رهبری در سطوح سلسله مراتبی: سطوح چندگانه از مدیریت و سطوح چندگانه تجزیه و تحلیل

We examined differences in leadership influence processes, perceptions, and multiple levels-of-analysis effects between close and distant charismatic and contingent reward leadership across three hierarchical levels in 13 Korean companies. Multi-source data revealed that followers' commitment to the leader mediated relationships between leadership and followers' attitudinal, behavioral, and performance outcomes in close situations, but not in distant relationships. Leadership at higher levels of management was positively related to leadership at the next lower level, which in turn related to follower outcomes at the lowest echelon. Multivariate within and between analysis indicated multiple-level effects differing by leader–follower distance and for the variables of interest.

Organizational leadership represents a linking process among various organizational members at different hierarchical levels (Likert, 1961). Close leadership between a focal leader and his/her immediate followers has been the subject of extensive research in various settings. But our understanding of distant leadership between a focal leader and his/her followers not reporting directly to him/her is much more limited (Antonakis and Atwater, 2002, Avolio et al., 2004 and Waldman and Yammarino, 1999), and suggests several gaps in the leadership literature. First, leadership scholars have tended to presume that organizational leadership at upper echelons represents distant leader–follower relationships. However, a leader's hierarchical level does not necessarily indicate the leader–follower distance, as evidenced by chief executive officers and their top management teams and U.S. presidents and their cabinet members (Shamir, 1995). Upper-echelon leadership perceived by close followers may not actually represent leadership at a distance and the perception of leadership and its effects may not be applicable to distant followers. Second, the extant literature on charismatic leadership has been criticized for focusing primarily on leaders' personal characteristics and thus failing to recognize leadership based on a social relationship between the leader and follower (Howell & Shamir, 2005). The identification of differences in leaders' behaviors and influence processes between close and distant situations needs to be complemented by explanations of why those behaviors and processes are relevant to those situations in terms of follower perception formation and subsequent attitude change. Lastly, leadership is by nature a multiple-level phenomenon occurring between an individual leader and individual followers, groups of followers, and/or collectives of the groups of followers (Dansereau & Yammarino, 1998). In particular, consideration of leader–follower distance requires us to reconceptualize previous multiple levels-of-analysis perspectives largely limited to close leadership situations and demands empirical testing of alternative possibilities regarding variability and other levels of analysis. Unfortunately, limited conceptual work (e.g., Waldman and Yammarino, 1999 and Yammarino, 1994) exists which incorporates a multiple levels-of-analysis perspective to examine organizational leadership across multiple levels of management. Very few empirical studies correctly applied a multiple levels-of-analysis perspective to such an examination regardless of the leadership approach (Yammarino, Dionne, Chun, & Dansereau, 2005). To begin to address these limitations in the current literature, this study examines various differences in charismatic and contingent reward leader influence processes and followers' leadership perceptions and outcomes across multiple levels of management through rigorous application of a multiple levels-of-analysis perspective. By integrating the literature on dual-mode information processing (Chaiken, 1980 and Petty and Cacioppo, 1986) with the literature on charismatic and contingent reward leadership, we develop and test a conceptual model of close and distant charismatic and contingent reward leadership in organizations. Our study thus contributes to the leadership literature by providing conceptual and empirical evidence of differences in the appropriateness of these leadership behaviors and relationships across managerial levels and leader–follower distance.

