Description
Transformational leadership enhances the motivation, morale, and performance of followers through a variety of mechanisms. These include connecting the follower's sense of identity and self to the project and the collective identity of the organization
STUDY ON THE RELATIONSHIP BETWEEN LEADER CORE SELFEVALUATIONS, TEAM FEEDBACK, LEADER EFFICACY, TRANSFORMATIONAL LEADERSHIP, TEAM EFFICACY, TEAM GOALS, TEAM ACTION AND TRANSITION PROCESSES, AND TEAM PERFORMANCE
This research attempts to explain how internal team leaders can help teams perform in high stress, dynamic environments. Specifically, the aim of the research was to examine a chain of relationships beginning with team leaders' core self-evaluations and ending in team performance. Several core research questions were examined, including: "Do leaders' core self-evaluations and self-efficacy serve as important antecedents of transformational leadership? Is transformational leadership an important antecedent of team efficacy? How does team efficacy lead to higher levels of team performance?" First, I reviewed the extant relevant literature. Second, I derived from this review a theoretical model, which proposes that both leader core self-evaluations and team feedback may influence leader self-efficacy and transformational leadership. Transformational leadership may then influence team efficacy, which in turn may influence performance through its impact on team goals and team action and transition processes. Third, using data collected from a sample of teams composed of first-year
MBA students involved in a week-long business simulation, I tested the model and demonstrated support for a number of hypothesized relationships. Fourth, I revised the model based on modification indices provided by covariance structure analysis. I found that leaders' core self-evaluations do influence their self-efficacy; this efficacy exhibited positive and significant direct relationships with team efficacy and with team processes. In addition, I found support for a chain of relationships linking team efficacy to team goals to team processes to team performance.
THE RELATIONSHIP BETWEEN LEADER CORE SELF-EVALUATIONS, TEAM FEEDBACK, LEADER EFFICACY, TRANSFORMATIONAL LEADERSHIP, TEAM EFFICACY, TEAM GOALS, TEAM ACTION AND TRANSITION PROCESSES, AND TEAM PERFORMANCE
By Narda Roxanne Quigley
Dissertation submitted to the Faculty of the Graduate School of the University of Maryland, College Park in partial fulfillment of the requirements for the degree of Doctor of Philosophy 2003
Advisory Committee: Professor Edwin A. Locke, Co-Chairman/Advisor Professor Paul E. Tesluk, Co-Chairman/Advisor Professor Kathryn M. Bartol Professor Paul D. Bliese Professor Paul J. Hanges
ACKNOWLEDGEMENTS Three very special groups of people offered me an incredible amount of support during my doctoral studies at the University of Maryland: the professors with whom I have had the privilege to work, both in the Department of Management and Organization and in the Department of Psychology; the doctoral students of the Department of Management and Organization; and my family and friends outside the University of Maryland community. It is thanks to these wonderful individuals that I have been able to achieve my goals during my time in the doctoral program. I owe special thanks to both Ed Locke and Paul Tesluk—the two professors with whom I have worked most closely over the past four years—for their unflagging belief in me from the first day I set foot on campus. From Ed I have learned to value the sometimes seemingly contradictory goals of both attention to detail and simplicity in research undertakings; intellectual curiosity and rigor; and the importance of meticulously grounded research. From Paul I have learned to value the importance of viewing research questions and methods from all possible angles; the practical use of having knowledge of many statistical tools; and the importance of departmental community and the effort that should be spent building it. I feel truly honored to have had the opportunity to work with and be mentored by each of them. A number of other professors have generously assisted me during my time in the doctoral program. Kay Bartol has served as a fantastic role model of professionalism, service, and accomplishment; it has been a true pleasure to work with her. Paul Bliese has been extremely generous with his time in meeting with me over the last year to discuss various statistical issues regarding both this dissertation and other research; and Paul
ii
Hanges has helped me clarify many research questions and ways of going about answering them over the years with unbelievable patience. Cindy Stevens, Ken Smith, and Hank Sims, at various points during my time during the program, have provided both professional and emotional support. I also owe thanks to a number of doctoral students—Lisa Dragoni and Wei Liu for being above and beyond even the best cohort that an incoming doctoral student could hope for; Sharyn Gardner for being an incredible mentor; Jen Marrone, Gosia Langa, and Jay Carson for being incredible mentees; Amanuel Tekleab and Riki Takeuchi for always being willing to answer my questions and smiling while they did so; Qing Cao for putting up with my mess in the office; and Abhishek Srivastava, Jon Eckhardt, and Vincent Duriau for their words of wisdom (and humor) over the years. Thank you. Many thanks are also due to a number of individuals outside the University of Maryland community. My parents have been wonderful sources of support; I owe thanks especially to both for providing me with the unlikely combination of unconditional love and high expectations that have paved the way for everything that I have been able to accomplish. Both have also provided true inspiration to me in the examples they have set. To Paran Quigley I owe thanks for helping me to stay grounded, remember myself as a non-graduate student, and keep everything in perspective. To Kay Ovington I owe thanks for countless hours of data entry, computer help, and the like—but more importantly, for her indefatigable support, encouragement, and friendship as I have made my way through the program. To Pat Thomas I owe thanks for the friendship, mentoring, and mail over the years (not to mention the countless dinners at Roy's Place). To Leslie Jansen and Susy Gallor I owe thanks for being both fantastic housemates and friends.
iii
Concerning the execution of this research, I owe special thanks to all of my committee members, particularly Ed and Paul, for their excellent guidance. Special thanks are also due to Joyce Russell, whose helpful cooperation in data collection and subsequent answers to my many questions regarding the ELM and MARKETPLACE during the last year has been an unbelievably wonderful blessing. Thanks, once again, to all.
iv
TABLE OF CONTENTS List of Tables List of Figures Chapter 1—Introduction Chapter 2—Literature Review Chapter 3—Model and Hypotheses Chapter 4—Research Methods Chapter 5—Results Chapter 6—Discussion References vi vii 1 9 34 53 76 86 120
v
LIST OF TABLES 1. Judge, Erez, & Bono's (1998) Meta-Analysis Findings with Respect to the Correlations Between the Four Core Traits 2. Sequence of Events, Measures Collected, and Measure Uses in MARKETPLACE Simulation 3. List of Concepts and Measures Used 4. Descriptive Statistics, Scale Reliabilities, and Agreement Statistics of Study Variables 5. All Study Variable Intercorrelations 77 78 60 65 13
vi
LIST OF FIGURES 1. The Four Core Trait Indicators of Core Self-Evaluations 2. Summary Diagram of All Empirical Findings Regarding Core SelfEvaluations to Date 3. Hypothesized Model 4. Transformational Leadership Ratings Matrix 5. Results of the Hypothesized Model 6. Revised Path Model 18 35 70 80 84 10
vii
Chapter 1 Introduction
Consider the following situation: a team of 28 men at the start of World War I attempt to make the first historic crossing of Antarctica, from the Weddell Sea to the Ross Sea, on the ship Endurance. As they sail south for their destination, they find that the pack ice in the Weddell Sea is unusually thick. Just a day away from their destination, their boat becomes hopelessly trapped in the ice; instead of achieving honor, recognition, and glory, the team now faces a grueling trial of survival. For 10 months the team drifts north with the ice, living on the boat, eating penguins and seals, and camping on ice floes when their ship eventually sinks. What possible chance could this team have to survive? As it happened, the explorer in charge of the expedition was Ernest Henry Shackleton, a man renowned for his good humor, charm, and energy. Described as an "obstinate boy" by his first sea captain, Shackleton grew up to be described by Thomas Orde-Lees (one of his 28 team members) as never "anything but the acme of good humor and hopefulness." Shackleton's positive view of even the most adverse of conditions allowed him to keep the team together; at one point, for example, he asked his crew to train all 60-odd sled dogs they had brought with them so as to keep up morale. All 28 members of Shackleton's team eventually survived the ordeal; this amazing feat has been attributed to Shackleton's legendary positive leadership during a time of intense adversity. Sir Raymon Preistley, a British geologist and explorer who was part of the Endurance team, said, "For scientific leadership, give me [British explorer] Scott, for swift and efficient travel, give me [Norwegian explorer] Amundsen; but when you are in
1
a hopeless situation, when there seems to be no way out, get on your knees and pray for Shackleton." A number of issues surface in regard to the above situation. First, for example, what was Shackleton's response (both emotional and intellectual) as he faced extreme adversity and seemingly hopeless conditions with his team, and why? Second, how did Shackleton's response to the situation influence the confidence his team members had in their ability to overcome the challenge at hand, the team processes they experienced, and their ability to come up with a viable approach to the situation? Finally, did the team's confidence in its own abilities and the team's processes have any impact on whether it successfully navigated the challenges it faced in the frozen Weddell Sea; what exactly was it that allowed the entire team's survival? A relevant theoretical perspective to begin thinking about the above questions associated with leadership is core self-evaluations theory (Judge, Locke, & Durham, 1997). This theory is rooted in appraisal theory (e.g., Arnold, 1960; Lazarus, 1991; Locke, 1969, 1976; Packer, 1985/1986), which broadly asserts that individuals subconsciously appraise objects, people, or events in relation to their perceived needs, values, and commitments—and that emotions are the form by which individuals experience these subconscious appraisals (Judge et al., 1997). Using an appraisal theory perspective, therefore, Shackleton encountered a particular situation, and then was faced with his own emotional response to the situation—which then had an impact on his team's confidence, interactions, and outcomes. Judge et al. (1997), based on Packer (1985/1986), suggested that appraisals occur on many different levels. At the deepest level, appraisals of specific situations are affected by "core evaluations," or the basic
2
conclusions we hold subconsciously about ourselves, other people, and the external environment. Judge et al. (1997) argued that these core evaluations could be considered to be traits, which they define as "stable and consistent ways of thinking, feeling, or acting exhibited by individuals" (p.155). Judge, Locke, Durham, and Kluger (1998) found that core evaluations of the self were associated with core evaluations of the world and others. Four indicator traits have been used to measure these core self-evaluations: self-esteem, generalized self-efficacy, locus of control, and emotional stability/ neuroticism. While core self-evaluations theory has never been applied in a leadership context, the broader consideration of personality factors such as the four indicator traits and their impact on leadership is nothing new. "Great man" theories of leadership, which asserted that leadership qualities were inherited (this last unnecessarily, since they conceivably could be acquired as well), were popular in the nineteenth and early twentieth centuries (Kirkpatrick & Locke, 1991) only to lose their popularity by the mid-twentieth century. In the late twentieth century, trait theories experienced a resurgence of interest; this may have been partly due to the understanding that leadership success was not merely a matter of possessing some trait or combination of traits. Rather, as Kirkpatrick and Locke (1991) theorized, it may be the combination of those traits with certain actions that give rise to more effective leadership. It is here where core self-evaluations may prove to be of interest to modern leadership theory. Transformational and charismatic leadership theories (e.g., Bass, 1985; Burns, 1978; Conger & Kanungo, 1987) have never been particularly clear about the respective contributions of traits and behaviors to those types of leadership. They seem to
3
acknowledge both the importance of specific leadership behaviors and specific traits. The first component of transformational leadership as defined by Bass's (1985) theory, for example, is charisma; the original Greek meaning of the word means "gift," which suggests a trait. However, a different component of Bass's transformational leadership— individual consideration—is very similar to the consideration dimension from the Ohio State—Michigan studies, which were clearly studies about leadership behavior. The incorporation of core self-evaluations as an antecedent to transformational leadership may be an interesting way to examine whether a theoretically relevant trait contributes to transformational leadership behaviors; if so, the relationship between leadership traits and behaviors (in the context of transformational leadership) may be somewhat clarified. Core self-evaluations may also be particularly relevant to examine in the context of transformational leadership, as both concepts have been linked to higher performance through motivational pathways. This leads me to my first research question:
RQ1: Do leader core self-evaluations serve as an antecedent to transformational leadership?
Regardless of whether core self-evaluations serve as an important antecedent to transformational leadership, transformational leadership in team settings has been widely overlooked both by those studying transformational leadership and those studying teams (Zaccaro & Klimoski, 2002). Indeed, the majority of leadership research has centered on organizational leadership—leadership at a much higher level than team leadership. As a
4
result, there is a dearth of knowledge about what the effects of transformational leadership in a small team setting might be. In this dissertation, I seek to begin addressing this issue. It is likely that transformational leadership helps teams transcend the immediate issues and problems they face, helping team members stay focused and motivated. Part of this may be due to the ability of the transformational leader to generate team member confidence in the team's ability to handle the task at hand (team/collective efficacy—e.g., Gibson, 1999; Lindsley, Brass, & Thomas, 1995) and achieve its goals, which may in turn positively influence team processes and team effectiveness. This leads me to my second research question:
RQ2: Does transformational leadership in a team setting influence how teams respond when faced with challenging tasks?
Team efficacy is another critical variable examined in this dissertation (Bandura, 1997). Pescosolido (2001) reviewed the extant work on team efficacy and noted that though much work has demonstrated links between team efficacy and performance, little work exists that has attempted to determine how this type of efficacy is built and what can be done to increase a group's efficacy levels. It seems reasonable, as noted above, to consider team efficacy in tandem with both team leader core self-evaluations and transformational leadership due to their conceptual links. In this dissertation, I also consider whether the team leader's self-efficacy plays a role in linking the leader's core self-evaluations, transformational leadership, and team efficacy. Although some research has found leader efficacy to be an immediate antecedent of team efficacy (Pescosolido,
5
2001), no prior research has examined whether transformational leadership may play a mediating role in that relationship. My third research question, therefore, is as follows:
RQ3: What are the critical antecedents of team efficacy; how important is the role of transformational leadership as a pathway through which the leader may influence team efficacy?
Although a fair amount of research has found team efficacy to be directly related to team performance (e.g., Gully, Beaubien, Incalaterra, & Joshi, in press; Prussia & Kinicki, 1996; Shea & Guzzo, 1987; Silver & Bufiano, 1996; Spink, 1990), questions still remain regarding how exactly team efficacy is translated into performance. Previous research has supported the mediating role of team goals (e.g., Durham, Knight, & Locke, 1997); substantial theoretical support also exists for the mediating role of team action and transition processes (Marks, Mathieu, & Zaccaro, 2001). However, no study to my knowledge has considered how team efficacy may influence performance through its impact on both team goals and team action and transition processes. This leads me to my fourth research question:
RQ4: What are the specific mechanisms through which team efficacy influences performance; do team goals and team action and transition processes play mediating roles?
6
This research contributes to the existing literature on leadership, teams, and core self-evaluations in a number of ways. First, as noted above, few empirical studies have examined the impact of leaders in small team settings in particular. As Zaccaro and Klimoski (2002, p. 5) noted, "although there exist a large theoretical and empirical literature on? leadership? we still know relatively little about how leaders create and direct team processes to achieve collective success." This is especially interesting given the clear importance of work teams in modern organizations; it seems that more attention should be devoted to understanding the interactions of leaders with other team members. Given the paucity of our knowledge in this area, the research in this dissertation has potential to contribute to the field. A second contribution of this research is in examining how core self-evaluations may influence transformational leadership. Judge and Colbert (2002) noted that existing research on leadership traits has not explored the "black box" between personality traits and leadership itself; the processes by which traits affect leadership have gone virtually unstudied. The model proposed in Chapter 3 begins to address this issue by hypothesizing a potential motivational pathway—leader efficacy—through which a leader's core selfevaluations may influence transformational leadership. Therefore, another contribution of this study is in how it illuminates the connection between leader traits and transformational leadership. A third contribution of this research is in delineating a set of potentially major antecedents of team efficacy. As noted above, little work exists that has attempted to determine how this type of efficacy is built and what can be done to increase a group's efficacy levels (Pescosolido, 2001). This research sheds further light on potential
7
antecedents of team efficacy; no prior research has examined team leader core selfevaluations, team leader efficacy, or transformational leadership as antecedents of team efficacy. A fourth contribution of this research is in the potential implications of the findings for selection and staffing purposes in organizations. If core self-evaluations do indeed impact the way leaders respond in the face of challenges to motivate team members and facilitate effective team processes, organizations with teams in difficult situations may want to specifically choose leaders on the basis of their core selfevaluations (which may be related to their ability to effectively handle the situation). It may be that leaders who are above average on core self-evaluations should be selected for teams assigned especially arduous tasks, if the trait helps them to cope with adversity and still be both high performers and satisfied with their jobs. Therefore, the purpose of this investigation is to examine core self-evaluations of leaders and the effects of this dispositional concept on leader efficacy, transformational leadership, team efficacy, team goals, team processes, and team performance. To this end, in Chapter 2, I review the relevant literature. In Chapter 3, I derive the theoretical model that is tested in this dissertation and provide support for the various hypotheses considered. In Chapter 4, I discuss the research methodology I utilized to test the hypotheses. In Chapter 5, I report the results of the study. In Chapter 6, I discuss theoretical, methodological, and practical implications, limitations, and future research directions.
8
Chapter 2 Literature Review
In this chapter, I establish the groundwork for the model proposed in Chapter 3. A review of the literature on core self-evaluations is followed by a discussion of leadership traits, transformational leadership, and emergent leadership. I then discuss team efficacy, team goals, and team processes. I close the chapter with a discussion of the scope of the research.
Core Self-Evaluations In the following review of the existing material on core self-evaluations, I first define core self-evaluations and discuss the significance of the four indicator traits and the connections between them. I then conclude with a review and summary of the findings in this area to date. Definition of core self-evaluations. As Judge, Bono, Erez, and Thoresen (2002) wrote, "Core self-evaluations is a higher order concept representing the fundamental evaluations people make about themselves, their environments, and the relationship between themselves and their environment" (p. 58), but in this dissertation I am only concerned with core evaluations of the self. As noted in the first chapter, the concept is manifested by four indicator traits as shown in Figure 1: self-esteem, emotional stability/neuroticism, locus of control, and generalized self-efficacy. These traits met three criteria that made them appropriate indicator traits for the core self-evaluations
9
Figure 1. The Four Trait Indicators of Core Self-Evaluations.
Self-Esteem Core SelfEvaluations
Locus of Control
10 Emotional Stability
Generalized Self-Efficacy
concept (Judge et al., 1997; Judge & Bono, 2001): evaluation-focus (the degree to which a trait represents evaluation, as opposed to description); fundamentality (as in Cattell's (1965) personality theory, fundamental traits underlie surface traits); and breadth or scope (Allport, 1961, noted that cardinal traits are broader in scope than secondary traits). The four core traits are conceptually related, as their definitions meet the above three criteria. Self-esteem is typically defined as the overall value that one places on oneself as a person (e.g., Harter, 1990). Neuroticism (emotional stability) represents the tendency to exhibit poor emotional adjustment and experience negative feelings such as fear, self-doubt, and depression (Judge et al., 1998; Barrick & Mount, 1991). Generalized self-efficacy is the most general type of self-efficacy, as it "encompasses individuals' judgments of their capacity to mobilize the motivation, cognitive resources, and courses of action needed to exercise general control over? their lives and deal successfully with life's challenges" (Judge et al., 1997). The last of the indicator traits, locus of control (Rotter, 1966), represents the perceived degree of control over the outcomes of one's actions (Judge et al. 1998). Rotter (1966) divided individuals into two camps based on their locus of control: individuals with an internal locus of control believe they control their own lives, while individuals with an external locus of control believe that they are controlled by luck, chance, fate, or powerful others. Empirical and theoretical links between traits. Judge et al. (2002) noted that though the indicator traits seem to be extremely prominent in the literature, few empirical investigations have included more than a single core trait. Indeed, even when two or more of the traits are included in the same study, they are usually treated as unrelated variables (e.g., Abouserie, 1994; Hesketh, 1984; Hojat, 1983; Horner, 1996; Tiggeman &
11
Winefield, 1984). Sometimes the traits are modeled as influences on each other, divorced from any theoretical framework. For example, Wambach and Panackal (1979) claimed an effect of neuroticism on locus of control, whereas Morelli, Krotinger, and Moore (1979) viewed locus of control as a cause of neuroticism. Judge et al. (1998) conducted a meta-analysis among 12 samples of roughly 15,000 individuals to investigate the relationship between the four core traits. Their results are summarized in Table 1. Note that the estimated true score correlations among the traits are substantial—when Judge et al. (1998) completed a factor analysis, they found that the core traits loaded strongly on a single underlying factor. Erez and Judge (2001) confirmed these results and found evidence that core self-evaluations is a higherorder factor that integrates the association among the four lower-level traits. Findings associated with core self-evaluations. Since Judge et al.'s (1997) theoretical piece discussing the nature of the proposed concept of core evaluations, a number of studies have advanced our understanding of the concept. The original concept they proposed included core evaluations of the self, other people, and the world in general (with trust vs. cynicism and just vs. unjust world as examples of two core evaluations in the domains of other people and the world). The individual studies that have been conducted to date examining core self-evaluations are reviewed below. Judge, Locke, et al. (1998) evaluated core evaluations of the self, the world, and other people and also considered the effects of these core evaluations on job and life satisfaction. They found that external core evaluations (those of the world and other people) did not explain further variance in job and life satisfaction than did core self-
12
Table 1. Judge, Erez, & Bono's (1998) Meta-Analysis Findings with Respect to the Correlations Between the Four Core Traits.
