literature(e.g., Mathieu et al., 2006).

We also extend the empowering leadership liter-ature to consider the critical role of team develop-mental phases, demonstrating that the influence ofleadership manifests directly through emergentteam processes and states. Empowering leaders en-courage team members to engage in role exchangesand collective investigation in the early role com-pilation phase of development, in which they learnabout their task environment and each others’ areasof expertise to develop team mental models of howto integrate their efforts, gain collective efficacyand commitment through psychological empower-ment, and foster routines to coordinate their behav-iors (Kozlowski et al., 1999; Pearsall et al., 2010).This time-consuming process puts them at a per-formance disadvantage compared to teams with di-rective leaders (who immediately focus on task per-formance) that rely on their leaders to provideexplicit within-team coordination. However, teamswith directive leaders are eventually overtaken byempowered teams as they enter the team compila-tion phase of development and adaptation, inwhich teams increasingly rely on their routinizedprocesses and shared cognitions to smoothly coor-dinate their efforts and knowledge and continu-ously improve their performance (e.g., Edwards etal., 2006; Marks et al., 2000).

By taking a longitudinal approach to examiningthese two leadership styles, we provide a morenuanced explanation for previous findings. A sim-ple cross-sectional analysis of the final perfor-mance in each condition would have revealed nosignificant difference between styles at that singlepoint in time. Similarly, a midpoint evaluationwould have suggested that directive leadership ismuch more effective than empowering. However,both approaches would have missed the complexrelationship between leadership and team memberinteractions that plays out over time and may ex-plain why previous work has been equivocal.These findings, therefore, help extend both theleadership and team development literatures andprovide a framework for understanding when and


at what stages of team development directing teammembers may positively impact team performance.

Finally, we extend situational leadership theoryby identifying four critical motivational, cognitive,and behavioral markers of team development andreadiness that help explain why the influence ofempowering leadership in the role compilationphase promotes higher continual performance im-provement in the team compilation phase of devel-opment. The current conceptualization of followerreadiness tends to focus on the maturity, commit-ment, and skill level of individual followers but toneglect the critical team processes and emergentstates that allow empowering leadership to be ef-fective at the team level. Teams with diverse exper-tise, working interdependently on complex tasks,require an investment in time and leader support todevelop the behavioral processes, such as learningand coordination, as well as the shared cognitiveunderstanding to effectively perform and adapt.This investment, made through the empowermentof team members during role compilation, takeslonger to pay off, but is critical for such teams’long-term success. However, because directiveleaders limit the emergence of these states and pro-cesses, these results also raise questions about thecentral tenet of situational leadership theory, thatteams with an initial directive leader will eventu-ally become ready for that leader to switch to anempowering style.

Managerial Implications

These findings also have number of implicationsfor managers placed in charge of teams. First, thetime-sensitive nature of the benefits of each leader-ship style provides a framework for managers tofollow depending on the timeline of their task orproject. For teams with short-term or emergent en-gagements and teams facing emergency situations(e.g., surgical, police, military, and flight teams), adirective style may be most appropriate, as teamsmust be able to immediately perform at a high leveland cannot afford the performance delays andlearning errors associated with empowered teams(e.g., Bobic & Davis, 2003; Sims et al., 2009; Yun etal., 2005). However, when teams have an extendedtimeline, as do project or software developmentteams, or must be able to adapt to complex andchanging environments over time, an initial andcontinuing empowering leadership style may bemost appropriate, as it encourages the develop-ment of shared cognitive structures, routines for

learning and coordination, and feelings of collec-tive competence and commitment during the rolecompilation phase that set the stage for higher long-term performance.

However, it is important that managers do notmisinterpret these findings to conclude that a di-rective leadership style is beneficial early, but thatone should switch to an empowering style as teamsprogress to the team compilation phase of develop-ment. Although there may be some advantage toemploying a combination of the two leadershipapproaches (e.g., Gratton & Erickson, 2007), ourresults suggest that the benefits of empoweringleadership in teams tended to manifest becauseteam members initially engaged in role identifica-tion and learning processes during the role compi-lation phase. Empowered teams, therefore, may notbe able to reap the benefits of improved perfor-mance over time without first suffering the initialperformance delays.

