In this paper I present a retrospective review of the development and history of the path-goal theory of leader effectiveness. 1 briefly describe the origin of the theory. The theory is then summarized. The various methodologies that have been used to test the theory and lessons learned from empirical testing are discussed. Two legacies of the theory are described: the substitutes for leadership theory and the 1976 theory of charismatic leadership. A reformulated 1996 path-goal theory of work unit leadership is presented.

The reformulated theory specifies leader behaviors that enhance subordinate empowerment and satisfaction and work unit and subordinate effectiveness. It addresses the effects of leaders on the motivation and abilities of immediate subordinates and the effects of leaders on work unit performance. The reformulated theory includes 8 classes of leader behavior, individual differences of subordinates, and contingency moderator variables which are related to each other in 26 propositions. The contingency moderators of the theory specify some of the circumstances in which each of the behaviors are likely to be effective or ineffective. It is argued that the essential underlying rationale from which the propositions are derived is strikingly parsimonious. The essence of the theory is the meta proposition that leaders, to be effective, engage in behaviors that complement subordinates' environments and abilities in a manner that compensates for deficiencies and is instrumental to subordinate satisfaction and individual and work unit performance. This meta proposition, and the specific propositions derived from it, are consistent with, and integrate, the predictions of current extant theories of leadership. Further, the propositions of the theory are consistent with empirical tests with empirical generalizations resulting from earlier task and person oriented research.

It is my hope that the 1996 theory will be subjected to empirical tests and that such tests will lead to a further improved theory to be formulated at some future time.

It has been over twenty five years since the original publication of the path-goal theory of leader effectiveness (House, 1971). Since that time there have been between 40 and 50 studies designed to test propositions of the theory. The results of these empirical investigations are mixed, some showing support and others failing to support the theory. Unfortunately, as Yukl (1994a) has noted, the theory has not been adequately tested. This state of affairs is largely a result of the use of inappropriate methods to test the theory. The use of inappropriate methods is partially due to the methodological precedents established in the original tests (House, 1971), as well as the prevailing norms in the 1970s and 1980s which were rather lenient with respect to methodological and conceptual rigor. Further, the boundary conditions of the theory were not adequately specified. Upon reflection I believe it is now possible to specify the boundary conditions within which the theory likely holds. I will discuss the above issues in greater detail below.

A number of lessons can be learned from a review of the history of path-goal theory of leadership. In some ways the history of path-goal theory reflects the history of the field of organizational behavior. In the following sections I will briefly describe the origin of the theory. The theory will then be summarized. The various methodologies that have been used to test the theory and lessons learned from empirical testing will then be discussed. I will then describe two legacies of the theory. A reformulated 1996 path-goal theory of work-unit leadership will then be presented. Finally I will conclude with a personal closing comment.

ORIGIN

The path-goal theory of leader effectiveness was developed to reconcile prior findings and anomalies resulting from empirical investigations of the effects of leader task orientation and leader person orientation on subordinate satisfaction and performance. Prior to the introduction of the theory, the leadership literature was dominated by concerns with, and research on, task and person orientation. The most frequently used measures were the Ohio State leader initiating structure and leader consideration scales (Stogdill & Coons, 1957). The findings were mixed. Some studies showed positive relationships between these two variables and leader, work-unit, or subordinate performance and satisfaction. Some studies found either no such relationships, or a positive relationship between only one of the two leader behaviors and dependent variables. Further, several studies showed negative relationships between leader initiating structure and various indicators of subordinate satisfaction (Bass, 1990; Korman, 1966).

The theory was stimulated by Evans' (1970) paper in which the relationships between the Ohio State measures of leader consideration and leader initiating structure and follower perceptions of path-goal relationships (instrumentalities and expectancies) were assessed. Evans found support for the hypothesis that the leader behaviors would be positively related to follower path-goal perceptions in one organization, but not in a second organization. At the time I read the paper by Evans, I was struggling to make sense of a set of findings that indicated that the same leader behaviors had different effects from sample to sample. The findings by Evans suggested to me that the effects of the two leader behaviors are likely contingent on the organizational context in which the leaden and followers worked.

More specifically, at the time I read Evans' 1970 paper, I was thinking about some of my own recently computed findings that showed a positive relationship between leader initiating structure and satisfaction of subordinate white collar professional employees in research and engineering departments of large manufacturing organizations. Such a relationship was not found in prior studies. Rather, the literature at that time included only reports of negative relationships between leader initiating structure and subordinate satisfaction (Korman, 1966). Thus I was faced with an anomaly for which I had no explanation until I reed the paper by Evans. Evans' paper suggested to me that the relationship between structure and subordinate satisfaction and motivation is contingent on the degree to which subordinates needed clarification of the behaviors required of them in order to perform effectively. Once I began thinking in terms of such contingencies and the effect of leaders on subordinate motivation, a number of hypotheses came to mind. I called Evans and asked him how he felt about my publishing a paper entitled, "Path-Goal Theory of Leadership." He replied that his paper did not present a theory, and encouraged me to develop one. Thus the theory was born.

THE THEORY

The Scope Of The Theory

The scope of path-goal theory reflects the dominant paradigm of the study of leadership through about 1975. Path-goal theory is a dyadic theory of supervision. It concerns relationships between formally appointed superiors and subordinates in their day-to-day functioning. It is concerned with how formally appointed superiors affect the motivation and satisfaction of subordinates. It is a dyadic theory of supervision in that it does not address the effect of leaders on groups or work units, but rather the effects of superiors on subordinates.

Consistent with the dominant leadership paradigm of the time, path-goal theory is primarily a theory of task and person oriented supervisory behavior. Also consistent with the dominant paradigm, it does not concern the leadership of entire organizations, emergent-informal leadership, leadership as it affects several levels of managers and subordinates in organizations, political behavior of leaders, strategic leadership of organizations, or leadership as it relates to change.

Essential Notions

In the initial version of the theory it was asserted that "the motivational function of the leader consists of increasing personal payoffs to subordinates for work-goal attainment and making the path to these payoffs easier to travel by clarifying it, reducing roadblocks and pitfalls, and increasing the opportunities for personal satisfaction en route." (House, 1971, p. 324).

In a later version of path-goal theory House and Mitchell (1974) advanced two general propositions:

Leader behavior is acceptable and satisfying to subordinates to the extent that the subordinates see such behavior as either an immediate source of satisfaction or instrumental to future satisfaction. (p. 84).

Leader behavior is motivational, i.e., increases effort, to the extent that ( 1) such behavior makes satisfaction of subordinate's needs contingent on effective performance and (2) such behavior complements the environment of subordinates by providing coaching, guidance, support and rewards necessary for effective performance. (p. 84).

The essential notion underlying the path-goal theory is that individuals in positions of authority, superiors, will be effective to the extent that they complement the environment in which their subordinates work by providing the necessary cognitive clarifications to ensure that subordinates expect that they can attain work goals and that they will experience intrinsic satisfaction and receive valent rewards as a result of work goal attainment. To the extent that the environment does not provide for clear causal linkages between effort and goal attainment, and between goal attainment and extrinsic rewards, it is the leaders function to arrange such linkages. To the extent that subordinates do not perceive such linkages when they do indeed exist, it is the leaders function to clarify such perceptions. Finally, to the extent that subordinates lack support or resources required to accomplish work goals, it is the leaders function to provide such support and resources. Thus, consistent with Katz and Kahn's (1978) definition of leadership, the role of the leader is to provide the necessary incremental information, support, and resources, over and above those provided by the formal organization or the subordinate's environment, to ensure both subordinate satisfaction and effective performance. According to the theory, leaders are justified in their role by being instrumental to the performance and satisfaction of subordinates.

INDEPENDENT VARIABLES: LEADER BEHAVIORS

The independent variables of the theory are leader behaviors. The seminal paper in which the theory was advanced (House, 1971 ) made assertions about two general classes of leader behavior: path-goal clarifying behavior and behavior directed toward satisfying subordinate needs. These behaviors were described generally but not defined operationally as part of the theory. However, in the tests reported in that paper, The Ohio State measures of leader initiating structure and consideration (Stogdill, 1965) were used as approximate measures of path-goal clarifying behavior and behavior directed toward satisfying subordinate needs. Subsequently, House and Mitchell (1974) defined four kinds of behavior in more specific terms:

1. Directive path-goal clarifying leader behavior is behavior directed toward providing psychological structure for subordinates: letting subordinates know what they are expected to do, scheduling and coordinating work, giving specific guidance, and clarifying policies, rules, and procedures. Directive behavior is one form of path-goal clarifying behavior. Nonauthoritarian and nonpunitive directive leader behavior was asserted in the seminal path-goal theory paper to reduce subordinate role ambiguity, clarify follower perceptions concerning the degree to which their effort would result in successful performance (goal attainment), and the degree to which performance would be extrinsically rewarded with recognition by the leader through pay, advancement, job security and the like.
2. Supportive leader behavior is behavior directed toward the satisfaction of subordinates needs and preferences, such as displaying concern for subordinates' welfare and creating a friendly and psychologically supportive work environment. Supportive leader behavior was asserted to be a source of self confidence and social satisfaction and a source of stress reduction and alleviation of frustration for subordinates (House & Mitchell, 1974). Supportive leader behavior was asserted to increase performance of subordinates to the extent that it increases the net positive valences associated with goal-directed effort (House, 1971). Thus supportive leader behavior was expected to increase performance when such behavior was contingent on goal-directed effort.
3. Participative leader behavior is behavior directed toward encouragement of subordinate influence on decision making and work unit operations: consulting with subordinates and taking their opinions and suggestions into account when making decisions. Participative leader behavior was asserted to have four effects: first, to clarify path-goal relationships concerning effort and work-goal attainment and work-goal attainment and extrinsic rewards; second, to increase congruence between subordinate goals and organizational goals, because under participative leadership subordinates would have influence concerning their assigned goals and therefore would select goals they highly value; third, to increase subordinate autonomy and ability to carry out their intentions thus leading to greater effort and performance; fourth, to increase the amount of pressure for organizational performance by increasing subordinate involvement and commitment and by increasing social pressure of peers.
4. Achievement oriented behavior is behavior directed toward encouraging performance excellence: setting challenging goals, seeking improvement, emphasizing excellence in performance, and showing confidence that subordinates will attain high standards of performance. Achievement oriented leader behavior was asserted to cause subordinates to strive for higher standards of performance and to have more confidence in their ability to meet challenging goals.