Trait Self-Esteem Generalized Self-Efficacy Locus of Control Neuroticism
1 1.00 .86 .58 -.62
2 .70 1.00 .59 -.54
3 .44 .45 1.00 -.47
4 -.51 -.45 -.36 1.00
Notes. K = 12 and n = 15,888 for locus of control—self-esteem, locus of control—neuroticism, and self-esteem—neuroticism correlations. K = 11 and n = 14,777 for correlations involving generalized self-efficacy. Correlations below the diagonal are corrected for measurement and sampling error; correlations above the diagonal are uncorrected. A 95% confidence interval for each correlation excluded zero.
13
evaluations and concluded that how individuals view the external world might, in fact, be a function of how they view themselves. They also found that core self-evaluations were both directly and indirectly related to job and life satisfaction. Positive core selfevaluations were found to have consistent effects on job satisfaction through their effects on job perceptions and also independently of the attributes of the job itself. Judge, Erez, et al. (1998) also investigated the theoretical relationship between core self-evaluations and job performance. They conducted a meta-analysis for correlations among the four core traits over 12 samples of roughly 15,000 individuals. In support of their previous work, as noted above, they found that the traits are highly correlated and comprise a common factor. They also presented a logical argument linking positive core self- evaluations to performance through the impact core self-evaluations have on motivation. In addition, they discussed the implications for selection of core selfevaluations research, covering legal, measurement, applicant reaction, and faking issues. Judge et al. (2000) returned to the consideration of job satisfaction as the primary dependent variable. They expanded upon Judge, Locke, et al., (1998), however, in that they considered both perceived job characteristics and objective job complexity in a procedure that included two studies. The first study found direct relationships between core self-evaluations and perceived job characteristics, between core self-evaluations and job complexity, and between core self-evaluations and job satisfaction. Furthermore, core self-evaluations had an indirect effect on perceived job characteristics through actual job complexity. The authors suggested that individuals with high core self-evaluations may seek out and attempt more complex jobs; they may be more likely to exert greater effort with respect to goal-setting activities and higher task involvement; they may be less
14
likely to withdraw from complex jobs when failure is experienced; and they may have better coping mechanisms, all of which may lead to increased attempts to attain and keep more complex jobs. In the second study covered in Judge et al. (2000), the constancy of the core-self evaluations concept and its relation to job complexity and job satisfaction over time was examined. Core self-evaluations as assessed in childhood and adolescence was related to job complexity and job satisfaction later in life, suggesting that core selfevaluations are indeed dispositional in nature. Erez and Judge (2001) examined the relationship between core self-evaluations, goal-setting, motivation, and performance in three studies. The first of these was to confirm that the four indicator traits loaded on one higher order factor; confirmatory factor analysis supported this hypothesis. The second study demonstrated that more positive core self-evaluations were related to higher levels of task motivation and performance. In addition, motivation (which was measured through amount of time spent on attempting to solve anagrams and also by three direct questions assessing motivation) was found to partially mediate the relationship between core self-evaluations and performance. The final study replicated the above findings in a sample of insurance agents; also, in this study, the effect of core self-evaluations on performance (as measured both by objective number of sales and rated performance) was found to be mediated by both goal-setting and job behavior (activity level—how active the agents were in making sales). The studies also answered several questions regarding the validity of the overall concept. Core self-evaluations appeared to have incremental validity above and beyond conscientiousness in this study, suggesting that core self-evaluations may be an important dispositional predictor of job performance. In addition, when the four core
15
traits were investigated as "one nomological network, the overall construct [concept] proved to be a more consistent predictor of job behaviors than when the individual traits were used in isolation." (Judge & Erez, 2000, p. 1277). Therefore, Judge and Erez (2000) provided some support for Judge, Erez, et al. (1998) in that core self-evaluations may influence performance through having an effect on motivation. Srivastava, Locke, & Judge (2002) revisited the question of why individuals with more positive levels of core self-evaluations exhibit higher levels of satisfaction in a lab setting. They found that core self-evaluations affect the level of task complexity that individuals choose, which was in turn related to perceptions of task characteristics and task satisfaction. Srivastava et al. (2002) also suggested that core self-evaluations might be related to a constellation of behaviors on actual jobs (for example, showing initiative, asking for more responsibility, seeking opportunities for learning, assuming leadership positions, etc.) that would lead to increased satisfaction. Durham, Locke, and Judge (2002), picking up on this last point and on Erez and Judge (2001), examined the relationship of core evaluations to behavior on the job. The authors hypothesized that core self-evaluations would influence job behaviors both directly and indirectly through perceptions of job characteristics. More positive core selfevaluations were found to influence self-reported job behaviors such as seeking greater job challenge; showing persistence in the face of setbacks; using independent judgment; seeking raises and promotions by asking managers how to get them; and motivating subordinates. However, core self-evaluations were not directly related to job behaviors when managers assessed the behaviors. The authors suggested that this last finding may have been due to the fact that managers were not able to monitor employees at all times;
16
therefore, employees may be more aware of their own behaviors than managers are. In addition, all managers probably did not use the same standards in evaluating employee behavior, which may have influenced results. Durham et al. (2002) noted that consideration of their results and the results of Erez and Judge (2001) in tandem yields some interesting conclusions, as each study considered different mediators between core self-evaluations and job behaviors. The conclusion that can be reached is that core self-evaluations may affect behavior on the job indirectly by a variety of paths and possibly directly as well. In addition, core selfevaluations affect both performance and satisfaction directly and indirectly, suggesting that this dispositional trait is worthy of more exploration in different realms. Figure 2 summarizes the findings regarding core self-evaluations to date. The research that has involved core self-evaluations to date provides some compelling evidence of the viability and utility of the theory in predicting motivation, satisfaction, and performance. Another potentially fruitful realm in which to examine core self-evaluations may be in their relationship to leadership. Two streams of leadership theory are particularly relevant: trait theories of leadership and transformational leadership theories. In the next section I discuss the findings and relationship between these two sets of theories; I close the section with a brief discussion of emergent leadership and its relevance in the context of work teams.
17
Figure 2. Summary Diagram of All Empirical Findings Involving Core Self-Evaluations to Date. Motivation, Goal-Setting Productivity
Job Behaviors
Self-Esteem Perceived Job Characteristics Generalized Self Efficacy Performance Ratings 18 Locus of Control Core SelfEvaluations Emotional Stability
Objective Job Complexity
Job/Task Satisfaction
Chosen Level of Task Complexity
Trait and Transformational Theories of Leadership Various theories of leadership abound, of course, ranging from trait theories (e.g., Stogdill, 1948) to behavioral theories (e.g., the Ohio State and University of Michigan leadership studies of the 1940s-50s: e.g., Fleishman, 1953; Halpin & Winer, 1957; Hemphill & Coons, 1957; Katz & Kahn, 1952; Katz, Maccoby, & Morse, 1950; etc.) to contingency theories (e.g., Fiedler, 1964, 1967) to theories of transformational leadership (e.g., Bass, 1985; Conger & Kanungo, 1987) to theories of shared leadership (e.g., Seers, 1996). As a clarifying point, the primary type of leadership upon which I focus in this dissertation is transformational leadership, due to this type of leadership's widely acknowledged impact on follower motivation. Theories of both charismatic and transformational leadership (e.g., Bass, 1985; Burns, 1978; Conger & Kanungo, 1987; House, 1977) have all specifically emphasized the motivating influence these types of leaders may have on followers, suggesting that characteristics, rhetoric, and behaviors of the leader may motivate followers to do extraordinary things. A secondary theoretical approach to leadership used in this dissertation is trait theory. In Chapter 3, I will argue that a team leader's core self-evaluations are an important individual antecedent related to the rise of transformational leadership. As noted in Chapter 1, the approach I will take is consistent with that taken by Kirkpatrick and Locke (1991), who suggested that traits might be a precondition for effective business leadership; effective leaders must take certain actions, however, in order to have a true impact (e.g., role modeling, setting goals, coaching subordinates, etc.). Therefore, in the next few paragraphs, I focus on trait and transformational theories of leadership.
19
Trait Theories of Leadership. As noted in the introduction, trait theories of leadership have been among the oldest and most widely discussed theories of leadership in the twentieth century (cf., Kirkpatrick & Locke, 1991; Stogdill, 1948). Prior to 1948 (the date of a seminal review by Stogdill) trait theories were popular, but often failed to show results. Stogdill's review pointed out these many inconsistencies, and the field of leadership as a whole took a new direction as theorists began to focus more on leadership behaviors and, eventually, leadership styles. Recently, however, trait theories have again become part of the leadership literature, as scholars have been considering the possibility that some individuals may have traits that raise the likelihood of certain effective leadership behaviors (Kirkpatrick and Locke, 1991). Many traits have been hypothesized to influence leader effectiveness, some of which are conceptually related to core self-evaluations. Kirkpatrick and Locke (1991), for example, suggested that traits such as drive, leadership motivation, honesty and integrity, cognitive ability, and knowledge of the business, among others, may be linked to leadership effectiveness. They also suggest that self-confidence and emotional stability are critical traits. For example, managers who are confident in their abilities may be more likely to inspire commitment among and be trusted by their employees. In addition, emotional stability is critical—especially when resolving interpersonal conflicts or facing stressful events and challenges. These two traits are closely related to core selfevaluations—an individual with positive core-self-evaluations is probably more likely to be more confident in his/her abilities, and emotional stability is one of the four indicator traits. A number of trait theorists have identified self-confidence, internal locus of control, and emotional stability as potentially important traits for leaders to have; many of
20
these are reviewed in more detail in Chapter 3. Existing empirical evidence does indicate that examining core self-evaluations in a leadership context is a logical direction to take. Transformational Leadership Theory. While trait theories of leadership experienced a resurgence of interest in the last decade, research on transformational leadership has increased exponentially. Three seminal works are widely credited with initially advancing theories of charismatic and transformational leadership: Burns (1978), House (1977), and Bass (1985). The stream of literature that grew surrounding this new area would eventually overtake research considering all other major theories of organizational leadership combined; as Judge and Bono (2000) noted, out of all articles in the decade between 1990 and 2000 available on the PsycINFO database, 207 articles cited transformational or charismatic leadership theory, and 190 cited other once popular leadership theories (least-preferred coworker theory, situational leadership theory, leadermember exchange/vertical dyad linkage theory, normative decision/Vroom-Yetton theory, behavioral theories, path-goal theory, implicit leadership theory, and romance of leadership). Theories of transformational leadership have certainly captured the attention of researchers in the field of organizational behavior. The approach highlighted in this dissertation emerges from Bass's (1985) theory of transformational leadership. In this theory, there are four dimensions of transformational leadership: idealized influence, inspirational motivation, intellectual stimulation, and individual consideration. Idealized influence comprises both the charismatic and the role-modeling aspect of transformational leadership. This is often considered the most prototypic and single most important dimension of transformational leadership (Judge & Bono, 2000). Inspirational motivation involves the articulation of a
21
clear, inspiring, and appealing vision to followers; intellectual stimulation involves the stimulation of follower creativity and thought by the questioning of assumptions and of the status quo. Finally, individual consideration involves attending to and supporting the individual needs of followers, much like the consideration dimension from the Ohio State—Michigan studies. Unlike the consideration dimension, however, individual consideration is more related to follower development than participative decision-making (Bass, 1995). In the original Bass (1985) piece, both transformational leadership and transactional leadership were discussed. In this research, however, I focus only on transformational leadership; I am most concerned with leaders who are able to motivate followers to pursue a common, inspiring goal and/or vision, rather than with leaders who establish exchange relationships with followers and seek to monitor that exchange (Burns's 1978 and Bass's 1985 transactional leadership type). A substantial amount of evidence has accumulated in support of the effectiveness of transformational leadership. A meta-analysis by Lowe, Kroeck, and Sivasubramaniam (1996) suggested that transformational leadership behaviors are related to subjective ( = .73) and objective ( = .30) measures of leadership effectiveness. Fuller, Patterson, Hester, and Stringer (1996) found similar results, with a correlation of .34 between transformational leadership and leadership effectiveness. Finally, transformational leadership seems to be effective in a variety of cultures using a variety of methods (Bass, 1997). Given this information, examining transformational leadership more closely in work team settings may be a potentially fruitful endeavor.
22
Before ending this section on leadership, I review one final topic: the issue of emergent leadership. This particular type of leadership is particularly important in a variety of organizational situations; I review these in more detail below. Emergent leadership and team development. An underlying assumption in much of the literature reviewed above is that teams operate with a designated individual who begins his/her life with the team as the appointed leader. Many teams in organizational contexts, however, begin existence without the designation of a formal leader. The phenomenon of emergent leadership may be observed in these teams. Schneider and Goktepe (1983) defined emergent leaders as group members who exert significant influence over other members of the group although no formal authority has been vested in them; these leaders, as Taggar et al. (1999, p. 901) noted, "may be just as important to the facilitation of team task completion as are designated leaders." Indeed, the manner in which a leader comes to power (either formally or through an emergent process) may be "unimportant in comparison to the behaviors of the leader" (Firestone, Lichtman, & Colamonosca, 1983). The literature on emergent leadership, much like the literature reviewed above, has therefore placed emphasis on the appropriateness of specific leadership behaviors in certain situations. Evidence suggests that those individuals who behave in different leader-like ways to help the group achieve its goals, given different situations, will be perceived by peers as being more leader-like. For example, Zaccaro, Foti, and Kenny (1991) found that emergent team leaders (who were rated highest on a measure of perceived leadership by their peers) were more adept than other team members at perceiving team requirements and selecting appropriate behavior to these demands
23
(Taggar et al., 1999). In addition, Pescosolido (2001) suggested that different behaviors of an emergent leader may be important during different phases of a group's development; emergent leaders may have more of an impact on certain aspects of group functioning earlier rather than later in a group's development. Seers and colleagues (Seers, 1989; Seers, Petty, & Cashman, 1995) postulated another related approach to how leaders emerge in team situations. They suggest that the process of leadership emergence in teams is that of negotiating roles and relationships between team members. The role an individual assumes on a team, then, depends on his/her abilities, the needs of other group members, and the team task to be completed. Some group members may possess inherent characteristics and exhibit certain behaviors that are perceived by other group members as indicative of leadership (Taggar et al., 1999). From this perspective, different people may emerge as leaders, depending on what roles are perceived as important leadership roles given the specifics of the situation. In this dissertation, I focus on teams that, at their inception, are initially leaderless. Furthermore, the perspective I assume is that certain individuals will surface as emergent, transformational leaders. I suggest in chapter 3 that these individuals will have an impact on team efficacy, team goals, and team processes.
Team Efficacy In the above section, I reviewed findings with respect to trait, transformational, and emergent leadership theories. In Chapter 3, I hypothesize connections between these areas and the development of a team emergent state (Marks, Mathieu, & Zaccaro, 2001)—team efficacy—and team processes. In this section, I focus on team efficacy. I
24
define it, discuss how it is different from the concept of self-efficacy, and then review some of the major findings associated with the concept. The concept of team efficacy has been considered for the last two decades, though research exploring the topic has increased considerably in the last decade. In the framework of social cognitive theory (cf., Bandura, 1991; Bandura, 1997), efficacy beliefs play a key role at all levels of analysis. At the individual level of analysis, selfefficacy is defined as "beliefs in one's capabilities to organize and execute the courses of action required to produce given attainments" (p. 3). It has been described as taskspecific self-confidence (Locke & Latham, 1990); a wide range of studies have confirmed a strong relationship between self-efficacy and performance (e.g., Gist & Mitchell, 1992; Stajkovic & Luthans, 1998). These definitions and findings clearly pertain to the individual level of analysis. However, efficacy beliefs can exist at higher levels of analysis; as early as 1982, Bandura called for a "broad a comprehensive research effort" and the development of suitable tools for measuring team efficacy, which he referred to at that point as "groups' perceptions of their efficacy to achieve varying levels of results" (p. 14). Since that time, a number of studies have assessed the impact of group efficacy on group performance (e.g., Shea & Guzzo, 1987; Silver & Bufiano, 1996; Spink, 1990; etc.) and Bandura has defined team efficacy as the "group's shared belief in its conjoint capabilities to organize and execute the courses of action required to produce given levels of attainment" (Bandura, 1997, p. 477). The critical difference between self-efficacy and team efficacy is the level of analysis. This distinction is important to make, as both self-efficacy and team efficacy are a part of the model presented in Chapter 3. Self-efficacy, as noted above, is used to
25
describe a single individual's assessment of his/her capabilities, acting alone as a single unit. Team efficacy is an analogous concept defined at the team level of analysis (Bandura, 1997; Lindsley et al., 1995). The assumption is that teams can be treated (metaphorically) as social entities that are capable of acting together as a unit (Lindsley et al., 1995). Individuals are typically used as "informants to estimate the group's or organization's collective efficacy" (Lindsley et al., 1995, p. 648); of course, cognitions reside only in the individual, so groups and organizations may possess beliefs only in the sense of shared cognitions. Lindsley et al. (1995) pointed out that the individual-level cognitions of group or organization members are quite distinguishable from their beliefs regarding the group's capabilities. These group-based beliefs fundamentally develop from an individual's ability to cognitively assess the capabilities of other group members and their ability to work together based on observation and experience. Therefore, team efficacy is distinguishable from individual-level self-efficacy in that it is an emergent property of groups. It is also distinguishable from the average level of team member self-efficacy within a group. This concept would refer to the average overall extent to which different team members felt confident about their own individual capabilities to achieve a given level of task performance, rather than the extent to which group members felt confident about the ability of the group as a whole. As noted above, there is an emerging literature on team efficacy. This work has focused on a number of different issues, including the most appropriate way to measure the concept (e.g., Gibson, Randel, & Earley, 1999), the most appropriate way to conceptualize and distinguish it separately from related concepts such as team potency (e.g., Guzzo et al., 1993), and its influences on a variety of organizational phenomenon.
26
Research has demonstrated a strong link between group efficacy and group motivation and performance (e.g., Bandura, 1997; Durham, Knight, & Locke, 1999; Gully, Beaubien, Incalaterra, & Joshi, in press; Prussia & Kinicki, 1996), and Jex and Bliese (1999) found that team efficacy acts as a buffer of stressor-strain relationships. There is a good deal of promise associated with the continued incorporation of group efficacy into models of team performance. Given the promising results so far, I include group efficacy in Chapter 3 as a major aspect of the model to be presented. In the next section, I review the literature on team goals.