Finally owing to the critical role of team learningin empowered teams, managers who seek to adoptan empowering leadership style should also focustheir efforts on facilitating a team learning and in-formation-sharing orientation (e.g., Burke et al.,2006; Bunderson & Sutcliffe, 2003). They can do soby encouraging risk taking and experimentation,and the development of a psychologically safe en-vironment (e.g., Edmondson, 1999; Edmondson,Bohmer, & Pisano, 2001). Recently, Walumbwa andSchaubroeck (2009) found that leaders who en-gaged in active, ethical leadership created feelingsof psychological safety in their followers, leading tohigher levels of voice behavior. This suggests thatempowering leaders who genuinely value the in-puts of their team members and allow them to failwithout repercussion may be best able to cultivatean autonomous learning environment. However, italso raises the possibility that leaders who attemptto fake an empowering manner, without having anyreal willingness to sacrifice their own authority,may be thought of as disingenuous, and the teammay never develop the levels of safety so necessaryfor psychological empowerment and learning.

Limitations and Directions for Future Research

This study also has a number of limitations. First,the scope of this study was restricted to relativelyshort-term teams put together for a specific com-plex task (Sundstrom, 1999). In addition, the teamsin this study enjoyed very high levels of interde-pendence, which, coupled with a comparatively


short life cycle, may limit the generalizability ofour findings to action and project teams. Althoughthese types of teams are increasingly common inorganizations, it is possible that the relationshipswe examined in this study may evolve differentlyin long-standing teams or in teams with lower lev-els of interdependence. For example, in teams withlow interdependence, the role specification andtask assignments related to directive leadershipmay lead to even greater performance benefits,whereas the performance improvements associatedwith empowering leadership may never fully ma-terialize. Similarly, creative teams and teams facinggreater task ambiguity may not develop or benefitfrom the shared cognitions and enhanced coordi-nation associated with empowering leadership.However, those types of teams may still gain higherlevels of creative engagement and performancefrom feelings of engagement and commitment(Zhang & Bartol, 2010).

Second, because this study was conducted in alaboratory, research would benefit from establish-ing the external validity of our proposed relation-ships in a field setting. Although controlled set-tings offer the clear advantage of manipulatingvariables of interest and reducing causality con-cerns, certain features of our task might have biasedour results. For example, teams in this study oper-ated under a functional structure with distributedexpertise that required high levels of informationand knowledge exchange for the teams to be suc-cessful. A different task, employing a divisional orparallel team structure, for example, might haverequired less knowledge exchange among teammembers, activating different team learning pro-cesses and behaviors (Ilgen et al., 2005).

Also, these teams were assigned new, inexperi-enced leaders without any particular technical ex-pertise. Although such situations may be commonin many organizations (e.g., recent MBAs are oftenassigned to lead a product development team with-out having worked their way up through their neworganization), we acknowledge that our leader se-lection may not mirror the leader assignment pro-cess in organizations, in which leaders’ technicalskills and/or expertise are the basis of their projectassignments. It may be that high levels of skill andexperience are particularly beneficial for directiveleaders but less important for empowering ones(e.g., Vroom & Jago, 1988), and characteristics of theleaders as well as the teams may moderated theperformance trajectories of each leadership style inthis study. Further investigation of the important

role of leader ability in team development would bevaluable for extending our results and boundingthe external validity of our findings.

The long-term implications of our longitudinalresults are also somewhat ambiguous. Although theperformance of empowered teams increasedsteadily during the team compilation phase, andthe performance of teams with directive leaderswas flatter, it is impossible to state definitively howlong these trajectories would have continued. Al-though there is no reason to believe that the perfor-mance of teams with directive leaders would beginto improve in future rounds of task execution, it isnot clear how much longer teams with empoweringleaders would have continued to learn and im-prove, and it may be that empowering teams’ per-formance would have begun to plateau shortly afterour study ended.

