A Hierarchical Taxonomy of Leadership Behavior: Integrating a Half Century of Behavior Research Gary Yukl Angela Gordon University at Albany, Tom Taber State University of New York, Albany, NY A major problem leadership research and has been lack of agreement about which theory behavior categories are relevant and meaningful. It is difficult to integrate findings from five decades of research unless the many diverse leadership behaviors can be integrated in a parsimonious and meaningful conceptual An emerging solution is a framework. hierarchical taxonomy with three metacategories (task, relations, and change behavior).
Confirmatory factor analysis of a behavior A in effective major problem in research leadership has been and theory on the lack of are agreement about which behavior categories description questionnaire found more support for this taxonomy than for alternative models. INTRODUCTION A large amount of the empirical research on effective leadership has sought to identify the types of behaviors that enhance individual and collective performance. The most common research method has been a survey field study with a behavior description questionnaire. In the past half century, hundreds of survey studies have examined the correlation between leadership ehavior and various indicators of leadership effectiveness (Bass, 1990; Yukl, 2002). Other methods (e. g. , laboratory experiments, field experiments, critical incidents) have been used much less frequently to identify effective types of leadership behavior. relevant and meaningful for leaders. It is very difficult to compare and integrate the results from studies that use different sets of behavioral There has been a bewildering categories. proliferation of taxonomies on leadership behavior Sometimes (see Bass, 1990; Yukl, 2002). different terms have been used to refer to the same type of behavior.
At other times, the same term has been defined differently by various theorists. What is treated as a general behavior category by one theorist is viewed as two or three distinct categories by another theorist. What is a key concept in one taxonomy is absent from another. Different taxonomies have emerged from different research disciplines, and it is difficult to translate from one set of concepts to another. Task And Relations Behavior The early leadership research emphasized two general, broadly-defined behavior categories (&dquo;metacategories&dquo;) that are best described as relations-oriented behavior and task-oriented behavior.
Examples include consideration and initiating structure (Fleishmen, 1953; Halpin & Winer, 1957) in early research on leader behavior, and concern for people and concern for production in the managerial grid model (Blake & Mouton, 1982). For three decades, research on leader behavior was dominated by a focus on these two broadly-defined categories of behavior. Many studies were conducted to see how measures of 16 consideration and initiating structure were correlated with criteria of leadership effectiveness, such as subordinate satisfaction and performance.
A meta-analysis of this survey research found that both behaviors have a positive but weak correlation with subordinate performance (Fisher & Edwards, 1988). Subsequent research on specific types of task and relations behavior found correlations with unit performance that were sometimes stronger but still not consistent across situations (Yukl, 2002). Importance Of Leading Change In their preoccupation with task and relations behaviors, the early scholars mostly ignored change-oriented leadership. Only recently have researchers become interested in the way leaders initiate and implement change in organizations.
It is important to clarify the distinction task-oriented, relationsamong oriented, and change-oriented behaviors, because all three types of behaviors may be relevant for understanding effective leadership in different situations. The importance of leading change is suggested by some organization theories (e. g. , Miller & Friesen, 1984; Tushman & Romanelli, 1985), but they do not describe the specific types of change behaviors that are required. Theories of transformational and charismatic leadership (e. g. , Bass, 1985; Conger &
Kanungo, 1998; House, Arthur, 1993) include some change-oriented behaviors, and there is growing evidence that these behaviors are related to effectiveness of leaders (e. g. , Lowe, Kroeck, & Sivasubramaniam, 1996). However, the high level of confounding among specific transformational 1977; Shamir, House, & behaviors makes it difficult to determine which ones are most important in a particular situation (Yukl, 1999). Evidence For The Three Metacategories The prior theories of leadership do not make a clear distinction among task, relations, and change behavior.
The first evidence that changeoriented leadership is a distinct type of behavior comes from two studies conducted during the 1990s. In the first study, Ekvall and Arvonen (1991) developed a behaviok description with items from earlier questionnaire questionnaires such as the LBDQ (Fleishman, 1953; Stogdill, Goode, & Day, 1962) and some aspects of change-oriented The 36-item questionnaire was administered to subordinates who described 346 Swedish managers, 229 Finnish managers, and 123 American managers.
Most of the leaders described were middle-level managers in private companies. There was strong support for a threefactor solution in each national sample, and the factors were labeled production centered, employee centered, and change centered. The latter factor included promoting change and growth, providing creative solutions, encouraging creative thinking by others, experimenting with new ways of doing things, making risky decisions when necessary, and planning for the future. Scales were formed using the best items from the factor analysis.
Change-oriented behavior correlated the strongest with subordinate ratings of the manager’s competence, whereas employeecentered behavior correlated highest with subordinate satisfaction with the manager. In the second study, Yukl (1998) administered leader behavior questionnaires to 318 direct reports of managers in charge of 48 organizational units (division, agency, district office, plant) of varying size from 15 private and public sector organizations. Most of the managers occupied middle or upper-level management positions.
The leader behavior questionnaire included representative items from the Managerial Practices Survey (MPS), an instrument used for multi-source feedback workshops (Yukl, Wall, & Lepsinger, 1990). The questionnaire also included some items adapted from the MLQ (Bass & Avolio, 1990). Some new items were written to describe aspects of change-oriented behavior not represented in these earlier questionnaires. An exploratory factor analysis produced a clear factor structure for task-oriented behavior, relationshiporiented behavior, and change-oriented behavior.
The latter factor included identifying external threats and opportunities, envisioning new possibilities, proposing innovative strategies, and encouraging innovative thinking by followers. were to Scales created measure each metacategory. The scale scores for task, relations, and change behavior were all correlated significantly with subordinate satisfaction with the leader and organizational commitment. These two studies made a good start at identifying a distinct category of change-oriented leadership and showing that it is relevant for new items on leadership. 17 leadership effectiveness.