Team Goals Much research has been conducted in the last three decades regarding the operation of goals at the individual level. Specific, difficult goals—if accepted—lead to higher levels of performance than easy goals or no goals (Locke & Latham, 1990). Although the majority of goal-setting research has been conducted at the individual level, goal-setting at the team level has been receiving more attention as of late (cf. Locke & Latham, 1990; O'Leary-Kelly, Martocchio, & Frink, 1994; Weldon & Weingart, 1993). Basic relationships between goal level and performance at the individual level have been replicated at the team level (cf. Locke, Durham, Poon, & Weldon, 1997; O'Leary et al., 1994). A number of studies have shown that specific, difficult group goals result in higher levels of performance than when there are no goals or easy goals present (e.g., Becker, 1978; Buller & Bell, 1986; Ivancevich, 1974; Klein & Mulvey, 1989, 1995; Latham & Locke, 1975; Latham & Yukl, 1975; Lawrence & Smith, 1955; Pearson, 1987; Weingart, 1992; Weingart & Weldon, 1991; Welden, Jehn, & Pradhan, 1991).
27
Research examining goals at the team level has been increasingly focused on discovering the mediators and moderators of the goal effect (Durham, Knight, & Locke, 1997; Guzzo & Dickson, 1996; Weldon & Weingart, 1993). As Durham, Locke, Poon, and McLeod (2000) noted, a puzzling theoretical issue in the goal-setting literature currently exists. Research to date has indicated that goals may influence performance in a number of different ways: through direct effects, through both direct effects and effects moderated by strategy or tactics (i.e., interaction effects; Chesney & Locke, 1991; Durham et al., 1997), and through effects mediated by strategy (DeShon & Alexander, 1996; Durham et al., 2000). The approach taken in this dissertation, which will be discussed more in Chapter 3, focuses on the direct effect team goals have on performance. In the next section, I review the literature on team process and denote what processes may potentially be influenced by leadership and team efficacy.
Team Processes Team process research has historically been characterized by the use of a wide and inconsistent array of process variables (Marks, Mathieu, & Zaccaro, 2001); different scholars have varied widely in the extent to which they have used and ways in which they have operationalized different types of variables to represent team process (Weingart, 1997). In the last few decades, however, team processes have occupied a critical role in many theoretical models of team effectiveness (e.g., Gist, Locke, & Taylor, 1987; Guzzo & Shea, 1992; Hackman, 1987, etc.). A number of taxonomies of group process have been proposed (e.g., Argote & McGrath, 1993; Fleishman & Zaccaro, 1992; Neiva, Fleishman, & Rieck, 1978; Prince & Salas, 1993). Many of these models described
28
processes as being the critical mediating mechanisms through which member and organizational characteristics are related to effectiveness criteria (Marks et al., 2001). Despite this proliferation of research on process, however, few approaches to the topic included a temporal dimension with task completion as the centerpiece (the few that exist are Kozlowski, Gully, Nason, & Smith, 1999; Marks et al., 2001; McGrath, 1991); this type of approach is central to the work teams accomplish in organizations. Marks and colleagues (2001) proposed a temporally based framework and taxonomy of team processes to clarify the "process muddle." I use this particular model as the foundation for the hypotheses involving team process presented in Chapter 3. Marks et al. defined team process "as members interdependent acts that convert inputs to outcomes through cognitive, verbal, and behavioral activities directed toward organizing taskwork to achieve collective goals" (p. 357). In their taxonomy, team processes are divided into three categories. Transition processes are those that involve mission analysis, formulation, and planning; action processes involve monitoring progress toward goals, systems monitoring, team monitoring and backup behavior, and team coordination; and interpersonal processes involve conflict management, motivation and confidence building, and affect management. Teams use these different processes simultaneously, though some processes happen more frequently in different phases of a team's work on a task. The two types of phases in each performance episode that Marks et al. (2001) considered were action phases and transition phases. During action phases, teams are engaged in acts that contribute directly to goal accomplishment (e.g., taskwork), while during transition phases, teams focus primarily on evaluation and/or planning activities to guide their accomplishment of a team goal or objective. Marks et al. (2001) posited that
29
while interpersonal processes occur through both phases of the episodic framework, action processes most often occur in the action phase, while transition processes most often occur in the transition phase. In Chapter 3, I focus on these action and transition processes that occur in newly formed teams. Though interpersonal processes are surely related to team viability and member satisfaction, it is beyond the scope of this dissertation to include interpersonal processes and those aspects of team effectiveness. The final section in this chapter will cover the scope of this research. This section is intended to bridge the material I reviewed in this chapter and the theoretical model to be discussed in Chapter 3.
Scope of the Research This section specifies the exact scope of the research for this dissertation. First, I discuss the types of teams and tasks for which this research is most relevant; I then discuss the specific role of leadership; and I close the chapter with a discussion of the levels of analysis examined. Team and task type. Scholars have identified many different types of teams in organizational settings (e.g., Sundstrom et al., 1990—identified advice and involvement teams, production and service teams, project and development teams, and action and negotiation teams; Cohen & Bailey, 1997—identified work teams, parallel teams, project teams, and management teams, etc.). The model discussed in Chapter 3 is intended to generalize to newly formed, autonomous work teams (whose members have no shared history of working together) who are working on an unfamiliar, complex task that must be completed during a specific time period. Moreover, the teams I focus on have just
30
undergone a critical midpoint transition period, wherein they received important feedback regarding past progress and future plans for task completion. The nature of the task itself is an important aspect of group functioning; a number of authors incorporate task structure into theoretical models of group functioning (e.g., Hackman, 1987; Campion, Medsker, & Higgs, 1993). Hyatt and Ruddy (1997) noted that tasks are especially critical to consider in group contexts as they determine what the nature of team effectiveness is for any given situation and can affect the importance of many work group characteristics. The general task type to be examined in this dissertation is a complex information processing and problem-solving task that requires a high level of interaction between team members. This type of task is most congruent with Tesluk, Mathieu, Zaccaro, and Marks's (1997) description of tasks that produce an "intensive" work arrangement, where team members must "work together closely to diagnose and solve problems in performing the task" (p. 203). In addition, these types of dynamic and complex tasks require well-orchestrated teamwork, which may be best spearheaded by a transformational leader. Role of leadership. As noted above, the type of leader examined in this dissertation is not a designated leader, but rather an emergent one. As reviewed in the section in this chapter on emergent leadership, these types of leaders often have a great deal of influence over group members. Emergent leaders of this type of team may fill a wide variety of roles—from being a structure initiator/direction setter to being a facilitator of team process to being a team motivator (e.g., Hackman & Walton, 1986; Zaccaro & Marks, 1999). The focus in this dissertation, however, is on emergent, transformational leaders. This type of leadership and its relationship to the development
31
of team efficacy and team processes will be highlighted more in Chapter 3; the dissertation focuses on the role of emergent leaders due to the fact that they have been linked to the development of team efficacy and subsequent team performance (e.g., DeSouza & Klein, 1996; Pescosolido, 2001; Zaccaro et al., 2001). Levels of analysis. I focus on concepts at two levels of analysis in this dissertation: the individual leader level and the team level. The concepts in the pathway from leader core self-evaluations to transformational leadership (described in detail in Chapter 3) are at the individual leader level. Team efficacy, team goal, team action and transition processes, and team performance are at the team level. Issues of aggregation (e.g., Bliese, 2000; Chan, 1998; Kozlowski & Klein, 2000) arise only with respect to team processes, as the other team-level variables were either assessed through consensus methods or—in the case of team performance—assessed with objective task feedback. I consider team action and transition processes to be an emergent property of groups that arises out of individual-level perceptions—in other words, individuals' combined perceptions and their agreement regarding these perceptions define the group-level characteristic. Therefore, I will be considering both individual leader level and team level variables in the model to be presented in Chapter 3. It is important to note that although I am focusing on two levels of analysis (leader and team level), typical methodological issues associated with cross-level hypotheses (e.g., Klein, Bliese, Kozlowski, Dansereau, Gavin, Griffin, Hofmann, James, Yammarino, & Bligh, 2000) do not present the usual statistical challenges in this dissertation. Although I do consider the effect of leader-level variables on team-level variables—and vice versa—due to the fact that there is only one leader per team, typical
32
cross-level issues such as non-independence in the individual-level data (Bliese, 2000; Bliese & Hanges, 2001) are not as critical.
33
Chapter 3 Model and Hypotheses
In Chapter 2, I developed the groundwork for the model that is tested in this dissertation. I reviewed a number of different theories and findings relating to core selfevaluations, leadership, team efficacy, and team processes, and closed the chapter with a discussion of the scope of the research. In this chapter, I build on that foundation by presenting an overview of the model, followed by the underlying theory and support for each of the specific hypotheses in the model.
Model Overview The model to be presented in this chapter, depicted in Figure 3, incorporates elements of core self-evaluations theory, transformational leadership theory, social cognitive theory, team process theory, and goal setting theory. As noted in Chapter 2, the model is intended to generalize to a common situation found in organizations: that of newly formed work teams, lacking a shared history, that nonetheless face one overarching complex task (which may incorporate a number of subtasks and which may require different inputs at different points in time from leaders and team members—e.g., Kozlowski et al., 1996; Marks et al., 2001) and that have received critical midpoint feedback. Work teams in similar situations have received some recent attention in the literature (e.g., Lester, Meglino, & Korsgaard, 2002); these situations may be particularly conducive to the examination of the effects of team leader core self-evaluations due to the high stress levels involved.
34
Figure 3. Hypothesized Model H4b
H3b
Leader CSE's
H1 Leader's Efficacy Transformational Leadership H2a H3a
H4a Team Efficacy H5a
Team Goals
H6
Team Performance Team Processes
35
Midpoint Feedback
H1
H6
H2b H5b
H6b
In the model, leader core self-evaluations and critical midpoint feedback are hypothesized to impact team efficacy through their effects on leader self-efficacy and transformational leadership, which in turn is hypothesized to have a direct impact on team efficacy. Team efficacy, in turn, is expected to indirectly influence team performance through influencing team goals and processes. In the next sections, each set of relationships in the model will be discussed.
Team Leader Core Self-Evaluations and Midpoint Feedback
Leadership Efficacy
In the following paragraphs, I first discuss the connection between core selfevaluations and leadership efficacy. Then, I discuss the role feedback may play in the specific situation on which this dissertation focuses: teams operating in a dynamic environment that have reached a critical midpoint in their existence. There is both theoretical and empirical reason to believe that a team leader's core self-evaluations should be related to his/her subsequent self-efficacy for leading the team to successful completion of the task. Theoretically, Bandura (1997) suggested that selfefficacy (which is defined as task-specific self-confidence; here, it is defined as the confidence individuals have in their ability to meet the role requirements expected of a leader) primarily arises from four sources of information: enactive mastery experiences, vicarious experiences, verbal persuasion and other types of social influences, and "physiological and affective states from which people partly judge their capableness, strength, and vulnerability to dysfunction" (Bandura, 1997, p. 79). This last source may be one logical link to the team leader's core self-evaluations. When a team faces a complex situation, the team leader is likely to formulate a sense of his/her efficacy from
36
his/her own initial cognitive and affective reactions to the task, as he/she has had no prior experience with the task (either personal or vicarious). The basis for these reactions would be the leader's appraisal of the situation, which in turn is rooted in the leader's appraisal of him or herself. As noted in Chapters 1 and 2, core self-evaluations theory has its roots in appraisal theory (e.g., Arnold, 1960; Lazarus, 1991; Locke, 1969, 1976; Packer, 1985/1986), which suggests that appraisals of specific situations are affected by deeper and more fundamental evaluations of oneself (Judge et al., 1997). Judge et al. (1998, p. 30-31) suggested that individuals "who consider themselves worthy and able to cope with life's exigencies bring a 'positive frame' to the events and situations they encounter, whereas people who do not see themselves as worthy and able bring a negative frame to the same situations." By this logic, leaders of newly formed teams who face a complex task will be likely to perceive the situation at least partly through the lens of their own core self-evaluations. Another logical link between the team leader's core self-evaluations and his/her efficacy may be through generalized self-efficacy. Generalized self-efficacy, as noted in Chapter 2, is the belief an individual has in his/her ability to successfully manage life and the challenges he/she encounters; it is one of the four core traits included in the core selfevaluations concept. As Bandura (1997) noted, judgments of task-based efficacy are likely to be related to more general assessments of efficacy. Therefore, it is likely that more positive core self-evaluations will be related to more positive assessments of taskbased efficacy. In addition, there is some empirical evidence to support the existence of a relationship between a leader's core self-evaluations and his/her leadership efficacy.
37
Recent studies have suggested that managers with more positive core self-evaluations may perform better when facing challenges, which may arise out of a heightened sense of efficacy. Both Judge, Thoresen, and Pucik (1999) and Wanberg and Banas (1997) found that managers with a more positive view of themselves were more likely to cope effectively with changes induced by organizational transformations. In the case of Judge et al. (1999), coping with change mediated the relationship between positive self-concept and job performance, suggesting that core self-evaluations motivate leaders to take task relevant actions. Another stream of literature has examined the relationships between the core indicator traits and self-efficacy, providing further empirical evidence linking the concepts. Chen, Gully, Whiteman, and Kilcullen (2000) found support for a relationship between generalized self-efficacy and task-oriented self-efficacy. Therefore, leadership efficacy may arise out of the team leader's core selfevaluations (Judge et al., 1997). A leader who is more likely to perceive him or herself in a positive manner may be more likely to put a positive frame on the challenging task (Judge et al., 1998), thus leading to higher levels of self-efficacy (Bandura, 1997). A leader's efficacy is not simply an outgrowth of more global personality factors, however. Some of the leader's estimation of his/her efficacy as a leader is most likely related to the feedback that the team has received. As noted above, both enactive mastery experience and verbal persuasion (both of which can be considered as forms of feedback) are also important antecedents of self-efficacy (Bandura, 1997). Teams that have reached critical midpoints have typically received some type of performance feedback, be it associated with the objective quality of task completion or verbal feedback from an external manager. To the extent that the leader identifies with his/her team and feels
38
responsible for the successes and failures of the team during the first half of task completion, the leader is likely to view these successes and failures as reflections of the effectiveness of his/her leadership. Moreover, any feedback that is received at the critical midpoint is likely to be particularly linked to a leader's assessment of his/her effectiveness. Often verbal feedback received at the midpoint is specifically directed toward team leaders, as they may be the liaison with external managers. Additionally, as Gersick (1988) noted, feedback received at the midpoint may be a particularly salient indication of whether the team's trajectory is on track. Teams—and their leaders—tend to be in an evaluative mode during this phase of their existence and, as such, may be more influenced by the feedback they receive. Leaders of teams that receive very positive feedback at the midpoint, then, should experience both reinforcement and an increase in their existing levels of selfefficacy, while leaders of teams that receive more negative feedback at the midpoint should experience a decrease in their existing levels of efficacy. To conclude, a team leader's efficacy shortly after his/her team receives midpoint feedback on a complex task is likely to be related to two main antecedents: his/her own core self-evaluations and the feedback the team has received on its performance. Hypothesis 1: Team leader core self-evaluations and positive midpoint team feedback are positively related to leadership efficacy.
Leader Efficacy
Transformational Leadership
Existing theory provides support for the idea that the team leader's efficacy may be related to behaviors indicative of transformational leadership. Though there is little (if
39
any) research on the link between leadership efficacy and transformational leadership, support for this relationship follows from social cognitive theory. Research has shown that high levels of efficacy seem to be associated with higher levels of performance on all types of tasks in many different realms (cf., Bandura, 1997). Part of this effect is associated with the influence that self-efficacy has on personal choice. Individuals who feel highly efficacious regarding a particular task will be more likely to choose to perform that particular task, set high performance goals, and in turn exhibit higher performance (Bandura, 1997; Locke & Latham, 1990). Transformational leadership is traditionally associated with challenging the status quo and instilling confidence in followers that they can achieve higher levels of performance (Eden, 1990; House & Howell, 1992; Kirkpatrick & Locke, 1991). It follows that the leader's own efficacy may be an important antecedent of transformational leadership, as individuals with low levels of efficacy are not likely to take the initiative in challenging situations and persuade others to do the same. More support for this relationship can be found in examining Bandura's (1997) discussion of the important role of efficacy in social processes. He wrote: ?those who have a firm belief in their efficacy, through ingenuity and perseverance, figure out ways to exercise some measure of control over social systems containing limited opportunities and many constraints. Given a social environment with surmountable barriers, people who have a high sense of efficacy will be able to exercise more control over it, and will view it as more changeable, than will self-doubters who give up in the face of difficulty. (p. 483)
40
Though Bandura was not writing in reference to transformational leadership specifically, certain elements of the statement above point to a potential connection between leadership efficacy and transformational leadership. At its core, leadership is a social influence process (e.g., Yukl & Van Fleet, 1992), and transformational leadership has been identified as being particularly effective in times of crisis and change in organizations. Therefore, though there has been little literature linking self-efficacy to transformational leadership, one can draw the conclusion that higher levels of efficacy may lead to higher levels of leadership performance. Therefore, though there is no existing literature to my knowledge that draws the specific connection between leadership efficacy and transformational leadership, I hypothesize that there should be a strong, positive relationship between the two. Hypothesis 2a: The team leader's efficacy is positively related to the extent to which the leader exhibits transformational leadership.
The Mediating Role of Leader Efficacy Though research on transformational leadership has considered personality traits closely related to core self-evaluations—such as self-confidence—as antecedents of transformational behaviors (e.g., Bass, 1990; House, 1977) and research on trait theories of leadership share conceptual links with research on transformational leadership (e.g., Lord, De Vader, & Alliger, 1986), more process-oriented models of how traits and behaviors are linked are rare in either body of work. Judge, Bono, Ilies, and Gerhardt (2002) noted that trait theories of leadership would greatly benefit from "process models that illuminate the dispositional source of leadership" (p. 36). Judge and Colbert (2002)
41
also noted a "lack of research on the processes by which personality traits affect leadership" (p.4). Though Judge and Colbert (2002) focused primarily on the mediating role of process behaviors such as idea generation, integration of others' ideas, amount of talking, and expressive communication style, a leader's efficacy at leading the group task in question may also be an important mediator between his/her traits and the expression of transformational leadership. Hypothesis 2b: The team leader's efficacy mediates the relationship between core self-evaluations and the extent to which the leader exhibits transformational leadership.
