One purpose of this article is to describe what has been learned about effective leadership behavior in organizations.
Extensive research on leadership behavior during the past half century has yielded many different behavior taxonomies and a lack of clear results about effective behaviors. One purpose of this article is to describe what has been learned about effective leadership behavior in organizations. A hierarchical taxonomy with four meta-categories and 15 specific component behaviors was used to interpret results in the diverse and extensive literature and to identify conditions that influence the effectiveness of these behaviors. Limitations and potential extensions of the hierarchical taxonomy are discussed, and suggestions for improving research on effective leadership behavior are provided.
The essence of leadership in organizations is influencing and facilitating individual and collective efforts to accomplish shared objectives. Leaders can improve the performance of a team or organization by influencing the processes that determine performance. An important objective in much of the leadership research has been to identify aspects of behavior that explain leader influence on the performance of a team, work unit, or organization. To be highly useful for designing research and formulating theories, leader behavior categories should be observable, distinct, measurable, and relevant for many types of leaders, and taxonomies of leader behaviors should be comprehensive but parsimonious.
Thousands of studies on leader behavior and its effects have been conducted over the past half century, but the bewildering variety of behavior constructs used for this research makes it difficult to compare and integrate the findings (Bass, 2008; Yukl, in press). The behavior taxonomies guiding past research have substantial differences in the number and type of behaviors they include. Some taxonomies have only a few broadly defined behavior meta-categories, whereas other taxonomies have a larger number of narrowly defined behavior categories. Some taxonomies are intended to cover the full range of leader behaviors, whereas others include only the behaviors identified in a particular leadership theory. Some taxonomies describe leader behaviors used to motivate individual subordinates, whereas other taxonomies describe behaviors used to lead groups or organizations. Some taxonomies include other types of constructs along with behaviors, such as leader roles, skills, and values. Additional confusion is created by lack of consistency in the use of category labels. Sometimes different terms are used to refer to the same type of behavior, and sometimes the same term is used for different forms of behavior.
The primary purpose of this article is to review what has been learned about effective leadership behavior from research conducted over more than half a century. To integrate results from a large number of studies with many different ways of classifying and measuring leadership behavior, it was first necessary to develop a comprehensive behavior taxonomy. The article begins by describing how decades of behavior research provides the basis for a hierarchical taxonomy with four broad meta-categories and 15 specific component behaviors. Next is a brief overview of research on the effects of widely used behavior categories, followed by a more detailed description of what has been learned about the relevance of each specific behavior in the hierarchical taxonomy. Several conditions that influence the effects of the behaviors are described, and the need for more research on them is explained. The article ends with a summary and suggestions for improving future research.
Research on Behavior Taxonomies
The method used most often to identify categories of leadership behavior is factor analysis of behavior description questionnaires. This method is most useful when clear, relevant items are selected for the initial questionnaire and respondents are able to remember the leader’s past behavior and provide accurate ratings. Unfortunately, the selection of behavior items for a questionnaire is usually influenced by preconceptions about effective leadership or the desire to develop a measure of key behaviors in a leadership theory. The sample of respondents is seldom systematic, and the accuracy of most behavior questionnaires is seriously reduced by respondent biases and attributions. Finally, the basic assumptions of factor analysis (high correlation among examples from the same category) do not apply very well when a behavior category includes several alternative ways to achieve the same objective and a leader needs to use only one or two of them. The limitations of this method may help to explain the substantial differences among leader behavior taxonomies.
Another common method for identifying distinct behavior categories is to have subject matter experts sort behavior descriptions into categories based on similarity of purpose and content, but this method also has limitations. The selection of categories may be biased by prior assumptions and implicit leadership theories, and disagreements among subject matter experts are not easily resolved. A behavior taxonomy is more likely to be useful if it is based on multiple methods and is supported by research on the antecedents and outcomes of the behaviors.
From 1950 to 1980 most of the research on leadership behavior was focused on explaining how leaders influence the attitudes and performance of individual subordinates. In the early survey research, factor analysis of leadership behavior questionnaires found support for two broadly defined behavior categories involving task-oriented and relations-oriented behaviors. Different labels were used for these meta-categories, including initiating structure and consideration (Fleishman, 1953; Halpin & Winer, 1957), production-centered and employee-centered leadership (Likert, 1961), instrumental and supportive leadership (House, 1971), and performance and maintenance behavior (Misumi & Peterson, 1985). The specific behaviors defining the two meta-categories varied somewhat from one taxonomy to another, and some relevant behaviors were not adequately represented in any of these taxonomies. Finding the two meta-categories was a good start, but researchers failed to conduct systematic follow-up research to build on the initial discoveries.
Leadership behaviors directly concerned with encouraging and facilitating change did not get much attention in the early leadership research. Change behaviors are more relevant for executives than for the low-level leaders studied in much of the early research, and they are more important for the dynamic, uncertain environments that have become so common for organizations in recent decades. In the 1980s one or two specific change-oriented behaviors were included in questionnaires used to measure charismatic and transformational leadership, but leading change was still not explicitly recognized as a distinct meta-category. Researchers in Sweden and the United States (Ekvall & Arvonen, 1991; Yukl, 1999; Yukl, Gordon, & Taber, 2002) eventually found evidence for the construct validity of a leading-change meta-category. The classification of change-oriented behavior as a distinct and meaningful meta-category provided important new insights about effective leadership.
