ایجاد ارتباط بین رفتار رهبر و اجماع رهبری به عملکرد تیم : یکپارچه سازی اجماع مستقیم و مدل های پراکندگی ترکیب گروه

We sought to establish whether mean level within-team assessments of a leader's transformational behavior and the extent of perceived variability (i.e., consensus) among team members’ ratings around this mean level are separate yet related indicators of leader quality. To this end, using data from 108 work teams in a multinational field setting, we explored the relationship between managers’ transformational leadership behavior and team performance as moderated by the extent of variability among team members’ leadership behavior ratings. Recognizing that the relationship between leader behavior and team performance is indirect, we further examined whether team empowerment served as a mediating mechanism through which transformational leadership is evoked. Study results, based on ordinary least squares (OLS) regression analyses and bootstrapped estimates, were consistent with the hypothesized conceptual scheme of moderated mediation, in that the joint effects of transformational leadership behavior and consensus about transformational leadership were found to have an indirect effect on team performance through team empowerment. These findings broaden the focus of transformational leadership theory by illustrating that, within a team context, the transformational leadership–performance linkage is more nuanced than previously believed.

In an effort to understand the antecedents of job performance, leadership research has traditionally emphasized the relationship between managers and subordinates as individuals (Day, Gronn, & Salas 2004). With the widespread shift to team-based organizations, however, there is mounting evidence that a manager's leadership behavior in supervising subordinates as a group influences the performance in and of teams ( Kozlowski and Ilgen, 2006 and Mathieu et al., 2008). Recognizing this shift in emphasis, Chen, Kirkman, Kanfer, Allen, and Rosen (2007), among others, have observed that a focus on group methods of leadership raises a series of yet unanswered questions. For example, what compromises may occur when managing both individuals and teams as a whole? Do efforts to lead and motivate teams affect individual leader–follower interactions, and what are the likely consequences for one-on-one relationships between team members? Are the behaviors necessary for building and leading high performing teams the same as for motivating performance at the individual level? Such questions have prompted leadership researchers (e.g., Chen & Kanfer 2006) to explore the interplay between simultaneously managing team members as individuals and as a group in toto. Illustrative of these researchers’ efforts are studies investigating leadership behavior as a group-level phenomenon. Bass, Avolio, Jung, and Berson (2003), for instance, were interested in examining how leader behaviors directed at teams as a whole relate to unit performance. In doing so, they relied on a direct consensus model of composition (Chan 1998). Consistent with the direct consensus model, individual and team effects are partitioned using group mean scores to measure the absolute level of a leader's behavior. Tests of within-group agreement (e.g., rwg; James, Demaree, & Wolf 1984) are typically then computed to determine whether creating aggregate scores from individual level data is empirically justifiable; unexplained variance is regarded as noise or measurement error (Kozlowski & Klein 2000). In contrast to this view, it has been increasingly argued that unexplained variance among team members’ assessments may convey relevant unit-level information rather than simply reflect error variance (Chan, 1998 and Harrison and Klein, 2007). Exploring this possibility, leadership researchers have recently considered what are termed dispersion models of group composition. In a dispersion composition model, within-group variance (or, alternatively, within-group agreement or consensus) is treated as a meaningful higher level construct rather than a statistical prerequisite for aggregation (Chan 1998). Within the leadership domain, the extent to which team members agree about the quality of their managers’ leadership behavior has been dubbed leadership consensus (Feinberg, Ostroff, & Burke 2005; see also Lindell & Brandt 2000). In a dispersion composition model, leadership consensus is considered high when team members’ perceptions of a leader's behavior are homogenous. Conversely, leadership consensus is considered low when team members’ perceptions of a leader's behavior are heterogonous or dissimilar. In either instance, the degree of consensus is, by definition, a group-level measure of the amount of variability in team members’ perceptions of the quality of their managers’ leadership behavior. As discussed below, evidence from a handful of studies suggests that team members’ consensual perceptions about the quality of their managers’ leadership behavior may be an important explanatory variable for understanding the influence of leadership within teams. The primary aim of the present study was to extend transformational leadership research by integrating a direct consensus model of composition and a dispersion model of group composition within a single framework. To this end, the present study contributes to the leadership literature in two ways. First, it answers recent calls for research exploring whether differences in leadership consensus moderate the relationship between the general (i.e., average) behavior exhibited by a leader and team performance (e.g., Cole and Bedeian, 2007 and Dawson et al., 2008). The current investigation focuses on transformational leader behaviors for two principal reasons: (a) they “dominate” current thinking about leadership research and (b) they are at the core of contemporary leadership theory (Judge & Piccolo 2004, p. 762). Second, recognizing that the relationship between leaders’ transformational behavior and team performance may not be direct, we cast team empowerment as a mediating mechanism in an attempt to replicate prior research (e.g., Chen et al. 2007). In doing so, the present study responds to Schaubroeck, Lam, and Cha's (2007) challenge that leadership researchers go beyond investigating transformational leadership's main effects (at the team level) and to examine not only the boundary conditions under which transformational leadership is most likely to be effective, but also the mediating processes by which transformational leadership exerts its effect on workplace outcomes. Thus, by simultaneously considering leadership consensus and team empowerment in a hypothesized model, we broaden the focus of transformational leadership theory by illustrating that the transformational leadership team performance linkage is more nuanced than previously believed.