Transformational Leadership
Team Efficacy
As noted in Chapter 2, team efficacy can be defined as a "group's shared ability in its conjoint capabilities to organize and execute the courses of action required to produce given levels of attainment" (Bandura, 1997, p. 477). Though Chen and Bliese (2002) and Pescosolido (2001) have both noted the lack of research on the antecedents of team efficacy, in previous research leadership has been cited as a potential source of team efficacy. Zaccaro, Blair, Peterson, and Zazanis (1995, p. 315), for example, wrote that leadership functions "are directed at fusing a capable team from disparate individuals, and? building perceptions among individual members of their combined and collective abilities." In addition, Chen and Bliese (2002), Dusig (2000), and Sosik, Avolio, and Kahai (1997) found that leadership is, in fact, an important antecedent of team efficacy. As Lindsley et al. (1995) noted, team efficacy typically arises from common exposures of members to objective stimuli (such as the objective constraints of the situation) and the processes of social influence and comparison. It is this latter category that leader
42
behaviors come into play, as leaders play a special part in the social influence process in groups. By definition, leaders are individuals who exert influence over other group members, in addition to exerting influence over task objectives and strategies (Yukl & Van Fleet, 1992). Transformational leadership in particular may be an especially relevant source of team efficacy. Transformational leadership theorists have highlighted how transformational leaders develop, intellectually stimulate, and inspire followers to achieve their purpose, mission, or vision (Bass, 1985; Howell & Avolio, 1992). One of the ways in which these activities are accomplished is through coaching followers, which often involves verbal persuasion to convince them that they are capable of accomplishing goals. Indeed, Pescosolido (2001) suggested that verbal persuasion (an activity often associated with transformational and charismatic leaders) may be the key way leaders with high efficacy may raise team efficacy. Empowering leadership (often considered to be part of transformational leadership—Bass, 1985) may also be related to higher levels of team efficacy, as Kumpfer, Turner, Hopkins, and Librett (1993) found. Some research has begun to examine the connection between transformational leadership and team potency, a term closely related to team efficacy. Though there is a difference between the two terms (potency is more a general sense of confidence that the group can accomplish its goals, while efficacy is more related to the specific task at hand; Guzzo, Yost, Campbell, & Shea, 1993), the findings with respect to potency can easily be applied in an efficacy context. Guzzo et al. (1993) argued that transformational leadership directly influences group potency by boosting the confidence of team members and developing their belief in the ability of the team to succeed. In addition,
43
Sivasubramaniam, Murry, Avolio, and Jung (2002) suggested that transformational leadership ?may enhance the potency of groups or teams by making participation in a group's efforts more meaningful and tied to the collective identity of the group (Shamir, House, & Arthur, 1993)?to the degree that transformational leadership builds personal identification with a group, along with a sense of confidence, a group's level of potency and performance is expected to be higher. (p. 72)
In an empirical test, Sivasubramaniam et al. (2002) indeed found transformational leadership to be significantly related to group potency. Therefore, there is a fair amount of evidence in support of the idea that transformational leadership would assist in the development of team efficacy and that it would be significantly and positively related to the concept. Hypothesis 3a: Transformational leadership is positively related to team efficacy.
The Mediating Role of Transformational Leadership Although Pescosolido (2001) found that leadership efficacy directly influenced team efficacy, the potential mediating effects of transformational leadership were not considered. While there is good reason to believe that a leader's efficacy would be positively related to team efficacy, leaders are likely to convey their efficacy to their fellow team members through their behaviors. To the extent that these behaviors help motivate followers and are indicative of the confidence the leader has, they may be the
44
way in which the level of a leader's efficacy is made clear to the rest of the team—which then is likely to result in higher levels of team efficacy. Hypothesis 3b: Transformational leadership mediates the relationship between team leader efficacy and team efficacy.
Team Efficacy
Team Goals
The above few sections have dealt with a chain of antecedents leading to team efficacy, beginning with the team leader's core self-evaluations. In the next few sections, I discuss how higher levels of team efficacy may translate into performance. First, I discuss the influence team efficacy may have on goals; I will discuss other team processes that may be influenced by team efficacy in the following section. I will close the chapter with a discussion of hypothesized connections between team performance, team goals, and team processes. A number of research teams have found team efficacy to be related to one major team process: goal-setting (e.g., Durham, Knight, & Locke, 1997; Silver & Bufiano, 1996; Weldon & Weingart, 1993). Gist (1987) argued that group perceptions of team efficacy should be related to group performance, and Prussia and Kinicki (1996) and Whitney (1994) both found support for this relationship. Scholars have found this efficacy-performance relationship to be mediated by goals, both at the individual and group levels (Mulvey & Klein, 1998; Prussia and Kinicki, 1996). Additionally, there is good theoretical reason to believe that team efficacy is related to team goal-setting; both Bandura's social cognitive theory (1986, 1998) and Locke and Latham's (1991) goalsetting theory support this connection. Bandura (1982) argued that "perceived collective
45
efficacy will influence what people choose to do as a group, how much effort they put into it, and their staying power when the group efforts fail to produce results" (p. 143). As Mulvey and Klein (1998) noted, the logical conclusion of this statement is that team efficacy should influence team goals and the team's commitment to those goals. Therefore, Hypothesis 4a: Team efficacy is positively related to team goals.
The Mediating Role of Team Efficacy in the Transformational Leadership Relationship
Team Goals
Although a critical characteristic of the transformational leader is having a vision and effectively communicating that vision to his/her followers, directive leadership (wherein leaders are involved on a more micro level with their employees) is usually the type of leadership that is associated with the setting of specific team goals (Yukl, 1998). In the case of transformational leaders, it is more likely that team members be inspired by the vision, motivation, and ideas put forth by the transformational leader, and as such take more of an active role in setting their team's own immediate performance goals. Because team members are the individuals setting those team goals, it is likely that goal levels will be dependent on team efficacy, and that transformational leadership will only indirectly influence team goals through influencing team efficacy. Hypothesis 4b. The relationship between transformational leadership and team goals is mediated by team efficacy.
46
Team Efficacy Team Processes As reviewed in Chapter 2, there is an emerging literature regarding both the antecedents and consequences of team efficacy. I postulate here that higher levels of team efficacy will be positively related to both action and transition team processes (Marks et al., 2001). Theoretically, as members feel more confident in their team's capabilities, "they are more motivated to work hard for the team, persist in the face of? obstacles, and are willing to accept more difficult challenges" (Zaccaro et al., 2001, p. 467). Despite the potential to interweave literature on team efficacy and group processes that Zaccaro and colleagues' words suggest, Marks (1999) noted that the majority of the literature on team efficacy has focused on establishing links to performance, rather than considering team efficacy as a potential determinant of team processes. Connections between processes other than goal-setting that a team might use to productively approach task work (e.g., information collection, strategy formulation, etc.) are notably absent in the literature linking team efficacy to performance. Marks (1999) did find that highly efficacious teams seemed to change their strategies more than less efficacious teams when faced with a challenging situation. Other team processes, however, may also be important to consider. Marks et al. (2001) suggested that teams progress through transition phases, wherein "teams focus primarily on evaluation and/or planning activities to guide their accomplishment of a team goal or objective" (p. 364). During this time, processes such as mission analysis, goal specification, and strategy formulation and planning typically occur. Marks et al. (2001) categorize a second important set of processes as action processes, which consist of activities leading directly to team goal accomplishment. These activities typically include monitoring progress
47
toward goals, systems monitoring, team monitoring and backup responses, and coordination activities. Though no study to my knowledge has ever directly examined the relationship between team efficacy and action or transition processes, Edmondson (1999) found that team efficacy was related to team learning behavior, which consisted of processes such as feedback seeking, experimentation, and discussion of errors. It may be that teams with higher levels of efficacy are more likely to scrutinize their own actions in an attempt to improve in the future. Similarly, teams with higher levels of efficacy may be more likely to formulate strategy and plan for the future; additionally, they may be more likely to carefully examine their progress towards their goals and monitor themselves. Given the empirical evidence that does exist on the relationship between team efficacy and team processes, and given the logical reasoning that teams that are more confident in their own ability to achieve certain performance levels will be more likely to take action to achieve those performance levels, it is likely that team efficacy is positively related to important team processes. Therefore, Hypothesis 5a: Team efficacy is positively related to team action and transition processes that are geared towards helping the team reach its goals.
The Mediating Role of Team Efficacy in the Transformational Leadership Processes Relationship
Team
Transformational leaders motivate team members to be inspired through effectively presenting a vision for the future; however, as noted above, transformational leaders are typically not directly involved on a micro level enforcing specific goal-setting
48
or micro-managing exactly what processes teams utilize. Moreover, transformational leaders inspire team members to identify with and care more about the team itself; transformational leaders also typically build the team's confidence through verbal persuasion and encouragement. As team members become more engaged in the activities of the team and as their team efficacy reaches new heights due to the efforts of the transformational leader, it is likely that teams will exhibit subsequent upswings in the extent to which constructive team action and transition processes are utilized. Therefore, transformational leadership is likely to indirectly influence team action and transition processes through influencing team efficacy. Hypothesis 5b. The relationship between transformational leadership and team action and transition processes is mediated by team efficacy.
Team Goals and Team Action and Transition Processes
Performance
As Durham, Knight, and Locke (1997) noted, goal-setting is a well-established motivational technique for individuals and, to a certain extent, for groups as well (Locke & Latham, 1990; O'Leary-Kelley et al., 1994; Weldon & Weingert, 1993). Both individuals and teams that set higher goals exhibit higher levels of performance attainment; additionally, goals have been found to mediate the relationship between efficacy and performance, both at the individual and group level (Durham et al., 1997; Locke & Latham, 1990). Therefore, it is likely that higher team goals will be related to higher levels of team performance. Action and transition processes may also influence performance, though there is no direct empirical evidence to date to support this contention. However, some of the
49
literature on goal-setting has alluded to the importance of other team processes in predicting performance. Durham et al. (1997, p. 206), for example, wrote that the effects of goal-setting are "most reliable when the goals are specific and difficult, when there is commitment to the goals, when feedback is available regarding the goal process, high self-efficacy, and knowledge regarding how the goals can be achieved." According to Durham et al. (1997), this knowledge really consists of the knowledge of the strategies needed to perform the task so as to attain the goal, which supports the idea that effective strategy formulation (a transition process) is a critical group process to examine. For newly formed teams that are operating in highly turbulent and unfamiliar environments, or on an unknown and complex task, the importance of developing and using a strategy to cope effectively with the demands of the situation given the team's constraints is doubly critical. The team processes of information collection, analysis, strategy development, and strategy implementation should all also be positively related to objective team performance. It is necessary for teams to have knowledge of critical information in order to be effective; to the extent that teams make an effort to collect this information, the more positive the resulting performance should be. Similarly, teams that actively analyze, develop strategies for approaching the task, and implement those strategies accordingly should experience more positive performance than teams that do not actively engage in these processes. It is theoretically possible, of course, for a group that does not exhibit any of the above processes to do well; however, it is highly unlikely, as work teams that do not engage in these types of activities on complex tasks would be relying on a great
50
deal of luck. These processes, then, are hypothesized to have a positive impact on team performance. Hypothesis 6a: Team goals and team action and transition processes are positively related to team performance.
The Mediating Roles of Team Goals and Team Process in the Team Efficacy Performance Relationship The relationship between team efficacy and objective team performance has been solidly established by a number of authors, though this relationship is not quite as strong as the relationship between individual-level self-efficacy and individual performance (Collins & Parker, 2002). Gist and Mitchell (1992), Peterson, Mitchell, Thompson, and Burr (2000), Silver and Bufiano (1996), and Zander and Medow (1964), among others, have all found evidence that suggests that higher levels of group efficacy are related to higher levels of performance. In addition, Bandura (1997) and Lindsley et al. (1995) suggested that efficacy may exhibit a direct relationship with performance. The team efficacy-performance relationship, however, may be partly an indirect one. It is likely that team efficacy impacts performance partially through its influence on team goals, as goals provide an immediate focal point on which to concentrate motivation (Locke & Latham, 1990). It is also likely that team efficacy impacts performance partially through its influence on team action and transition processes. Teams that are more confident in their ability to successfully complete the task should be more likely to focus effort on how to do it; they should be more involved in analyzing feedback results, seeking further information, and adjusting team strategy, which are all critical processes
51
necessary to achieve high levels performance when dealing with complex tasks in dynamic environments. Hypothesis 6b: The relationship between team efficacy and team performance is mediated by team goals and team action and transition processes.
Conclusion In this chapter, I presented a model that suggested that a team leader's core selfevaluations and midpoint feedback the team receives influence team efficacy, goals, processes, and performance through influencing leadership efficacy and transformational leadership. Team leader core self evaluations and midpoint feedback the team receives were hypothesized to be related to leader efficacy, which in turn was hypothesized to be related to transformational leadership. Transformational leadership, in turn, was hypothesized to be an antecedent of team efficacy. Finally, team efficacy was proposed to be an antecedent of both team action and transition processes and team goal level; the latter two variables were predicted to have a direct impact on team performance. This general conceptualization of the effects of leader core self-evaluations is consistent with the findings suggesting that core self-evaluations may have an impact on job performance through their impact on individual motivation. In Chapter 4, I discuss the research methods, including sample, experimental task, procedure, and aggregation and analytic strategy.
52
Chapter 4 Research Methods
This chapter presents data collection and analysis methods for this research. In the first section, I describe research design, including a description of the task; in the second section, I describe the sample, in addition to discussing the schedule of the collection of measures; and in the third section, I describe analytic procedures.
Research Design In order to test the hypotheses proposed in Chapter 3, a number of methodological issues had to be addressed. It was important for the study to feature a task as its centerpiece that was complex, engaging, provided objective performance assessments, and involved both ongoing and critical midpoint feedback. Additionally, the research required access to an appropriate sample of individuals whose results would be generalizeable. This study was developed with these design issues in mind. As a broad overview of when and how the data were collected, team members' core self-evaluations were assessed prior to the start of the simulation. Objective performance data was provided by the task itself, which was a business simulation that translated teams' decisions into performance. Survey data regarding leader efficacy, transformational leadership, team efficacy, team goals, and team action and transition processes were gathered from participants immediately following the critical midpoint of the task. Examining survey data collected at this point in time was particularly useful for several reasons. First, collecting data after teams had worked together during the first half
53
of task completion allowed for development of team level constructs through shared history of interaction and task feedback. Second, central concepts and relationships in the model hypothesized in Chapter 3 involved the response to critical midpoint feedback; it was therefore important to assess variables of interest after critical midpoint feedback had been received. Third, due to the fact that leaders were emergent rather than formally designated, it was important to allow team members time to gravitate toward roles in which they felt comfortable; assessing variables after the critical midpoint allowed team members the entire first half of task completion to do exactly that. Also, for the purposes of this study, individuals were placed on teams such that team composition in terms of gender and race/ethnicity was balanced across groups (demographic information had been provided to the individual who split the participants up into teams; the project director ensured that there were at least two women and two racial or ethnic minorities—but not more than three of either—in each group of five participations). Core self-evaluations, however, were allowed to randomly vary. Individuals were required to participate as part of the University of Maryland's MBA program's orientation activities for first-year students. Additionally, participants were given one incentive for participating: a summary of the findings tailored to their specific team and implications of those findings. Description of the task. MARKETPLACE (Cadotte, 2003), the task that was used in this study, is a complex, computer-based interactive team decision-making business simulation. The simulation is based on the history of the personal computer industry. During the course of the simulation, participating teams worked to build an entrepreneurial firm, experiment with strategies, make decisions regarding the major
54
business functions (e.g., marketing, manufacturing, logistics, human resources, finance, and accounting), and compete with other teams in an extremely realistic and detailed virtual business world. The simulation required each participant to imagine being a member of the top management team of a start-up computer firm. In their role as top management team members, participants made a series of strategic decisions over a period of eight "quarters" (each quarter representing a single decision cycle). Teams received performance feedback at the end of each quarter following Quarter 3 (the first quarter that teams actually sold products). Teams were pitted against each other in four "universes" of 10 teams each; teams within universes were directly competing for the same customers, though each team could choose to capture any number of five market segments of business users. These five segments ranged from the "Cost Cutter" segment (a large segment that looked for very easy-to-use computers for basic office applications) to the "Mercedes" segment (the smallest segment overall that looked for a high-performance computer to use in sophisticated engineering and manufacturing applications). Each quarter of play in MARKETPLACE involved new information and a set of decisions that built upon each other. The simulation unfolded to follow the normal lifecycle of a start-up business, though there was a natural break-point halfway through the simulation where more performance feedback and extra funding from "venture capitalists" was received. Quarters 1 and 2 essentially involved "setting up shop." In Quarter 1, team members got organized, assigned corporate responsibilities, named the company, and purchased market research. Team members also sold stock to themselves
55
during this period as owners in order to raise capital for initial operation. In Quarter 2, market research arrived, and teams analyzed the data and decided on an overall business strategy, including corporate and performance goals, target markets, and strategic direction. Several tactical decisions also had to be made during this period, which focused on brand design, plant location, production capacity, and sales office location. Teams finished Quarter 2 by answering the first survey I distributed, which included team efficacy and team goal measures (used in my analysis to derive an ability measure as a control; specifics are discussed in the below section regarding the measures). Quarter 3 involved testing the market, as fledgling firms tried out marketing strategies (including brands, prices, ad copy, media campaigns, and sales staffing). In addition, teams decided on hiring policies and production processes for the quarter, then scheduling production for the quarter. Once again, teams purchased market research in order to discover customer reactions to marketing decisions and to find out what the competition has done. Lastly, in this quarter, teams forecasted market demand and simulated their production operations given the supply chain decisions teams had made prior to that point. At the end of this quarter, MARKETPLACE made available cash flow and income statements and balance sheets in order for teams to evaluate the potential financial impact of the firm's first quarter of total business operation. In Quarter 4, team members received the market data they purchased in Quarter 3 regarding the test market and received manufacturing and accounting data from the first quarter of sales. Once again, at the end of this quarter, MARKETPLACE made available cash flow and income statements and balance sheets in order for teams to evaluate their
56
performance. Teams could then adjust the firm's strategies and tactics if necessary after examining this data. Quarter 5 was a pivotal quarter—the halfway point—with the critical midpoint feedback occurring here. In this quarter, teams had to obtain capital from venture capitalists in order to substantially expand their position in the market (which would occur by investing in research and development, new sales outlets, factory capacity, logistics, and/or employee recruitment and retention programs). Prior to making their pitch to the venture capitalists, teams received a review of marketing, human resource, manufacturing, and accounting data from Quarter 4. Then they formulated a one-year business plan to obtain outside funding. The plan had to include an assessment of market opportunities, a review of performance to date, and a strategy for the second year of business operations. It had also to include coordinated marketing, manufacturing, human resources, and financial strategies, in addition to a tactical plan that included the sequencing and timing of events to achieve the firm's goals. Finally, the plan had to include historical financial statements from the first four quarters (year 1) in addition to pro-forma statements for the second four quarters (year 2). The simulation was designed such that at this point teams provided this information as a formal presentation during the middle of the day allocated for Quarter 5 play to a panel of venture capitalists (volunteers who were chosen from the Robert H. Smith School of Business community of alumni, executives, and faculty), who on the basis of what the teams presented offered them advice and some amount of "virtual" funding. Each team presented to one panel of venture capitalists; there were eight panels of seven venture capitalists, each of which viewed five team presentations (each set of
57
five teams was in the same universe). After each team completed its presentation, venture capitalists completed a 6-item questionnaire that reflected their impressions of each team's potential and gave these questionnaire responses to the teams. Teams then met with the venture capitalists and had a chance to receive additional funding. All the venture capitalists had the same amount to offer in terms of funding; they were making their investment decisions based on the teams that they believed were in the best position to perform well during the remainder of the simulation. Although teams and venture capitalists negotiated over the funding level and price of stock, the virtual money each team could receive from the venture capitalist panel was capped at five million dollars; the funding cap ensured that the simulation could continue into the second year of play without completely eliminating any teams. It is important to note here that the amount of capital that teams received from the venture capitalists was not a pure measure of whether the venture capitalists were willing to invest money in any given team. Teams approached their negotiations with the venture capitalists with different strategies in mind. Some teams were, indeed, interested in obtaining as much capital as possible, no matter what their stock price. Other teams were more cautious about diversifying ownership; rather than sell a lot of stock, these teams were interested in driving a hard bargain for the price of their stock. Finally, some teams were simply not as interested in selling their stock to receive funding from the venture capitalists, no matter how high their stock price. Due to the many strategies utilized by different teams, the amount of capital received by each team was, again, not a pure measure of the interest venture capitalists had in investing in that team's stock.