One limitation of both studies was the failure to identify distinct component behaviors for each metacategory. A half century of research on leadership behavior has taught us the dangers of relying exclusively on behavior constructs that are very broad and abstract (Yukl, 1998). The specific behaviors provide a much better basis for developing contingency theories of leadership effectiveness (Yukl, 2002). A hierarchical taxonomy provides a way to reconcile the three-factor solution with the many specific behaviors already found relevant for effective leadership in several types of research.
Research Objectives This paper describes a hierarchical taxonomy and research conducted to verify it. The purpose of the research was to evaluate whether the three metacategories provide a basis for developing an integrative taxonomy of leadership behavior. consulting with subordinates about ways to apply new technology to make major improvements in productivity. Several criteria were used in selecting the specific behavior components to include in the proposed hierarchical taxonomy. First, each behavior must be directly observable.
It cannot be defmed only in terms of attributions or outcomes. Second, each behavior must be potentially applicable to all types of leaders in organizations. Third, each behavior must have primary relevance for one metacategory, even though it could have secondary relevance for the other metacategories. Fourth, each behavior must be grounded in prior theory and research on effective leadership. Prior measures of leadership behavior that provide evidence for the construct validity of the component behaviors include the following: such as when ~ THE HIERARCHICAL TAXONOMY
The theoretical basis for the distinction among the three metacategories is the primary objective of the behavior. The primary objectives of task behavior include high efficiency in the use of resources and personnel, and high reliability of operations, products, and services. The primary objectives of relations behavior include strong commitment to the unit and its mission, and a high level of mutual trust and cooperation among The primary objectives of change members. behavior include major innovative improvements (in processes, products, or services), and ~ ~
C-K Scale: Conger-Kanungo Leadership Scale (Conger & Kanungo, 1998) LBDQ-12: Leader Behavior Description Questionnaire (Stogdill, Goode, & Day, 1962) LOS: Leader Observation Scale (Luthans & Lockwood, 1984) ~ ~ ~ ~ ~ adaptation to external changes. Because a ~ leader’s behavior may have ~ multiple objectives, it is more accurately described in terms of three independent dimensions than in terms of three mutually exclusive behavior categories. For example, providing recognition for significant contributions to the unit reflects a primary concern for the person but also a secondary concern for the mission.
Sometimes a leadership behavior involves all three objectives, Leadership Practices Inventory (Kouzes 1995) MBS: Managerial Behavior Survey (Yukl & Nemeroff, 1979) MPS: Managerial Practice Survey (Yukl, Wall, & Lepsinger, 1990) MLI: Multifactor Leadership Inventory & Schriesheim, 1998) (Castro MLQ: Multifactor Leadership Questionnaire (Bass & Avolio, 1990) SMP: Survey of Management Practices (Wilson, O’Hare & Shipper, 1990) TLI: Transformational Leadership Inventory (Podsakoff, MacKenzie, Morrman, & Fetter, 1990) LPI: & Posner,
The proposed behaviors in each metacategory shown in Table 1. A description of each component behavior and the prior evidence for it are described next. are 18 Table 1. Hierarchical Taxonomy of Leader Behavior. Table 2. Similar Task Behaviors in Earlier Measures. TASK BEHAVIORS Specific task behaviors include: (1) shortplanning, (2) clarifying responsibilities and performance objectives, (3) monitoring operations and performance. Similar leadership behaviors in term earlier measures are indicated in Table 2. is difficult to observe.
Nevertheless, there are some observable aspects such as writing plans, preparing written budgets, developing written schedules, and meeting with others to determine how to accomplish a task. Planning is most observable when a manager takes action to implement plans, a process that often involves clarifying responsibilities and objectives (Yukl, Short-Term Planning Planning means deciding what to do, how to 2002). A number of empirical studies have identified behavior similar to short-term planning (see Table 2).
Evidence that planning is relevant for effective leadership is provided by research on a do it, who will do it, and when it will be done. Because planning is largely a cognitive activity that seldom occurs as a single discrete episode, it 19 managerial competencies (e. g. , Boyatzis, 1982). Several survey studies have found a positive correlation between planning and an independent criterion of managerial effectiveness (e. g. , Carroll & Gillen, 1987; Kim & Yukl, 1995; Morse & Wagner, 1978; Shipper & Wilson, 1992; Yukl, Wall, & Lepsinger, 1990).
Clarifying Responsibilities Clarifying is the communication of plans, policies, and role expectations. The purpose of clarifying behavior is to guide and coordinate work activity and make sure people know what to do and how to do it. Clarifying includes setting specific task objectives, and these objectives direct effort toward performance of important duties and responsibilities, encourage a search for efficient ways to do the work, and facilitate evaluation of performance by providing a benchmark against which to compare it.
Clarifying is a core component of initiating structure. It is also the primary component of instrumental (directive) behavior in the path-goal theory of leadership (House & Mitchell, 1974). Although research on the consequences of using initiating structure was inconclusive (Fisher & Monitoring Operations and Performance Monitoring involves gathering information about the operations of the manager’s organizational unit, including the progress of the work, the performance of individual subordinates, the quality of products or services, and the success of projects or programs.
Monitoring can take many forms, including observation of work operations, reading written reports, watching computer screen displays of performance data, inspecting the quality of samples of the work, and holding progress review meetings with an individual or group. Evidence that monitoring is a distinct and meaningful behavior is provided by research using observation of managers (Luthans & Lockwood, 1984; Mintzberg, 1973) and by studies involving factor analysis of behavior description questionnaires (e. g. , Bass & Avolio, 1990; Yukl et al. , 1990).