Implicit Assumptions and Boundary Conditions

The initial version of the theory made four implicit assumptions. First, it was assumed that individuals choose the level of effort they will devote to their tasks on the basis of the degree to which they expect to receive, or experience, valued outcomes as a result of their effort. Thus, the theory makes a strong self interest driven assumption about the nature of subordinates' work motivation. Second, the theory assumed that the propositions of valence-expectancy theory of motivation (Vroom, 1964) were adequate to account for individual work motivation. Valence-expectancy theory on which path-goal theory of leadership rests implicitly assumes that individuals cognitively calculate work outcomes contingent on the level of effort they put forth and that they consciously choose the level of effort to be expended which will maximize the attainment of valent outcomes. Thus path-goal theory of leadership made a strong rationality assumption about individual work motivation. In the reformulated theory advanced below, we define the first two assumptions as boundary conditions for the path-goal clarifying propositions of the reformulated theory.

The initial theory further assumed that the experience of role ambiguity is stressful and unpleasant and that reducing ambiguity will lead to subordinate satisfaction and effective performance. Role ambiguity is experiencing lack of clarity about what is expected of one, how one will be evaluated, and criteria for evaluation. Stinson and Johnson (1975) and Yukl (1994a) note that some people like jobs in which duties and responsibilities are not defined in detail and there is ample opportunity to define their own work role. They therefore argue that path-goal theory rests on a questionable assumption that role ambiguity is stressful. What Stinson and Johnson, and Yulk, are talking about has little to do with role ambiguity as defined in path-goal theory. Rather, they are concerned with latitude for description, not ambiguity about evaluation criteria and process. I continue to assume that role ambiguity, as defined in the original path-goal theory, is unpleasant and stressful. Substantial evidence and managerial implications of this assumption were reviewed in House (1970). Original data in support of this assumption were presented in the seminal path-goal theory paper and in two additional papers (House & Rizzo, 1972; Rizzo, House, & Lirtzman, 1970). An abundance of empirical evidence in support of this assumption has been subsequently reviewed by Jackson and Schuler (1985) whose meta analysis of over 30 studies clearly shows that the experience of role ambiguity is dissatisfying and stressful.

Yukl (1994a, p. 290) also notes that the theory assumes that reduction of role ambiguity will result in increased expectancies and that sometimes role clarification may make it clear to individuals that successful task performance and goal attainment is more difficult than the individual initially expected (1994a, p. 290). Propositions 11, 12 and 13 of the reformulated theory explicitly recognizes and deals with this possibility.

The mixed findings with respect to empirical tests of the theory are in part likely due to the strong self interest and rationality assumptions of the theory. Clearly individuals engage in behaviors that are not self interest driven. One example of such behavior is organizational citizenship behavior (Organ, 1988). The rationality assumption has been shown to hold only under rather limited conditions. It is likely that the propositions concerning path-goal clarifying behavior hold and are most predictive when it is possible to rather accurately assess the probability of attaining valued outcomes, contingent on high, medium, or low levels of effort. Thus, the propositions concerning path-goal clarifying behavior are most likely invalid when subordinates are under conditions of substantial stress (Fielder & Garcia, 1987) or uncertainty (Simon, 1987). Such conditions make it impossible to formulate accurate, confident, and rational expectations of rewards contingent on effort expended. It is most likely that propositions concerning path-goal clarifying behavior hold under conditions of certainty or risk, and when subordinates are not highly stressed. Under such conditions probabilities can be assessed rationally. Therefore, these conditions satisfy the underlying rationality assumptions of the theory. These are the boundary conditions of the path-goal clarifying propositions of the theory.

METHODOLOGY USED

Leader Behavior Scales

The original tests of the theory relied on convenience samples of white collar employees that I had in my possession at the time I formulated the theory. The leader behavior measurements consisted of precursors to the Ohio State Form XII leader consideration and leader initiating structure scales (Stogdill, 1965). The consideration and initiating structure scales were assumed to measure leader supportive person oriented behavior and nonauthoritarian directive path-goal clarifying leader behavior, respectively. By use of the convenience data collected with these scales, I was able to demonstrate the plausibility of the validity of the theory. The leader initiating structure and consideration scales used in the seminal studies, as pointed out in the original path-goal theory paper, are only approximate measures of the leader behavior constructs of the theory. Unfortunately, by setting the precedent of using approximate measures, I stimulated much unintended mischief by subsequent scholars. This mischief has to do with inappropriate measurement of the constructs of the theory. I discuss this issue below.

Confusion Over Scales: Measurement Issues

Unfortunately, several subsequent tests of path-goal theory employed the subscales of the Ohio State Leader Behavior Description Questionnaire (LBDQ) (Fleishman, 1957), the Supervisory Behavior Description Questionnaire (SBDQ) (Fleishman, 1972), or items selected from these scales. It is likely that these scales were selected due to the fact that Martin Evans and I had both used versions of the LBDQ, due to confusion concerning the various versions of previously published Ohio State Scales, and due to the availability of these scales in the public domain.

Schriesheim and Von Glinow (1977) point out that the dimensions of these questionnaires "...differ substantially from the constructs of the theory ..." (p. 399). They state that these scales "...have been found to measure very different kinds of behavior ...contain punitive, autocratic, and production oriented items ...which are extraneous to the measurement of the theory's leadership constructs." (p. 399, emphasis by the authors). Schriesheim and Von Glinow go on to note that "...nevertheless nearly all tests of the theory continue to use the Ohio State Leadership scales." (p. 399)

The leader initiating structure scales in the SBDQ and the pre Form XII versions of the LBDQ do not capture leader coaching, goal clarification, path clarification, the use of contingent rewards, or a number of other work facilitating behaviors included in the path-goal clarification construct of the theory. Further, the inclusion of production emphasis, autocratic, and punitive items in these scales are inconsistent with the path-goal clarification construct of the theory. In contrast the initiating structure scale used in the seminal path-goal studies (House, 1971 ) did not include any items descriptive of punitive, production emphasis, or autocratic leader behavior. Thus, it can be argued theoretically that tests of the theory based on the SBDQ and the pre Form XII versions of the LBDQ are not valid. Since this issue is crucial to assessing the validity of the theory, I review the relevant empirical evidence below.

Empirical Evidence Relevant to Measurement Issues

Schriesheim, House, and Kerr (1976) have shown empirically how the differential findings using the various versions of the Ohio State initiating structure scale can be explained in terms of the version of the scale used and the occupational level of the subordinates of the leaders studied. Schriesheim and Von Glinow (1977) reviewed prior tests of the theory. They argued that form XII of the LBDQ is an approximate measure of the theoretical path-goal clarification construct because it includes a number of path clarification items but does not include autocratic, production emphasis, or punitive items. They then showed that tests based on the form XII LBDQ scales are more supportive of the theory than tests based on SBDQ and the pre Form XII versions of the LBDQ. Specifically, tests of 5 of 7 hypotheses based on form XII of the LBDQ and various subsets of items selected from this questionnaire were supported. In contrast, only 3 of 9 tests of path-goal hypotheses based on items from the SBDQ or the pre Form XII LBDQ were supported.

Schriesheim (1976) developed scales to specifically measure path-goal clarifying behavior and supportive behavior in a manner consistent with path-goal theory. Unfortunately his scales have not been used to test the theory. Further, and not surprisingly, original data reported by Schriesheim and Von Glinow (1977) showed that scales that correspond to the theory produce results more consistent with the theory than the SBDQ or pre Form XII versions of the LBDQ scales.

The theory asserts that when the task demands of followers are ambiguous, nonauthoritarian leader directive behavior will be a source of clarification and therefore instrumental to both follower performance and satisfaction. The conclusion to be reached from the above empirical evidence is that the inclusion of punitive, production emphasis, and autocratic items in the SBDQ or pre Form XII versions of the LBDQ scales offsets the positive effects of directive path-goal clarifying behavior. Consequently, tests of this proposition using the SBDQ and pre Form XII versions of the LBDQ scales cannot be considered valid tests of the theory.

The Ohio State leader consideration scales are also problematic. These scales include items that describe participative as well as supportive leader behavior. When tasks are unambiguous, supportive leader behavior is predicted by path-goal theory to have a positive effect on follower satisfaction and motivation. However, participative leader behavior is not necessarily called for under such conditions, and may be inappropriate. Thus, these items may obfuscate the effects of supportive leader behavior.

A further problem concerns the often found positive correlation between structure and consideration scales of the Form XII LBDQ (Stogdill, 1965). When the two measures of leader behavior are significantly correlated, the prediction should concern the partial correlation of one of the leader behaviors with the dependent variable, holding the effect of the other leader behavior constant. The need for this procedure stems from the hypothesis that each of the leader behaviors will have unique effects.

This procedure was rarely followed in the studies with which I am familiar. Since several opposite predictions of the theory are made for initiating structure and consideration, failure to control for the confounding effects of the second leader behavior on the first completely invalidates the test. Evidence for this assertion was provided in an early study by House and Dessler (1974). Failure to use appropriate partial correlations runs throughout the literature on path-goal theory and is a fatal flaw of many of the tests of the theory.

The issue of appropriate measurement is important for future development of the field of organizational behavior. The use of existing approximate measures of constructs should be seriously questioned. Publication of empirical research which follows this norm should be allowed only in exceptional circumstances.