58
Receiving feedback and capital from the venture capitalists was considered to be the critical midpoint of the simulation. Teams had received not just objective performance results from Quarters 1-4, but also outside assessments of past performance and expected future potential. The situation provided an opportunity for teams to reflect upon whether they were on track, based on previous performance trajectory, to meet their performance goals that had been set in Quarter 2. Teams finished Quarter 5 by allocating the newly procured funding to different investment opportunities of their choice (e.g. research and development, new sales outlets, etc.) and then responding to the second survey for the purposes of this research, which included leader efficacy, transformational leadership, team efficacy, team goal, and team action and transition processes. Quarters 6-8 were a time in which the team continuously monitored and adjusted different areas of the firm (such as marketing, manufacturing, etc.). Competition also heightened during this period, as other firms began introducing new technology, more reliable products, better prices, etc. Teams during this time also had the opportunity to explore new relationships through cross-licensing and supply outsourcing. At the end of Quarter 8, the team prepared another report to assess strategy and performance in year 2 of operations and to give an assessment of how well the firm was prepared to take on the challenges of the future. Table 2 provides a summary overview of the sequence of events in the simulation, in addition to describing what variables were collected when and what were their uses. As noted above, I considered a number of parameters when choosing the appropriate research design for this study. MARKETPLACE, as the centerpiece task, provided an excellent context in which to test my theoretical model. The simulation was
59
Table 2. Sequence of Events, Measures Collected, and Measure Uses in MARKETPLACE Simulation Quarter 1 Team Activities Measures Collected Before Quarter 1: Core SelfEvaluations (collected pre-ELM) Independent variable Measure Use
• • • •
•
2
team gets organized, assigns corporate responsibilities, sets broad goals team names the company team purchases initial market research team members sell stock to themselves during this period as owners in order to raise capital for initial operation no performance feedback yet available market research arrives team engages in data analysis team makes decisions regarding overall business strategy, including corporate goals, target markets, and strategic direction team makes tactical decisions: brand design, plant location, production capacity, and sales office location no performance feedback yet available
• • •
End of Quarter 2: Team Efficacy Used to create ability residual Used to create ability residual
Team Goal
•
•
3
•
team tests market: tries its marketing strategy (including brands, prices, ad copy, media campaigns, and sales staffing) • team decides on hiring policies and production processes for the quarter, 60
End of Quarter 3: Team Performance (profit) Used to create ability residual— and provided to teams as feedback
and then schedules production for the quarter • team purchases market research in order to discover customer reactions to marketing decisions and to find out what the competition has done • team forecasts market demand and simulates its production operation given the supply chain decisions • cash-flow and income statements available at end of quarter, in addition to a balance sheet to help team members evaluate potential financial impact of the firm's first quarter of total business operation 4
• • •
team receives market data purchased in Quarter 3 regarding the test market team receives manufacturing and accounting data from the first quarter of sales team uses data to evaluate performance and adjust the firm's strategies and tactics if necessary
End of Quarter 4: Team Performance (profit) Used to create ability residual— and provided to teams as feedback
5
•
team receives review of marketing, human resource, manufacturing, and accounting data from Quarter 4 • team must formulate a oneyear business plan to obtain outside funding. Plan includes: assessment of market opportunities, review of performance to date, strategy for the Year 2 (coordinated marketing,
End of Quarter 5: VC Rating Independent variable Independent variable
Leader Efficacy
Transformational Independent Leadership variable—and used to determine leader
61
• • • •
manufacturing, human resources, and financial strategies), tactical plan (with sequencing and timing of events to achieve the goals), historical financial statements from year 1 in addition to proforma statements for year 2 team physically presents this information to venture capitalists VC panel offers team advice and some amount of funding team may negotiate with the panel team allocates funding to different investment opportunities (e.g. research and development, new sales outlets, etc.) team monitors and adjusts different areas of the firm (such as marketing, manufacturing, etc.) competition heightens during this period as other firms begin introducing new technology, more reliable products, better prices, etc. team has the opportunity to explore new relationships through cross-licensing and supply outsourcing
Team Efficacy
Independent variable
Team Goal
Independent variable Independent variable
Team Processes
6-8
• •
End of Quarter 8: Team Performance (profit) Dependent variable
•
62
meant to introduce teams to a constantly changing environment to which they had to adapt in order to survive. Also, the critical midpoint of the simulation was very clearly defined. Variance in performance at the end of the simulation was assured due to the fact that teams were directly competing against each other. Indeed, some teams finished with extremely high performance ratings, while others drifted into the final quarter bankrupt (though they did not have to stop play).
Sample Dr. Joyce Russell, coordinator of the Smith School's full-time MBA program Educational Learning Module (ELM)1 project, granted me access to the 2002 ELM. The ELM took place during the last week of August 2002; all first year full-time MBA's were included in the project. As such, 198 individual participants competed in 38 teams of five members and two teams of four members. The sample was appropriate for a number of reasons. First, teams were made up of first-year MBA's who devoted a full, uninterrupted week to the ELM project, playing two quarters each day. Teams worked full days in the hopes of gaining an edge over the competition. In addition, teams were divided so as to ensure there were at least two women and two racial/ethnic minorities (but not more than three of either), including foreign students, on each team; it was likely that differences in performance of these teams would be a result of leadership or team process factors rather than as a result of any demographic differences.
1
The ELM is designed to be the first activity first-year MBA students complete when they arrive on campus; it is intended to give them both an opportunity to receive some hands-on business decision-making experience and to gain a valuable experience in team interaction. MARKETPLACE is the simulation program used in the ELM.
63
Measures Dependent Variable Team Performance. Table 3 provides a summary of concepts and measures used in this study. Teams received financial information regarding their operating profit at the end of every quarter following Quarter 3. Operating profit was the most critical and salient measure of whether a team's strategies and decisions were effective; it was generated by the simulation after team decisions had been inputted. Therefore, Quarter 8 operating profit (the profit with which teams ended the simulation) was used as the basis of the team performance measure. However, in order to control for team ability, a standardized residual score was created from a regression that considered Quarter 8 performance as a dependent variable and the team ability control variable, described below, as the independent variable; it was this standardized residual score that was the actual dependent variable used in all the analyses.2 More information regarding the mean and standard deviation of this variable, in addition to zero-order correlations with other study variables, can be found in Chapter 5. I also examined Quarter 8 rank (a number between 1 and 10 reflecting each team's relative final position, based on operating profit, with respect to the other teams in the same universe) as a measure of final performance, using a similar procedure to that which is described above to control for ability. Due to the strong similarity of the results attained for each measure of performance, however, I will focus exclusively on Quarter 8 profit as the team performance measure in Chapter 5.
2
All analyses were also run with a separate ability control measure and the Quarter 8 profit measure as the dependent variable; these analyses yielded virtually identical results. For purposes of parsimony, I only report the use of and results associated with the standardized residual final performance measure as the dependent variable.
64
Table 3. List of Concepts and Measures Used
Concept Team Performance
Measures Standardized residual of regression with team profit in Quarter 8 as dependent variable and ability (i.e., standardized residual of first half performance regressed on team efficacy and goal assessed after Quarter 2) as the independent variable I am confident I get the success I deserve in life. Sometimes I feel depressed. (r) When I try, I generally succeed. Sometimes when I fail I feel worthless. (r) I complete tasks successfully. Sometimes, I do not feel in control of my life. (r) Overall, I am satisfied with myself. I am filled with doubts about my competence. (r) I determine what will happen in my life. I do not feel in control of my success in my life. (r) I am capable of coping with most of my problems. There are times when things look pretty bleak and hopeless to me. (r) This team acts like a unified team. This team seems to have strong leadership. I might invest some money in this team. This team seems to have a clear strategy for the future. This team was responsive in answering questions. This team's presentation was effective. During Quarter 5 of the simulation, I had a high degree of confidence in my ability to? ?steer this team in a successful direction. ?get the team to develop viable strategies. ?inspire others to be motivated to do well. ?build this team's sense of spirit and cohesiveness. ?get the people on this team to be excited about working together.
Core Self-Evaluations
Venture Capitalist Feedback
Leader Efficacy
Transformational Leadership
Leadership behavior ratings matrix—please see Figure 4
65
Team Efficacy
Please take a moment to discuss and agree upon what your team's confidence levels are. What is your team's confidence level (ranging from 0% to 100%) in being able to attain the following performance rankings in your universe by the end of the game? Top 9 or better: _______ Top 8 or better: _______ Top 7 or better: _______ Top 6 or better: _______ Top 5 or better: _______ Top 4 or better: _______ Top 3 or better: _______ Top 2 or better: _______ Top 1 or better: _______
Team Goal
Please take a moment to discuss and agree upon what your team's goal is in terms of end ranking within your universe. At the conclusion of this simulation, our team's goal is to be ranked ______ out of the 10 teams in our universe (1 = top performing team; 10 = lowest performing team).
Team Processes
During Quarter 5 of the simulation, the members of my team? ?set specific goals for our team to accomplish. ?collected and analyzed all information we could before making decisions. ?monitored changing environmental conditions. ?developed and implemented a comprehensive strategy. ?adjusted our strategy in keeping with what we had learned. ?coordinated activities among team members. Standardized residual of regression with first half (Quarters 3 and 4) profit as dependent variable and team efficacy and team goal in Quarter 2 as independent variables
Team Ability
66
Control Team Ability. Performance prior to the critical midpoint— an average of operating profit information provided by the simulation in Quarters 3 and 4—was used as the basis for the measure of team ability. In order to control for motivational variables that may have been affecting performance, I conducted a regression analysis, with profit prior to the critical midpoint (an average of Quarter 3 and Quarter 4 profits) as the dependent variable and Quarter 2 team efficacy and team goals as the independent variables3. Because the residuals from these regressions represented observed team performance early in the simulation, controlled for team efficacy and goals, they served as measures of actual team ability. These residuals were used as independent variables in the regression to create the Quarter 8 performance residuals—the standardized final performance residuals—to control for team ability.
Independent Variables Venture Capitalist Feedback. As noted in the description of the critical midpoint transition (Quarter 5), venture capitalists completed a six-item, Likert-type assessment of each team immediately after viewing each team's presentation of their past performance and future strategies and goals. Sample items were as follows (1 = strongly disagree; 5 = strongly agree): "This team acts like a unified team"; "I might invest some money in this team"; and "This team seems to have a clear strategy for the future." The inter-item reliability for the six-item measure as assessed by Cronbach's alpha was .82; the withingroup agreement as assessed by rwg (James, Demaree, & Wolff, 1984; 1993) was .86;
3
In creating the rank-based ability measure, I used rank prior to the critical midpoint, as determined by an average of Quarter 3 and Quarter 4 profit.
67
Lindell's (Lindell & Brandt, 1999) alternative within-group agreement statistic was .83; ICC(1) (Bartko, 1976; James, 1982; McGraw & Wong, 1996) was .45; and ICC(2) (Bartko, 1976; James, 1982) was .79. These agreement and reliability statistics were at levels that allowed me to meaningfully average venture capitalists' individual assessments to reflect an overall venture capitalist assessment for each team. Team Leader Core Self-Evaluations. Core self-evaluations was measured by selfreport using the 12-item Core Self-Evaluations Scale, developed by Judge, Erez, Bono, & Thoreson (2003). Judge et al. (2003) found substantial support for the validity of the scale; it is reliable and displays acceptable levels of internal consistency and test-retest reliability. In addition, inter-source (self-significant other) level of agreement is comparable to that of other personality measures. Sample items are as follows (assessed on a 1-5 Likert scale; 1 = strongly disagree, 5 = strongly agree): "I am confident I get the success I deserve in life"; "Sometimes I feel depressed" (r); "When I try, I generally succeed"; and "Overall, I am satisfied with myself." Cronbach's alpha for the scale was .80, suggesting adequate inter-item reliability. Leaders were emergent rather than formally designated. I first used the transformational leadership scale score (discussed below) collected after the critical midpoint transition to identify which team member was considered to be the team leader by his/her teammates during the critical midpoint period. I then utilized that individual's core self-evaluations score as the team leader's core self-evaluations score. Team Leader Efficacy. I developed a measure for this concept that assessed the extent to which individuals felt confident in assuming different aspects of the team leadership role. Just as was noted for the assessment of the leader's core self-evaluations,
68
I was only interested in the leader efficacy scale scores of the individual who was recognized by his/her teammates as being the emergent leader during the critical midpoint transition period. The 5-item, Likert-type scale was tailored to refer to the level of efficacy experienced by the leader during the critical midpoint. Sample items are as follows (1-5 Likert scale, 1 = strongly disagree, 5 = strongly agree): "During Quarter 5 of the simulation, I had a high degree of confidence in my ability to?" 1) "?steer this team in a successful direction"; 2) "?get the team to develop viable strategies"; 3) "?inspire others on this team to be motivated to do well." The scale demonstrated a high level of inter-item reliability; Cronbach's alpha was .90. Team Leader's Transformational Leadership. All participants completed a measure assessing their perceptions of their teammates' transformational leadership. For each team, I created a matrix (see Figure 1) with space for team member names horizontally across the top and three items listed vertically down the left-hand column. The three items were intended assess transformational leadership as conceptualized by Bass's (1985) four dimensions: idealized influence, inspirational motivation, intellectual stimulation, and individual consideration. Each participant rated all his/her team members on the extent to which they exhibited the transformational leadership behaviors listed. Team members assessed each other following the critical midpoint period (end of Quarter 5); the emergent leader for the team was defined as the individual who received the highest score from his/her teammates. I used that individual's score to serve as the leader's transformational leadership rating, and—as noted above—I used that individual's core self-evaluations and leadership efficacy scores. Team members were
69
Figure 4. Transformational Leadership Ratings Matrix
Directions: Please write each of your fellow team members' names on the diagonal lines across the top of the grid. Then, rate each team member on the extent to which he/she exhibited the behaviors listed below in the first two quarters of the simulation. Do not rate yourself.
1=never; 2=very infrequently; 3=sometimes; 4=often; 5=always Inspired team members by demonstrating confidence in the team and showing optimism. Encouraged other team members to be creative problem solvers. Demonstrated sensitivity to and support of the needs and concerns of other team members.
70
instructed with the following directions to fill out the matrix: "Please write each of your fellow team members' names on the diagonal lines across the top of the grid. Then rate each team member on the extent to which he/she exhibited the behaviors listed below during Quarter 5 of the simulation." Sample items team members used to rate each other are as follows (1-5 Likert, 1 = Never, 5 = Always): "Encouraged others to be creative problem solvers" and "Inspired team members by demonstrating confidence in the team and showing optimism." Cronbach's alpha for this 3-item measure was .89, suggesting a high level of inter-item reliability. Additionally, rwg was .97, Lindell's alternative rwg was .92, ICC(1) was .52, and ICC(2) was .85, suggesting outstanding levels of within-team agreement and inter-rater reliability for this measure that justified the aggregation of this measure across raters. In order to ensure that the transformational leadership measure was stable, I also collected the same measure at the end of the simulation. 70% of the individuals who received the highest transformational leadership scores following the midpoint feedback also received the highest transformational leadership scores at the conclusion of the simulation, suggesting that the measure indeed was stable and was being consistently utilized by raters. Therefore, each individual's transformational leadership score reflected an aggregation of their peer's ratings. Team Efficacy. A scale was created to assess this measure based on the recommendations of Locke, Frederick, Lee, and Bobko (1984) and others (e.g., Gibson, Randel, & Earley, 1999; Gist & Mitchell, 1992; etc.). This measure was tied directly to the team's belief in its capabilities to achieve 10 specific performance rankings. Teams were asked to indicate their confidence (using a 100 point scale, where 0 = certainly cannot be achieved and 100 = certainly can be achieved) as to whether their team will
71
finish in the top 9 or better, top 8 or better, top 7 or better, etc. all the way up to top 1 in their respective 10-team universes, based on their assessments of the team's capabilities. Teams completed this scale by consensus—directions were as follows: "Please take a moment to discuss and agree upon what your team's confidence levels are. Please write your team's assessment of confidence below." The team efficacy score that was used to test hypotheses was assessed at the end of the midpoint transition period (end of Quarter 5). Cronbach's alpha of this scale was .94, suggesting high levels of inter-item agreement. Due to the fact that the scores were captured through group consensus, intra-team agreement and reliability measures could not be calculated. The team efficacy scale was also captured following Quarter 2 of the simulation; that initial efficacy score was used in the calculation of the team's ability measure. Cronbach's alpha for the post-Quarter 5 collection of team efficacy was .94. Cronbach's alpha for team efficacy post-Quarter 2 was .92. Team End Goal. Teams were asked to provide a response to one question regarding their team's end goal after the critical midpoint transition period. The team completed this item by consensus. Directions were as follows: "Please take a moment to discuss and agree upon what your team's goal is in terms of end ranking within your universe." The item itself read as follows: "At the conclusion of this simulation, our team's goal is to be ranked ____ out of the 10 teams within our universe (1 = top performing team; 10 = lowest performing team)." Team end goal was also assessed in the same manner at the end of Quarter 2; this initial goal measure was used as a control to calculate the team ability measure.
72
Team Processes. Team members provided individual responses to questions developed for this dissertation that were intended to reflect Marks et al.'s (2001) action and transition processes, discussed in Chapters 2 and 3. At the end of the critical midpoint transition period (end of Quarter 5), team members assessed the extent to which they believed that their team had engaged in action and transition processes reflected by a six-item, Likert-type scale. Sample items (1 = strongly disagree to 5 = strongly agree) are as follows: "The members of my team?" (1) "?set specific goals for our team to accomplish"; (2) "collected and analyzed all information we could before making decisions"; (3) "developed and implemented a comprehensive strategy"; and "coordinated activities among team members." Cronbach's alpha was .83, suggesting an acceptable level of inter-item agreement; rwg was .98; Lindell's alternative rwg measure was .88; ICC(1) was .26, and ICC(2) was .64. These statistics suggested that the items could be meaningfully aggregated into a score that reflected the team's action and transition processes.
Analytic Strategy I used path analysis through covariance structure analysis in Bentler's (1989) EQS program to test the strength of the relationships among leader core self-evaluations, venture capitalist feedback, leader efficacy, transformational leadership, team efficacy, team processes, team goals, and team performance (hypotheses 1, 2a, 3a, 4a, 5a, and 6a). Data from the 40 groups were input into EQS in the form of a correlation matrix with accompanying standard deviations. Model structural paths were evaluated for significance, and goodness of fit of the overall model was assessed by several fit indices:
73
the comparative fit index (CFI; .90 and above is considered to be indicative of acceptable model fit; Bentler, 1989), the LISREL goodness of fit index (GFI; .90 and above is considered to be indicative of acceptable model fit; Joreskog & Sorbom, 1989), the Bentler-Bonett Normed Fit Index (NFI; .90 and above is considered to be indicative of acceptable model fit; Bentler, 1989), and the Bentler-Bonett Non-Normed Fit Index (NNFI; .90 and above is considered to be indicative of acceptable model fit; Bentler, 1989). Because the model tested 16 parameters with a sample size of 40, which falls short a great deal of the rule-of-thumb requirement that there should be at least 5 cases per parameter tested in a structural model, I also tested the model using traditional path analysis in SPSS regression. This regression revealed virtually identical results, suggesting that the parameter estimates derived from structural equations modeling were stable. Klein, Conn, and Sorra (2001) advocated the use of both regression and covariance structure modeling in their study, which also considered a particularly small sample size (n = 39); they also noted that in cases of such small sample size, fit statistics for the overall model should be interpreted very cautiously. Chi-squared difference tests (James, Mulaik, & Brett, 1982) were conducted to determine the degree to which hypothesized mediators indeed acted as mediators, as proposed by hypotheses 2b, 3b, 4b, 5b, and 6b. I followed Baron and Kenny's (1986) guidelines for testing mediation in this set of analyses in conjunction with a modelcomparison approach. I compared the chi-squared statistic of my hypothesized model with chi-squared statistics of models that included the hypothesized mediated paths in addition to direct paths. If the difference in chi-squared statistics was significant, the
74
model with the direct path was a better fit with the data, indicating either partial mediation or no mediation. The results from these mediation tests also mirrored the results from traditional SPSS regression mediation tests, once again supporting the stability of the model in EQS.