Monitoring indirectly focuses attention on aspects of performance that are measured and it facilitates the effective as or Edwards,1988; Podsakoff, MacKenzie, Aheame, & Bommer, 1995; Wofford & Liska, 1993), research on A positive clarifying has found stronger results. relationship between clarifying and was of other behaviors such recognizing clarifying. Two observational studies found that leaders who did more monitoring were more effective (Komaki, 1986; Komaki, Desselles, & Bowman, 1989). In the survey studies, monitoring was related to leader effectiveness for some samples but not others (Kim & Yukl, 1995; Yukl, Wall, & Lepsinger, use 990). RELATIONS BEHAVIORS managerial effectiveness studies, although not for There is found in several all situations (Bauer & Green, 1998; Kim & Yukl, 1995; Yukl & Van Fleet, 1982; Yukl, Wall, & Lepsinger,1990). ample evidence from lab and field experiments as well as survey studies that setting specific, challenging goals results in higher performance as long as the goals are accepted (see Locke & Latham, 1990). Specific relations behaviors include: (1) supporting, (2) developing, (3) recognizing, (4) Similar consulting, and (5) empowering. behaviors in earlier measures are leadership indicated in Table 3. 0 J Table 3. Similar Relations Behaviors in Earlier Measures. defined as showing and concern for the consideration, acceptance, needs and feelings of other people. Supporting is the core component of consideration (Fleishman, 1953; Stogdill, Goode, & Day, 1962) and Supporting Supporting is supportive leadership (Bowers & Seashore, 1966; Mitchell, 1974). Supporting is also a component of individualized consideration, as defined by Bass and Avolio (1990) and Podsakoff et al. (1990). Studies involving factor analysis of behavior description questionnaires indicate that House & upporting is a distinct and meaningful aspect of leadership behavior (see Table 3). Supportive leadership helps to build and maintain effective interpersonal relationships. There is strong evidence that supporting is related to follower satisfaction with the leader (Bass, 1990; Yukl, 1998). However, only a weak, inconsistent relationship has been found between supporting and follower performance (e. g. , Fisher Lepsinger, 1990). Supporting is more likely to be effective when combined with other relevant leadership behaviors. & & Edwards, 1988; Kim &
Yukl, 1995; Yukl, Wall, 21 Developing The core component of developing is include showing someone a better way to do a task, asking questions that help someone learn how to perform a task better, helping someone learn from a mistake, and explaining how to solve a complex problem rather than just providing the solution. Developing also includes providing opportunities to develop skills and confidence (e. g. , special assignments, challenging new responsibilities) and facilitating skill learning (making it easier for subordinates to attend courses or workshops).
Developing was identified as distinct and meaningful leadership behavior in studies involving observation of managers (Luthans & Lockwood, 1984), and in some studies involving factor analysis of behavior description questionnaires (see Table 3). In the MLQ (Bass & Avolio, 1990), some developing items are included in the individualized consideration scale. The descriptive research provides evidence that effective managers take a more active role in developing the skills and confidence of subordinates Cohen, 1984; (Bradford & In the survey research, McCauley, 1986). eveloping was correlated with follower performance in some studies but not others (Javidan, 1992; Kim & Yukl, 1995; Yukl, Wall, & coaching. Examples that effective leaders provide extensive praise and recognition to subordinates for their achievements and contributions. A rare field experiment by Wikoff, Anderson, and Crowell (1983) found that praise by the supervisor increased subordinate performance significantly. Most survey studies on the consequences of recognizing have found a positive correlation with subordinate satisfaction.
However, results for effects on performance are less consistent in the survey studies (e. g. , Kim & Yukl, 1995; Lowe, Kroeck & Sivasubramaniam, 1996; Podsakoff & Todor, 1985; Podsakoff, Todor, Grover, & Huber, 1984; Yukl et al. , 1990). Consulting The key component of consultation is involving followers in making important decisions. Consultation with individuals or the group is one form of participative leadership. This type of leadership behavior is included in the pathgoal theory of leadership (House & Mitchell, 1974) and the normative leadership models proposed by Vroom and Yetton (1973) and by Vroom and Jago (1988).
Evidence that consultation is a distinct and meaningful form of leadership behavior is provided by studies factor of behavior involving analysis questionnaires (see Table 3). The potential benefits of consultation include better decisions and greater acceptance of decisions by people who will implement them or be affected by them. Hundreds of studies have evaluated the consequences of participative leadership, and reviews of this literature generally conclude that there is only a weak, inconsistent relationship with follower satisfaction and performance (e. g. , Leana, Locke, & Schweiger, 1990; Sagie & Koslowsky, 2000).
Only a small number of studies examined the effects of consulting as a separate type of participative leadership, and here again the result were inconsistent (Yukl & Nemeroff, 1979; Yukl et al. , 1990). The effectiveness of consultation may depend on aspects of the situation such as the type of task, the distribution of relevant information, and the values of followers (Vroom & Yetton, Lepsinger, 1990). Recognizing Recognizing involves giving praise and showing appreciation to others for effective performance, significant achievements, and important contributions to the organization. Recognition is often given along with tangible rewards.
The combination of recognition and contingent rewards has been given a variety of labels in the leadership literature (e. g. , contingent reward behavior, positive reward behavior). Several studies involving factor analysis of behavior description questionnaires have found evidence that recognizing (either alone or combined with rewarding) is a distinct type of leadership behavior (see Table 3). Our taxonomy emphasizes recognition because it is easier to provide than tangible rewards, it is more personal, and it is relatively independent of the formal reward system of the organization.
Descriptive studies in organizations (Kouzes & Posner, 1995; Peters & Austin, 1985) suggest 1973; Yukl, 2002). Empowering Empowering includes providing more autonomy subordinates. and discretion to Evidence that empowering is a delegating and 22 distinct form of leadership behavior is provided by factor analysis of behavior description questionnaires (see Table 3), and by studies in which delegation and consultation were found to have distinct antecedents and consequences (Leana, 1987; Yukl & Fu, 1999). Empowering is likely to result in more commitment by a subordinate to implement decisions effectively.