Proxies For Theoretical Moderators

The theory predicts that followers whose jobs are satisfying, but which have unclear performance demands, will view nonauthoritarian leader directive behavior as satisfying and instrumental for performance. In contrast, followers whose jobs are dissatisfying, but which have unambiguous performance demands, will view leader directive behavior as over controlling and dissatisfying. Several authors have grouped respondents into white and blue collar categories, or have grouped followers according to their organizational level to test the above predictions. The assumption of such grouping procedures is that white collar and higher level employees have more satisfying, yet more ambiguous, task demands. While it is understandable that one might assume blue collar employees to have less satisfaction and mote routine and boring jobs than white collar employees, it is risky to make this assumption since many blue collar workers are skilled craft-persons or high level technicians doing challenging work. Further, many blue collar workers are quite satisfied when doing routine work and even highly repetitive tasks.

The use of occupational or organizational level as a moderator is also problematic. Subordinates' level of ability should increase as a function of level unless one assumes that promotion is random and incompetents are promoted as frequently as capable individuals. The high level of ability at higher organizational levels should thus lessen the instrumentality of leader directiveness. That is, leader initiating structure should theoretically be less instrumental to high ability individuals at high levels.

However, ambiguity of role and task demands and satisfaction increase and routineness decreases with increases in level, thus making initiating structure theoretically more instrumental. Consequently, there are multiple and contradictory moderating effects of level, thus making the use of level an inappropriate moderator to test the theory.

The same rationale holds for the moderating effect of level on relationships between supportive leader behavior and dependent variables. Stress and challenge may increase with level thus requiring consideration from the leader. However, satisfaction with the job, job conditions, and compensation increase along with ability to cope, thus offsetting the need for supportive leadership.

This analysis shows that findings based on the use of surrogates to measure the constructs of the theory have resulted in tests that have multiple interpretations and are not adequate to assess the validity of the theory.

INTERVENING VARIABLES

Another problem with tests of the theory to date is that its intervening variables have seldom been assessed. The following five variables are the intervening motivational variables of the theory: intrinsic valence of behavior, expectancy that effort leads to accomplishment, intrinsic valence of goal accomplishment, expectancy that goal accomplishment leads to valent rewards, and the valence of rewards available to followers. The theory asserts that leaders have a direct influence on these variables and that these variables in turn influence subordinate satisfaction, effort, and performance. To my knowledge there have been no tests of the effects of leader behavior on follower valences. Further, the only test of the effects of leader behavior on follower expectancies is that of House and Dessler (1974) which yielded rather strong support for the theory based on two independent samples.

Most of the tests of the theory have assessed the effects of observed leader behavior on followers' satisfaction and performance. While the original theory predicted effects of leader behavior on these variables, adequate operationalization of these predictions requires that other potential sources of variance in satisfaction and performance be controlled. To my knowledge, none of the reported studies have employed such controls.

Because there are so many additional intervening variables that may effect performance and satisfaction, the prevailing literature does not include adequately controlled tests of the prediction of path-goal theory, with the exception of tests which use satisfaction with supervision as a dependent variable. When performance is measured, other causes of performance should be controlled in order for tests of the theory to be adequate. Almost all of the tests of the theory are based on concurrent variation rather than longitudinal tests. In the seminal paper, I suggested experimental as well as correlational tests of the theory (House, 1971, p. 337).

Further, almost all of the studies rely on self support data with respect to the moderator variables. Such moderator variables may reflect social desirability response bias. That is, respondents may be unwilling to describe their jobs, themselves, or their environment in socially undesirable terms. To do so would imply that they are stuck in bad jobs, are "undesirable" persons, or work for undesirable organizations which in turn reflect badly on themselves. Adequate tests of the theory should include independent measures of moderator variables. Exceptions to this concern individual differences. Again, the point to be stressed here is that there have been many cognitive gaps between the theory and its tests.

The conclusion to be drawn from this analysis is that methodological problems associated with prior tests of the theory render these tests not directly applicable to the theory. Consequently, there are so many possible interpretations of the empirical findings that it is impossible to assess die validity of the theory at this time. This is the basis of Yukl's (1994a) assertion that the theory has not yet been adequately tested.

LESSONS LEARNED

Clearly, the lesson learned from the above observations is that measures that only approximate constructs of a theory should not be used to test the theory. Rather, with each new theory advanced it will likely be necessary to develop and validate measures specifically designed to test the theory.

Further, the few longitudinal tests of the theory do not control for extraneous situational variables or correlated measurement error. While these are demanding criteria, it is important that they be met if the field of organization behavior is to establish valid empirical foundations. The only way to test for causal effects of leader behavior is to conduct controlled laboratory or quasi field experimentation. To my knowledge, Ralph Katz (1977) is the only one to have performed a controlled laboratory experiment to test the theory. Katz findings supported the theory. Laboratory experimentation can be used to test for causal inferences. Quasi field experimentation can be used to test for causal inference and assess effect sizes as well.

The effects of historical context present another lesson learned from the history of path-goal theory. At the time path-goal theory was developed, valence-expectancy theory of motivation (Vroom, 1964) was the prevailing motivational theory of the day. Path-goal theory of leadership took as its underlying axioms the propositions of valence-expectancy theory. Since then we have come to realize that individuals are not nearly as rational or cognitively calculating as valence-expectancy theory would have us believe. Viewing path-goal theory in this historical context merely reminds us that theories of the day reflect other theories of the day. While this is to be expected, and is understandable, it is not widely appreciated in the organizational behavior community and is worth pointing out.

For 25 years, from the early 1950s to the mid-1970s, students of leadership were trapped in the limited person and task orientation paradigm of leadership. This paradigm, coupled with the prevailing rationality assumptions underlying motivation theory, resulted in several theories that ignore the effects of nonconscious motives, affect, symbolic leader behavior, and leader behavior that appeals to emotions of followers. Several leadership scholars have become aware of the importance of these variables which were largely overlooked or ignored until the mid-1970s (Bass, 1985; Bennis & Nanus, 1985; Burns, 1978; Conger & Kanungo, 1987; House, 1977; Sashkin, 1988). We still do not have theories of leadership as it relates specifically to major organizational change, political behavior, or strategic competitive organizational performance. Clearly, social scientists need to escape the boundaries of prevailing paradigms and to question prevailing wisdom.

LEGACY

Path-goal theory has given us a two-fold legacy. First, the framework for analysis of leadership in terms of substitutes for leadership offered by Kerr and Jermier (1978) grew out of early work conducted by House, Filley and Kerr (1971). In the speculative discussion of that paper we advanced notions of organizational formalization and occupational level as moderators (substitutes) of the effects of leader behaviors. These notions were further elaborated in the original statement of path-goal theory (House, 1971, p. 326). Substitutes theory is an extension of path-goal theory in that it elaborates in substantial detail many of the moderating variables suggested by path-goal theory. The evidence relative to substitutes theory is mixed (Podsakoff, Mackenzie, & Fetter, 1993). Yet it is widely cited in the organizational behavioral literature and represented in most organizational behavior textbooks.

Second, path-goal theory led to the formulation of the 1976 theory of charismatic leadership (House, 1977). In contrast to earlier leadership theory which primarily addressed the effects of leaders on follower cognitions and behaviors, charismatic leadership theory primarily addresses the effects of leaders on followers' valences, emotions, nonconscious motivation, and self esteem. Charismatic theory has enjoyed considerable support from a number of studies using a wide variety of methods and samples. (See Yukl, 1994b for a review of the empirical evidence and House & Shamir, 1993 for the most recent version of charismatic theory).

Charismatic theory grew out of path-goal theory notions as a result of lengthy discussions between David Berlew and me. About 2 years after the seminal path-goal theory paper was published, I began a rather ambitious consulting project together with David Berlew. Having been a student of David McClelland's about 15 or so years earlier, Dave Berlew had naturally mastered McClelland's theory of personality. According to this theory, the psychological nature of human beings can be explained fairly well by the operation of three motives: achievement, affiliation, and power. These motives are conceived as nonconscious motivators that can be aroused by a select set of stimuli relevant to each motive.

I had read this literature prior to meeting Dave. I was impressed with the achievement motivation training that had been conducted by McClelland and Winter (197 D in India. Dave and I had many long discussions concerning the McClelland theory of personality. Dave believed that effective leaders articulate visions and empower followers by building their sense of self-efficacy.

From my discussions with him, I concluded that effective leaders also arouse motives that are relevant to particular followers' tasks. Thus, I speculated that effective military combat leaders arouse the power motive; effective leaders of social groups arouse the affiliative motive; and effective leaders of salespersons, profit center managers, entrepreneurial firms, and scientists and engineers arouse the achievement motive.

Motive arousal is equivalent to powerfully enhancing valence (attraction) of particular kinds behaviors and outcomes. As a result of motive arousal, the intrinsic valence of selected behaviors and outcomes is substantially increased. From this line of reasoning, and discussions with Berlew, I developed the theoretical notion that path-goal theory needed to be supplemented with a set of propositions concerning leaders who empower followers and arouse motives to enhance intrinsic valences.

If an image of such a leader is formed in the mind's eye, that image is likely to be strikingly similar to the stereotypic charismatic leader. Leaders who enhance follower selfesteem and arouse follower motives appeared to me to be similar to charismatic leaders as commonly perceived. I learned a great deal from my conversations with Dave Berlew. He was a major influence on my thinking and the stimulus for the development of the 1976 theory. Thus, the 1976 theory of charismatic leader was conceived. It had yet to be nurtured and brought to birth.

Later, while at the University of Toronto, I was visited by an elderly gentleman who was very high up in the government of the People's Republic of China. His position was something like the equivalent to that of the head of the National Science Foundation in the United States. When he met me, he stated, "I've been looking forward to meeting you because there are so few Marxists in the field of organizational behavior." I asked, "Whatever led you to believe that I'm a Marxist?" He said, "The path-goal theory. It is a theory of the people! In your theory it is the needs and the conditions of the people that determine the behavior of the leaders. According to the theory leaders are justified only to the extent to which they are instrumental to follower satisfaction and performance. It is clearly a Marxist theory." I now wonder what he would say about the 1976 theory of charismatic leadership, which is clearly a theory about how leaders change people rather than respond to them.