Further, although our results suggest that teamswith initial directive leaders will not develop thereadiness to be empowered, we have little idea ofwhether initially empowered teams will benefitfrom future directive leadership. It may be thatteams have asymmetric capacities to adapt to newleaders, much as they differentially respond tochanges in structure (e.g., Moon et al., 2004) orrewards (e.g., Johnson, Hollenbeck, Humphrey, Il-gen, Jundt, & Meyer, 2006). Further research fo-cused on longer-term teams with multiple types ofleaders over time would help to resolve the extentof the influence of leadership on team developmentand performance.

Finally, although we examined the influence ofleadership over the multiple phases of team devel-opment described by Kozlowski et al. (1999), wecannot definitively identify the beginning or end ofeach phase. Teams develop through a “continuousseries of phases, with partial overlap at transitions”(Kozlowski et al., 1999: 248), and the timing of ourmeasures was intended to capture the essence ofthat transition rather than to conclusively, dis-cretely demarcate each phase. Therefore, it wouldbe valuable for future research to examine the lon-gitudinal benefits of empowering leadership over alonger-term task or series of tasks in which theduration of each developmental phase varies.

REFERENCES

Ahearne, M., Mathieu, J., & Rapp, A. 2005. To empoweror not to empower your sales force? An empiricalexamination of the influence of leadership empow-erment behavior on customer satisfaction and per-


formance. Journal of Applied Psychology, 90: 945–955.

Arnold, J. A., Arad, S., Rhoades, J. A., & Drasgow, F.2000. The empowering leadership questionnaire:The construction and validation of a new scale formeasuring leader behaviors. Journal of Organiza-tional Behavior, 21: 249–269.

Barling, J., Weber, T., & Kelloway, E. K. 1996. Effects oftransformational leadership training on attitudinaland financial outcomes: A field experiment. Journalof Applied Psychology, 81: 827–832.

Bauer, D. J., Preacher, K. J., & Gil, K. M. 2006. Conceptu-alizing and testing random indirect effects and mod-erated mediation in multilevel models: New proce-dures and recommendations. PsychologicalMethods, 11: 142–163.

Blanchard, K. H. 2007. Leading at a higher level. UpperSaddle River, NJ: Prentice Hall.

Bliese, P. D. 2000. Within-group agreement, non-inde-pendence, and reliability: Implications for data ag-gregation and analysis. In K. J. Klein & S. W. J.Kozlowski (Eds.), Multilevel theory, research andmethods in organizations: 349–381. San Francisco:Jossey-Bass.

Bliese, P. D., & Ployhart, R. E. 2002. Growth modelingusing random coefficient models: Model building,testing, and illustrations. Organizational ResearchMethods, 5: 362–387.

Bobic, M. P., & Davis, W. E. 2003. A kind word for theoryX: Or why so many newfangled management tech-niques quickly fail. Journal of Public Administra-tion Research and Theory, 13: 239–264.

Bunderson, J. S., & Sutcliffe, K. M. 2003. When to put thebrakes on learning? Harvard Business Review,81(2): 20–21.

Burke, C. S., Stagl, K. C., Klein, C., Goodwin, G. F., Salas,E., & Halpin, S. M. 2006. What type of leadershipbehaviors are functional in teams? A meta-analysis.Leadership Quarterly, 17: 288–307.

Burke, C. S., Stagl, K. C., Salas, E., Pierce, L., & Kendall,D. 2006. Understanding team adaptation: A concep-tual analysis and model. Journal of Applied Psy-chology, 91: 1189–1207.

Chen, G., Kirkman, B. L., Kanfer, R., Allen, D., & Rosen,B. 2007. A multilevel study of leadership, empower-ment, and performance in teams. Journal of AppliedPsychology, 92: 331–346.