It can improve decision quality when a subordinate has more expertise in how to do the task than the manager, especially if there is need for a quick response to a changing situation. The results from survey research on the relationship between delegation and subordinate performance have been inconsistent and difficult to interpret, which may reflect problems in the criterion measures (e. g. , Bauer & Green, 1996; Leana,1987; Schriesheim, Neider, & Scandura, 1998; Yukl et al. , 1990). In a survey study with Table 4. Similar ndependent, objective criteria, Miller and Toulouse (1986) found that the amount of delegation by top executives in 97 small businesses was related to their profitability and sales growth. Descriptive research on effective management also supports the effectiveness of delegation and empowerment when used in appropriate situations (see Bass, 1990; Yukl, 2002). CHANGE BEHAVIORS Specific change behaviors include: ( 1 ) external monitoring, (2) envisioning change, (3) encouraging innovative thinking, and (4) taking personal risks to implement change. Similar leadership behaviors in earlier measures are indicated in Table 4.
Change Behaviors in Earlier Measures. External Monitoring One of the most important activities of executives is to monitor the external environment and identify threats and opportunities for the organization. This change-oriented behavior is also called &dquo;environmental scanning. &dquo; Most leaders of business organizations need to be sensitive to a wide range of information, including the concerns of customers and clients, the availability of suppliers and vendors, the actions of competitors, market trends, economic conditions, and government policies, technological developments. The information may be gathered in a variety of ways (e. . , reading government 23 reports and industry publications, attending professional and trade meetings, talking to customers and suppliers, examining the products and reports of competitors, conducting market research, and developing an external network of information sources). After the information is gathered, it must be analyzed and interpreted. Identifying problems or opportunities is a separate process from proposing solutions or new strategies. Interpreting events and explaining why change is needed is a key behavior in theories of change management (Kotter, 1996; Nadler et al. , 1995).
Prior research involving factor analysis of survey questionnaires (see Table 4) provides some evidence that external monitoring can be treated as a distinct type of leadership behavior. However, these studies do not tell us whether gathering and analyzing information should be differentiated from interpreting information for followers. Some field studies provide evidence that external monitoring and interpretation of events is related to effective leadership. Bourgeois (1985) studied 20 companies and found that profitability was greater when executives had an accurate perception of the amount of industry volatility in markets and technology.
Grinyer, Mayes, and McKieman (1990) found that the leaders of highperforming companies did more external monitoring (e. g. , environmental scanning, consultation with key customers) than leaders of low-performing companies and were quicker to recognize and exploit opportunities revealed by it. More external monitoring is needed when the organization is highly dependent on outsiders validity of this type of behavior (see Table 4). Evidence that visioning is relevant for effective leadership is provided by survey field studies (see Lowe et al. , 1996), laboratory experiments (e. g. Howell & Frost, 1989; Kirkpatrick & Locke, 1996), and descriptive studies (e. g. , Bennis & Nanus, 1985; Kouzes & construct Posner, 1995). Encouraging Innovative Thinking Encouraging innovative thinking by others can be differentiated from proposing innovations yourself, and a leader can use various combinations of the two behaviors. Evidence that encouraging innovative thinking is distinct and meaningful is provided by studies involving factor analysis of questionnaires on transformational leadership (see Table 4). The behavior is similar to &dquo;intellectual stimulation&dquo; in the MLQ (Bass & Avolio, 1990), the TLI (Podsakoff et al. 1990), and the MLI (Castro & Schriesheim, 1998). Evidence that this type of behavior is relevant for effective leadership comes primarily from the survey studies on transformational leadership (see see meta-analysis by Lowe et al. , 1996). Additional evidence is provided by a field experiment (Barling, Weber, & Kelloway, 1996). (e. g. , clients, customers, suppliers, subcontractors, joint venture partners), the environment is rapidly changing, or the organization faces severe competition or serious threats from outside enemies (Ginter & Duncan, 1990). ision of a better future is a common element in most theories of transformational and charismatic leadership. A vision is more effective in influencing follower commitment to a proposed strategy or change if it is relevant for follower values and ideals, it is communicated with enthusiasm and confidence, and it is perceived as feasible. Several studies that included factor analyses of behavior description questionnaires found evidence supporting the Envisioning Change Articulating an inspiring
Taking Personal Risks Undertaking major change is risky, especially when the need for change is not yet obvious to most people and there is a lot of vested interest in maintaining the status quo. The risks entailed by pushing for change when possible there is strong resistance include loss of job, diminished reputation, derailed career, and personal rejection by colleagues. Evidence that this behavior is distinct and meaningful is provided by studies involving factor analysis of behavior description questionnaires (see Table 4). Empirical research provides evidence that risk taking and personal sacrifice relevant for effective leadership by combat officers (e. g. , Frost, Fiedler, & Anderson, 1983; Yukl & Van Fleet, 1982), but the studies did not directly involve initiation of change. More attribution of charisma likely for a leader who takes risks and makes personal sacrifices to pursue a vision or innovative strategy (Conger & Kanungo, 1998; Shamir et al. , 1993), but the effects of risk taking on successful implementation of change have not been directly investigated. is 24
RESEARCH EVIDENCE FOR THE TAXONOMY A survey field study was conducted to evaluate the proposed hierarchical taxonomy of leadership behaviors. No prior study has included all of these behaviors at the same time. Confirmatory factor analysis was used to determine whether the three metacategories explain more of the variance in the specific behaviors than alternative models of grouping the specific behaviors into metacategories. Sample samples were used in this The first sample included 174 middle managers who were participating in training workshops conducted by a consulting company.
Managers who volunteered to participate in the study were asked to rate the behaviors of their immediate boss over the past several months. Most of the respondents were managers who described the behavior of a boss who was a middle The questionnaire was manager or executive. and respondents were not completely anonymous, asked to provide demographic information that could be used to identify them. Thus, we did not have accurate information on age, gender, Two types of research. Measures Of Leader Behavior Two different versions of the leadership questionnaire were used in the research.