75
Chapter 5 Results
The following report of results begins with the results of the covariance structure analysis of the original model and mediation analyses. I then present a revised version of the model and discuss results of exploratory analyses.
Descriptive Statistics Table 4 presents the descriptive statistics, scale reliabilities, rwgs, and ICCs of the study variables; Table 5 presents the intercorrelations among study variables. In both cases, team goal is reverse scored (a score of 10 indicates the team's goal was to end in first place within its universe). Also, team performance is a standardized variable representing the residual of a regression of ability on Quarter 8 (end) profit.4 Additionally, the results reported here are associated exclusively with the use of final (Quarter 8) profit as the basis for the dependent variable. As noted in the measures section, all analyses were also run with Quarter 8 rank as the dependent variable. Results for each set of analyses were virtually identical; therefore, only results associated with Quarter 8 profit are reported here.
4
The use of this residual, as noted in the previous chapter, was intended to control for any differences in ability across teams. In the following chapter, I report only the results associated with this performance measure. Analyses were also run, however, using a separate variable for ability and the raw final profit scores for each team; due to the similarity of the two sets of results, and the relative simplicity associated with reporting the results associated with the residual performance measure, I have reported only the results associated with the residual performance measure.
76
Table 4. Descriptive Statistics, Scale Reliabilities, Rwgs, and ICCs
Number N 1. Leader CSE's 2. VC Feedback 3. Leader Efficacy 4. Trans. Leadership 5. Team Efficacy 77 6. Team Goals 7. Team Processes 8. Team Performance 40 40 40 8.60 4.07 0.00 2.16 .53 .99 1 6 --.82 --.98 -38 40 40 40 40 M Lindell 5.46 3.68 3.92 4.31 76.26 SD .67 .35 .62 .34 20.77 12 6 5 3 9 .80 .82 .86 .89 .94 rwg -.86 -.97 --
rwg ICC(1) -.83 -.92 --.88 --
ICC(2) of Items -.45 -.52 --.26 --.79 -.85 --.64 --
Notes. Team goal is reverse scored (scores of 10 indicate goals of first place). Team performance is a standardized variable representing the residual of a regression of ability on Quarter 8 (end) profit.
Table 5. All Study Variable Intercorrelations.
1 1. Leader CSE's 2. VC Feedback 3. Leader Efficacy 4. Transformational Leadership 5. Team Efficacy 6. Team Goals 7. Team Processes 78 Notes. * p < .05 ** p < .01 Team Performance --.03 .40* .18 .16 .20 .39* .29
2 -.33* .27 .38* .33* .37* .32*
3
4
5
6
7
8
-.39* .43** .23 .70** .26 -.24 .11 .28 .19 -.58** .52** .16 -.54** .20 --8. .37* --
Team goal is reverse scored (a score of 10 indicates the team's goal was to end in first place within its universe. Team performance is a standardized variable representing the residual of a regression of ability on Quarter 8 (end) profit.
Results I tested the hypothesized model presented in Chapter 3 by analyzing the covariance matrix using EQS; the results of this test are depicted in graphically in Figure 5 below. Hypothesis 1 suggested that the team leader's core self-evaluations and feedback the team receives at critical points would simultaneously have an impact on the leader's own efficacy to lead effectively. This hypothesis was supported; paths leading from the leader's core self-evaluations and team feedback to leader efficacy were both positive and significant (.38, p < .01, and .35, p < .01, respectively). The percentage of variance explained in leader efficacy was 26% (R2 = .26). Hypothesis 2a postulated that the leader's efficacy would be related to transformational leadership behaviors. This hypothesis was also supported; the path leading from leader efficacy to transformational leadership was significant (.39, p < .01). The percentage of variance explained in transformational leadership was 16%. Hypothesis 2b postulated that leader efficacy would mediate between both the leader's core self-evaluations and venture capitalist feedback and subsequent transformational leadership. In order to test for the presence of mediation, I separately tested two models with direct paths added between core self-evaluations and transformational leadership and between venture capitalist feedback and transformational leadership, respectively. Neither direct path model proved to be a better fit with the data ( of freedom, NS;
2 2
= .05 on one degree
= 1.01 on one degree of freedom, NS; respectively). Therefore,
hypothesis 2b was also supported.
79
Figure 5. Results of the Hypothesized Model.
Leader CSE's 80 VC Feedback
.38** Leader's Efficacy .35**
.39** Transformational Leadership
.24
.58*** Team Efficacy .52***
Team Goals
.00
Team Performance Team Processes
.38*
Notes. Note that the values are standardized parameter estimates *** p < .001 ** p < .01 * p < .05
Hypothesis 3a concerned the relationship between the leader's transformational leadership and team efficacy. This hypothesis was not supported (.24, NS); transformational leadership was not significantly related to team efficacy (with R2 = .06). Hypothesis 3b, which suggested that transformational leadership would mediate the relationship between leader efficacy and team efficacy, was also not supported, as there was no evidence of a direct effect between transformational leadership and team efficacy. Hypothesis 4a suggested that team efficacy would be related to team goal. This hypothesis was supported; team efficacy was significantly related to team goal (.58, p < .001, R2 = .34). Hypothesis 4b suggested that team efficacy would mediate the relationship between transformational leadership and team goal. To begin, transformational leadership had to be significantly related to team efficacy (hypothesis 3a); because hypothesis 3a was not supported, hypothesis 4b also was not supported. Hypothesis 5a suggested that team efficacy would be related to team action and transition processes. This hypothesis was supported; team efficacy was significantly related to team action and transition processes (.52, p < .001, R2 = .27). Hypothesis 5b suggested that team efficacy would mediate the relationship between transformational leadership and team action and transition processes; once again, because transformational leadership was unrelated to team efficacy (hypothesis 3a), hypothesis 5b was not supported. Hypothesis 6a suggested that both team goals and team action and transition processes would be related to higher levels of team performance. This hypothesis was partially supported; team action and transition processes were significantly related to team performance (.38, p < .05), but team goal was not (.00, NS). Team action and
81
transition processes accounted for 14% of the variance in performance (R2 = .14). Hypothesis 6b suggested that team goal and team action and transition processes would mediate the relationship between team efficacy and team performance; although team efficacy was related to both team action and transition processes and team goals (hypotheses 4a and 5a), only team action and transition processes was related to performance. When an alternative model was tested that included a direct path from team efficacy to team performance, a chi-squared difference test revealed no model improvement (
2
= .11 on one degree of freedom, NS). Therefore, it is reasonable to
conclude that team action and transition processes do, in fact, serve as a mediating mechanism between team efficacy and team performance; hypothesis 6b was partially supported. On the whole, the hypothesized model provided a poor fit with the data, with (20) = 47.34; p
Transformational leadership enhances the motivation, morale, and performance of followers through a variety of mechanisms. These include connecting the follower's sense of identity and self to the project and the collective identity of the organization
STUDY ON THE RELATIONSHIP BETWEEN LEADER CORE SELFEVALUATIONS, TEAM FEEDBACK, LEADER EFFICACY, TRANSFORMATIONAL LEADERSHIP, TEAM EFFICACY, TEAM GOALS, TEAM ACTION AND TRANSITION PROCESSES, AND TEAM PERFORMANCE
This research attempts to explain how internal team leaders can help teams perform in high stress, dynamic environments. Specifically, the aim of the research was to examine a chain of relationships beginning with team leaders' core self-evaluations and ending in team performance. Several core research questions were examined, including: "Do leaders' core self-evaluations and self-efficacy serve as important antecedents of transformational leadership? Is transformational leadership an important antecedent of team efficacy? How does team efficacy lead to higher levels of team performance?" First, I reviewed the extant relevant literature. Second, I derived from this review a theoretical model, which proposes that both leader core self-evaluations and team feedback may influence leader self-efficacy and transformational leadership. Transformational leadership may then influence team efficacy, which in turn may influence performance through its impact on team goals and team action and transition processes. Third, using data collected from a sample of teams composed of first-year
MBA students involved in a week-long business simulation, I tested the model and demonstrated support for a number of hypothesized relationships. Fourth, I revised the model based on modification indices provided by covariance structure analysis. I found that leaders' core self-evaluations do influence their self-efficacy; this efficacy exhibited positive and significant direct relationships with team efficacy and with team processes. In addition, I found support for a chain of relationships linking team efficacy to team goals to team processes to team performance.
THE RELATIONSHIP BETWEEN LEADER CORE SELF-EVALUATIONS, TEAM FEEDBACK, LEADER EFFICACY, TRANSFORMATIONAL LEADERSHIP, TEAM EFFICACY, TEAM GOALS, TEAM ACTION AND TRANSITION PROCESSES, AND TEAM PERFORMANCE
By Narda Roxanne Quigley
Dissertation submitted to the Faculty of the Graduate School of the University of Maryland, College Park in partial fulfillment of the requirements for the degree of Doctor of Philosophy 2003
Advisory Committee: Professor Edwin A. Locke, Co-Chairman/Advisor Professor Paul E. Tesluk, Co-Chairman/Advisor Professor Kathryn M. Bartol Professor Paul D. Bliese Professor Paul J. Hanges
ACKNOWLEDGEMENTS Three very special groups of people offered me an incredible amount of support during my doctoral studies at the University of Maryland: the professors with whom I have had the privilege to work, both in the Department of Management and Organization and in the Department of Psychology; the doctoral students of the Department of Management and Organization; and my family and friends outside the University of Maryland community. It is thanks to these wonderful individuals that I have been able to achieve my goals during my time in the doctoral program. I owe special thanks to both Ed Locke and Paul Tesluk—the two professors with whom I have worked most closely over the past four years—for their unflagging belief in me from the first day I set foot on campus. From Ed I have learned to value the sometimes seemingly contradictory goals of both attention to detail and simplicity in research undertakings; intellectual curiosity and rigor; and the importance of meticulously grounded research. From Paul I have learned to value the importance of viewing research questions and methods from all possible angles; the practical use of having knowledge of many statistical tools; and the importance of departmental community and the effort that should be spent building it. I feel truly honored to have had the opportunity to work with and be mentored by each of them. A number of other professors have generously assisted me during my time in the doctoral program. Kay Bartol has served as a fantastic role model of professionalism, service, and accomplishment; it has been a true pleasure to work with her. Paul Bliese has been extremely generous with his time in meeting with me over the last year to discuss various statistical issues regarding both this dissertation and other research; and Paul
ii
Hanges has helped me clarify many research questions and ways of going about answering them over the years with unbelievable patience. Cindy Stevens, Ken Smith, and Hank Sims, at various points during my time during the program, have provided both professional and emotional support. I also owe thanks to a number of doctoral students—Lisa Dragoni and Wei Liu for being above and beyond even the best cohort that an incoming doctoral student could hope for; Sharyn Gardner for being an incredible mentor; Jen Marrone, Gosia Langa, and Jay Carson for being incredible mentees; Amanuel Tekleab and Riki Takeuchi for always being willing to answer my questions and smiling while they did so; Qing Cao for putting up with my mess in the office; and Abhishek Srivastava, Jon Eckhardt, and Vincent Duriau for their words of wisdom (and humor) over the years. Thank you. Many thanks are also due to a number of individuals outside the University of Maryland community. My parents have been wonderful sources of support; I owe thanks especially to both for providing me with the unlikely combination of unconditional love and high expectations that have paved the way for everything that I have been able to accomplish. Both have also provided true inspiration to me in the examples they have set. To Paran Quigley I owe thanks for helping me to stay grounded, remember myself as a non-graduate student, and keep everything in perspective. To Kay Ovington I owe thanks for countless hours of data entry, computer help, and the like—but more importantly, for her indefatigable support, encouragement, and friendship as I have made my way through the program. To Pat Thomas I owe thanks for the friendship, mentoring, and mail over the years (not to mention the countless dinners at Roy's Place). To Leslie Jansen and Susy Gallor I owe thanks for being both fantastic housemates and friends.
iii
Concerning the execution of this research, I owe special thanks to all of my committee members, particularly Ed and Paul, for their excellent guidance. Special thanks are also due to Joyce Russell, whose helpful cooperation in data collection and subsequent answers to my many questions regarding the ELM and MARKETPLACE during the last year has been an unbelievably wonderful blessing. Thanks, once again, to all.
iv
TABLE OF CONTENTS List of Tables List of Figures Chapter 1—Introduction Chapter 2—Literature Review Chapter 3—Model and Hypotheses Chapter 4—Research Methods Chapter 5—Results Chapter 6—Discussion References vi vii 1 9 34 53 76 86 120
v
LIST OF TABLES 1. Judge, Erez, & Bono's (1998) Meta-Analysis Findings with Respect to the Correlations Between the Four Core Traits 2. Sequence of Events, Measures Collected, and Measure Uses in MARKETPLACE Simulation 3. List of Concepts and Measures Used 4. Descriptive Statistics, Scale Reliabilities, and Agreement Statistics of Study Variables 5. All Study Variable Intercorrelations 77 78 60 65 13
vi
LIST OF FIGURES 1. The Four Core Trait Indicators of Core Self-Evaluations 2. Summary Diagram of All Empirical Findings Regarding Core SelfEvaluations to Date 3. Hypothesized Model 4. Transformational Leadership Ratings Matrix 5. Results of the Hypothesized Model 6. Revised Path Model 18 35 70 80 84 10
vii
Chapter 1 Introduction
Consider the following situation: a team of 28 men at the start of World War I attempt to make the first historic crossing of Antarctica, from the Weddell Sea to the Ross Sea, on the ship Endurance. As they sail south for their destination, they find that the pack ice in the Weddell Sea is unusually thick. Just a day away from their destination, their boat becomes hopelessly trapped in the ice; instead of achieving honor, recognition, and glory, the team now faces a grueling trial of survival. For 10 months the team drifts north with the ice, living on the boat, eating penguins and seals, and camping on ice floes when their ship eventually sinks. What possible chance could this team have to survive? As it happened, the explorer in charge of the expedition was Ernest Henry Shackleton, a man renowned for his good humor, charm, and energy. Described as an "obstinate boy" by his first sea captain, Shackleton grew up to be described by Thomas Orde-Lees (one of his 28 team members) as never "anything but the acme of good humor and hopefulness." Shackleton's positive view of even the most adverse of conditions allowed him to keep the team together; at one point, for example, he asked his crew to train all 60-odd sled dogs they had brought with them so as to keep up morale. All 28 members of Shackleton's team eventually survived the ordeal; this amazing feat has been attributed to Shackleton's legendary positive leadership during a time of intense adversity. Sir Raymon Preistley, a British geologist and explorer who was part of the Endurance team, said, "For scientific leadership, give me [British explorer] Scott, for swift and efficient travel, give me [Norwegian explorer] Amundsen; but when you are in
1
a hopeless situation, when there seems to be no way out, get on your knees and pray for Shackleton." A number of issues surface in regard to the above situation. First, for example, what was Shackleton's response (both emotional and intellectual) as he faced extreme adversity and seemingly hopeless conditions with his team, and why? Second, how did Shackleton's response to the situation influence the confidence his team members had in their ability to overcome the challenge at hand, the team processes they experienced, and their ability to come up with a viable approach to the situation? Finally, did the team's confidence in its own abilities and the team's processes have any impact on whether it successfully navigated the challenges it faced in the frozen Weddell Sea; what exactly was it that allowed the entire team's survival? A relevant theoretical perspective to begin thinking about the above questions associated with leadership is core self-evaluations theory (Judge, Locke, & Durham, 1997). This theory is rooted in appraisal theory (e.g., Arnold, 1960; Lazarus, 1991; Locke, 1969, 1976; Packer, 1985/1986), which broadly asserts that individuals subconsciously appraise objects, people, or events in relation to their perceived needs, values, and commitments—and that emotions are the form by which individuals experience these subconscious appraisals (Judge et al., 1997). Using an appraisal theory perspective, therefore, Shackleton encountered a particular situation, and then was faced with his own emotional response to the situation—which then had an impact on his team's confidence, interactions, and outcomes. Judge et al. (1997), based on Packer (1985/1986), suggested that appraisals occur on many different levels. At the deepest level, appraisals of specific situations are affected by "core evaluations," or the basic
2
conclusions we hold subconsciously about ourselves, other people, and the external environment. Judge et al. (1997) argued that these core evaluations could be considered to be traits, which they define as "stable and consistent ways of thinking, feeling, or acting exhibited by individuals" (p.155). Judge, Locke, Durham, and Kluger (1998) found that core evaluations of the self were associated with core evaluations of the world and others. Four indicator traits have been used to measure these core self-evaluations: self-esteem, generalized self-efficacy, locus of control, and emotional stability/ neuroticism. While core self-evaluations theory has never been applied in a leadership context, the broader consideration of personality factors such as the four indicator traits and their impact on leadership is nothing new. "Great man" theories of leadership, which asserted that leadership qualities were inherited (this last unnecessarily, since they conceivably could be acquired as well), were popular in the nineteenth and early twentieth centuries (Kirkpatrick & Locke, 1991) only to lose their popularity by the mid-twentieth century. In the late twentieth century, trait theories experienced a resurgence of interest; this may have been partly due to the understanding that leadership success was not merely a matter of possessing some trait or combination of traits. Rather, as Kirkpatrick and Locke (1991) theorized, it may be the combination of those traits with certain actions that give rise to more effective leadership. It is here where core self-evaluations may prove to be of interest to modern leadership theory. Transformational and charismatic leadership theories (e.g., Bass, 1985; Burns, 1978; Conger & Kanungo, 1987) have never been particularly clear about the respective contributions of traits and behaviors to those types of leadership. They seem to
3
acknowledge both the importance of specific leadership behaviors and specific traits. The first component of transformational leadership as defined by Bass's (1985) theory, for example, is charisma; the original Greek meaning of the word means "gift," which suggests a trait. However, a different component of Bass's transformational leadership— individual consideration—is very similar to the consideration dimension from the Ohio State—Michigan studies, which were clearly studies about leadership behavior. The incorporation of core self-evaluations as an antecedent to transformational leadership may be an interesting way to examine whether a theoretically relevant trait contributes to transformational leadership behaviors; if so, the relationship between leadership traits and behaviors (in the context of transformational leadership) may be somewhat clarified. Core self-evaluations may also be particularly relevant to examine in the context of transformational leadership, as both concepts have been linked to higher performance through motivational pathways. This leads me to my first research question:
RQ1: Do leader core self-evaluations serve as an antecedent to transformational leadership?
Regardless of whether core self-evaluations serve as an important antecedent to transformational leadership, transformational leadership in team settings has been widely overlooked both by those studying transformational leadership and those studying teams (Zaccaro & Klimoski, 2002). Indeed, the majority of leadership research has centered on organizational leadership—leadership at a much higher level than team leadership. As a
4
result, there is a dearth of knowledge about what the effects of transformational leadership in a small team setting might be. In this dissertation, I seek to begin addressing this issue. It is likely that transformational leadership helps teams transcend the immediate issues and problems they face, helping team members stay focused and motivated. Part of this may be due to the ability of the transformational leader to generate team member confidence in the team's ability to handle the task at hand (team/collective efficacy—e.g., Gibson, 1999; Lindsley, Brass, & Thomas, 1995) and achieve its goals, which may in turn positively influence team processes and team effectiveness. This leads me to my second research question:
RQ2: Does transformational leadership in a team setting influence how teams respond when faced with challenging tasks?