Chen, G., Sharma, P. N., Edinger, S. K., Shapiro, D. L., &Farh, J. L. 2011. Motivating and de-motivating forcesin teams: Cross-level influences of empowering lead-

ership and relationship conflict. Journal of AppliedPsycology, 96: 541–557.

Cohen, S. G., Chang, L., & Ledford, G. E., Jr. 1997. Ahierarchical construct of self-management leader-ship and its relationship to quality of work life andperceived work group effectiveness. Personnel Psy-chology, 50: 275–308.

Cotton, J. L., Vollrath, D. A., Lengnick-Hall, M. L., &Froggatt, K. L. 1990. Fact: The form of participationdoes matter—A rebuttal to Leana, Locke, andSchweiger. Academy of Management Review, 15:147–153.

Cox, J. F., & Sims, H. P., Jr. 1996. Leadership and teamcitizenship behavior: A model and measures. InM. M. Beyerlein & D. A. Johnson (Eds.), Advances ininterdisciplinary studies of work teams, vol. 3:1–14. Greenwich, CT: JAI.

Davison, R. B., Hollenbeck, J. R., Barnes, C. M., Sleesman,D. J., & Ilgen, D. R. 2012. Coordinated action inmulti-team systems. Journal of Applied Psychology,97: 808–824.

Durham, C. C., Knight, D., & Locke, E. A. 1997. Effects ofleader role, team-set goal difficulty, efficacy, andtactics on team effectiveness. Organizational Be-havior and Human Decision Processes, 72: 203–231.

Edmondson, A. C. 1999. Psychological safety and learn-ing behavior in work teams. Administrative ScienceQuarterly, 44: 350–383.

Edmondson, A. C., Bohmer, R. M., & Pisano, G. P. 2001.Disrupted routines: Team learning and new technol-ogy implementation in hospitals. AdministrativeScience Quarterly, 46: 685–716.

Edwards, B. D., Day, E. A., Arthur, W., & Bell, S. T. 2006.Relationships among team ability composition, teammental models, and team performance. Journal ofApplied Psychology, 91: 727–736.

Ellis, A. P. J. 2006. System breakdown: The role of sharedmental models and transactive memory in the rela-tionship between acute stress and tem performance.Academy of Management Journal, 49: 576–589.

Ensley, M. D., Hmieleski, K. M., & Pearce, C. L. 2006. Theimportance of vertical and shared leadership withinnew venture top management teams: Implicationsfor the performance of startups. Leadership Quar-terly, 17: 217–231.

Espinosa, J. A., Lerch, F. J., & Kraut, R. E. 2004. Explicitversus implicit coordination mechanisms and taskdependencies: One size does not fit all. In E. Salas &S. M. Fiore (Eds.), Team cognition: Understandingthe factors that drive process and performance:


107–129. Washington, DC: American PsychologicalAssociation.

Evans, M. 1970. The effects of supervisory behavior onthe path-goal relationship. Organizational Behav-ior and Human Performance, 5: 277–298.

Fiedler, F. E. 1968. Personality and situational determi-nants of leadership effectiveness. In D. Cartwright &A. Zanders (Eds.), Group dynamics: Research andtheory: 362–380. New York: Harper & Row.

Gibson, C., & Vermeulen, F. 2003. A healthy divide:Subgroups as a stimulus for team learning behavior.Administrative Science Quarterly, 48: 202–239.

Goodson, J. R., McGee, G. W., & Cashman, J. F. 1989.Situational leadership theory: A test of leadershipprescriptions. Group & Organization Studies, 14:446–461.

Graen, G. B., & Scandura, T. A. 1987. Towards a psychol-ogy of dyadic organizing. In L. L. Cummings & B. M.Staw (Eds.), Research in organizational behavior,vol. 9: 175–208. Greenwich, CT: JAI.