The initial version included sets of items designed to measure the 12 specific behaviors in the proposed taxonomy. The definitions for the behaviors are shown in Table 5. Each behavior was represented by 4 to 5 items adapted from earlier scales. The items had a five-choice response format with anchors for each choice. The anchors emphasized magnitude rather than frequency ( 1-Not at all or not applicable, 5-To a great extent). The order of scale items was randomized within the constraint that all behaviors must appear in the first part of the questionnaire and no behavior could be concentrated in only one part of the questionnaire.
Exploratory factor analysis and item analysis were used to identify the best items for each scale, and these scales were used in the subsequent data analyses. analysis of the initial questionnaire, we developed a revised questionnaire to measure the 12 specific behaviors. Each item in the questionnaire had the on Based the results from the education, or job tenure. The second type of sample included 101 MBA students who had regular daytime jobs but attending management courses at night at a large northeastern university. Participation was voluntary, and students were assured that their were esponses would remain confidential and would not be seen by anyone except the researchers. About 75% of the respondents were professional, nonsupervisory employees who described the leadership behavior of a boss who was a first-level manager. The remainder of the students were managers who described the behavior of a boss who was a middle manager or executive. same five anchored response choices as in the earlier version, but there was also a response choice labeled: &dquo;Don’t know or not applicable&dquo; (which was scored as a &dquo;1 &dquo;).
Each behavior scale included four distinct items to ensure content validity and avoid obvious redundancy. The items in the same scale were grouped together under the scale name in order to improve respondent discrimination among the different behaviors. The order of scales was varied so that the task, relations, and change scales were evenly distributed within the questionnaire. This procedure was used to avoid suggesting any of behaviors into grouping specific metacategories. 25 Table 5. Definition of the 12 Specific Leadership Behaviors
Results The internal consistency reliability for each scale was determined with the Cronbach alpha statistic, and the values for all scales in both versions of the questionnaire exceeded the recommended lower bound for an acceptable estimate of internal consistency (alpha > . 70). The lowest alpha value for any scale was . 77, and most of the values were greater than . 80. The mean scale scores for the 59 MBA students who used the randomized version were compared to the corresponding scores for the 48 MBA students who used the grouped version, and there were no significant differences.
We used the students rather than the managers for this analysis of equivalence, because the 131 managers who filled out the randomized version were not from the same companies as the 37 managers who filled out the grouped version. Scale scores from both versions were used to ensure a sample large enough to do the confirmatory factor analysis. The scale intercorrelations for the combined sample (N 275) are shown in Table 6. The pattern of correlations is consistent with the of behaviors into proposed grouping but within the same metacategory metacategories, = 26 the values are low enough (less than . 0) to treat the behaviors as distinct. The intercorrelations among the 12 behavior scales were analyzed using Lisrel 8. 51 confirmatory factor analysis with maximum likelihood estimation (Joreskog & to test the fit of the nested of theoretical models. sequence Sorbom, 1996) Table 6. Correlation Matrix for Behavior Scales. Three nested factor models derived from the literature were evaluated in terms of the fit of each model to the sample correlation matrix. Model 1 represents the proposition that the variance in all twelve leadership behavior scales can be explained by a single common factor.
The common factor could represent an overall leadership activity dimension, or scale method variance, or a combination of both. Model 2 represents the proposition that leader behaviors can be classified as either task-oriented or relationship-oriented. The task metacategory was defmed as including clarifying, monitoring, and short-term planning. The relations metacategory was including supporting, recognizing, developing, consulting, and empowering. The change-oriented behaviors were unconstrained for as defined this model and were free to load on the other two factors.
Model 3 represents the the proposed hierarchical taxonomy, as defined earlier in Table 1. Following current practice (Joreskog, 1993; McDonald & Ho, 2002), multiple criteria were used to test the fit of the models to the matrix of intercorrelations. The Chi-squared statistic was computed to test the overall goodness of fit between the observed correlation matrix and the matrix reproduced from the factor model. A large 27 value of this statistic indicates a poor fit of the model to the data, because it indicates a significant difference between the observed correlation matrix and the correlations estimated by the theoretical model.
Unfortunately, this statistic is affected by sample size as well as model fit. A large sample often yields a significant value even when the model fits the data very well. For that reason, some authors have suggested evaluating Chi squared relative to its degrees of freedom. Marsh and Hocevar (1985) suggested that a ratio of less than 2. 0 indicates a reasonable model fit. Model fit was also evaluated with three indices that are not affected by sample size: the Goodness of Fit Index (GFI), the Comparative Fit Index (CFI), and the Root Mean Square Error of Approximation (RMSEA).
The GFI (Bentler, 1990) is an index of how well the theoretical model reproduces the observed correlations. The CFI (Bentler, 1990) is an index of how well the theoretical model fits the data compared to the null model (which hypothesizes no relationships between any variables). CFI and GFI values range from zero to 1. 0, and it has been proposed that an acceptable fit requires a value of at least . 90, with . 95 representing a very good fit (Bentler and Table 7. Nested The RMSEA (Steiger, 1990) is an estimate of the mean difference between each observed and reproduced correlation.
It has been proposed that an acceptable fit requires a value of . 08 or less, with . 05 representing a very good fit (Browne & Cudeck, 1993; Hu & Bentler, 1999). It must be noted, however, that all decision criteria for judging the adequacy of fit of a measurement model are only rules of thumb. The results from the comparison of models are shown in Table 7. For models 1 and 2 there was a poor fit to the data. Thus, we can reject the hypotheses that the intercorrelations among the twelve leader behaviors can be explained in terms of a single common factor or by a two-factor (task vs. relations) model.