Team efficacy is another critical variable examined in this dissertation (Bandura, 1997). Pescosolido (2001) reviewed the extant work on team efficacy and noted that though much work has demonstrated links between team efficacy and performance, little work exists that has attempted to determine how this type of efficacy is built and what can be done to increase a group's efficacy levels. It seems reasonable, as noted above, to consider team efficacy in tandem with both team leader core self-evaluations and transformational leadership due to their conceptual links. In this dissertation, I also consider whether the team leader's self-efficacy plays a role in linking the leader's core self-evaluations, transformational leadership, and team efficacy. Although some research has found leader efficacy to be an immediate antecedent of team efficacy (Pescosolido,
5
2001), no prior research has examined whether transformational leadership may play a mediating role in that relationship. My third research question, therefore, is as follows:
RQ3: What are the critical antecedents of team efficacy; how important is the role of transformational leadership as a pathway through which the leader may influence team efficacy?
Although a fair amount of research has found team efficacy to be directly related to team performance (e.g., Gully, Beaubien, Incalaterra, & Joshi, in press; Prussia & Kinicki, 1996; Shea & Guzzo, 1987; Silver & Bufiano, 1996; Spink, 1990), questions still remain regarding how exactly team efficacy is translated into performance. Previous research has supported the mediating role of team goals (e.g., Durham, Knight, & Locke, 1997); substantial theoretical support also exists for the mediating role of team action and transition processes (Marks, Mathieu, & Zaccaro, 2001). However, no study to my knowledge has considered how team efficacy may influence performance through its impact on both team goals and team action and transition processes. This leads me to my fourth research question:
RQ4: What are the specific mechanisms through which team efficacy influences performance; do team goals and team action and transition processes play mediating roles?
6
This research contributes to the existing literature on leadership, teams, and core self-evaluations in a number of ways. First, as noted above, few empirical studies have examined the impact of leaders in small team settings in particular. As Zaccaro and Klimoski (2002, p. 5) noted, "although there exist
7
antecedents of team efficacy; no prior research has examined team leader core selfevaluations, team leader efficacy, or transformational leadership as antecedents of team efficacy. A fourth contribution of this research is in the potential implications of the findings for selection and staffing purposes in organizations. If core self-evaluations do indeed impact the way leaders respond in the face of challenges to motivate team members and facilitate effective team processes, organizations with teams in difficult situations may want to specifically choose leaders on the basis of their core selfevaluations (which may be related to their ability to effectively handle the situation). It may be that leaders who are above average on core self-evaluations should be selected for teams assigned especially arduous tasks, if the trait helps them to cope with adversity and still be both high performers and satisfied with their jobs. Therefore, the purpose of this investigation is to examine core self-evaluations of leaders and the effects of this dispositional concept on leader efficacy, transformational leadership, team efficacy, team goals, team processes, and team performance. To this end, in Chapter 2, I review the relevant literature. In Chapter 3, I derive the theoretical model that is tested in this dissertation and provide support for the various hypotheses considered. In Chapter 4, I discuss the research methodology I utilized to test the hypotheses. In Chapter 5, I report the results of the study. In Chapter 6, I discuss theoretical, methodological, and practical implications, limitations, and future research directions.
8
Chapter 2 Literature Review
In this chapter, I establish the groundwork for the model proposed in Chapter 3. A review of the literature on core self-evaluations is followed by a discussion of leadership traits, transformational leadership, and emergent leadership. I then discuss team efficacy, team goals, and team processes. I close the chapter with a discussion of the scope of the research.
Core Self-Evaluations In the following review of the existing material on core self-evaluations, I first define core self-evaluations and discuss the significance of the four indicator traits and the connections between them. I then conclude with a review and summary of the findings in this area to date. Definition of core self-evaluations. As Judge, Bono, Erez, and Thoresen (2002) wrote, "Core self-evaluations is a higher order concept representing the fundamental evaluations people make about themselves, their environments, and the relationship between themselves and their environment" (p. 58), but in this dissertation I am only concerned with core evaluations of the self. As noted in the first chapter, the concept is manifested by four indicator traits as shown in Figure 1: self-esteem, emotional stability/neuroticism, locus of control, and generalized self-efficacy. These traits met three criteria that made them appropriate indicator traits for the core self-evaluations
9
Figure 1. The Four Trait Indicators of Core Self-Evaluations.
Self-Esteem Core SelfEvaluations
Locus of Control
10 Emotional Stability
Generalized Self-Efficacy
concept (Judge et al., 1997; Judge & Bono, 2001): evaluation-focus (the degree to which a trait represents evaluation, as opposed to description); fundamentality (as in Cattell's (1965) personality theory, fundamental traits underlie surface traits); and breadth or scope (Allport, 1961, noted that cardinal traits are broader in scope than secondary traits). The four core traits are conceptually related, as their definitions meet the above three criteria. Self-esteem is typically defined as the overall value that one places on oneself as a person (e.g., Harter, 1990). Neuroticism (emotional stability) represents the tendency to exhibit poor emotional adjustment and experience negative feelings such as fear, self-doubt, and depression (Judge et al., 1998; Barrick & Mount, 1991). Generalized self-efficacy is the most general type of self-efficacy, as it "encompasses individuals' judgments of their capacity to mobilize the motivation, cognitive resources, and courses of action needed to exercise general control over? their lives and deal successfully with life's challenges" (Judge et al., 1997). The last of the indicator traits, locus of control (Rotter, 1966), represents the perceived degree of control over the outcomes of one's actions (Judge et al. 1998). Rotter (1966) divided individuals into two camps based on their locus of control: individuals with an internal locus of control believe they control their own lives, while individuals with an external locus of control believe that they are controlled by luck, chance, fate, or powerful others. Empirical and theoretical links between traits. Judge et al. (2002) noted that though the indicator traits seem to be extremely prominent in the literature, few empirical investigations have included more than a single core trait. Indeed, even when two or more of the traits are included in the same study, they are usually treated as unrelated variables (e.g., Abouserie, 1994; Hesketh, 1984; Hojat, 1983; Horner, 1996; Tiggeman &
11
Winefield, 1984). Sometimes the traits are modeled as influences on each other, divorced from any theoretical framework. For example, Wambach and Panackal (1979) claimed an effect of neuroticism on locus of control, whereas Morelli, Krotinger, and Moore (1979) viewed locus of control as a cause of neuroticism. Judge et al. (1998) conducted a meta-analysis among 12 samples of roughly 15,000 individuals to investigate the relationship between the four core traits. Their results are summarized in Table 1. Note that the estimated true score correlations among the traits are substantial—when Judge et al. (1998) completed a factor analysis, they found that the core traits loaded strongly on a single underlying factor. Erez and Judge (2001) confirmed these results and found evidence that core self-evaluations is a higherorder factor that integrates the association among the four lower-level traits. Findings associated with core self-evaluations. Since Judge et al.'s (1997) theoretical piece discussing the nature of the proposed concept of core evaluations, a number of studies have advanced our understanding of the concept. The original concept they proposed included core evaluations of the self, other people, and the world in general (with trust vs. cynicism and just vs. unjust world as examples of two core evaluations in the domains of other people and the world). The individual studies that have been conducted to date examining core self-evaluations are reviewed below. Judge, Locke, et al. (1998) evaluated core evaluations of the self, the world, and other people and also considered the effects of these core evaluations on job and life satisfaction. They found that external core evaluations (those of the world and other people) did not explain further variance in job and life satisfaction than did core self-
12
Table 1. Judge, Erez, & Bono's (1998) Meta-Analysis Findings with Respect to the Correlations Between the Four Core Traits.
Trait Self-Esteem Generalized Self-Efficacy Locus of Control Neuroticism
1 1.00 .86 .58 -.62
2 .70 1.00 .59 -.54
3 .44 .45 1.00 -.47
4 -.51 -.45 -.36 1.00
Notes. K = 12 and n = 15,888 for locus of control—self-esteem, locus of control—neuroticism, and self-esteem—neuroticism correlations. K = 11 and n = 14,777 for correlations involving generalized self-efficacy. Correlations below the diagonal are corrected for measurement and sampling error; correlations above the diagonal are uncorrected. A 95% confidence interval for each correlation excluded zero.
13
evaluations and concluded that how individuals view the external world might, in fact, be a function of how they view themselves. They also found that core self-evaluations were both directly and indirectly related to job and life satisfaction. Positive core selfevaluations were found to have consistent effects on job satisfaction through their effects on job perceptions and also independently of the attributes of the job itself. Judge, Erez, et al. (1998) also investigated the theoretical relationship between core self-evaluations and job performance. They conducted a meta-analysis for correlations among the four core traits over 12 samples of roughly 15,000 individuals. In support of their previous work, as noted above, they found that the traits are highly correlated and comprise a common factor. They also presented a logical argument linking positive core self- evaluations to performance through the impact core self-evaluations have on motivation. In addition, they discussed the implications for selection of core selfevaluations research, covering legal, measurement, applicant reaction, and faking issues. Judge et al. (2000) returned to the consideration of job satisfaction as the primary dependent variable. They expanded upon Judge, Locke, et al., (1998), however, in that they considered both perceived job characteristics and objective job complexity in a procedure that included two studies. The first study found direct relationships between core self-evaluations and perceived job characteristics, between core self-evaluations and job complexity, and between core self-evaluations and job satisfaction. Furthermore, core self-evaluations had an indirect effect on perceived job characteristics through actual job complexity. The authors suggested that individuals with high core self-evaluations may seek out and attempt more complex jobs; they may be more likely to exert greater effort with respect to goal-setting activities and higher task involvement; they may be less
14
likely to withdraw from complex jobs when failure is experienced; and they may have better coping mechanisms, all of which may lead to increased attempts to attain and keep more complex jobs. In the second study covered in Judge et al. (2000), the constancy of the core-self evaluations concept and its relation to job complexity and job satisfaction over time was examined. Core self-evaluations as assessed in childhood and adolescence was related to job complexity and job satisfaction later in life, suggesting that core selfevaluations are indeed dispositional in nature. Erez and Judge (2001) examined the relationship between core self-evaluations, goal-setting, motivation, and performance in three studies. The first of these was to confirm that the four indicator traits loaded on one higher order factor; confirmatory factor analysis supported this hypothesis. The second study demonstrated that more positive core self-evaluations were related to higher levels of task motivation and performance. In addition, motivation (which was measured through amount of time spent on attempting to solve anagrams and also by three direct questions assessing motivation) was found to partially mediate the relationship between core self-evaluations and performance. The final study replicated the above findings in a sample of insurance agents; also, in this study, the effect of core self-evaluations on performance (as measured both by objective number of sales and rated performance) was found to be mediated by both goal-setting and job behavior (activity level—how active the agents were in making sales). The studies also answered several questions regarding the validity of the overall concept. Core self-evaluations appeared to have incremental validity above and beyond conscientiousness in this study, suggesting that core self-evaluations may be an important dispositional predictor of job performance. In addition, when the four core
15
traits were investigated as "one nomological network, the overall construct [concept] proved to be a more consistent predictor of job behaviors than when the individual traits were used in isolation." (Judge & Erez, 2000, p. 1277). Therefore, Judge and Erez (2000) provided some support for Judge, Erez, et al. (1998) in that core self-evaluations may influence performance through having an effect on motivation. Srivastava, Locke, & Judge (2002) revisited the question of why individuals with more positive levels of core self-evaluations exhibit higher levels of satisfaction in a lab setting. They found that core self-evaluations affect the level of task complexity that individuals choose, which was in turn related to perceptions of task characteristics and task satisfaction. Srivastava et al. (2002) also suggested that core self-evaluations might be related to a constellation of behaviors on actual jobs (for example, showing initiative, asking for more responsibility, seeking opportunities for learning, assuming leadership positions, etc.) that would lead to increased satisfaction. Durham, Locke, and Judge (2002), picking up on this last point and on Erez and Judge (2001), examined the relationship of core evaluations to behavior on the job. The authors hypothesized that core self-evaluations would influence job behaviors both directly and indirectly through perceptions of job characteristics. More positive core selfevaluations were found to influence self-reported job behaviors such as seeking greater job challenge; showing persistence in the face of setbacks; using independent judgment; seeking raises and promotions by asking managers how to get them; and motivating subordinates. However, core self-evaluations were not directly related to job behaviors when managers assessed the behaviors. The authors suggested that this last finding may have been due to the fact that managers were not able to monitor employees at all times;
16
therefore, employees may be more aware of their own behaviors than managers are. In addition, all managers probably did not use the same standards in evaluating employee behavior, which may have influenced results. Durham et al. (2002) noted that consideration of their results and the results of Erez and Judge (2001) in tandem yields some interesting conclusions, as each study considered different mediators between core self-evaluations and job behaviors. The conclusion that can be reached is that core self-evaluations may affect behavior on the job indirectly by a variety of paths and possibly directly as well. In addition, core selfevaluations affect both performance and satisfaction directly and indirectly, suggesting that this dispositional trait is worthy of more exploration in different realms. Figure 2 summarizes the findings regarding core self-evaluations to date. The research that has involved core self-evaluations to date provides some compelling evidence of the viability and utility of the theory in predicting motivation, satisfaction, and performance. Another potentially fruitful realm in which to examine core self-evaluations may be in their relationship to leadership. Two streams of leadership theory are particularly relevant: trait theories of leadership and transformational leadership theories. In the next section I discuss the findings and relationship between these two sets of theories; I close the section with a brief discussion of emergent leadership and its relevance in the context of work teams.
17
Figure 2. Summary Diagram of All Empirical Findings Involving Core Self-Evaluations to Date. Motivation, Goal-Setting Productivity
Job Behaviors
Self-Esteem Perceived Job Characteristics Generalized Self Efficacy Performance Ratings 18 Locus of Control Core SelfEvaluations Emotional Stability
Objective Job Complexity
Job/Task Satisfaction
Chosen Level of Task Complexity
Trait and Transformational Theories of Leadership Various theories of leadership abound, of course, ranging from trait theories (e.g., Stogdill, 1948) to behavioral theories (e.g., the Ohio State and University of Michigan leadership studies of the 1940s-50s: e.g., Fleishman, 1953; Halpin & Winer, 1957; Hemphill & Coons, 1957; Katz & Kahn, 1952; Katz, Maccoby, & Morse, 1950; etc.) to contingency theories (e.g., Fiedler, 1964, 1967) to theories of transformational leadership (e.g., Bass, 1985; Conger & Kanungo, 1987) to theories of shared leadership (e.g., Seers, 1996). As a clarifying point, the primary type of leadership upon which I focus in this dissertation is transformational leadership, due to this type of leadership's widely acknowledged impact on follower motivation. Theories of both charismatic and transformational leadership (e.g., Bass, 1985; Burns, 1978; Conger & Kanungo, 1987; House, 1977) have all specifically emphasized the motivating influence these types of leaders may have on followers, suggesting that characteristics, rhetoric, and behaviors of the leader may motivate followers to do extraordinary things. A secondary theoretical approach to leadership used in this dissertation is trait theory. In Chapter 3, I will argue that a team leader's core self-evaluations are an important individual antecedent related to the rise of transformational leadership. As noted in Chapter 1, the approach I will take is consistent with that taken by Kirkpatrick and Locke (1991), who suggested that traits might be a precondition for effective business leadership; effective leaders must take certain actions, however, in order to have a true impact (e.g., role modeling, setting goals, coaching subordinates, etc.). Therefore, in the next few paragraphs, I focus on trait and transformational theories of leadership.
19
Trait Theories of Leadership. As noted in the introduction, trait theories of leadership have been among the oldest and most widely discussed theories of leadership in the twentieth century (cf., Kirkpatrick & Locke, 1991; Stogdill, 1948). Prior to 1948 (the date of a seminal review by Stogdill) trait theories were popular, but often failed to show results. Stogdill's review pointed out these many inconsistencies, and the field of leadership as a whole took a new direction as theorists began to focus more on leadership behaviors and, eventually, leadership styles. Recently, however, trait theories have again become part of the leadership literature, as scholars have been considering the possibility that some individuals may have traits that raise the likelihood of certain effective leadership behaviors (Kirkpatrick and Locke, 1991). Many traits have been hypothesized to influence leader effectiveness, some of which are conceptually related to core self-evaluations. Kirkpatrick and Locke (1991), for example, suggested that traits such as drive, leadership motivation, honesty and integrity, cognitive ability, and knowledge of the business, among others, may be linked to leadership effectiveness. They also suggest that self-confidence and emotional stability are critical traits. For example, managers who are confident in their abilities may be more likely to inspire commitment among and be trusted by their employees. In addition, emotional stability is critical—especially when resolving interpersonal conflicts or facing stressful events and challenges. These two traits are closely related to core selfevaluations—an individual with positive core-self-evaluations is probably more likely to be more confident in his/her abilities, and emotional stability is one of the four indicator traits. A number of trait theorists have identified self-confidence, internal locus of control, and emotional stability as potentially important traits for leaders to have; many of
20
these are reviewed in more detail in Chapter 3. Existing empirical evidence does indicate that examining core self-evaluations in a leadership context is a logical direction to take. Transformational Leadership Theory. While trait theories of leadership experienced a resurgence of interest in the last decade, research on transformational leadership has increased exponentially. Three seminal works are widely credited with initially advancing theories of charismatic and transformational leadership: Burns (1978), House (1977), and Bass (1985). The stream of literature that grew surrounding this new area would eventually overtake research considering all other major theories of organizational leadership combined; as Judge and Bono (2000) noted, out of all articles in the decade between 1990 and 2000 available on the PsycINFO database, 207 articles cited transformational or charismatic leadership theory, and 190 cited other once popular leadership theories (least-preferred coworker theory, situational leadership theory, leadermember exchange/vertical dyad linkage theory, normative decision/Vroom-Yetton theory, behavioral theories, path-goal theory, implicit leadership theory, and romance of leadership). Theories of transformational leadership have certainly captured the attention of researchers in the field of organizational behavior. The approach highlighted in this dissertation emerges from Bass's (1985) theory of transformational leadership. In this theory, there are four dimensions of transformational leadership: idealized influence, inspirational motivation, intellectual stimulation, and individual consideration. Idealized influence comprises both the charismatic and the role-modeling aspect of transformational leadership. This is often considered the most prototypic and single most important dimension of transformational leadership (Judge & Bono, 2000). Inspirational motivation involves the articulation of a
21
clear, inspiring, and appealing vision to followers; intellectual stimulation involves the stimulation of follower creativity and thought by the questioning of assumptions and of the status quo. Finally, individual consideration involves attending to and supporting the individual needs of followers, much like the consideration dimension from the Ohio State—Michigan studies. Unlike the consideration dimension, however, individual consideration is more related to follower development than participative decision-making (Bass, 1995). In the original Bass (1985) piece, both transformational leadership and transactional leadership were discussed. In this research, however, I focus only on transformational leadership; I am most concerned with leaders who are able to motivate followers to pursue a common, inspiring goal and/or vision, rather than with leaders who establish exchange relationships with followers and seek to monitor that exchange (Burns's 1978 and Bass's 1985 transactional leadership type). A substantial amount of evidence has accumulated in support of the effectiveness of transformational leadership. A meta-analysis by Lowe, Kroeck, and Sivasubramaniam (1996) suggested that transformational leadership behaviors are related to subjective ( = .73) and objective ( = .30) measures of leadership effectiveness. Fuller, Patterson, Hester, and Stringer (1996) found similar results, with a correlation of .34 between transformational leadership and leadership effectiveness. Finally, transformational leadership seems to be effective in a variety of cultures using a variety of methods (Bass, 1997). Given this information, examining transformational leadership more closely in work team settings may be a potentially fruitful endeavor.