Gratton, L., & Erickson, T. J. 2007. Eight ways to buildcollaborative teams. Harvard Business Review,11(11): 101–109.

Hersey, P., & Blanchard, K. H. 1969. Management oforganizational behavior: Utilizing human re-sources. Englewood Cliffs, NJ: Prentice Hall.

Hersey, P., & Blanchard, K. H. 1982. Management oforganizational behavior: Utilizing human resources(4th ed.). Englewood Cliffs, NJ: Prentice Hall.

Hersey, P., & Blanchard, K. H. 1988. Management oforganizational behavior (5th ed.): 169–201. Engle-wood Cliffs, NJ: Prentice Hall.

Hinsz, V. B., Tindale, R. S., & Vollrath, D. A. 1997. Theemerging conceptualization of groups as informationprocessors. Psychological Bulletin, 121: 43–64.

House, R. J. 1971. A path-goal theory of leader effective-ness. Administrative Science Quarterly, 16: 321–338.

House, R. J. 1996. Path-goal theory of leadership: Les-sons, legacy, and a reformulated theory. LeadershipQuarterly, 7: 323–352.

Ilgen, D. R., Hollenbeck, J. R., Johnson, M., & Jundt, D.2005. Teams in organizations: From input-process-output models to IMOI models. In S. Fiske, D. L.Schacter, & C. Zahn-Wexler (Eds.), Annual reviewof psychology, vol. 56: 517–543. Palo Alto, CA: An-nual Reviews.

James, L. R. 1982. Aggregation bias in estimates of per-ceptual agreement. Journal of Applied Psychology,67: 219–229.

James, L. R., Demaree, R. G., & Wolf, G. 1984. Estimatingwithin group interrater reliability with and withoutresponse bias. Journal of Applied Psychology, 69:85–98.

Johnson, M., Hollenbeck, J., Humphrey, S., Ilgen, D.,Jundt, D., & Meyer, C. 2006. Cutthroat cooperation:Asymmetrical adaptation to changes in team rewardstructures. Academy of Management Journal, 49:103–119.

Kahai, S. S., Sosik, J. J., & Avolio, B. J. 2004. Effects ofparticipative and directive leadership in electronicgroups. Group & Organization Management, 29:67–105.

Katz, D., & Kahn, R. L. 1978. The social psychology oforganizations (2nd ed.). New York: Wiley.

Katzenbach, J. R., & Smith, D. K. 1993. The wisdom ofteams: Creating the high performance organiza-tion. Cambridge, MA: Harvard Business SchoolPress.

Kirkman, B. L., & Rosen, B. 1999. Beyond self-manage-ment: Antecedents and consequences of team em-powerment. Academy of Management Journal, 42:58–74

Kirkman, B. L., Rosen, B., Tesluk, T., & Gibson, C. 2004.The impact of team empowerment on virtual teamperformance: The moderating role of face-to-face in-teraction. Academy of Management Journal, 47:175–192.

Klein, K. J., Ziegert, J. C., Knight, A. P., & Xiao, Y. 2006.Dynamic delegation: Shared, hierarchical, and dein-dividualized leadership in extreme action teams.Administrative Science Quarterly, 51: 590–621.

Klein, K., & Kozlowski, S. W. J. 2000. Multilevel theory,research and methods in organizations. San Fran-cisco: Jossey-Bass.

Korsgaard, M. A., Schweiger, D. M., & Sapienza, H. J.1995. Building commitment, attachment, and trustin strategic decision-making teams: The role of pro-cedural justice. Academy of Management Journal,38: 60–84.

Kozlowski, S. W. J., Gully, S. M., Nason, E. R., & Smith,E. M. 1999. Developing adaptive teams: A theory ofcompilation and performance across levels and time.In D. R. Ilgen & E. D. Pulakos (Eds.), The changingnature of performance: Implications for staffing,personnel actions, & development: 240–292. SanFrancisco: Jossey-Bass.