This study examined various differences between close and distant charismatic and contingent reward leadership. Those differences were defined and investigated in terms of distinctive leader-influencing mechanisms and followers' leadership perceptions and multiple levels-of-analysis effects for close and distant leader–follower relationships. Hence, the primary purposes of study were two-fold, focusing on differences in substantive relationships among variables of interest and levels of analysis between close and distant leadership situations. 4.1. Levels of management issues By integrating the literature on dual-mode information processing of persuasion and attitude change with the literature on charismatic and contingent reward leadership, a conceptual model of close and distant charismatic and contingent reward leadership was developed. Two key points of that conceptualization are that: (a) two different attitude consequences in terms of strength emerge (strong attitude toward close leader and weak attitude toward distant leader); and (b) attitude strength moderates the attitude-mediating relationship between leadership and follower outcomes. Therefore, we proposed that a strong attitude toward the leader in close leadership situations would fully, or at least partially, mediate the relationship between leadership and follower outcomes; and a weak attitude toward the leader in distant leadership contexts would not mediate the leadership-outcome relationships. To complement the non-mediating distant leadership model, a cascading model of charismatic and contingent reward leadership also was proposed and tested. Finally, interaction effects of distant and close leadership on distant follower's outcomes were tested as an alternative to the cascading leadership model. Results support most of our hypotheses and the theoretical rationale of this study. However, further elaborated discussion is necessary to address several unexpected, yet important findings which may be Korean culture-specific or generalizable to the individualistic societies of the West. Many studies examining transformational-charismatic and transactional-contingent reward paradigms have been conducted in Western societies, especially in the U.S., yet the universality of the models has been assumed (Bass, 1997) even without sufficient empirical evidence in Eastern cultures. Few studies examining the leadership approach have been conducted in Korea (e.g., Shin and Zhou, 2003 and Shin and Zhou, 2007) or with Asian-American residents in the U.S. (e.g., Jung & Avolio, 1999). Thus, it was necessary to test the U.S.-developed models in an Asian context, especially given the pressures of globalization and westernization of management practices in Korea (Bae and Lawler, 2000 and Rowley and Bae, 2003). First, as hypothesized, personal identification and value internalization with the leader fully mediated the relationships between close charismatic leadership and follower outcomes. However, the mediating role of instrumental compliance was not fully demonstrated through single and multiple ratings. It is possible that instrumental compliance might not be developed sufficiently by followers in Korea where employees at the same hierarchical levels and with identical organizational tenures are often given the same amount of monetary rewards, regardless of their individual performance. This possibility seems even more likely considering that only followers, who had been involved in leader–follower relationships for more than 3 months to allow sufficient acquaintanceship, were included in the current study. Another plausible explanation for this result is related to a cultural orientation, power distance. Power distance refers to the extent to which a society and individuals accept inequality in power distribution among members of that society (Hofstede, 1980 and Triandis, 1994). From this definition, people in a cultural domain characterized by high power distance are likely to accept power and status differences among people. Korean followers in this study might not have strong instrumental compliance with their contingent reward leaders; rather they might normatively comply with the leadership. Related to the cultural issue, social desirability bias might operate especially when instrumental compliance was reported from self-ratings, in that the Korean culture is oriented from a deep root of Confucianism which values saving one's face and socially accepted norms and behavior. There was another interesting finding regarding close contingent reward leadership. Staff members' instrumental compliance with their managers fully mediated the relationship between managers' contingent reward leadership and their job satisfaction. Kovach (1995) found that individuals' need structures differ across organizational levels: lower non-supervisory employees emphasize “good wage and job security” first, whereas middle and higher level employees prioritize “interesting work and full appreciation of work done.” This study implies that contingent reward leadership effectively dealing with extrinsic rewards and thereby lower-order needs would be more appropriate at lower levels of management, as our findings demonstrated. This result appears more generalizable in Western cultures, in that instrumental compliance still operates as a mediator even in a high power distance society. 4.2. Levels of analysis issues The second purpose of this study was to examine various levels-of-analysis effects between close and distant leadership. The dynamics in the substantive relationships among variables for close and distant leadership were rigorously tested at individual, dyad, group, and collective levels of analysis using single- and multi-source data via Multivariate WABA. Various multiple-level effects were found, differing by leader–follower distance and for different variables involved in the leadership process. Regarding close charismatic leadership, between-dyads and within-group effects were found for the relationships between leadership and follower outcomes. Specifically, the relationships between the department head's charismatic leadership and managers' job satisfaction and helping behavior held only at the between-dyads level of analysis and were not replicated at the group level of analysis. In contrast, the relationship of the department head's charismatic leadership with managers' performance was found at both between-dyads level of analysis and within-groups level of analysis. For example, while the relationships of charismatic leadership with job satisfaction and helping behavior were solely based on one-to-one leader–follower relationships independent of group membership, the one-to-one dyadic relationship of charismatic leadership with performance developed into charismatic dyadic relationships within the groups. This finding suggests that managers' job satisfaction and helping behavior depend on only the department head's charismatic leadership, but