The results indicated that the three-factor TRC model fits the data significantly better than does the traditional two factor model, and most of the fit indices were adequate. Nevertheless, the moderately large Chi-square value associated with Model 3 indicates that an additional path or paths could be added from the latent factors to the observed scales in order to reproduce the intercorrelations among the scales Bonett, 1980; Hu and Bentler, 1999). more closely. Confirmatory Factor Analyses of Leader Behavior Models. Note: GFI = Goodness of fit index; CFI = comparative fit index; RMSEA = Root mean square rror of approximation. Examination of the modification indices associated with Model 3 indicates that adding an additional path from the task factor to the developing scale yields a significant improvement in the fit of the model to the data. Each of the fit indices improved, and the ratio of Chi-squared to degrees of freedom droped below 2. 0. Paths should be added in confirmatory factor analysis only if there is a theoretical rationale to support their addition (McDonald & Ho, 2002). The additional path is appropriate, because most behaviors can contribute developing to task and relations objectives. imultaneously We also explored the possibility of adding other paths to Model 3, but they yielded only small improvements in fit. The factor loadings for the modified Model 3 are shown in Figure 1, and they all are statistically significant. 28 Figure 1 Path Coefficients for Modified Model 3. 29 DISCUSSION The proposed hierarchical taxonomy includes most of the specific behaviors found to be relevant for effective leadership in research conducted over the past half century, even though the labels are not always the same. We found the 12 specific behaviors measured by the questionnaire can be grouped into he three proposed metacategories in terms of their The results from the primary objective. confirmatory factor analysis showed that the model representing the proposed hierarchical taxonomy provided a better fit with the data than any of the alternative models we tested. Our hierarchical taxonomy offers a It provides a number of advantages. and meaningful conceptual parsimonious framework that shows how the behaviors are interrelated. It combines the parsimony of a few, broadly defined metacategories with the greater explanatory power of specific component behaviors that can be related to the requirements for a particular situation.
It helps to integrate findings from prior research, and it can be used to derive more comprehensive theories of effective leadership. The use of objectives as the basis for identifying the metacategories does not sort all specific behaviors into mutually exclusive categories, because actual behaviors often involve more than one type of objective. For example, consultation can be used either to improve the efficiency of procedures for performing the current task, or for eliciting innovative ideas for new tasks.
Encouraging innovative thinking (intellectual stimulation) can be used either to develop the cognitive skills of subordinates or to encourage a more creative, open-minded view of change. A more intensive analysis would reveal whether the behaviors differ in important ways when used for different purposes. In the process, we could gain a better understanding of best practices for managers in different situations. The taxonomy identifies behaviors that are potentially relevant for effective leadership, but it is not assumed that they are equally relevant in all situations, or that every behavior is relevant in every situation.
In future research, it will be desirable to relate the leadership behaviors to criteria of leadership effectiveness. Because many of the behaviors emphasize leader influence on collective processes rather than on dyadic processes, it is especially desirable to conduct studies that include objective measures of unit performance. The studies should also measure aspects of the situation (e. g. , environmental uncertainty, competitive strategy, and external threats opportunities) that determine the relative importance of the different behaviors. In this way, we can make progress in developing contingency theories of flexible, adaptive leadership.
The research reported here is promising, but it has some limitations. One possible limitation involves the validity of the behavior scales. Our study demonstrated that respondent perceptions of leader behavior can be described by the proposed hierarchical taxonomy, but it was not our objective to verify the accuracy of the individual scales. Questionnaires that to recognize and remember require respondents many different types of behavior observed over a period of months or years are very susceptible to response biases and attributions (Yukl, 1998). Some behaviors may be observed more easily and rated more accurately than others.
To minimize measurement problems, we selected only behaviors that have some prior evidence of construct validity. However, our scales are not identical to the ones used in earlier questionnaires. We attempted to select the best examples of each component behavior from a diverse set of existing measures, and we also created some new items to improve content validity. Moreover, we used a response format that emphasized magnitude rather than frequency. Only more intensive research with multiple methods and multiple raters for each leader can assess how accurately each of our scales measures the intended behavior construct.
Another possible limitation involves the range of behaviors included in the research. The questionnaire included a representative selection of behaviors for which there was some prior evidence of construct validity, but to limit the length of the questionnaire, we did not include all leadership behaviors identified in prior research. It may be desirable in future research to include additional behaviors to determine if the three metacategories provide an adequate basis for classifying them as well as the behaviors currently in the taxonomy. 30 involves use of the metacategories in leadership research.
In the past research there has been an overemphasis on metacategories. For example, much of the research during the 1960s and 1970s used of consideration and initiating measures rather that examining results for structure, specific components of these broadly-defined behaviors. The metacategories are useful for organizing specific behaviors with a similar objective, but they should not be used as a substitute for the specific behaviors. Researchers should examine results for the specific behaviors as well as the metacategories in the same study.
The utility of the metacategories will depend on the extent to which they are able to improve the prediction of A final caveat Bowers, D. G. , & Seashore, S. E. (1966). Predicting organizational effectiveness with a four-factor theory of leadership. Administrative Science Quarterly, 11, 238-263. Bradford, D. L. , & Cohen, A. R. (1984). Managing for excellence: The guide to developing high performance organizations. New York: John Wiley. Browne, M. W. , & Cudeck, R. (1993). Alternative ways of assessing model fit. In K. A. Bolen & J. S. Long (Eds. ), Testing structural equation model fit (pp. 136-162).
Newbury Park, CA: Sage. Carroll, S. J. , Jr. , & Gillen, D. J. (1987). Are the classical management functions useful in describing managerial work? Management Review, 12, 38-51. Castro, Academy of leadership effectiveness or the explanation of why some leaders are more effective than others in a given situation. REFERENCES Barling, J. , Weber, T. , & Kelloway, E. K. (1996). Effects of transformational leadership training on attitudinal and financial outcomes: A field experiment. Journal of Applied Psychology, 81, 827-832. Bass, B. M. (1985). Leadership and performance beyond expectations. New York: Free Press.