In most of the early research on leadership behavior the focus was on describing how leaders influence subordinates and internal activities in the work unit. Leader behavior descriptions were usually obtained from subordinates who had little opportunity to observe their leaders interacting with people outside the work unit. Thus, it is not surprising that few leadership studies examined external (“boundary-spanning”) behavior, and only a few leader behavior taxonomies included any external behaviors (e.g., Stogdill, Goode, & Day, 1962). However, in the late 1970s and early 1980s, descriptive research on managers found that it is important to influence bosses, peers, and outsiders as well as subordinates (Kaplan, 1984; Kotter, 1982; Mintzberg, 1973), and later research on teams found that boundary-spanning behavior is important for effective team performance (e.g., Ancona & Caldwell, 1992; Joshi, Pandey, & Han, 2009; Marrone, 2010). The importance and uniqueness of external leadership behavior provides justification for classifying it as a separate meta-category.
Hierarchical Behavior Taxonomy
The hierarchical taxonomy proposed in this article describes leadership behaviors used to influence the performance of a team, work unit, or organization. The four meta-categories and their component behaviors are shown in Table 1. Each meta-category has a different primary objective, but the objectives all involve determinants of performance. For task-oriented behavior the primary objective is to accomplish work in an efficient and reliable way. For relations-oriented behavior the primary objective is to increase the quality of human resources and relations, which is sometimes called “human capital.” For change-oriented behavior the primary objectives are to increase innovation, collective learning, and adaptation to the external environment. For external leadership behavior the primary objectives are to acquire necessary information and resources, and to promote and defend the interests of the team or organization. In addition to these differences in primary objectives, each meta-category includes unique specific behaviors for achieving the objectives. The relevance of each component behavior depends on aspects of the situation, and the effect is not always positive for the primary objective or for other outcomes.
Table 1: Hierarchical Taxonomy of Leadership Behaviors
Facilitating collective learning
The proposed taxonomy builds on the extensive factor analysis research by Yukl and colleagues (2002), and it also reflects findings in other taxonomic research linking specific behaviors to the performance of a team or organization. The three meta-categories in the Yukl and colleagues (2002) taxonomy were retained, but another component on task-oriented behavior (problem solving) was added, consulting and delegating were combined into a broader relations-oriented component (empowering), and taking risks to promote change was included in a broader change-oriented component (advocating change). The new taxonomy also includes a fourth meta-category (external behavior). Two of the component behaviors (networking and representing) were not included in the questionnaire used for the Yukl and colleagues (2002) research, and the third component (external monitoring) was in their questionnaire but it was included in the change-oriented meta-category.
Overview of Research on Effects of Leader Behavior
Much of the research on effects of leader behavior has been guided by popular leadership theories that emphasized one or two broadly defined behaviors. Early leadership theories such as path-goal theory (House, 1971), leadership substitutes theory (Kerr & Jermier, 1978), situational leadership theory (Hersey & Blanchard, 1977), and the managerial grid (Blake & Mouton, 1964) emphasized task-oriented and relations-oriented behavior, and these meta-categories were used in much of the research conducted from 1960 to 1980. Reviews and meta-analyses of results from hundreds of studies concluded that both meta-categories are related to independent measures of leadership effectiveness (DeRue, Nahrgang, Wellman, & Humphrey, 2011; Judge, Piccolo, & Ilies, 2004).
Since the 1980s, much of the research on the effects of leadership behavior has been based on theories of transformational and charismatic leadership (Avolio, Bass, & Jung, 1999; Bass, 1985; Conger & Kanungo, 1987; House, 1977; Shamir, House, & Arthur, 1993). As in the earlier research, most of these studies reported results only for composite scores on behavior meta-categories included in the theory. Reviews and meta-analyses of this research found that transformational leadership was related to indicators of leadership effectiveness in a majority of studies, but results were inconsistent for transactional leadership and charismatic leadership (De Groot, Kiker, & Cross, 2000; Judge & Piccolo, 2004; Lowe, Kroeck, & Sivasubramaniam, 1996; Wang, Oh, Courtright, & Colbert, 2011; Yukl, 2013).
The research on effects of broadly defined behaviors has limitations that make the results difficult to interpret. The limitations include differences in the way behavior is defined and measured from study to study, use of composite scores based on diverse component behaviors that do not have the same effects, the exclusion of other relevant behaviors likely to be confounded with the measured behaviors, and over-reliance on weak research methods such as same-source survey studies. The results found for independent measures of leadership effectiveness were much weaker than results found for same-source measures, especially when objective performance measures were used (Burke et al., 2006; Kaiser, Hogan, & Craig, 2008).
The popularity of survey research on meta-categories may have inhibited research on effects of specific behaviors, because the number of studies on them is much smaller. The research on effects of specific leadership behaviors included several types of studies. Some studies used a behavior description questionnaire, but other studies used behavior descriptions from observation, diaries, or critical incidents. Several multiple-case studies used interviews, records, and other data collection methods to investigate how leader decisions and actions influenced performance for a team or organization, and the behavior of effective and ineffective leaders was usually compared. A few studies used laboratory or field experiments in which leader behavior was manipulated to assess the effects on subordinate performance. The findings in this research provide evidence that each of the 15 specific behaviors in the proposed taxonomy is relevant for effective leadership.
Effectiveness of Specific Leader Behaviors
In this section, the relevance of each specific component behavior is briefly explained, and the research linking it to