Because leaders direct many of their behaviors to team members en bloc rather than to individual team members, researchers have increasingly settled on conceptualizing leadership phenomenon as shared, unit-level variables (i.e., a direct consensus approach). In doing so, they have neglected the possibility that effective leadership may also depend on a leader's ability to create a consensus or a shared reality among a team's members (Feinberg et al. 2005). This possibility suggests an alternative to predominate conceptual and analytic methods in mainstream leadership research. Marion and Uhl-Bien (2001) see this predominance as creating a classical problem of reductionism, wherein researchers offer a simpler explanation for a complex phenomenon. They argue that by overemphasizing managers as leaders and by underemphasizing the larger social system (e.g., interactions among team members) in which leadership occurs, researchers risk providing incomplete explanations for their findings (also see Uhl-Bien & Marion 2009). Marion and Uhl-Bien are essentially suggesting that this imbalance in emphasis is undermining that which we seek to understand most. Stated differently, because leadership researchers have opted to treat followers’ (i.e., subordinates’) response variability as measurement error rather than conveying meaningful information, contemporary research (especially that employing transformational leadership theory) tends to study a component of leadership (i.e., managers’ behavior) while ignoring the contextual factors necessary for the emergence of effective leadership. This raises the question of whether traditional leadership models are too simplistic and, thereby, inadequate for understanding the dynamic and emergent nature of leadership in and of teams. In response to concerns that current leadership thinking is inadequate, we suggest that the exchange between leaders and their team members in toto may shape members’ leadership attributions. In this regard, we investigated whether within-team mean level assessments of a leader's behavior and the extent of perceived variability among team members’ ratings around this mean level are separate yet related indicators of leadership effectiveness. We recognize that even though team members may hold similar perceptions about the quality of their manager's leadership behavior, variance among their perceptions may exist and, as our results suggest, this variance is of theoretical interest. Indeed, our results are noteworthy, in that, the alternative perspective they support more accurately reflects the transformational leadership process; that is, a social influence process that is characterized by a complex interplay between the absolute level of leader behavior and team members’ consensus regarding leadership within teams. For instance, on initial impression the finding related to the negative slope for the low consensus condition (see Fig. 2) may appear inconsistent with received theory. This result, however, makes a unique contribution by expanding recent theoretical advancements on leadership consensus. More specifically, from a theoretical perspective, social influence theory posits that consensus only matters on issues of relevance (Bliese and Halverson, 1998 and Festinger, 1950). Therefore, when absolute levels of transformational leadership are low, dissensus (i.e., low consensus) among team members is less likely to impact team functioning because the leadership behaviors in question are infrequently exhibited and, thus, less salient to team members. It moreover suggests that leaders may be capable of encouraging empowerment beliefs if they develop high quality relationships with at least some team members (Liden et al. 2006). On the other hand, low consensus is more likely to negatively influence a team's functioning as the frequency of exhibited leadership behavior increases from low to moderate to high. Leaders that are rated moderate to high in their transformational behavior, but who fail in creating aligned perceptions within their teams, may be perceived by members as developing idiosyncratic individual relationships (e.g., “we–they” distinctions and subgroup categorizations); hence, to the extent that leadership consensus is low and absolute levels of transformational behavior are high (increased salience), a team's empowerment is shown to decrease in magnitude as members struggle with interpersonal tensions and conflict. Our results make a further meaningful contribution by illustrating that the effects of transformational leadership are more subtle than previously believed. Our data demonstrate the extent to which team empowerment translates the effect of transformational leadership on team performance is contingent on team members’ consensus about their managers’ transformational leadership behavior—thereby yielding a pattern of moderated mediation. This finding broadens the focus of transformational leadership research; it also presents a more elaborate view of how managers’ transformational behavior influences performance in team contexts. On the basis of the current results, for example, one may conclude that “the attribution of transformational leadership to an individual depends on both the leader exhibiting a set of positive behaviors as well as also fostering consensus among subordinates in their perceptions of the leader” (Feinberg et al. 2005, p. 483). 