their performance co-varies with the department head's charismatic leadership and other group members as well. This is an important finding for at least three reasons. First, dyadic view of leadership, which is more conceivable in individualistic societies (Yammarino & Jung, 1998), also was found in a collectivistic society, Korea. This indicates that dyadic leadership approach, such as individualized leadership (Dansereau et al., 1995), may be generalized to Eastern cultures as well. Second, a continuing criticism of the dyadic leadership approach is the issue of how differentiated dyadic relationships affect overall performance by the leader's work unit (Schriesheim et al., 1999 and Yukl, 2001). However, the current finding implies that the dyadic relationship under charismatic leaders emphasizing collective orientation can develop into a group-level phenomenon over time where the unit members are motivated to collaborate and produce higher overall performance within the work units. Third, building on this finding, we may be able to reconcile or integrate the individualized leadership approach based on a dyadic view of leadership (Dansereau et al., 1995) with charismatic leadership theories valuing a collective orientation (Shamir et al, 1993). Therefore, leading a team as a whole and establishing personal relationships by focusing on individual differences within the team at the same time appear a critical essence of transformational-charismatic leadership within the team context. In fact, Kark and Shamir (2002) proposed dual effects of transformational leadership not only utilizing idealized influence and inspirational motivation for collective purpose but also demonstrating intellectual stimulation and individually considerate behaviors for individual team members. Contrary to the findings in close charismatic leadership at upper levels of management between department heads and managers, staff members' performance at lower levels of management neither co-varied with manager's charismatic leadership at the dyad level of analysis nor with other staff members' performance and manger's charismatic leadership at the group level of analysis. The relationship between manager's charismatic leadership and staff members' performance was solely based on individual differences (To conserve space, these tabular results are not presented here but are available upon request from the first author.). It seems possible that since staff members might be naïve entry-level employees with short tenure in organizations and their work, they could not form established relationships with their leader and coworkers and thus their performance did not co-vary with leadership and other staff members' performance. The close leader–follower context is conducive for a contingent reward leader to identify each immediate follower's unique needs and provide each follower with extrinsic rewards correspondent to his/her needs, contingent on each follower's performance. The leader controls rewards to a specific follower, whereas the follower also controls his/her performance to the focal leader. The two parties may form a unique independent dyadic relationship by exerting mutual control (Yammarino et al., 1998). This theoretical proposition was supported in the relationships between contingent reward leadership and helping behavior and performance at both upper and lower levels. For the distant charismatic and contingent reward leadership approach, we expected that distant leadership would be an attributional phenomenon where peripheral/heuristic information processing may be a primary route for distant followers to evaluate leadership. A limited number of leader-related peripheral cues are passed and shared among the distant followers through social information processing in follower–follower relationships. Thus, we hypothesized that the attributional phenomenon in distant situations would be a department-level property. Contrary to this expectation, the results for the department level of analysis indicated that the leadership phenomena involving all variables of interest were based on individual differences. What might explain these unexpected results? When followers work closely together, they are more likely to engage in social information processing (Meindl, 1990 and Salancik and Pfeffer, 1978). Frequent interaction appears a prerequisite for the social influence and the contagion process by which certain collective-level properties may be created. Unlike managers, entry-level staff members might not have enough opportunities to interact with other members outside their work units but within their department. Hence, the social information processing might mainly operate only inside their units within the department. In fact, the within-eta correlations for contingent reward leadership and three bases of commitment assessed by staff members were significantly greater in a practical sense than the between-eta correlations for the variables. Thus, it seems necessary to speculate why the within-department level views of contingent reward leadership and corresponding attitudes did not co-vary with staff members' outcomes, yielding these individual-level effects. Recall that the staff members' attitudes toward the department head were weak and the relationships between distant charismatic and contingent reward leadership and staff members' outcomes were weaker than those in close leadership situations. These results imply that follower outcomes might not co-vary with the department-level leadership perceptions and attitudes. 4.3. Implications for practice The theoretical model and empirical findings of the current study provide several practical implications. It is critical to recognize the importance of building positive leader images in upper echelons. Because the information processing of distant organizational members is based on a peripheral route, and hence followers' attitudes toward the distant leader may be temporary and susceptible to change, distant followers are likely to be vulnerable to symbolic impression management (Gardner and Avolio, 1998 and Sosik et al., 2002). Salancik and Meindl (1984) also demonstrated how CEO's symbolic actions as a part of impression management can have a positive impact on organizational performance. Through pep talks, campaign-like political speeches, sagas, storytelling, and symbolic slogans, distant charismatic leaders can provide an ideological vision and value that can serve as a shared organizational value and develop inter-group cohesion manifested as group-wide and organization-wide phenomena (Waldman & Yammarino, 1999). This represents the role of a charismatic leader in upper echelons not only as the definer of organizational culture (Schein, 1990) but also as the communicator of the shared values for followers at a distance in organizations. Social identification as a basis of organizational commitment is the influence process that distant charismatic leaders need to arouse