From this story, one can see how path-goal theory led to charismatic theory. D.O. Hebb, a famous psychologist, stated that "A good theory is one that holds together long enough to get you to a better theory." (Hebb, 1969, p. 21). Clearly, path-goal theory held together long enough (in my mind) to set the stage for charismatic theory. Whether charismatic theory is a better theory is still an open question. However, our recent research, and that of at least 20 other investigators, much to my pleasant surprise, shows rather strong support for the theory (Yukl, 1994b).

THE 1996 PATH-GOAL THEORY OF WORK UNIT LEADERSHIP

The substantial amount of empirical research conducted to test path-goal theory suggests that the theory is in need of reformulation. In the light of this evidence, I suggest a number of propositions as a reformulated 1996 path-goal theory of leadership. The reformulated theory is a theory of work unit leadership. It specifies leader behaviors that enhance subordinate empowerment and satisfaction and work unit and subordinate effectiveness. It addresses the effects of leaders on the motivation and abilities of immediate subordinates and the effects of leaders on work unit performance.

Axioms

The axioms of the theory are propositions assumed to be true for the sake of studying the consequences that follow from them. The following axioms of the theory provide the foundation for subsequent more specific propositions.

1. Leader behavior is acceptable and satisfying to subordinates to the extent that the subordinates see such behavior as either an immediate source of satisfaction or as instrumental to future satisfaction. This proposition is the first proposition advanced by House and Mitchell (1974) in the second statement of the path-goal theory of leadership.
2. Leader behavior will enhance subordinate goal oriented performance to the extent that such behavior (a) enhances the motivation of work unit members, (b) enhances task relevant abilities of work unit members, (c) provides guidance, (d) reduces obstacles, and (e) provides resources required for effective performance.
3. Leader behavior will enhance subordinate motivation to the extent that such behavior (a) makes satisfaction of subordinate's needs and preferences contingent on effective performance, (b) makes subordinate's tasks intrinsically satisfying, (c) makes goal attainment intrinsically satisfying, (d) makes rewards contingent on goal accomplishment, and (e) complements the environment of subordinates by providing psychological structure, support, and rewards necessary for effective performance.
4. Leader behavior will enhance subordinate task relevant abilities to the extent that the leader engages in subordinate development efforts or serves as a role model from which followers can learn appropriate task relevant behavior.
5. Leader behavior will enhance work unit performance to the extent that such behavior (a) facilitates collaborative relationships among unit members, (b) maintains positive relationships between the unit and the larger organizations in which it is embedded, (c) ensures that adequate resources are available to the work unit, and (d) enhances the legitimacy of the work unit in the eyes of other members of the organization of which the work unit is a part.

The seminal theory focused on the effects of leaders on subordinates' motivation, satisfaction, and performance. Proposition 4 broadens the theory to include effects of leaders on subordinates' ability to perform effectively. Proposition 5 broadens the scope of the theory to include effects of leaders on work unit performance as well as performance of individual subordinates.

Independent Variables: Leader Behaviors

The theory specifies ten classes of leader behaviors that are theoretically acceptable, satisfying, facilitative, and motivational for subordinates. In this section I define the behaviors specified by the theory and advance propositions concerning the effective exercise of these behaviors. Included in these propositions are specifications of the theoretical conditions under which each class of leader behavior is likely to be most functional or dysfunctional.

Path-Goal Clarifying Behaviors

A number of leader behaviors are capable of making subordinates needs and preferences contingent on effective performance by subordinates under a select set of conditions. These include (a) clarifying subordinates performance goals, (b) clarifying means by which subordinates can effectively carry out tasks, (c) clarifying standards by which subordinate's performance will be judged, (d) clarifying expectancies that others hold for subordinates to which the subordinate should and should not respond, and (e) judicious use of rewards and punishment, contingent on performance. These behaviors are referred to as path-goal clarify behaviors in that they metaphorically clarify subordinates' paths to goal accomplishment.

The acceptability and motivational effect of path-goal clarifying behaviors depends on the tasks performed by subordinates. According to the original path-goal theory, path-goal clarifying behaviors will have the most positive effect on subordinates when subordinates' role and task demands are ambiguous and intrinsically satisfying. According to the original path-goal theory, it was assumed that under such conditions path-goal clarifying behavior by superiors will be seen as helpful and instrumental to task performance. Thus:

Proposition 1: When the task demands of subordinates are satisfying but ambiguous, path-goal clarifying behavior by superiors will be a source of clarification and subordinate satisfaction and therefore will be motivational.

The acceptability and motivational effect of path-goal clarifying behaviors depends on subordinates perceptions of their abilities to perform effectively and to resolve task and role ambiguity independently of their superiors. Where subordinates perceive their task relevant ability as high, path-goal clarifying behavior is likely to have little positive effect on motivation of subordinates and to be perceived as excessively controlling.

Proposition 2: The higher the degree of subordinates self-perceived ability relative to task demands, the less subordinates will view path-goal clarifying behavior by superiors as acceptable.

In contrast, when subordinates' task and role demands are unambiguous and not intrinsically satisfying, subordinates will see path-role clarifying behavior as redundant and over controlling. Further, when subordinates tasks are dissatisfying, path-goal clarifying behavior will be seen as a means by superiors to induce followers to work harder at distasteful tasks, thus, consistent with the seminal path-goal theory,

Proposition 3: When the task demands of subordinates are unambiguous and dissatisfying, path-gal clarifying behavior will be dissatisfying to subordinates, will be seen as over controlling, will be resented and resisted and therefore demotivational.

Path-goal clarifying behaviors can be enacted in a nonauthoritarian directive manner or in a participative manner. Participative and directive leadership are defined above. Whether nonauthoritarian directive leadership or participative leadership will be motivational to subordinates will depend first and foremost on subordinates' level of personal involvement in their work. When individuals are highly involved in their work, they take personal responsibility for work quality, take pride in their work, and exercise initiative and creativity to ensure work is accomplished. Consequently, when highly involved in their work, individuals desire to have influence over decisions that affect their tasks or themselves at work. Thus:

Proposition 4: When subordinates are highly personally involved in a decision or a task and the decision or task demands are ambiguous and satisfying, participative leadership will have a positive effect on the satisfaction and motivation of subordinates.

Whether nonauthoritarian directive leadership or participative leadership will be most effective in providing path-goal clarification for subordinates with ambiguous and satisfying tasks who are not highly ego involved in their work will depend on the level of subordinates preference for independence and self directed behavior.

There are a number of personality traits associated with preference or motivation for independence and self directed behavior: need for independence, (Abdel-Halim, 1981; Vroom, 1959), authoritarianism (Vroom, 1959), achievement motivation (McClelland, 1985), internal locus of control (Mitchell, Smyser & Weed, 1975; Runyon, 1973), to name only a few. Individuals with strong preferences for independence and self direction find participative leadership to be valent, and individuals with strong preferences for dependence and direction from others find directive leadership to be valent (Abdel-Halim, 1981; Runyon, 1973; Tannenbaum & Allport, 1956; Vroom, 1959). Thus, the reformulated theory asserts that:

Proposition 5: Whether nonauthoritarian directive leadership or participative leadership will be most effective in providing path-goal clarification for subordinates who are not highly ego involved in their work will depend on the level of subordinates' preference for independence and self directed behavior. Specifically:

Proposition 5a: Individuals with a low preference for independence and self direction will find nonauthoritarian directive leadership to be valent. Therefore, when task demands are ambiguous and satisfying, for individuals with a low preference for independence and self direction, directive leadership will be motivational.

Proposition 5b: Individuals with a high preference for independence and self direction will find participative leadership to be valent. Therefore, when task demands are ambiguous and satisfying, for individuals with a strong preference for independence and self direction, participative leader behavior will be motivational.

As stated above tests of the hypotheses of the original path-goal theory concerning the effects of path-goal clarifying behavior have yielded mixed results. The failure to confirm these hypotheses are likely due to the boundary conditions also described above. Thus:

Proposition 6: Propositions 1 through 5 will be most predictive when it is possible to accurately assess the probability of attaining valued outcomes, contingent on high, medium, or low levels of effort, and will be less predictive when it is impossible to make such assessments accurately.

This proposition suggests that the effects of path-goal clarifying behavior of superiors cannot be predicted from the theory when subordinates are under conditions of substantial stress, or non-reduceable uncertainty. Such conditions make it impossible to formulate accurate and rational expectations of rewards contingent on effort expended. It is most likely that the theory holds under conditions of certainty or risk, and when subordinates are not highly stressed. Under such conditions probabilities can be assessed rationally. Therefore, these conditions satisfy the underlying rationality assumptions of the theory. These represent boundary conditions for the above propositions.

Achievement Oriented Leader Behavior

Another class of leader behavior specified by path-goal theory is achievement oriented behavior. Achievement oriented leader behavior is defined above. Achievement oriented leader behavior is not merely performance or goal emphasis. Through achievement oriented leader behavior leaders stress pride in work and self evaluation based on personal accomplishment.

The effect of leader achievement oriented behavior will depend on the achievement motivation of subordinates. Achievement motivation is a nonconscious concern for personal involvement in competition against some standard of excellence and unique accomplishment (McClelland, 1985). Individuals who are highly achievement motivated are motivated to make accomplishments through their own personal efforts rather than through influencing others or delegation of responsibility for achievement. Individuals with high achievement motivation set goals that are challenging, pursue them persistently and vigorously, take intermediate levels of calculated risk, assume responsibility for goal attainment, anticipate obstacles, establish strategies for goal accomplishment and for overcoming obstacles, and seek and use feedback information (McClelland, 1985).

