Leadership Styles and Behaviors
Unlike the limited amount of work on knowledge, skills, and problem-solving capability, there has been an extensive research on differences in leadership styles and behaviors (Hughes et al., 1999; Yukl, 1998). Much of this emerged following the disappointing conclusions reported by Mann and Stogdill's reviews of leadership traits. This led to a stream of research on the people versus production styles of leaders, as well as on initiation of structure and consideration generated in research conducted at the University of Michigan and Ohio State University (Bass, 1990). At Ohio State University researchers measured nine behavioral constructs, which initially included initiation, membership, representation, integration, organization, domination, communication, ^Rcognition, and production orientation. Leaders were rated on how frequently they displayed behaviors associated with each construct (Hemphill & Coons, 1957; Stogdill, 1963). Factor analyses resulted in a clustering of constructs into four categories labeled consideration, initiation of structure, production emphasis, and sensitivity (Bass, 1990). These early results led to the development of two-factor theories of leadership, which dominated the literature well into the 1980s— for example in Fiedler's (1967) contingency model, Blake and Mouton's (1964) managerial grid, Hershey and Blanchard's (1969) situational leadership theory, and more recent work by Graen and his colleagues on leader-member exchange (LMX) theory (Fiedler, 1967; Graen, Novak, & Sommerkamp, 1982). Preceding this period, work by Hemphill (1949) focused on examining what the situation demanded of leaders. The out-| growth of focusing on the situation led to other contin-" gency models such as Vroom and Yetton's (1973) normative decision-making model, and House and Mitchell's (1974) path-goal theory.
(Non)Contingent Rewards and Punishment
Other research on leadership styles included how leaders used rewards and punishment to influence follower motivation and performance. Podsakoff and Todor (1985) examined the relationship between the use of contingent reward and punishment behaviors on follower motivation. They reported that group cohesion, drive, and productivity were all related positively to leader-contingent reward behavior. Contingent punishment was also positively related to group drive and Productivity, whereas noncontingent reward and punishment Produced equivocal results. Their findings supported earlier
Integrating Several Streams of Leadership Research 285
arguments by Hunt and Osborn (1980), as well as House and Mitchell's (1975) path-goal theory of leadership, indicating that noncontingent versus contingent rewards were less likely to produce positive motivational effects.
Podsakoff, Todor, and Skov (1982) demonstrated that the impact of using contingent or noncontingent rewards depends on the nature of the context. For example, both low and high performers were equally dissatisfied with the use of noncontingent punishment; however, there were no effects on performance. The use of contingent rewards has been associated with higher follower satisfaction, advancement opportunities, and performance over a large number of samples, levels, and cultures (Bass, 1998; Lowe, Kroeck, & Sivasubramaniam, 1996).
Fieldler's Contingency Theory
Additional work on leadership styles and behaviors based on Fiedler's (1967) contingency model of leadership has generated considerable controversy over the last 30 years (Schriesheim, Tepper, & Tetrault, 1994). Part of the controversy stems from Fielder's measurement of relational- versus task-focused leadership, using what he called the least preferred coworker (LPC) scale. According to Fiedler's theory, leaders are categorized according to their scores on the LPC scale as being more task oriented than people oriented. Fiedler then classified the context in terms of those situations being more or less favorable using the following three dimensions: leader-member relations, task structure, and position power. Fiedler argued that task-oriented leaders were more effective in highly favorable and unfavorable situations, whereas relationship-oriented leaders were more effective in the middle range.
A third aspect of the controversy concerns Fiedler's insistence that leader effectiveness is based on changing the situation versus the leader. Fiedler argued in favor of changing the context to match the leader's preferred style. Unfortunately, research on the leader-match process has produced both support (see Peters, Hartke, & Pohlmann, 1985; Strube & Garcia, 1981) and discrepancies for his model (e.g., Jago & Ragan, 1986).
Schriesheim et al. (1994) examined 147 empirical studies that used Fiedler's contingency model and concluded that both high- and low-LPC leaders demonstrated effective performance depending on the context. Schriesheim et al. (1994) concluded that not all of the predictions in Fiedler's model held up; however, altering the situation may indeed be one way of enhancing the impact of leadership on performance. Fielder's early emphasis on the context balanced off the emphasis on behavior and attributes in the literature.