by engaging in various forms of symbolic actions toward distant followers who tend to be receptive to those behaviors. Additionally, our study suggests the important role of intermediate leaders as “linking pins” (Likert, 1961), who can demonstrate similar leadership behaviors which distant leaders also display. Another essence of leadership for distant leaders to keep in mind is to understand the influence processes of cascading leadership and develop close followers' full potential to serve as their surrogates (Waldman & Yammarino, 1999). Close followers' attitudes toward their immediate leaders tend to be persistent over time and resistant to counterargument. Developing close followers who are dependable and loyal is an area a leader at upper echelons has to focus on to indirectly lead from a distance. Two areas a distant charismatic leader needs to manage simultaneously are personalized relationships with intermediate leaders, who are also immediate followers, based on their personal identification and value internalization, and socialized relationships with distant followers building on their social identification. 4.4. Limitations and future research paths Several limitations of this study provide suggestions for future research. First, as several results may be attributable to culture-specific characteristics, those findings can be an inherent limitation as well as a unique contribution. Generalizability of the empirical evidence from the current study's Korean sample should be validated across various work-settings in different national cultures. For example, perceived social distance may be higher in a high-power distance society, and thus the hypothesized relationship could be better evidenced, given the hierarchical rank-based partition of close and distant situations in our study. However, it is unlikely to expect frequent interaction in the distant relationship between department heads and staff members, as demonstrated in our study, even in low-power distance societies. Nonetheless, systematic comparison in a cross-cultural study including multiple cultures with common research questions is warranted. The effectiveness of charismatic and contingent reward leadership may vary depending on the cultural orientation of leader and followers (Jung & Avolio, 1999). Conceptualization and empirical testing for multiple levels of analysis also may be different from culture to culture (Yammarino & Jung, 1998). Second, subjective measures using paper-and-pencil instruments may raise concerns for whether the obtained findings may be methodological artifacts. Although such possibilities cannot be ruled out fully, they can be minimized. In this study, we followed the principle that leadership is assessed by followers and followers' outcomes are measured by corresponding leaders. However, followers' bases of commitment, as mediators, were measured by self-report, and thus the obtained results of relationships between leadership and bases of commitment may be subject to common-source bias. It was necessary to measure the bases of commitment through self-report because self-evaluation regarding those attitudes were the focus of theoretical interest. When the theoretical constructs deal with self-evaluation (i.e., evaluation of personal identification with leader), self-report measures are useful and essential (Howard, 1994 and Maurer and Tarulli, 1994). Furthermore, the focal issue in the mediating relationships is the linkage between bases of commitment rated by followers and outcomes rated by leaders, not the relations between leadership rated by followers and bases of commitment rated by follower. Empirically, the results of CFAs confirmed a five-factor model of the independent and mediating variables (charisma, contingent reward, personal identification, value internalization, and instrumental compliance), both of which were rated by followers (RMSEA = .05, TLI = .95, CFI = .96). Moreover, this five-factor model was superior to two-factor model where charisma, personal identification, and value internalization represent a factor and the other factor includes contingent reward and instrumental compliance (RMSEA = .10, TLI = .82, CFI = .84; ∆ χ2 (df) = 731.4 (9), p < .01). Finally, the relationships observed in the current study were generally consistent with previous research and theories. Although a serious common-method variance problem, due to paper-and-pencil subjective measures, appears unlikely, the necessity of using different methods and measures for further study is apparent. Third, three hierarchical levels were involved in testing the hypotheses of interest, where staff member's outcomes might be influenced by the manager's close leadership as well as the leadership of a department head. Due to the matched-report procedure for multiple levels of analysis across three hierarchical levels of management, an unequal number of individual-level raw scores for each level of management occurred. To fully incorporate the control issue into hypothesis testing, the raw scores had to be transformed, but these transformations make multiple levels of analysis issues not testable. Nonetheless, the lack of controlling potential exogenous effects is an inherent limitation of this study and future research involving multiple hierarchical levels should address this issue. Fourth, leadership processes develop over time, and as such this notion suggests another implication for multiple levels of analysis (Dansereau, Yammarino, & Kohles, 1999). For example, individual-level phenomenon can become dyadic agreements, and between-dyads effects may become within-group level effects over time. A cross-sectional study like the current research cannot capture those longitudinal transformations in levels of analysis effects. Finally, to obtain matched-reports from a leader and immediate followers and to simultaneously ensure anonymity, the department heads and managers in the present study were asked to randomly select three followers for participation. Although they were instructed to randomly select their followers, department heads and managers may have selected only their better performers who might share many personal characteristics with the leaders, resulting in potential artifacts for the findings regarding cascading leadership and levels of analysis. The empirical evidence in our study nevertheless showed many individual-difference effects across dyad and group-level analyses from single- and multiple-source ratings, implying that these potential artifacts might not be an issue. Nonetheless, future research adopting the matched-report procedure should consider this issue and potentially address it by including all followers of a leader (not just a subset of followers). Regardless of these limitations, we hope that this study and its results demonstrate the value of considering multiple levels of management and multiple levels of analysis simultaneously in leadership theory and research in organizations.