Nonconscious motivation such as achievement motivation predicts spontaneous behavior in the absence of stimuli, strength of motive arousal in the presence of stimuli, and long term (as long as 16 years) global behavior patterns such as patterns of friendship, leadership, family relationships, and leisure activities (McClelland & Boyatzis, 1982; McClelland, Koestner, & Weinberger, 1989; Spangler, 1992).

For subordinates who have a moderate to high level of achievement motivation, achievement oriented leader behavior arouses subordinates achievement motivation. Occupations in which the achievement motive has been found to be most predictive of performance are technical jobs, sales persons, scientists and engineers, and owners of entrepreneurial firms. Individuals who are highly achievement motivated respond to achievement stimuli such as tasks for which one can assume personal responsibility, tasks which when performed well reflect upon the competence of the individual, tasks that require moderate levels of risk and therefore are challenging, and tasks that provide opportunities for development and performance feedback.

Achievement motivated individuals do not obtain satisfaction from, and usually become frustrated by, tasks that rely on others for effective performance. Consequently, a high level of achievement motivation is dysfunctional for higher level managers whose effectiveness depends on effective delegation (House, Spangler, & Woycke, 1991; Spangler & House, 1991).2 Thus:

Proposition 7: Achievement oriented leader behavior will be effective when enacted by superiors who manage subordinates who have individual responsibility and control over their work.

Proposition 8: Achievement oriented leader behavior will be most motivational for subordinates who are moderately or highly achievement motivated.

Proposition 9: Achievement oriented leader behavior will enhance the valence of performance and increase the intrinsic satisfaction of moderately to highly achievement motivated subordinates.

Work Facilitation

Leader behaviors that facilitate work consists of planning, scheduling, and organizing work; personally coordinating the work of subordinates; providing mentoring, developmental experiences, guidance, coaching, counseling and feedback to assist subordinates in developing the knowledge and skills required to meet expectancies and performance standards; reducing obstacles to effective performance of subordinates by eliminating roadblocks, bottlenecks, providing resources; and authorizing subordinates to take actions and make decisions necessary to perform effectively. The following discussion and propositions specify the conditions under which work facilitation leader behaviors are likely to be effective.

The ability of a work unit leader to effectively plan, schedule, and organize work and to coordinate work through formal mechanisms depends on the degree to which the technology is understood and the work demands are predictable. Technological uncertainty, that is unclear and unknown cause and effect relationships concerning the activities that lead to effective performance, mitigates against one's ability to plan, schedule, and organize work and to coordinate work through pre-arranged coordinative systems.

Similarly, unpredictable changing competitive and environmental conditions have the same effect. Carl Von Clausewitz (Hartwick Case, 1993), the philosopher of military strategy, articulates the problem of adhering to a plan of action. He asserts that uncertainty and chance are the province of war. Similarly, uncertainty and chance are the province of intensive competition. According to Von Clausewitz, under conditions of uncertainty and chance, "... we do not gain all our experience at once, but by degrees: thus our determinations continue to be assailed incessantly by fresh experience, and the mind, if we may use the expression, must always be "under arms." (p. 6)

Conditions of uncertainty and unpredictability therefore require a personal rather than formal planned coordination of work. Thus:

Proposition 10: When the work of the unit is free of technological uncertainty, and the demands imposed upon the work unit are predictable, leader planning, scheduling, organizing, and the establishment of formal pre-arranged coordination mechanisms will facilitate the work of the unit members.

Proposition II: When the work of the unit is characterized by technological uncertainty or the external demands imposed upon the unit are unpredictable, personal coordination of the work by the leader or reciprocal coordination by members of the work unit will facilitate work unit goal accomplishment.

Which of these two modes of coordination will be most effective will depend on the level of ability of work unit members. Thus:

Proposition 11a: When work unit members do not have task relevant knowledge and experience, personal coordination of uncertain work by the leader will facilitate work unit goal accomplishment.

Proposition 11b: When work unit members have substantial task relevant knowledge and experience, coordination of uncertain work by reciprocal coordination among work unit members will facilitate work unit goal accomplishment.

Thus:

Proposition 11c: Under the conditions specified in Proposition lib, work unit effectiveness will be enhanced by delegation of responsibility for reciprocal coordination to work unit members.

Similarly, the degree to which it is necessary to provide mentoring, developmental experiences, guidance, coaching, counseling, and feedback for current performance effectiveness depends on the task relevant knowledge and experience of work unit members. It is possible that the process of clarifying path-goal relationships described above will result in making subordinates aware that effective performance is more difficult than they had believed. When this occurs, the role of the leader is to facilitate the development of subordinates or remove obstacles to their effective performance.

Proposition 12: When work unit members lack task relevant knowledge and experience, developmental efforts on the part of superiors will enhance work unit effectiveness.

A similar rationale holds with respect to supervisory behavior directed toward reduction of obstacles.

Proposition 13: When subordinates lack the necessary task relevant knowledge and experience, supervisory efforts to reduce obstacles faced by subordinates will facilitate work unit accomplishment.

Proposition 13a: When subordinates have the necessary task relevant knowledge and experience, supervisory delegation of authority to subordinates to reduce work related obstacles will facilitate work unit accomplishment.

Supportive Leader Behavior

Supportive leader behavior is described above. Supportive leader behavior provides psychological support for subordinates. Such behavior is especially needed under conditions in which tasks or relationships are psychologically or physically distressing. Supportive relationships increase the quality of relationships between superiors and subordinates (Graen & Cashman, 1975) and decreases subordinate stress. Under conditions of stress individuals are unable to use their intelligence and rely on experience. Consequently, intelligence becomes negatively related to performance (Fledler & Garcia, 1987). Therefore, as a consequence, supportive leader behavior individuals are better able to maximize the application of their intelligence. The following propositions are consistent with the propositions concerning supportive leader behavior stated in the seminal path-goal theory. These propositions have been supported in a number of studies (House & Dessler, 1974; Katz, 1977; Schriesheim & Yon Glinow, 1977).

Proposition 14: When subordinates' tasks or work environment are dangerous, monotonous, stressful or frustrating, supportive leader behavior will lead to increased subordinate effort and satisfaction by enhancing leader subordinate relationships and self-confidence, lowering stress and anxiety, and compensating for unpleasant aspects of the work.

Proposition 15: When tasks are intrinsically satisfying or environmental conditions are not stressful supportive leader behavior will have little effect on follower satisfaction, motivation, or performance.

These propositions have been supported in a number of studies (Downey, Sheridan, & Slocum, 1975, 1976; Fulk & Wendler, 1982; Greene, 1974; House, 1971; House & Dessler, 1974; Schriesheim & Von Glinow, 1977; Stinson & Johnson, 1975).

Interaction Facilitation

Leader behavior that facilitates collaborative and positive interaction consists of resolving disputes, facilitating communication, giving the minority a chance to be heard, emphasizing the importance of collaboration and teamwork, and encouraging close satisfying relationships among members. These behaviors are of special relevance when the work of group members is interdependent. Thus:

Proposition 16: Leader behavior directed toward interaction facilitation will increase work unit cohesiveness and reduce voluntary absenteeism and attrition.

Proposition 16a: Leader behavior directed toward interaction facilitation will increase work unit effectiveness when the work of the unit members is interdependent and the norms of the work group encourage unit members performance.

Proposition 16b: Leader behavior directed toward interaction facilitation will be unnecessary, will increase social nontask related communication, but will not increase work unit effectiveness when the work of the unit members is not interdependent.

Group Oriented Decision Process

Another class of work unit leader behavior concerns the manner by which decisions that affect the group are made. According to Maier (1963) the effectiveness of decisions are determined by the degree to which decisions meet physical and economic requirements, referred to as decision quality, and the degree to which decisions are acceptable to individuals who influence the implementation of decisions. A substantial program of experimental research conduced by Maier (1963,1967) demonstrates that the use of the group oriented decision making process substantially increases decision acceptance and quality. Group decision making is a special case of participative leadership requiring some leader skills that are different from participative leadership between superiors and subordinates as dyads.

The group decision process consists of a number of specific behaviors by group or work unit leaders: posing problems, not solutions to the group, searching for and identifying mutual interests of group members with respect to solving problems, encouraging all members of the group to participate in discussion, ensuring that the participation is balanced so that all contribute and no single individual dominates the discussion, searching for alternatives, delaying evaluation of alternatives until the group members have exhausted their ability to generate alternatives, encouragement of the group to evaluate the advantages and disadvantages of each alternative, and combining the advantages into a creative solution. When problems can be segmented into parts for analysis, effective group leaders also allocate parts of the problems to individuals or subgroups who have special expertise with respect to the problem topic. The research by Meier (1967) and numerous others suggests the following proposition:

Proposition 17: When mutual interests among work unit members with respect to solving problems or making effective decisions exists or can be established, the group decision process will increase both decision quality and decision acceptance.

Proposition 18: When decisions require acceptance by group members for implementation, inclusion of group members in the decision process whose acceptance is required will increase decision acceptance.

Proposition 19: When group members have expertise relevant to the technical or economic quality of decisions, inclusion of group members in the decision process who have relevant expertise will increase decision quality.

Proposition 20: A boundary condition for the successful application of propositions 17, 18 and 19 is that a mutual interest in making effective decisions exists or can be established among the group members involved.

Representation and Networking

Work units require resources to perform the tasks for which they are responsible. The ability of work units to acquire necessary resources depends on their relative power within their organizations and on their legitimacy in the eyes of those upon whom they are dependent. Work units on whom others depend for resources, performance, or information enjoy a relatively high degree of power and therefore are able to obtain the resources necessary to perform their functions and reward work unit members for effective performance (Mintzberg, 1983; Pettigrew, 1973; Pfeifer, 1981). Work units that do not control resources, information, or performance of other units must rely on their perceived legitimacy in order to require such resources. Effective representation of work units contributes to their perceived legitimacy. Consequently, a critical function of leaders of such work unit leaders is work unit representation.