Kozlowski, S. W. J., Gully, S. M., Salas, E., & Cannon-Bowers, J. A. 1996. Team leadership and develop-ment: Theory, principles, and guidelines for trainingleaders and teams. In M. M. Beyerlein, D. Johnson, &


S. T. Beyerlein (Eds.), Interdisciplinary studies ofwork teams: 253–291. Greenwich, CT: JAI.

Leana, C. R. 1987. Power relinquishment versus powersharing: Theoretical clarification and empirical com-parison of delegation and participation. Journal ofApplied Psychology, 72: 228–233.

Leana, C. R., Locke, E. A., & Schweiger, D. M. 1990. Factand fiction in analyzing research on participativedecision making: A critique of Cotton, Vollrath,Lengnick-Hall and Jennings. Academy of Manage-ment Review, 15: 137–146.

LePine, J. A. 2003. Team adaptation and post-changeperformance: Effects of team composition in terms ofmembers’ cognitive ability and personality. Journalof Applied Psychology, 88: 27–39.

MacKinnon, D. P., Fairchild, A. J., & Fritz, M. S. 2007.Mediation analysis. In M. I. Posner & M. K. Rothbart(Eds.), Annual review of psychology, vol. 58: 593–614. Palo Alto, CA: Annual Reviews.

Manz, C. C., & Sims, H. P. Jr. 1986. Beyond imitation:Complex behavioral and affective linkages resultingfrom exposure to leadership training models. Jour-nal of Applied Psychology, 71: 571–578.

Manz, C. C., & Sims, H. P. Jr. 1987. Leading workers tolead themselves: The external leadership of self-managing work teams. Administrative ScienceQuarterly, 32: 106–129.

Marks, M. A., Mathieu, J. E., & Zaccaro, S. J. 2001. Atemporally based framework and taxonomy of teamprocesses. Academy of Management Review, 26:356–376.

Marks, M. A., Zaccaro, S. J., & Mathieu, J. E. 2000. Per-formance implications of leader briefings and teaminteraction training for adaptation to novel environ-ments. Journal of Applied Psychology, 6: 971–986.

Mathieu, J. E., Gilson, L. L., & Ruddy, T. M. 2006. Em-powerment and team effectiveness: An empiricaltest of an integrated model. Journal of Applied Psy-chology, 91: 97–108.

Mathieu, J. E., Heffner, T. S., Goodwin, G. F., Cannon-Bowers, J. A., & Salas, E. 2005. Scaling the quality ofteammates’ mental models: Equifinality and norma-tive comparisons. Journal of Organizational Be-havior, 26: 37–56.

Mathieu, J. E., Maynard, M. T., Rapp, T., & Gilson, L.2008. Team effectiveness 1997–2007: A review ofrecent advancements and a glimpse into the future.Journal of Management, 34: 410–476.

Mathieu, J., Ahearne, M., & Taylor, S. R. 2007. A longi-tudinal cross-level model of leader and salespersoninfluences on sales force technology use and perfor-

mance. Journal of Applied Psychology, 92: 528–537.

McIntyre, R., & Salas, E. 1995. Measuring and managingfor team performance: Emerging principles fromcomplex environments. In R. Guzzo & E. Salas (Eds.),Team effectiveness and decision making in organ-izations: 9–45. San Francisco: Jossey-Bass.

Mohammed, S., Klimoski, R., & Rentsch, J. R. 2000. Themeasurement of team mental models: We have noshared schema. Organizational Research Methods,3: 123–165.

Moon, H., Hollenbeck, J. R., Humphrey, S. E., Ilgen, D. R.,West, B. J., Ellis, A. P. J., & Porter, C. O. L. H. 2004.Asymmetric adaptability: Dynamic team structuresas one-way streets. Academy of Management Jour-nal, 47: 681–696.

Muczyk, J. P., & Reimann, B. C. 1987. The case for direc-tive leadership. Academy of Management Execu-tive, 1(4): 301–311.