Bass, B. M. (1990). Handbook of leadership: A survey of theory and research. New York: Free Press. Bass, B. M. , & Avolio, B. J. (1990). Multifactor leadership questionnaire. Palo Alto, CA: Consulting Psychologists Press. Bauer, T. N. , & Green, S. G. , (1996). The development of leader-member exchange: A longitudinal test. Academy of Management Journal, 39, 1538-1567. Bennis, W. G. , & Nanus, B. (1985). Leaders: The strategies for taking charge. New York: Harper & Row. Bentler, P. M. (1990). Comparative fit indices in structural models. Psychological Bulletin, 107, 238-246. Bentler, P. M. (2000).
Rites, wrongs, and gold in model testing. Structural Equation Modeling, 7, 82-91. Blake, R. R. , & Mouton, J. S. (1982). Management by grid principles or situationalism: Which? Group and Organization Studies, 7, 207-210. Bourgeois, L. J. (1985). Strategic goals, perceived uncertainty, and economic performance in volatile environments. Academy of Management Journal, S. L. , & Schriesheim, C. A. (1998). Transformational leadership: A summary of the behavioral dimensions of the construct and an assessment of the validity of new scales measuring each dimension. Paper presented at the Academy of Management Meetings, Chicago.
Conger, J. A. , & Kanungo, R. (1998). Charismatic leadership in organizations. Thousand Oaks, CA: Sage Publications. Ekvall, G. , & Arvonen, J. (1991). Change-centered An extension of the two-dimensional model. Scandinavian Journal of Management, 7, 17-26. Fleishman, E. A. (1953). The description of supervisory behavior. Personnel Psychology, 36, 1-6. Fisher, B. M. , & Edwards, J. E. (1988). Consideration and initiating structure and their relationships with leader effectiveness: A meta-analysis. Academy of Management Best Papers Proceedings, 201-205. Frost, D. E. , Fiedler, F. E. & Anderson, J. W. (1983). The role of personal risk taking in effective leadership. Human Relations, 36, 185-202. Ginter, P. M. , & Duncan, W. J. (1990). Macroenvironmental analysis for strategic leadership: , 91-100. management. Long Range Planning, 23 Grinyer, P. H. , Mayes, D. , & McKiernan, P. (1990). The sharpbenders: Achieving a sustained improvement in performance. Long Range Planning, 23, 116- Halpin, 125. A. W. , & Winer, B. J. (1957). A factorial study of the leader behavior descriptions. In R. M. Stogdill & A. E. Coons (Eds. ), Leader behavior: Its description and measurement.
Columbus, OH: Bureau of Business Research, Ohio State University. House, R. J. (1977). A 1976 theory of charismatic In J. G. Hunt & L. L. Larson (Eds. ), The cutting edge (pp. 189-207). Carbondale, IL: Southern Illinois University Press. House, R. J. , & Mitchell, T. R. (1974). Path-goal theory of leadership. Contemporary Business, 3(Fall), 81- leadership. Leadership: 3, 548-573. Boyatzis, R. E. (1982). The competent manager. New 98. York: John Wiley. 31 A laboratory study of charismatic leadership. Organizational Behavior and Human Decision Processes, 43, 243 -269. Hu, L. , & Bentler, P. M. (1999).
Cutoff criteria for fit in indexes covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling, 6, 1-55. Javidan, M. (1992). Managers as leaders: Developing a profile of effective leadership in top management. In K. Clark, M. B. Clark, & D. P. Campbell (Eds. ), Howell, J. M. , & Frost, P. (1989). Luthans, F. , & Impact of Leadership (pp. 47-58). Greensboro, NC: Center For Creative Leadership. Joreskog, K. G. (1993). Testing structural equation models. In K. A. Bollen & J. S. Long (Eds. ), Testing structural equation models (pp. 294-316). Sage: Newbury Park.
Joreskog, K. G. , & Sorbom, D. (1996). LISREL 8: User’s reference guide. Chicago: Scientific Kim, H. , & Software International. Yukl, G. (1995). Relationships of selfreported and subordinate-reported leadership behaviors to managerial effectiveness and advancement. Leadership Quarterly, 6, 361-377. Kirkpatrick, S. A. , & Locke, E. A. (1996). Direct and indirect effects of three core charismatic leadership components on performance and attitudes. Journal Lockwood, D. L. (1984). Toward an observation system for measuring leader behavior in natural settings. In J. G. Hunt, D. Hosking, C. A. Schriesheim, & R.
Stewart (Eds. ), Leaders and International on managers: perspectives managerial behavior and leadership (pp. 117-141). New York: Pergamon Press. Marsh, H. W. , & Hocevar, D. (1985). Application of confirmatory factor analysis to the study of selfconcept : First and higher-order factor models and their invariance across groups. Psychological Bulletin, 97, 562-582. McCauley, C. D. (1986). Developmental experiences in managerial work (Tech. Rep. No. 26). Greensboro, NC: Center For Creative Leadership. Miller, D. , & Toulouse, J. (1986). Chief executive personality and corporate strategy and structure in small firms.
Management Science, 32, 1389-1409. McDonald, R. P. , & Ho, M. H. R. (2002). Principles and practice in reporting structural equation analyses. Psychological Methods, 7, 64-82. Mintzberg, H. (1973). The nature of managerial work. New York: Harper & Row. Morse, J. J. , & Wagner , F. R. (1978). Measuring the process of managerial effectiveness. Academy of , 23-35. Management Journal, 21 Nadler, D. A. , Shaw, R. B. , Walton, A. E. , & Associates Discontinuous (1995). change: Leading organizational transformation. San Francisco: Jossey-Bass. Peters, T. J. , & Austin, N. (1985). A passion for excellence: The leadership difference.