Group representation includes presentation of the group in a favorable manner and communicating the importance of its work to other members of the organization of which the group is a part. According to Yukl (1994a), such representation is enhanced by effective networking of work unit leaders. Networking involves maintaining positive relationships with influential others. Also according to Yukl (1994a), positive relationships are developed by entering into exchanges with others and being an effective trading partner, keeping in touch with network members, joining groups that offer opportunities to make contacts, participating in organization wide social functions and ceremonies, giving others unconditional favors, showing appreciation for favors and the work of others, and showing positive regard for others. This discussion of representation and networking suggests the following propositions:

Proposition 21: Work unit legitimacy and ability to obtain resources will be enhanced by active representation and networking by work unit leaders.

Proposition 22: Active representation and networking by work unit leaders will have a more positive effective on work units with relatively lower inter-organizational power compared to other work units.

Value Based Leader Behavior

Since the mid-1970s there has been developed a body of leadership literature concerning leaders who accomplish extraordinary follower commitment, identification with leader or organizational goals, and performance above and beyond the call of duty. Theoretically such effects are accomplished by appealing to subordinates' cherished values and nonconscious motives and by engaging their self perceived identities, enhancing their self efficacy and sense of consistency, and making their self-worth contingent on their contribution to the leaders' mission and the collective (House & Shamir, 1993). This genre of leadership is referred to as value based leadership. Value based leader behaviors include:

• Articulation of a vision of a better future for followers, to which the followers are claimed to have a moral right.
• Display of passion for the vision, and significant self sacrifice in the interest of the vision and the collective.
• Demonstration of self-confidence, confidence in the attainment of the vision, and determination and persistence in the interest of the vision.
• Selectively arousal of the nonconscious motives of followers that are of special relevance to the attainment of the vision.
• Taking extraordinary personal and organizational risks in the interest of the vision and the collective.
• Communication of high performance expectations of followers and confidence in their ability to contribute to the collective effort.
• The use of symbolic behaviors that emphasize the values inherent in the collective vision.
• Frequent positive evaluation of followers and the collective.

It is the central argument of the value based leadership paradigm that, under a select set of conditions, the above behaviors are generic to the leadership of individuals, small groups, work units, formal or informal organizations, social or revolutionary movements, political parties, societies, or nation states. Theories of the value based leadership have been the subject of approximately 50 empirical studies. Empirical evidence demonstrates that value based leader behavior has powerful effects on follower motivation and work unit performance, with effect sizes in the range of .40 to .50, and generally above .50. (See Bass & Avolio, 1994; House & Shamir, 1993 for a more elaborated statement of this genre of theory and citations of the relevant empirical evidence.)

Conditions for the Exercise of Value Based Leadership

Several authors have argued that value based leader behavior will only be effective under a select set of conditions. All scholars who have attempted to explain value based leadership agree that it must be based on the articulation of an ideological goal. However, since ideological goals often challenge the status quo, their expression is often suppressed. Opportunity to articulate such a goal, whether in stressful or non-stressful situations, can thus be considered as one of the situational requirements for a person to emerge as a value based leader. It is perhaps lack of such opportunity that accounts for the absence of value based leaders, under condition of suppression of democracy, of protest movements in totalitarian countries.

Previously I have argued that there are some roles in society which do not lend themselves to ideological value orientation (House, 1977). These are generally roles requiring highly routine, nonthinking effort in exclusively economically oriented organizations.

It is hard to conceive of clerks or assembly line workers in profit-making finns as perceiving their rules as ideologically oriented. However, the same work when directed toward an ideological goal could lend itself to charismatic value based leadership. For example in World War II, "Rosie the Riveter" expressed the ideological contribution of an assembly Free worker. And such menial effort as stuffing envelopes frequently are directed toward ideological goals in political and religious organizations.

This line of reasoning implies that whenever the roles of followers can be authentically described or defined as providing an opportunity for moral involvement, a leader can have a strong influence on the motivational states of followers by stressing ideological values and engaging in the value based behaviors described above.

Based on a study of U.S. presidential leadership House, Spangler, and Woycke (1991) argued that charismatic value based leadership is also required, or is at least more effective, in situations that require a combination of highly involved and active leadership plus emotional commitment and extraordinary effort by both leaders and followers. Examples of such conditions would be highly competitive environments in which competitor tactics change frequently and rapidly; conditions of environmental uncertainty and change; conditions that are stressful to members of the collective; or conditions in which members of the collective feel unfairly treated, persecuted, or oppressed. Under conditions requiring routine but reliable performance in the pursuit of pragmatic goals, value based leadership is less likely to be required and may even be dysfunctional.

The values inherent in the leader's vision is also a relevant consideration. Value based leadership gives meaning to efforts and goals by connecting them to the deeply held values of subordinates. The values advocated by value based leaders must, of necessity, not be in fundamental conflict with strongly internalized values of work unit members. For example if a CEO of a health management organization were to urge physicians to reduce patient care costs, in the interest of the economic performance of the organizations, at the expense of patient care health, physicians with deeply internalized the Hippocratic oath would likely resist such cost reduction and not perceive the CEO's urging as legitimate.

Weber (1947), House (1977), and Bums (1978), argued that stressful situations facing followers facilitate the emergence and effectiveness of value based leadership. Theoretically, under such conditions there is a felt need on the part of followers for a courageous leader who will challenge the established order and offer a radical, or at least innovative, solution to the stressful conditions that followers experience. If a leader emerges who expresses sentiments that are deeply held by followers, and a solution which followers believe will solve the crisis or eliminate the stressful conditions, such a leader is likely to be viewed as a value based leader. Two empirical studies have demonstrated that crisis or stressful uncertainty indeed facilitates the emergence of value based leadership (House, Spangler, & Woycke, 1991; Pillai & Meindl, 1991). It can also be argued that value based leaders will be more likely to emerge and be effective under ambiguous and uncertain conditions. Under such conditions followers theoretically have a need for the reduction of stressful uncertainty. Waldman, Rameriz and House (1996) found that CEO value based leadership was predictive of the economic performance of organizations facing high uncertainty but not predictive of performance of organizations in relatively certain environments. This discussion suggests the following proposition:

Proposition 23: Five conditions that facilitate the emergence and effectiveness of value based leaders are (a) the opportunity for the leader to communicate an ideological vision, (b) an opportunity for substantial moral involvement on the part of both the leader and subordinates, (c) exceptional effort, behavior, and sacrifices required of both the leaders and subordinates, (d) values inherent in the leader's vision that are compatible with the deeply internalized values of work unit members, and (e) the experience of severe threat, crisis, stress, feelings of unfair treatment, persecution, or oppression induced by sources other than the leader.

Shamir, House, and Arthur (1993) argue that value based leadership is also more likely to be relevant under conditions that do not favor transactional leadership, conditions that Mischel (1973) refers to as weak psychological situations. Transactional leadership involves negotiation between superiors and subordinates concerning the subordinates obligations in return for specific performance effort or accomplishments. Transactional leadership relies on contingent rewards as inducement for performance. Transactional leadership can only be exercised when leaders have ability to link extrinsic rewards to individual performance. Cognitive dissonance theory (Festinger, 1980) suggests that in the absence of extrinsic incentives followers are more likely to look for self-related justifications for their efforts. Cognitive dissonance theory also suggests that when leaders engage in specific transactions with subordinates, and make rewards contingent on specific performance outcomes, such transactional leadership undermines the leaders ability to foster an ideological orientation toward work. Under transactional leadership work becomes motivated toward the satisfaction of subordinates' self interest and this motivation undermines work unit member moral involvement in work and motivation toward making contributions to the work unit as a collective. Thus:

Proposition 24: The emergence and effectiveness of value based leadership will be enhanced to the extent that:

1. Extrinsic rewards cannot be, or are not made, contingent on individual performance.
2. There are few situational cues, constraints, and reinforcers to guide behavior and provide incentives for specific performance.
3. The leader refrains from the use of extrinsic rewards contingent on subordinate performance.

Finally, the relationship to the values inherent in the leader's vision and those of the larger organization are also relevant. The vision and powerful motivational ability of value based leaders is often a double edged sword. For example, a value based leader may emerge as a result of dissatisfaction of work unit members with the conditions under which they work, or strong disagreements between work unit members and the dominant coalition of their organization. As a result, the leader may have a vision that represents the values of the work unit members and is inconsistent with the values held by the dominant coalition or the culture of the larger organization. Under such conditions value based leadership is likely to result in inter-group conflict between the work unit managed by the value based leader and either other work units or the dominant coalition of the organization. Thus:

Proposition 25: When the values inherent in the vision of a value based leader are in conflict with the dominant coalition of the larger organization or the prevailing culture of the organization, value based leadership will induce substantial inter-group conflict, or conflict between the leader's work unit and the dominant coalition of the organization.

Shared Leadership

It is not necessary that the above behaviors be performed only by formally appointed work unit leaders. Bowers and Seashore (1966) studied the relationship between a number of leader behaviors similar to those specified in the reformulated 1996 path-goal theory of work unit effectiveness: supportive leadership, goal emphasis, work facilitation, and interaction facilitation. They measured the extent to which each of these behaviors was enacted by both superiors and peers in 40 agencies of a leading life insurance company. They found correlations ranging from .49 to .82 between the degree to which superiors and peers enacted specific behaviors. Despite the fact that these correlations likely reflect a fair amount of common method-common source hits, it is clear that the exercise of leader behaviors can be shared by members of work units as well as conducted by formal work unit managers. Further, Bowers and Seashore (1966) found that "peer leadership" often had a higher correlation with agency performance than leadership exercised by the formal manager of the agency. Finally they found that the highest correlations between manager and peer leadership measures consisted of measures of the same behavior: manager and peer work facilitation, manager and peer goal emphasis, manager and peer interaction facilitation, and manager and peer support, suggesting that the manager sets the example of appropriate peer leader behavior. The findings presented by Bowers and Seashore (1976) suggest the final proposition of the reformulated 1996 path-goal theory of leadership:

Proposition 26: When the work of work unit members is interdependent, encouragement by the leader of collaborative shared responsibility for the exercise of leader behaviors will enhance work unit cohesiveness and performance.