286 Leadership Models, Methods, and Applications
Leader Style, Relationships, and Leader-Member Exchange Theory
The roots of LMX theory can be traced to the work of Dansereau, Graen, and Haga (1975), which was originally referred to as vertical dyad linkage (VDL) theory. Graen et al. (1982) extended this work into what is now called LMX theory by focusing on exchanges and relationships that were not necessarily vertical.
Reviews by Gerstner and Day (1997) showed that the LMX scale was correlated with a broad range of variables, including follower satisfaction, performance, and turnover. However, controversy also surrounds this constructs measurement. Schriesheim, Castro, and Cogliser (1999) pointed to problems with how the LMX construct was defined, measured, and analyzed. They also criticized LMX research for not examining the level of analysis to assess relationships. However, there were important findings produced by LMX research, which has led to new streams of research focusing on individualized leadership, trust-building in teams, and cross-cultural research (Graen & Uhl-Bien, 1995; Schriesheim etal., 1999).
A main concern about LMX theory is how it has changed over time in terms of what constitutes LMX. Schriesheim etal. (1999) indicated that Graen and his colleagues have continued to define LMX as the quality of exchange between a leader and followers; however, what constitutes the quality of that exchange has varied. For example, Schriesheim et al. examined 13 studies published by Graen and his colleagues over a period of 10 years, concluding that there were 18 sub-dimensions describing the quality of LMX (trust, competence, motivation, assistance and support, understanding, latitude, authority, information, influence in decision making, communications, confidence, consideration, talent, delegation, innovativeness, expertise, control of organizational resources, and mutual control). Schriesheim et al. reviewed 37 dissertations and research papers reporting that there were U different theoretical definitions associated with LMX and 35 different sub content elements, and concluded that "a decade after the inception of LMX theory, there was still so much disagreement as to the basic definition of the construct as well as no clear or consistent direction provided about where or how to proceed in developing the theory" (p. 76). Alternatively, we might also describe LMX as changing with the times. From 1972 to 2001 organizations became flatter, more networked, technologically connected, and arranged in strategic alliances configured in B2B models. Leaders and followers now interact more at a distance through technology, and followers instead of leaders are often the experts in
work processes. How these global changes are affecting the exchanges between leaders and followers remains an inter esting domain for future research.
Taken together, the research on leadership styles and behaviors has identified a number of styles that consistently show up differentiating more or less effective leadership. Recently, this literature was significantly extended on the order of a giant leap as leadership research began to examine more the behaviors and styles of charismatic and transformational leaders.
Transformational, Charismatic, and Visionary Theories
Much effort in leadership research before the late 1980s did not focus on what constituted charismatic or inspirational leadership. A giant step toward understanding these profound forms of leadership was taken in the 1980s based on work by House (1977) and Bums (1978). House and Shamir (1993) highlighted the need to integrate charismatic, transformational, and visionary theories of leadership because all overlap with each other and appear to evolve in the same direction. A distinguishing characteristic of these theories builds on the relationship between leaders and followers discussed earlier in LMX theory. Charismatic leaders transform the needs, values, and aspirations of followers from individual to collective interests. They ask followers to consider the greater good of their group, organization, community, or society above and beyond their own self-interests (Bass, 1985; Bennis & Nanus, 1985; Conger & Kanungo, 1987; Tichy & Devanna, 1986). Earlier theories covered in this review focused attention on the tangible exchanges that occur between leaders and followers, as opposed to examining how trusting a leader motivates followers to extraordinary efforts and performance. The focus shifted to symbolic leadership, building identification with the leader's cause or vision, challenging followers to think differently, inspiring followers to extraordinary efforts, and building enough confidence in followers for them to lead (House & Shamir, 1993). Charismatic theories highlighted the importance of behaviors that were originally discussed by House (1977) and Burns (1978), and later by Bass (1985), Bennis and Nanus (1985), Tichy and Devanna (1986), and Conger and Kanungo (1987). Leaders who are transformational or charismatic have produced higher levels of effort, satisfaction, and performance (Avolio, 1999; Bass, 1998).