New York: of Applied Psychology, 81,36-51. Komaki, J. (1986). Toward effective supervision: An operant analysis and comparison of managers at work. Journal of Applied Psychology, 71, 270-278. Komaki, J. , Desselles, M. L. , & Bowman, E. D. (1989). Definitely not a breeze: Extending an operant model of effective supervision to teams. Journal of 522-529. Applied Psychology, 74, Kotter, J. P. (1996). Leading change. Boston, MA: Harvard Business School Press. Random House. Podsakoff, P. M. , MacKenzie, S. B. , Ahearne, M. , & Bommer, W. H. (1995). Searching for a needle in a Kouzes, J. M. , & Posner, B. Z. (1995). The leadership aystack: Trying to identify the illusive moderators of leadership behaviors. Journal of Management, 21, 423-470. Podsakoff, P. M. , MacKenzie, S. B. , Morrman, R. H. , & Transformational leader Fetter, R. (1990). behaviors and their effects on follower’s trust in leader, satisfaction, and organizational citizenship behaviors. Leadership Quarterly, 1, 107-142. Podsakoff, P. M. , & Todor, W. D. (1985). Relationships between leader reward and punishment behavior and group processes and productivity. Journal of challenge: How to get extraordinary things done in organizations. San Francisco: Jossey-Bass. Leana, C. R. (1987).
Power relinquishment versus power sharing: Theoretical clarification and empirical comparison of delegation and participation. Journal of Applied Psychology, 72, 228-233. Leana, C. R. , Locke, E. A. , & Schweiger, D. M. (1990). Fact and fiction in analyzing research on participative decision making: A critique of Cotton, Vollrath, Froggatt, Lengnick-Hall, and Jennings. Academy of Management Review, 15, 137-146. Locke, E. A. , & Latham, G. P. (1990). A theory of goal setting and task performance. Englewood Cliffs, NJ: Prentice Hall. 55-73. Managment, 11, Podsakoff, P. M. , Todor, W. D. , Grover, R. A. , & Huber, V. L. 1984). Situational moderators of leader reward and punishment behavior: Fact or fiction? Organizational Behavior and Human Performance, Lowe, K. B. , Kroeck, K. G. , & Sivasubramaniam, N. Effectiveness correlates of (1996). transformational and transactional leadership: A meta-analytic review of the MLQ literature. Leadership Quarterly, 7, 385-425. 34, 21-63. Sagie, A. , & Koslowsky, in M. (2000). Participation empowerment organizations. and Thousand Oaks, CA: Sage. Schriesheim, C. A. , Neider, L. L. , & Scandura, T. A. (1998). Delegation and leader-member exchange: Main effects, moderators, and measurement issues. 98-318. Academy of Management Journal, 41, 32 Shamir, B. , House, R. J. , & Arthur, M. B. (1993). The motivational effects of charismatic leadership: A 1-17. self-concept theory. Organization Science, 4, Shipper, F. , & Wilson, C. L. (1992). The impact of managerial behaviors on group performance, stress, and commitment. In K. Clark, M. B. Clark, & D. P. Campbell (Eds. ), Impact of Leadership (pp. 119129). Greensboro, NC: Center For Creative Leadership. Steiger, J. H. (1990). Wilson, C. L. , O’Hare, D. , & Shipper, F. (1990). Task cycle theory: The processes of influence. In K. E. M. B. Clark (Eds. ), Measures of Clark &
Leadership (pp. 185-204). West Orange, NJ: Leadership Library of America. Wofford, J. C. , & Liska, L. Z. (1993). Path-goal theories of leadership: A meta-analysis. Journal of Management, 19, 858-876. Yukl, G. (1998). An evaluative essay on current Structural model evaluation and modification: An interval estimation approach. Multivariate Behavioral Research, 25, 173-180. Stogdill, R. M. , Goode, O. S. , & Day, D. R. (1962). New leader behavior description subscales. Journal of Psychology, 54, 259-269. Sundstrom, E. , DeMeuse, K. P. , & Futrell, D. (1990). Work teams: Applications and effectiveness.
American Psychologist, 45, 120-133. Tushman, M. L. , & Romanelli, E. (1985). Organizational evolution: A metamorphosis model of convergence and reorientation. In L. L. Cummings & B. M. Staw (Eds. ), Research in Organizational Behavior (Vol. 7, pp. 171-222). Greenwich, CT: JAI Press. Vroom, V. H. , & Jago, A. G. (1988). The new leadership: organizations. Hall. decision in Managing participation Englewood Cliffs, NJ: Prentice Vroom, V. H. , & Yetton, P. W. (1973). Leadership and making. Pittsburgh Press. Pittsburgh: University of Wikoff, M. , Anderson, D. C. , & Crowell, C. R. (1983).
Behavior management in factory setting: Increasing work efficiency. Journal of Behavior Management, 4, 97-128. Organizational conceptions of effective leadership. European Journal of Work and Organizational Psychology, 8, 33-48. Yukl, G. (2002). Leadership in organizations (5th ed. ). Upper Saddle River, NJ: Prentice Hall. An evaluation of conceptual Yukl, G. (1999). weaknesses in transformational and charismatic leadership theories. Leadership Quarterly, 10, 285305. Yukl, G. , & Fu, P. (1999). Determinants of delegation and consultation by managers. Journal of Behavior, 20, 219-232. Organizational Yukl, G. A. , & Nemeroff, W. (1979).
Identification and measurement of specific categories of leadership behavior: A progress report. In J. G. Hunt & L. L. Larson (Eds. ), Crosscurrents in leadership (pp. Carbondale, IL: Southern Illinois 164-200). University Press. Yukl, G. , Wall, S. , & Lepsinger, R. (1990). Preliminary report on validation of the managerial practices survey. In K. E. Clark & M. B. Clark (Eds. ), Measures of leadership (pp. 223-238). West NJ: Leadership Library of America. Orange, Yukl, G. A. , & Van Fleet, D. (1982). Cross-situational, multi-method research on military leader effectiveness. Organizational Behavior and Human Performance, 30, 87-108.