DISCUSSION

The saga of the journey from path-goal theory to then value based leadership theory and to the 1996 theory of work unit leadership informs us of two kinds of traps into which social scientists, or for that matter, any scientific discipline, might fall. First, we too often use existing measures merely because they are available. We justify the use of such measures by citing their prior use in published papers, as if that demonstrates both validity and appropriateness for the topic under investigation. Thus we get trapped in our measurement system and apply it blindly to new questions for which it is inappropriate.

Second, we often become trapped in our own paradigms. Path-goal theory reflects the fact that the leadership scholars of the time were largely trapped in a paradigm of task and person oriented behavior with respect to leadership and a paradigm of cognitive orientation with respect to motivation. Little attention was paid to the importance of nonconscious motives or valence. To my knowledge, there has not been a single test of hypotheses that specify the effects of leader behavior on follower valences even though both intrinsic and extrinsic valences are major variables of the theory.

The propositions advanced here are relevant to 8 classes of leader behaviors that are likely to enhance work unit performance and member satisfaction when exercised under the conditions specified. However, it is unlikely that any one leader will have the ability to engage in all of the behaviors all, or even most, of the time. Effective leaders likely select those behaviors with which they are most comfortable, based on their personality and repertoire of abilities. The specific combinations of leader behaviors most effective for a given individual will likely depend on that individual's social skills and abilities. Those behaviors with which leaders are not comfortable, or for which leaders do not have the necessary abilities or social skills, but which are nevertheless required in specific situations can be shared with, or delegated to, work unit members.

The contingency moderators specified in the theory are intended to specify some of the circumstances in which each of the behaviors are likely to be effective or ineffective. It is possible that work unit effectiveness can be achieved in ways which are not considered in the present theory. No claim is made that the theory includes an exhaustive set of leader behaviors or that the propositions exhaust the conditions under which the various behaviors can be exercised.

It is also likely that some of the behaviors are substitutable for each other. For example, articulation of a vision coupled with role modeling of appropriate behaviors may be substitutable for the path-goal clarifying behaviors described above. Or, leader interaction facilitation or peer supportiveness may be substitutable for, or make unnecessary, supportive leadership. Some of the moderating variables specified by the theory are also likely substitutable for each other. For example, subordinate level of self perceived abilities and subordinate task relevant knowledge may substitute for task structure.

At present, in the absence of additional empirical evidence, I believe the theory would become overly complex by including speculative propositions concerning the interaction among leader behaviors or among the moderating variables of the theory. Hopefully, future empirical research will clarify how such interactions occur.

Current managerial literature emphasizes empowerment of subordinates. The reformulated theory specifies several ways such empowerment can be accomplished. Needed path-goal clarification establishes delegation for authority and responsibility. Work facilitation enhances subordinates' development and ability to work autonomously. Supportive leadership enhances psychological security. Achievement oriented leader behavior arouses achievement oriented behavior and encourages subordinates to take intermediate level calculated risks. Group decision process allows subordinates to influence decision making. Interaction facilitation empowers followers to engage in reciprocal coordination and inter dependent action. Work unit representation enhances the legitimacy of work units and the resources available to work unit members. Value based leadership strengthens subordinate' self efficacy and conviction in the appropriateness of their actions. Value based leadership strengthens collective identification and the motivation for work unit members to contribute to collective goals. Thus, the reformulated theory could well be entitled a theory of work unit empowerment. The advantage of this theory over the frequently found exhortations for empowerment in the managerial literature is that the theory specifies not only empowerment behaviors, but also the conditions under which such behaviors will theoretically be effective. Similar to the original path-goal theory, the reformulated theory asserts that leader behavior is justified only to the extent that it is satisfying and instrumental to the performance of subordinates.

A PARSIMONIOUS INTEGRATIVE PROPOSITION

It can be argued that the reformulated theory lacks parsimony in that it includes ten classes of leader behavior, individual differences of subordinates, and task moderator variables which are related to each other in 26 propositions. However, it can also be argued that the essential underlying rationale from which the propositions are derived is strikingly parsimonious. The essence of the theory is the meta proposition that leaders, to be effective, engage in behaviors that complement subordinate' s environments and abilities in a manner that compensates for deficiencies and is instrumental to subordinate satisfaction and individual and work unit performance. This meta proposition, and the specific propositions relating leader behavior to responses of subordinates, decision effectiveness, superior-subordinate relationships, and work unit behavior are consistent with, and integrate the predictions of, current extant theories of leadership (Bass, 1985; Bennis & Nanus, 1985; Burns, 1978; Conger & Kanungo, 1987; Fiedler & Garcia, 1987; Graen & Cashman, 1975; Graen & Scandura, 1987; House, 1977; House & Shamir, 1993; Kerr & Jermier, 1978; Sashkin, 1988; Vroom & Jago, 1988; Wofford, 1982). Further, the propositions of the theory are consistent with empirical generalizations resulting from task and person oriented research (Bass, 1990; Bowers & Seashore, 1967; Liken, 1977). That this proposition provides the basis for identification and integration of multiple leader behaviors, moderators, and leader effects into a coherent theory, and for the integration of extant theories of leadership as they apply to work unit behavior, illustrates the underlying parsimony of the theoretical rationale for the theory.

The reformulated theory, while broader than the original path-goal theory, remains somewhat limited in scope. It does not concern emergent-informal leadership, leadership as it affects several levels of managers and subordinates in organizations, political behavior of leaders, strategic leadership of organizations, or leadership as it relates to change. These limitations reflect the limitations of current knowledge about effective leadership. Hopefully, future empirical research and theoretical developments will provide additional useful information about leadership not addressed in the theory presented here.

A PERSONAl CLOSING COMMENT

Since I have the liberty of being somewhat autobiographical in this essay, I would like to take this opportunity to share with the reader four assertions that have impressed me and have helped me to formulate the philosophy of science that guides my theoretical and empirical efforts:

• ... all theories, no matter how good at explaining a set of phenomena, are ultimately incorrect and consequently will undergo modification over time (Paraphrased from Mackenzie & House, 1977 [1969], p. 13).
• "A theory which cannot be mortally endangered cannot be alive" (from personal communication to J.R. Platt by W. Ruston, 1964).
• "The fate of the better theories is to become explanations that hold for some phenomena in some limited conditions" (Statement originally by Mackenzie, repeated in Mackenzie & House, 1977, p. 13).
• "A good theory is one that holds together long enough to get you to a better theory" (D.O. Hebb, 1969, p. 21).

The virtue of internalizing the spirit of these assertions is that if one does so one will never be compelled to defend one's own theory, which inevitably will be shown to be false. I believe the 1971 path-goal theory of leadership has lead to better theories, namely the 1976 theory of charismatic leadership, the reformulated 1996 path-goal theory of work unit leadership, and the value based theory of leadership. Hopefully, the 1996 theory will be subjected to empirical tests and a further improved theory will be formulated at some future time.

Acknowledgments: The 1996 Path-Goal Theory of Work Unit Leadership presented in this paper was developed and the first draft of this paper was written while the author was at the Australian Graduate School of Management, The University of New South Wales, during the summer of 1994. The author is indebted to Lex Donaldson, Geoffry Eagleson, Boris Kabanoff, Phillip Yetton and Robert Wood for stimulating discussions of many of the issues presented in this paper.

NOTE

1. This is not to imply that highly achievement motivated individuals are not ambitious, do not have high aspirations, or are not concerned with bringing about outstanding achievements. Rather, I am stressing that achievement motivated individuals, as defined by the McClelland's (1985) theory of motivation, do not receive satisfaction for achievement unless they are personally involved in the achievement and can attribute the achievement to themselves rather than others.

[*] Direct all correspondence to: Robert J. House, Phd. The Wharton School of Management, University of Pennsylvania, Room 23. Fels Building. 3814 Walnut Street. Philadelphia. PA 19104-6197.

REFERENCES

Abdel-Halim, A.A. (1981). Personality and task moderators of subordinate responses to perceived leader behavior. Human Relations, 34, 73-88.

Bass, B.M. (1985). Leadership and performance beyond expectations. New York: Free Press.

Bass, B.M. (1990). Bass and Stogdill's handbook of leadership.' theory, research and managerial applications, 3rd ed. New York: Free Press.

Bass, B.M., & Avolio, BJ. (1993). Transformational leadership: a response to critiques. In M. Chemmers & R. Ayman (Eds.), Leadership theory and research: perspectives and directions (pp. 49-80). New York: Academic Press.

Bennis, W., & Nanus, B. (1985). Leaders: the strategies for taking charge. New York: Harper & Row.

Bowers D.G., & Seashore, S.E. (1966). Predicting organizational effectiveness with a four-factor theory of leadership. Administrative Science Quarterly, 11,238-263.

Burns, J.M. (1978). Leadership. New York: Harper & Row.

Conger, J.A., & Kanungo, R.A. (1987). Toward a behavioral theory of charismatic leadership in organizational settings. Academy of Management Review, 12, 637-647.

Downey, H.K., Sheridan, $.E., & Slocum, J.W., Jr. (1975). Analysis of relationships among leader behavior, subordinate job performance, and satisfaction: a path-goal approach. Academy of Management Journal, 18, 253-262.

Downey, H.K., Sheridan, J.E., & Slocum, J.W., Jr. (1976). The path-goal theory of leadership: a longitudinal analysis. Organizational Behavior and Human Performance, 16, 156-176.

Evans, G. (1970). The effects of supervisory behavior on the path-goal relationship. Organizational Behavior and Human Performance, 5, 277-298.

Festinger, E.A. (1980). Retrospections on social psychology. Oxford, UK: Oxford University Press.