9.1. Limitations and future research As in all research, the current study is not without potential limitations. Given that we measured the focal leadership variables, team empowerment, and team performance in the same survey, concerns for common-method variance should be considered when interpreting our results. The controls employed (e.g., positive team-affective tone and contingent-reward) should help to minimize this concern. Further, research has demonstrated that artifactual interactions cannot be created, although true interactions can be attenuated (Evans 1985). Taken together with our CFA results (Williams et al. 1989), in which we demonstrated that common-method variance did not significantly affect our ability to test study hypotheses, we believe that any concerns relating to common-method bias are lessened. Finally, method effects were further minimized in that we varied item response anchors and formats and collected data for each focal variable from multiple informants (Podsakoff, MacKenzie, Lee, & Podsakoff 2003). The cross-sectional design with which data were collected also limits us from drawing strong causal inferences. Whereas prevailing theory and published results indicate that the order of relationships tested is plausible, we nevertheless acknowledge that alternative relationship patterns may exist and should be explored in future research, including field experiments. Despite the inclusion of robust study covariates, a further limitation is that other controls were not considered (e.g., length of exposure to a leader). We also note that there is potential for bias in the current study's performance ratings. Because we cannot demonstrate that our perceptual measure is a valid predictor of more “objective” performance, one could argue that different results might be obtained for other performance measures (e.g., peer or supervisor ratings). Whereas the use of self-ascriptive ratings does not invalidate our findings (see, e.g., Heidemeier and Moser, 2009 and LeBreton et al., 2003), future studies that include additional, more objective performance measures would provide confidence in their robustness. A remaining limitation is that when within-team response rates fall below 100% (as in the present case), the estimation of team level properties is complicated by missing data (Allen et al., 2007 and Newman and Sin, 2009). Timmerman, 2005 and Allen et al., 2007, however, have examined relationships between team level (i.e., absolute means based on aggregated data and consensus scores based on SD) variables with various patterns of member nonresponse. Results from both studies demonstrate that random and not-at-random missing data attenuated team level relationships. These findings intimate that our results may actually underestimate the magnitude of the observed correlations between leadership behavior, leadership consensus, and performance. Beyond addressing study limitations, the present analysis suggests other interesting directions for future research. For example, our conceptual scheme is not exhaustive in considering all possible moderator variables. We focused on team members’ perceptual agreement about their managers’ leadership behavior, but did not consider consensus about team empowerment as a second moderator. Future research that expands our conceptual scheme to include moderated effects from team empowerment to team performance would be a further contribution. In a similar fashion, it would also be interesting to explore the within-team factors that predict the dispersion in team members’ performance judgments. It might also be worthwhile to consider alternative intervening variables as generative mechanisms between the joint effect of transformational leadership behavior and consensus on team performance. Research suggests that team identification and potency may be relevant in this respect (Kearney and Gebert, 2009 and Schaubroeck et al., 2007). By simultaneously investigating multiple mediators (cf. Preacher & Hayes 2008), scholars could provide a more detailed picture of team leadership processes and thus contribute to an improved explanation of work team performance. 9.2. Contributions to the literature and unanswered questions about leadership consensus To our knowledge, this is the first study to apply transformational leadership theory in an attempt to determine whether leaders’ behavior and leadership consensus variables interact in the prediction of team empowerment and performance. We developed an integrated conceptual scheme that is more elaborate than prior research has indicated and tested our hypotheses using a relatively large number of real teams in a multinational field setting. Further, we entered contingent-reward behavior as a study covariate because it is a basic proposition of transformational leadership theory that whereas contingent-reward behavior provides a “foundation for effective leadership and performance at expected standards,” transformational leadership provides “an enhancement engendering superlative leadership and performance beyond expectations” (Yammarino et al. 1998, p. 28). Including both forms of leadership behavior in our hypotheses tests (which is seldom done), thus, provides a more robust (and conservative) test of the proposed effects of transformational leadership (Judge & Piccolo 2004). In doing so, our findings make a contribution to the transformational leadership literature by demonstrating how and under what conditions team performance improvements may be realized. More broadly, our results also call into question the extent and magnitude of previously identified relationships between other team level leadership variables and outcomes. Such relationships may hinge on the extent of perceptual agreement among a team's members. This possibility builds upon the results obtained by Bliese and Britt, 2001, Bliese and Halverson, 1998, Cole and Bedeian, 2007 and Feinberg et al., 2005, all of whom concluded that leadership consensus constructs are worthy of continued examination. Integrating our results with those of the aforementioned