Podsakoff, MacKenzie, Moorman, and Fetter (1990) explored some of the internal mechanisms affected by transformational leadership. They examined how transactional and transformational leadership impacted the trust that followers
had in their leaders, as well as how trust influenced organizational citizenship behaviors. Their results provided evidence to support Bass's (1985} contention that transformational leadership activates higher order needs through the development of trust, leading followers to exhibit extra-role behaviors in addition to in-role behaviors that honor transactional agreements.
Bass and Avolio (1997) addressed some of the issues concerning the need for integrating various models of charismatic-transformational leadership, developing a model referred to as &fiill range theory of leadership. Bass and Avolio chose the label "full range" to expand the thinking in the field of what constitutes the broadest possible range of leadership beliefs, values, perspectives, and styles. As Yukl (1999) noted, "Although no single theory should be expected to include all aspects of leadership behavior, use of the label 'full range leadership theory' by Bass (1996) invites critical evaluation of
| « |
mpleteness" (p. 290). Yukl's criticism challenges the field of idership exactly as Bass and Avolio had intended in choosing the term 'full range.'
In a critique of the charismatic-transformational leadership literature, Yukl (1999) highlighted the importance of examining how transformational versus transactional leadership influences followers through instrumental, compliance, personal identification, and internalization citing Kelman's (1958, 1974) theoretical work as a basis. Yukl pointed out that "the theory would be stronger if the essential influence processes were identified more clearly and used to explain how each type of behavior affects each type of mediating variable and outcome" (p. 287). House and Shamir (1993) provided some useful suggestions for pursuing several lines of research recommended by Yukl. For example, if a follower associates his or her self-concept with a leader's vision or
^ values, one would expect higher levels of identification.
"Leaders who appeal to the ideological values of their followers would have followers whose images of themselves were linked to the leader's mission and vision. By implicating the follower's working self-concept, House and Shamir argued, a charismatic leader is making certain identities maintained by followers more salient, resulting in greater motivational potential.
Yukl (1999) described work on transformational leadership as following in the footsteps of earlier heroic theories of leadership (Calder, 1977). The effective leader is described as influencing followers to sacrifice and exert exceptional effort. Yukl argued that leadership should also be viewed as reciprocal or shared. More recent discussions of transformational leadership address some of Yukl's concerns (see Avolio, 1999; Sivasubramaniam et al., 2002). However, many leader-
Integrating Several Streams of Leadership Research 287
ship researchers still have difficulty viewing leadership as a collective phenomenon.
A final area of concern raised by Yukl (1999) relates to whether it is possible to have a simultaneous occurrence of transformational and charismatic leadership: "In fact, the developing and empowering behaviors associated with transformational leadership seem to make it less likely that followers will attribute extraordinary qualities to the leader" (p. 299). He argued that more empowered and developed followers are less likely to be dependent on their leader, apparently making the assumption that dependence is a necessary condition for charismatic leadership, which would contradict the theory and results in this area (Bass, 1998). Dependence is not a precondition for transformational leadership; nor do we suspect it is for leaders who are socialized charismatic leaders. Nevertheless, examining the power-dependence dynamics between charismatic-transformational leaders and their follower is an important area for leadership research to explore (see Xin& Tsui, 1996).
In sum, work on charismatic and transformational leadership has opened new and exciting areas for leadership researchers to pursue. Most of the work in this area is relatively new, and there is still much to be learned about these complex leadership phenomena.
Measurement Issues Pertaining to Leadership
Leadership has been measured primarily by survey methods (Beyer, 1999; Bryman, Stephens, & Campo, 1996), focusing more on the individual than on interactions between the individual and the situation (Beyer, 1999). Leadership theories, such as Fiedler's contingency model (Fiedler, 1967), include concepts that were developed using factor analysis or other quantitative methods. These techniques provide stable factors that are easy to replicate across multiple studies.