Fiedler, F.E., & Garcia, F.E. (1987). New approaches to effective leadership: cognitive resources and organizational performance. New York: Wiley.

Fleishman, E.A. (1957). Leader behavior description for industry. In R.M. Stogdill & A.E. Coons (Eds.), Leader behavior: its description and measurement (pp. 103-114). Columbus, OH: Ohio State University, Bureau of Business Research.

Fleishman, E.A. (1972). Examiners' manual for the supervisory behavior description questionnaire. Washington DC: Management Research Institute.

Fulk, J.F., & Wendler, ER. (1982) Dimensionality of leader-subordinate interactions: a path-goal investigation. Organizational Behavior and Human Performance, 30, 241-264.

Graen, G., & Cashman, J.F. (1975). A role-making model of leadership in formal organizations: a developmental approach. In J.G. Hunt & L.L. Larson (Eds.), Leadership frontiers (pp. 143165). Kent, OH: Kent State University Press.

Graen, G., & Scandura, T.A. (1987). Toward a psychology of dyadic organizing. In B.M. Staw & L.L. Cummings (Eds.), Research in organizational behavior, vol. 9 (pp. 175-208). Greenwich, CT: JAI Press.

Greene, C.N. (1974). The path-goal theory of leadership: a replication and analysis of causality. Proceedings, Academy of Management, Seattle, 47.

Hebb, D.O. (1969). Hebb on hocus-pocus: a conversation with Elizabeth Hall. Psychology Today, 3(6), 20-28.

House, R.J. (1970). Role conflict and multiple authority in complex organizations. California Management Review, (Summer) 12(4), 53-60.

House, R.J. (1971). A path-goal theory of leader effectiveness. Administrative Science Quarterly, 16, 321-338.

House, R.J. (1977) A 1976 theory of charismatic leadership. In J.G. Hunt & L.L. Larson (Eds.), Leadership: the cutting edge (pp. 189-207). Carbondale, IL: Southern Illinois University Press.

House, R.J., & Dessler, G. (1974). The path-goal theory of leadership: some post hoc and a priori tests. In J.G. Hunt & L.L. Larson (Eds.), Contingency approaches to leadership (pp. 29-62). Carbondale, IL: Southern Illinois University Press.

House, R.J., Filley, A.C., & Kerr, S. ( 1971 ). Relation of leader consideration and initiating structure to R and D subordinate satisfaction. Administrative Science Quarterly, 16, 19-30.

House, R.J., & Mitchell, T.R. (1974) Path-goal theory of leadership. Journal of Contemporary Business, 3, 81-97.

House, R.J., & Rizzo, J.R. (1972). Role conflict and ambiguity as critical variables in a model of organizational behavior. Organizational Behavior and Human Performance, 7, 467-505.

House, R.J., & Shamir, B. (1993). Toward the integration of transformational, charismatic, and visionary theories. In M. Chemmers & R. Ayman (Eds.), Leadership theory and research: perspectives and directions (pp. 81-107). Orlando, FL: Academic Press.

House, R.J., Spangler, D.W., & Woycke, J. (1991) Personality and charisma in the U.S. presidency: a psychological theory of leader effectiveness. Administrative Science Quarterly, 36, 364-396.

Jackson, S.E, & Schuler, R.S. (1985). A meta analysis and conceptual critique of research on role ambiguity andl role conflict in work settings. Organizational Behavior and Human Decision Processes, 36, 16-78.

Katz, D., & Kahn, R.L. (1978). The social psychology of organizations. New York: Wiley.

Katz, R.C. (1977). The influence of group conflict on leadership effectiveness. Organizational Behavior and Human Performance, 20, 256-286.

Kerr, S., & Jermier, J.M. (1978). Substitutes for leadership: their meaning and measurement. Organizational Behavior and Human Performance, 22, 375-403.

Korman, A.K. (1966). Consideration, initiating structure and organizational criteria--a review. Personnel Psychology, 19, 349-361.

Kousnes, J.M., & Posner, B.Z. (1987). The leadership challenge: how to get extraordinary things done in organizations. San Francisco: Jossey Bass.

Likert, R. (1977). Management styles and the human component. Management Review, 66, 23-28, 43-45.

MacKenzie, K.D., & House, R.J. (1977). Paradigm development in the social sciences: a proposed research strategy. Academy of Management Review, January, 7-24.

Maier, N.R.F. (1963). Problem-solving discussions and conferences: leadership methods and skills. New York: McGraw-Hill.

Maier, N.R.F. (1967). Assets and liabilities in group problem solving: the need for an integrative function. Psychological Review, 74, 239-249.

McClelland, D.C. (1985). Human motivation. Glenview, IL: Scott Foresman.

McClelland, D.C., & Boyatzis, R.E. (1982). Leadership motive pattern and long-term success in management. Journal of Applied Psychology, 67, 737-743.

McClelland, D.C., Koestner, R., & Weinberger, J. (1989). How do self-attributed and implicit motives differ? Psychological Review, 966 (4), 690-702.

McClelland, D.C., & Winter D.G. ( 197 l). Motivating economic achievement. New York: Free Press.

Mintzberg, H. (1983). Power in and around organizations. Englewood Cliffs, N J: Prentice Hall.

Mischel, W. (1973). Toward a cognitive social learning reconceptualization of personality. Psychological Review, 80, 252-283.

Mitchell, T., Smyser, C.R., & Weed, S.E. (1975). Locus of control: supervision and work satisfaction. Academy of Management Journal, 18(3), 623-631.

Organ, D.W. (1988). Organizational citizenship behavior. Lexington, MA: Lexington Books.

Pettigrew, A.M. (1973). The politics of organizational decision-making. London: Tavistock.

Pfeffer, J. (1981 ). Power in organizations. Boston: Pitman.

Pillai, R., & Meindl, J.R. (1991). The effects of a crisis on the emergence of charismatic leadership: a laboratory study. Presented at the Annual Meeting of the Academy of Management, Miami, FL [Best Paper Proceedings, August, 235-239].

Platt, J.R. (1964). Strong inference. Science, 146 (3642), 347 353.

Podsakoff, P.M., Mackenzie, S.B., & Fetter, R. (1993). Substitutes for leadership and management of professionals. Leadership Quarterly, 4( 1), 1-44.

Rizzo, J.R., House, R.J., & Lirtzman, S.I. (1970). Role conflict and ambiguity in complex organizations. Administrative Science Quarterly, June, 150-63.

Runyon K.E. (1973). Some interactions between personality variables and management styles. Journal of Applied Psychology, 57, 288-294.

Sashkin, M. (1988). The visionary leader. In J.A. Conger & R.A. Kanungo (Eds.), Charismatic leadership: the elusive factor in organizational effectiveness (pp. 122-160). San Francisco, CA: Jossey-Bass.

Schriesheim, C.A. (1976). Development, validation and application of new leadership behavior and expectancy research instruments. Unpublished manuscript, Ohio State University.

Schriesheim, C.A., & Von Glinow, M.A. (1977). The path-goal theory of leadership: a theoretical and empirical analysis. Academy of Management Journal, 20, 398-405.

Schriesheim, C.A., House, R.J., & Kerr, S. (1976). Leader initiating structure: a reconciliation of discrepant research results and some empirical tests. Organizational Behavior and Human Performance, 15, 297-321.

Shamir, B., House, R.J., & Arthur, M. (1993). The motivational effects of charismatic leadership: a self-concept based theory. Organization Science, 4, 1-17.

Simon, H.A. (1987) Making management decisions: the role of intuition and emotion. Academy of Management Executive, 1, 57-64.

Spangler, W.D. (1992). Validity of questionnaire and TAT measures of the need for achieveved two meta-analyses. Psychological Bulletin, 112 ( 1), 140-154.

Spangler, W.D., & House, R.J. (1991). Presidential effectivenss and the leadership motive profile. Journal of Personality and Social Psychology, 60, 439-455.

Stinton, J.E., & Johnson, T.W. (1975). The path-goal theory of leadership: a partial test and suggested refinement. Academy of Management Journal, 18, 242-252.

Stogdill, R.M., & Coons, A.E. (1957) Leader behavior: its description and measurement. Columbus OH: Ohio State University, Bureau of Business Research.

Stogdill, R.M. (1965). Manual for the job satisfaction and expectations scales. Columbus, OH: Ohio State University, Bureau of Business Research.

Tannenbaum A.S, & Allport F.H. (1956). Personality structure and group structure: an interpretive study of their relationship through an event-structure hypothesis. Journal of Abnormal and Social Psychology, 53, 272-280.

Von Clausewitz, C. (1993). Excerpts from "On War," the Hartwick Humanities in Management Institute, Hartwick College. New York: Oeonitqa.

Vroom, V.H. (1964). Work and motivation. New York: Wiley.

Vroom, V.H. (1959). Some personality determinants of the effects of participation. Journal of Abnormal and Social Psychology, 59, 322-327.

Vroom, V.H., & Jago, A.G. (1988). The new leadership: management practices in organizations. Englewood Cliffs, NJ: Prentice Hall.

Waldman, D.A., Ramirez, G.G., & House, R.J. (1996). CEO leadership and organizational performance under conditions of environmental certainty and uncertainty, Working Paper, The Reginald Jones Center for Strategic Management, The Wharton School.

Weber, M. (1947) [1924]. T. Parsons (Ed.). The theory of social and economic organization (translated by A.M. Henderson & T. Parsons). New York: Free Press.

Wofford, J.C. (1982). An integrative theory of leadership. Journal of Management, 8, 27-47.

Yukl, G. (1994a). Leadership in organizations, 3rd ed. Englewood Cliffs, NJ: Prentice Hall.

Yukl, G. (1994b). A retrospective on Robert House's 1976 theory of charismatic leadership and recent revisions. Leadership Quarterly, 4, 367-373.

~~~~~~~~

By Robert J. House[*] The Wharton School of Management