researchers, the mounting evidence provides broad support for Klein and House's (1995) assertion that the consideration of homogeneity or variability in team members’ leadership perceptions may not only refine and clarify a meso-level leadership paradigm, but also suggest new topics for leadership research. For example, a fundamental unanswered question is “What explains why some teams reach consensus about leadership behaviors and others do not?” A number of possible responses seem particularly promising. One such response focuses on implicit leadership theory (see, e.g., Howell & Shamir 2005). It has been specifically suggested, for example, that based on experiences with leaders and through socialization, followers develop cognitive structures or prototypes specifying the traits and abilities that characterize the ideal business leader. According to implicit leadership theory, managers are categorized as leaders on the basis of the perceived match between the attributes of a preexisting leader prototype that followers hold in memory and managers’ actual behavior or character. This approach would suggest that implicit leadership theories are the benchmark followers use to form an impression of their managers and, therefore, future researchers may wish to investigate how these prototypes facilitate or hinder the emergence of transformational leadership consensus in teams. A second response to this question may be that other team factors (e.g., composition, interdependence, task type) influence the emergence of consensus about leadership in teams. For instance, team interdependence increases the amount and intensity of interaction among members. Increased interaction and dependence among team members may strengthen perceptual agreement in high consensus teams, whereas in low consensus teams, it may cause existing intra-team conflicts to have an intensified effect on team and individual outcomes. One might also expect low consensus about a manager's transformational leadership behavior to have a greater effect on team performance in highly interdependent teams than in other types of work groups. Moreover, past research has shown that the type of task performed by a team can influence whether conflict helps, hinders, or has no impact on performance (Jehn 1995). Future research, therefore, may wish to account for the extent of team interdependence and type of tasks performed within and by teams. 9.3. Implications for research and practice Whereas researchers have used a number of indexes to assess within-team consensus, Roberson et al. (2007) were the first to consider how the operationalization of consensus may affect a study's findings. For example, consistent with prior research (Bedeian & Mossholder 2000), Roberson and her colleagues’ analyses demonstrated the weaknesses of coefficient of variation as an index of consensus. Further, rwg-based indices were outperformed by standard deviation in detecting true interaction effects in the simulated data, leading Roberson et al. to conclude that researchers would be best served by using standard deviation as a consensus measure. Thus, because of the subtle differences in consensus indices, future researchers should carefully scrutinize the pros and cons of alternative consensus measures (e.g., coefficient of variation, rwg-based indexes, and standard deviation) to identify the conditions under which a specific consensus index may be most effective (Roberson et al.). From an applied perspective, researchers and practitioners have long attempted to address the question of what enables some teams to perform at a high level and others to fall short. Thus, one practical implication of considering variability in the real-world use of leadership surveys is that it provides practitioners with a secondary indicator of a leader's effectiveness. In other words, because team members will individually perceive a leader's behaviors, they may also react to them differently. As a result, the use of aggregated responses as a surrogate for a manager's leadership behavior is an insufficient basis for summarizing a team's shared perceptions—potentially affecting leadership training and development decisions. Further, unpacking the variance that resides across team members’ perceptions may offer valuable insights in terms of intragroup dynamics. For example, a manager may receive a moderate to high rating on transformational behavior, but still fail to create aligned perceptions within a team (as indicated by a large standard deviation). In this scenario, a manager will likely be perceived by team members as developing idiosyncratic relationships, wherein “we–they” distinctions emerge and, thus, interpersonal tensions and conflict are likely. In addition, membership changes may be particularly damaging to a team because the shaping of newcomers’ expectations requires time and energy that might otherwise be directed towards activities such as task completion. As a worst-case scenario, the absence of consensus may lead to contentious interactions among team members, each seeking to optimize a personal point of view. Under such conditions, intra-team struggles may take their toll on a team's efficiency and effectiveness to the extent that a successful resolution of conflict consumes valuable resources (Levine & Moreland 1990) and is psychologically draining (Cole & Bedeian 2007). The successful handling of within-team dissent by team leaders appears to be a central factor in securing long-term team effectiveness. Actions such as sharing information, integrating new knowledge within team discussions, and constructively working through team member disagreements have been shown to be helpful in this regard (Behfar, Peterson, Mannix, & Trochim 2008).