The development of the MLQ (Avolio et al., 1999; Bass & Avolio, 1990) is used here as an example for both the utility and the challenges and limitations of using quantitative methodology. Bass (1985) developed the MLQ and included scales of transformational and transactional leadership. Extensive research on the MLQ over the last 15 years provided both support and criticism (e.g., Bycio, Hackett. & Allen, 1995) concerning its factor structure. These criticisms have led to modifications both in the number of scales used and in item wording. Avolio et al. conducted a comprehensive validation study of the MLQ on 14 diverse samples using confirmatory factor analysis and reported strong support for the original model offered by Bass (1985), as compared to eight other models.
288 Leadership Models, Methods, and Applications
Use of the MLQ demonstrates the utility of quantitative approaches to the measurement of leadership. Nevertheless, critics of quantitative research (e.g., Bryman et al., 1996) argue that these methods are not sufficient when used alone. Surveys typically fail to take into consideration how the context influences leadership, which is a major shortcoming of leadership research.
Implicit Leadership Theory and Leadership Measurement
Critics of survey methodology (e.g., Podsakoff & Organ, 1986) argue that questionnaires require raters to report a specific behavior, yet raters have to recall, weigh, and infer information to respond. In order to simplify this cognitive process, respondents fall back on their implicit models of leadership, which can bias their observations and responses to surveys (Lord & Maher, 1991). Eden and Leviatan (1975) argued that a respondent's reliance on attributions could lead to biases such as specific response patterns throughout a questionnaire that can distort evaluations.
In the last decade a growing number of studies have used qualitative research methods. Bryman et al. (1996) used qualitative methods to show that charisma was exhibited less frequently than instrumental (transactional) behaviors among British police officers. Beyer and Browning (1999) conducted intensive interviews and collected archival and ethnographic data to demonstrate the impact of a charismatic leader on the emergence of the U.S. semiconductor industry. Berson, Avolio, Shamir, and Popper (2001) analyzed the content of videotaped visions reporting a relationship between transformational leadership scores and the optimistic content of visions coded by raters.
Qualitative evidence can offer a more holistic perspective of leadership, helping to explain why differences in ratings exist, rather than simply showing differences. Qualitative strategies such as grounded theory (Glaser & Strauss, 1967) allow for theory development and the inclusion of multiple levels of analysis, including the context in which leadership is observed (Hunt & Ropo, 1995). Nevertheless, qualitative inquiry also has its disadvantages, such as replicating previous measurements (Reissman, 1993) and comparing data collected from different places.
The Utility of Triangulation for Assessing Leadership
Jick (1979) suggested triangulation using quantitative and qualitative measures of the same phenomenon as a method for boosting validity. Triangulation fosters the use of innovative methods, facilitates the examination of new aspects of theories, and allows synthesis and critical comparison between
different theories. Whereas quantitative methods allow for better generalizations based on systematic observation, qualitative measures are superior in the vividness and density of information that they provide on phenomena (Weiss, 1968). For example, Berson and Avolio (2000a) examined the utility of triangulation for measuring the relationship between visionary leadership and organizational performance. They reported that how frequently the vision was expressed was less important to organizational performance than was the vision's content.
Triangulating unobtrusive evidence with quantitative and qualitative data provides further support for using triangulation (Webb, Campbell, Schwartz, & Sechrest, 1966). Unobtrusive data can help confirm patterns in quantitative data that are more prone to measurement bias. For example, Berson and Avolio (2000b) used internal correspondence to confirm that a business unit whose managers were rated low in trust was facing shutdown. Interview data revealed that employees did not trust their managers, describing them as weak and helpless, even though followers were unaware of the impending shutdown. The triangulation of survey, interview, and unobtrusive data helped explain why these managers were rated so low on trust.
Leadership research has moved from relying solely on quantitative methods to adding in qualitative methods (e.g., Bryman et al., 1996; Hunt & Ropo, 1995). However, as Bryman et al. concluded, the use of qualitative methods is rare and is frequently done as an addendum to quantitative measurement. Triangulation may be an effective method for testing some of the more complex and controversial aspects of neo-chari striatic theories of leadership.