Muller, D., Judd, C. M., & Yzerbyt, V. Y. 2005. Whenmoderation is mediated and mediation is moderated.Journal of Personality and Social Psychology, 89:852–863.

Northouse, P. G. 2007. Leadership: Theory and prac-tice. Thousand Oaks, CA: Sage.

Pearce, C., Sims, H. P., Cox, J., Ball, G., Schnell, E., &Smith, K. 2003. Transactors, transformers and be-yond. Journal of Management Development, 22:273–307.

Pearsall, M. J., Ellis, A. P., & Bell, B. S. 2010. Building theinfrastructure: The effects of role identification be-haviors on team cognition development and perfor-mance. Journal of Applied Psychology, 95: 192–200.

Preacher, K. J., Rucker, D. D., & Hayes, A. F. 2007. Ad-dressing moderated mediation hypotheses: Theory,methods, and prescriptions. Multivariate Behav-ioral Research, 42: 185–227.

Sagie, A. 1996. Effects of leader’s communication styleand participative goal setting on performance andattitudes. Human Performance, 9: 51–64.

Sagie, A. 1997. Leader direction and employee participa-tion in decision making: Contradictory or compati-ble practices? Applied Psychology: An Interna-tional Review, 46: 387–452.

Sagie, A., Zaidman, N., Amichai-Hamburger, Y., Te’eni,D. S., & Schwartz, D. G. 2002. An empirical assess-ment of the loose-tight leadership model: Quantita-tive and qualitative analyses. Journal of Organiza-tional Behavior, 3: 303–320.


Seibert, S. E., Wang, G., & Courtright, S. H. 2011. Ante-cedents and consequences of psychological andteam empowerment in organizations: A meta-ana-lytic review. Journal of Applied Psychology, 96:981–1003.

Shrout, P. E., & Bolger, N. 2002. Mediation in experimen-tal and nonexperimental studies: New proceduresand recommendations. Psychological Methods, 7:422–445.

Sims, H. P., Faraj, S., & Yun, S. 2009. When should aleader be directive or empowering? How to developyour own situational theory of leadership. BusinessHorizons, 52(2): 149–158.

Somech, A. 2006. The effects of leadership style andteam process on performance and innovation infunctionally heterogeneous teams. Journal of Man-agement, 32: 132–157.

Sosik, J. J., Avolio, B. J., & Kahai, S. S. 1997. Effects ofleadership style and anonymity on group potencyand effectiveness in a group decision support systemenvironment. Journal of Applied Psychology, 82:89–103.

Spreitzer, G. M. 1995. Psychological empowerment inthe workplace: Dimensions, measurement, and vali-dation. Academy of Management Journal, 38:1442–1465.

Spreitzer, G. M. 2008. Taking stock: A review of morethan twenty years of research on empowerment atwork. In J. Barling & C. L. Cooper (Eds.), The Sagehandbook of organizational behavior: 73–88.Thousand Oaks, CA: Sage.

Srivastava, A., Bartol, K. M., & Locke, E. A. 2006. Em-powering leadership in management teams: Effectson knowledge sharing, efficacy and performance.Academy of Management Journal, 49: 1239–1251.

Sundstrom, E. 1999. The challenges of supporting workteam effectiveness. In E. Sundstrom & associates(Eds.), Supporting work team effectiveness: 3–23.San Francisco: Jossey–Bass.

Thomas, K. W., & Velthouse, B. A. 1990. Cognitive ele-ments of empowerment: An “interpretive” model ofintrinsic task motivation. Academy of ManagementReview, 15: 666–681.

Thompson, G., & Vecchio, R. P. 2009. Situational leader-ship theory: A test of three versions. LeadershipQuarterly, 20: 837–848.

Towler, A. J. 2003. Effects of charismatic influence train-ing on attitudes, behavior, and performance. Person-nel Psychology, 56: 363–381.

Vecchio, R. P. 1987. Situational leadership theory: Anexamination of a perspective theory. Journal of Ap-plied Psychology, 72: 444–451.

Vroom, V. H., & Jago, A. G. 1988. The new leadership:Managing participation in organizations. Engle-wood Cliffs, NJ: Prentice Hall.

Wagner, J. A. 1994. Participation’s effects on perfor-mance and satisfaction: A reconsideration of the ev-idence. Academy of Management Review, 19: 312–330.

Walumbwa, F. O., & Schaubroeck, J. 2009. Leader per-sonality traits and employee voice behavior: Medi-ating roles of ethical leadership and work grouppsychological safety. Journal of Applied Psychol-ogy, 94: 1275–1286.

Yukl, G. A. 1998. Leadership in organizations (4th ed.).Englewood Cliffs, NJ: Prentice Hall.

Yukl, G. A., & Lepsinger, R. 2004. Flexible leadership:Creating value by balancing multiple challengesand choices. San Francisco: Jossey-Bass.

Yukl, G. 2010. Leadership in organizations (7th ed.).Upper Saddle River, NJ: Prentice Hall.

Yukl, G., & Falbe, C. M. 1991. Importance of differentpower sources in downward and lateral relations.Journal of Applied Psychology, 76: 416–423.

Yun, S., Faraj, S., & Sims, H. P. 2005. Contingent leader-ship and effectiveness of trauma resuscitation teams.Journal of Applied Psychology, 90: 1288–1296.

Zaccaro, S. J., Rittman, A. L., & Marks, M. A. 2001. Teamleadership. Leadership Quarterly, 12: 451–483.

Zhang, X., & Bartol, K. M. 2010. Linking empoweringleadership and employee creativity: The influence ofpsychological empowerment, intrinsic motivation,and creative process engagement. Academy of Man-agement Journal, 53: 107–128.


APPENDIXPrescripted Leader Behavior Prompter Phrases

Directive Leader Suggested Phrases Empowering Leader Suggested Phrases

● OK, good performance so far.● Now, I need you to listen carefully to my instructions so we can do

better in the next rounds.● I expect you to stick to your roles and execute them with great

diligence.● Make sure you know what your own assets are and operate them

accordingly.● I want the intelligence players to send all their assets early in the

round in the upper/lower half of the grid.● Operation players, make sure you refuel when operating in the

upper grid.● Operation players, pay close attention to the mobile targets.● Our tactics for the next round is . . . you all need to follow this tactic.

● OK, well done so far, team.● I encourage all of you to communicate more in order to

improve our team performance.● OK, let’s discuss our performance and decide what we

want to do in order to improve it. Any ideas?● Let’s try to work together—we need ideas.● We need all the information you have on the table—tell

your team members what you see and think.● How about we try something more creative for the next

round? What do you think?● We need to work together as a team—it is up to us to find

a way to get more points.

Natalia M. Lorinkova ([email protected]) isan assistant professor in the Management and IS Depart-ment in the School of Business Administration at WayneState University. She received her Ph.D. in managementfrom the R. H. Smith School of Business, University ofMaryland. Her primary research interest is in the area ofleadership, with an emphasis on change and on devel-opmental and empowering leadership.

Matthew J. Pearsall ([email protected]) is anassistant professor in the Organizational Behavior De-partment in the Kenan-Flagler Business School at the

University of North Carolina at Chapel Hill. He receivedhis Ph.D. in management from the University of Arizona.His research interests include the development of emer-gent states in teams and the antecedents and outcomes ofshared cognition.

Henry P. Sims Jr. is professor emeritus at the R H. SmithSchool of Business, University of Maryland. He receivedhis Ph.D. from Michigan State University. His researchfocuses on leadership and teamwork, especially the issueof how empowering leadership and teamwork can lead toself-leadership at both the individual and team levels.


examining the differential longitudinal performance …leeds-faculty.colorado.edu/dahe7472/lorinkova...

Documents