22
Before ending this section on leadership, I review one final topic: the issue of emergent leadership. This particular type of leadership is particularly important in a variety of organizational situations; I review these in more detail below. Emergent leadership and team development. An underlying assumption in much of the literature reviewed above is that teams operate with a designated individual who begins his/her life with the team as the appointed leader. Many teams in organizational contexts, however, begin existence without the designation of a formal leader. The phenomenon of emergent leadership may be observed in these teams. Schneider and Goktepe (1983) defined emergent leaders as group members who exert significant influence over other members of the group although no formal authority has been vested in them; these leaders, as Taggar et al. (1999, p. 901) noted, "may be just as important to the facilitation of team task completion as are designated leaders." Indeed, the manner in which a leader comes to power (either formally or through an emergent process) may be "unimportant in comparison to the behaviors of the leader" (Firestone, Lichtman, & Colamonosca, 1983). The literature on emergent leadership, much like the literature reviewed above, has therefore placed emphasis on the appropriateness of specific leadership behaviors in certain situations. Evidence suggests that those individuals who behave in different leader-like ways to help the group achieve its goals, given different situations, will be perceived by peers as being more leader-like. For example, Zaccaro, Foti, and Kenny (1991) found that emergent team leaders (who were rated highest on a measure of perceived leadership by their peers) were more adept than other team members at perceiving team requirements and selecting appropriate behavior to these demands
23
(Taggar et al., 1999). In addition, Pescosolido (2001) suggested that different behaviors of an emergent leader may be important during different phases of a group's development; emergent leaders may have more of an impact on certain aspects of group functioning earlier rather than later in a group's development. Seers and colleagues (Seers, 1989; Seers, Petty, & Cashman, 1995) postulated another related approach to how leaders emerge in team situations. They suggest that the process of leadership emergence in teams is that of negotiating roles and relationships between team members. The role an individual assumes on a team, then, depends on his/her abilities, the needs of other group members, and the team task to be completed. Some group members may possess inherent characteristics and exhibit certain behaviors that are perceived by other group members as indicative of leadership (Taggar et al., 1999). From this perspective, different people may emerge as leaders, depending on what roles are perceived as important leadership roles given the specifics of the situation. In this dissertation, I focus on teams that, at their inception, are initially leaderless. Furthermore, the perspective I assume is that certain individuals will surface as emergent, transformational leaders. I suggest in chapter 3 that these individuals will have an impact on team efficacy, team goals, and team processes.
Team Efficacy In the above section, I reviewed findings with respect to trait, transformational, and emergent leadership theories. In Chapter 3, I hypothesize connections between these areas and the development of a team emergent state (Marks, Mathieu, & Zaccaro, 2001)—team efficacy—and team processes. In this section, I focus on team efficacy. I
24
define it, discuss how it is different from the concept of self-efficacy, and then review some of the major findings associated with the concept. The concept of team efficacy has been considered for the last two decades, though research exploring the topic has increased considerably in the last decade. In the framework of social cognitive theory (cf., Bandura, 1991; Bandura, 1997), efficacy beliefs play a key role at all levels of analysis. At the individual level of analysis, selfefficacy is defined as "beliefs in one's capabilities to organize and execute the courses of action required to produce given attainments" (p. 3). It has been described as taskspecific self-confidence (Locke & Latham, 1990); a wide range of studies have confirmed a strong relationship between self-efficacy and performance (e.g., Gist & Mitchell, 1992; Stajkovic & Luthans, 1998). These definitions and findings clearly pertain to the individual level of analysis. However, efficacy beliefs can exist at higher levels of analysis; as early as 1982, Bandura called for a "broad a comprehensive research effort" and the development of suitable tools for measuring team efficacy, which he referred to at that point as "groups' perceptions of their efficacy to achieve varying levels of results" (p. 14). Since that time, a number of studies have assessed the impact of group efficacy on group performance (e.g., Shea & Guzzo, 1987; Silver & Bufiano, 1996; Spink, 1990; etc.) and Bandura has defined team efficacy as the "group's shared belief in its conjoint capabilities to organize and execute the courses of action required to produce given levels of attainment" (Bandura, 1997, p. 477). The critical difference between self-efficacy and team efficacy is the level of analysis. This distinction is important to make, as both self-efficacy and team efficacy are a part of the model presented in Chapter 3. Self-efficacy, as noted above, is used to
25
describe a single individual's assessment of his/her capabilities, acting alone as a single unit. Team efficacy is an analogous concept defined at the team level of analysis (Bandura, 1997; Lindsley et al., 1995). The assumption is that teams can be treated (metaphorically) as social entities that are capable of acting together as a unit (Lindsley et al., 1995). Individuals are typically used as "informants to estimate the group's or organization's collective efficacy" (Lindsley et al., 1995, p. 648); of course, cognitions reside only in the individual, so groups and organizations may possess beliefs only in the sense of shared cognitions. Lindsley et al. (1995) pointed out that the individual-level cognitions of group or organization members are quite distinguishable from their beliefs regarding the group's capabilities. These group-based beliefs fundamentally develop from an individual's ability to cognitively assess the capabilities of other group members and their ability to work together based on observation and experience. Therefore, team efficacy is distinguishable from individual-level self-efficacy in that it is an emergent property of groups. It is also distinguishable from the average level of team member self-efficacy within a group. This concept would refer to the average overall extent to which different team members felt confident about their own individual capabilities to achieve a given level of task performance, rather than the extent to which group members felt confident about the ability of the group as a whole. As noted above, there is an emerging literature on team efficacy. This work has focused on a number of different issues, including the most appropriate way to measure the concept (e.g., Gibson, Randel, & Earley, 1999), the most appropriate way to conceptualize and distinguish it separately from related concepts such as team potency (e.g., Guzzo et al., 1993), and its influences on a variety of organizational phenomenon.
26
Research has demonstrated a strong link between group efficacy and group motivation and performance (e.g., Bandura, 1997; Durham, Knight, & Locke, 1999; Gully, Beaubien, Incalaterra, & Joshi, in press; Prussia & Kinicki, 1996), and Jex and Bliese (1999) found that team efficacy acts as a buffer of stressor-strain relationships. There is a good deal of promise associated with the continued incorporation of group efficacy into models of team performance. Given the promising results so far, I include group efficacy in Chapter 3 as a major aspect of the model to be presented. In the next section, I review the literature on team goals.
Team Goals Much research has been conducted in the last three decades regarding the operation of goals at the individual level. Specific, difficult goals—if accepted—lead to higher levels of performance than easy goals or no goals (Locke & Latham, 1990). Although the majority of goal-setting research has been conducted at the individual level, goal-setting at the team level has been receiving more attention as of late (cf. Locke & Latham, 1990; O'Leary-Kelly, Martocchio, & Frink, 1994; Weldon & Weingart, 1993). Basic relationships between goal level and performance at the individual level have been replicated at the team level (cf. Locke, Durham, Poon, & Weldon, 1997; O'Leary et al., 1994). A number of studies have shown that specific, difficult group goals result in higher levels of performance than when there are no goals or easy goals present (e.g., Becker, 1978; Buller & Bell, 1986; Ivancevich, 1974; Klein & Mulvey, 1989, 1995; Latham & Locke, 1975; Latham & Yukl, 1975; Lawrence & Smith, 1955; Pearson, 1987; Weingart, 1992; Weingart & Weldon, 1991; Welden, Jehn, & Pradhan, 1991).
27
Research examining goals at the team level has been increasingly focused on discovering the mediators and moderators of the goal effect (Durham, Knight, & Locke, 1997; Guzzo & Dickson, 1996; Weldon & Weingart, 1993). As Durham, Locke, Poon, and McLeod (2000) noted, a puzzling theoretical issue in the goal-setting literature currently exists. Research to date has indicated that goals may influence performance in a number of different ways: through direct effects, through both direct effects and effects moderated by strategy or tactics (i.e., interaction effects; Chesney & Locke, 1991; Durham et al., 1997), and through effects mediated by strategy (DeShon & Alexander, 1996; Durham et al., 2000). The approach taken in this dissertation, which will be discussed more in Chapter 3, focuses on the direct effect team goals have on performance. In the next section, I review the literature on team process and denote what processes may potentially be influenced by leadership and team efficacy.
Team Processes Team process research has historically been characterized by the use of a wide and inconsistent array of process variables (Marks, Mathieu, & Zaccaro, 2001); different scholars have varied widely in the extent to which they have used and ways in which they have operationalized different types of variables to represent team process (Weingart, 1997). In the last few decades, however, team processes have occupied a critical role in many theoretical models of team effectiveness (e.g., Gist, Locke, & Taylor, 1987; Guzzo & Shea, 1992; Hackman, 1987, etc.). A number of taxonomies of group process have been proposed (e.g., Argote & McGrath, 1993; Fleishman & Zaccaro, 1992; Neiva, Fleishman, & Rieck, 1978; Prince & Salas, 1993). Many of these models described
28
processes as being the critical mediating mechanisms through which member and organizational characteristics are related to effectiveness criteria (Marks et al., 2001). Despite this proliferation of research on process, however, few approaches to the topic included a temporal dimension with task completion as the centerpiece (the few that exist are Kozlowski, Gully, Nason, & Smith, 1999; Marks et al., 2001; McGrath, 1991); this type of approach is central to the work teams accomplish in organizations. Marks and colleagues (2001) proposed a temporally based framework and taxonomy of team processes to clarify the "process muddle." I use this particular model as the foundation for the hypotheses involving team process presented in Chapter 3. Marks et al. defined team process "as members interdependent acts that convert inputs to outcomes through cognitive, verbal, and behavioral activities directed toward organizing taskwork to achieve collective goals" (p. 357). In their taxonomy, team processes are divided into three categories. Transition processes are those that involve mission analysis, formulation, and planning; action processes involve monitoring progress toward goals, systems monitoring, team monitoring and backup behavior, and team coordination; and interpersonal processes involve conflict management, motivation and confidence building, and affect management. Teams use these different processes simultaneously, though some processes happen more frequently in different phases of a team's work on a task. The two types of phases in each performance episode that Marks et al. (2001) considered were action phases and transition phases. During action phases, teams are engaged in acts that contribute directly to goal accomplishment (e.g., taskwork), while during transition phases, teams focus primarily on evaluation and/or planning activities to guide their accomplishment of a team goal or objective. Marks et al. (2001) posited that
29
while interpersonal processes occur through both phases of the episodic framework, action processes most often occur in the action phase, while transition processes most often occur in the transition phase. In Chapter 3, I focus on these action and transition processes that occur in newly formed teams. Though interpersonal processes are surely related to team viability and member satisfaction, it is beyond the scope of this dissertation to include interpersonal processes and those aspects of team effectiveness. The final section in this chapter will cover the scope of this research. This section is intended to bridge the material I reviewed in this chapter and the theoretical model to be discussed in Chapter 3.
Scope of the Research This section specifies the exact scope of the research for this dissertation. First, I discuss the types of teams and tasks for which this research is most relevant; I then discuss the specific role of leadership; and I close the chapter with a discussion of the levels of analysis examined. Team and task type. Scholars have identified many different types of teams in organizational settings (e.g., Sundstrom et al., 1990—identified advice and involvement teams, production and service teams, project and development teams, and action and negotiation teams; Cohen & Bailey, 1997—identified work teams, parallel teams, project teams, and management teams, etc.). The model discussed in Chapter 3 is intended to generalize to newly formed, autonomous work teams (whose members have no shared history of working together) who are working on an unfamiliar, complex task that must be completed during a specific time period. Moreover, the teams I focus on have just
30
undergone a critical midpoint transition period, wherein they received important feedback regarding past progress and future plans for task completion. The nature of the task itself is an important aspect of group functioning; a number of authors incorporate task structure into theoretical models of group functioning (e.g., Hackman, 1987; Campion, Medsker, & Higgs, 1993). Hyatt and Ruddy (1997) noted that tasks are especially critical to consider in group contexts as they determine what the nature of team effectiveness is for any given situation and can affect the importance of many work group characteristics. The general task type to be examined in this dissertation is a complex information processing and problem-solving task that requires a high level of interaction between team members. This type of task is most congruent with Tesluk, Mathieu, Zaccaro, and Marks's (1997) description of tasks that produce an "intensive" work arrangement, where team members must "work together closely to diagnose and solve problems in performing the task" (p. 203). In addition, these types of dynamic and complex tasks require well-orchestrated teamwork, which may be best spearheaded by a transformational leader. Role of leadership. As noted above, the type of leader examined in this dissertation is not a designated leader, but rather an emergent one. As reviewed in the section in this chapter on emergent leadership, these types of leaders often have a great deal of influence over group members. Emergent leaders of this type of team may fill a wide variety of roles—from being a structure initiator/direction setter to being a facilitator of team process to being a team motivator (e.g., Hackman & Walton, 1986; Zaccaro & Marks, 1999). The focus in this dissertation, however, is on emergent, transformational leaders. This type of leadership and its relationship to the development
31
of team efficacy and team processes will be highlighted more in Chapter 3; the dissertation focuses on the role of emergent leaders due to the fact that they have been linked to the development of team efficacy and subsequent team performance (e.g., DeSouza & Klein, 1996; Pescosolido, 2001; Zaccaro et al., 2001). Levels of analysis. I focus on concepts at two levels of analysis in this dissertation: the individual leader level and the team level. The concepts in the pathway from leader core self-evaluations to transformational leadership (described in detail in Chapter 3) are at the individual leader level. Team efficacy, team goal, team action and transition processes, and team performance are at the team level. Issues of aggregation (e.g., Bliese, 2000; Chan, 1998; Kozlowski & Klein, 2000) arise only with respect to team processes, as the other team-level variables were either assessed through consensus methods or—in the case of team performance—assessed with objective task feedback. I consider team action and transition processes to be an emergent property of groups that arises out of individual-level perceptions—in other words, individuals' combined perceptions and their agreement regarding these perceptions define the group-level characteristic. Therefore, I will be considering both individual leader level and team level variables in the model to be presented in Chapter 3. It is important to note that although I am focusing on two levels of analysis (leader and team level), typical methodological issues associated with cross-level hypotheses (e.g., Klein, Bliese, Kozlowski, Dansereau, Gavin, Griffin, Hofmann, James, Yammarino, & Bligh, 2000) do not present the usual statistical challenges in this dissertation. Although I do consider the effect of leader-level variables on team-level variables—and vice versa—due to the fact that there is only one leader per team, typical
32
cross-level issues such as non-independence in the individual-level data (Bliese, 2000; Bliese & Hanges, 2001) are not as critical.
33
Chapter 3 Model and Hypotheses
In Chapter 2, I developed the groundwork for the model that is tested in this dissertation. I reviewed a number of different theories and findings relating to core selfevaluations, leadership, team efficacy, and team processes, and closed the chapter with a discussion of the scope of the research. In this chapter, I build on that foundation by presenting an overview of the model, followed by the underlying theory and support for each of the specific hypotheses in the model.
Model Overview The model to be presented in this chapter, depicted in Figure 3, incorporates elements of core self-evaluations theory, transformational leadership theory, social cognitive theory, team process theory, and goal setting theory. As noted in Chapter 2, the model is intended to generalize to a common situation found in organizations: that of newly formed work teams, lacking a shared history, that nonetheless face one overarching complex task (which may incorporate a number of subtasks and which may require different inputs at different points in time from leaders and team members—e.g., Kozlowski et al., 1996; Marks et al., 2001) and that have received critical midpoint feedback. Work teams in similar situations have received some recent attention in the literature (e.g., Lester, Meglino, & Korsgaard, 2002); these situations may be particularly conducive to the examination of the effects of team leader core self-evaluations due to the high stress levels involved.
34
Figure 3. Hypothesized Model H4b
H3b
Leader CSE's
H1 Leader's Efficacy Transformational Leadership H2a H3a
H4a Team Efficacy H5a
Team Goals
H6
Team Performance Team Processes
35
Midpoint Feedback
H1
H6
H2b H5b
H6b
In the model, leader core self-evaluations and critical midpoint feedback are hypothesized to impact team efficacy through their effects on leader self-efficacy and transformational leadership, which in turn is hypothesized to have a direct impact on team efficacy. Team efficacy, in turn, is expected to indirectly influence team performance through influencing team goals and processes. In the next sections, each set of relationships in the model will be discussed.
Team Leader Core Self-Evaluations and Midpoint Feedback
Leadership Efficacy
In the following paragraphs, I first discuss the connection between core selfevaluations and leadership efficacy. Then, I discuss the role feedback may play in the specific situation on which this dissertation focuses: teams operating in a dynamic environment that have reached a critical midpoint in their existence. There is both theoretical and empirical reason to believe that a team leader's core self-evaluations should be related to his/her subsequent self-efficacy for leading the team to successful completion of the task. Theoretically, Bandura (1997) suggested that selfefficacy (which is defined as task-specific self-confidence; here, it is defined as the confidence individuals have in their ability to meet the role requirements expected of a leader) primarily arises from four sources of information: enactive mastery experiences, vicarious experiences, verbal persuasion and other types of social influences, and "physiological and affective states from which people partly judge their capableness, strength, and vulnerability to dysfunction" (Bandura, 1997, p. 79). This last source may be one logical link to the team leader's core self-evaluations. When a team faces a complex situation, the team leader is likely to formulate a sense of his/her efficacy from
36
his/her own initial cognitive and affective reactions to the task, as he/she has had no prior experience with the task (either personal or vicarious). The basis for these reactions would be the leader's appraisal of the situation, which in turn is rooted in the leader's appraisal of him or herself. As noted in Chapters 1 and 2, core self-evaluations theory has its roots in appraisal theory (e.g., Arnold, 1960; Lazarus, 1991; Locke, 1969, 1976; Packer, 1985/1986), which suggests that appraisals of specific situations are affected by deeper and more fundamental evaluations of oneself (Judge et al., 1997). Judge et al. (1998, p. 30-31) suggested that individuals "who consider themselves worthy and able to cope with life's exigencies bring a 'positive frame' to the events and situations they encounter, whereas people who do not see themselves as worthy and able bring a negative frame to the same situations." By this logic, leaders of newly formed teams who face a complex task will be likely to perceive the situation at least partly through the lens of their own core self-evaluations. Another logical link between the team leader's core self-evaluations and his/her efficacy may be through generalized self-efficacy. Generalized self-efficacy, as noted in Chapter 2, is the belief an individual has in his/her ability to successfully manage life and the challenges he/she encounters; it is one of the four core traits included in the core selfevaluations concept. As Bandura (1997) noted, judgments of task-based efficacy are likely to be related to more general assessments of efficacy. Therefore, it is likely that more positive core self-evaluations will be related to more positive assessments of taskbased efficacy. In addition, there is some empirical evidence to support the existence of a relationship between a leader's core self-evaluations and his/her leadership efficacy.
37
Recent studies have suggested that managers with more positive core self-evaluations may perform better when facing challenges, which may arise out of a heightened sense of efficacy. Both Judge, Thoresen, and Pucik (1999) and Wanberg and Banas (1997) found that managers with a more positive view of themselves were more likely to cope effectively with changes induced by organizational transformations. In the case of Judge et al. (1999), coping with change mediated the relationship between positive self-concept and job performance, suggesting that core self-evaluations motivate leaders to take task relevant actions. Another stream of literature has examined the relationships between the core indicator traits and self-efficacy, providing further empirical evidence linking the concepts. Chen, Gully, Whiteman, and Kilcullen (2000) found support for a relationship between generalized self-efficacy and task-oriented self-efficacy. Therefore, leadership efficacy may arise out of the team leader's core selfevaluations (Judge et al., 1997). A leader who is more likely to perceive him or herself in a positive manner may be more likely to put a positive frame on the challenging task (Judge et al., 1998), thus leading to higher levels of self-efficacy (Bandura, 1997). A leader's efficacy is not simply an outgrowth of more global personality factors, however. Some of the leader's estimation of his/her efficacy as a leader is most likely related to the feedback that the team has received. As noted above, both enactive mastery experience and verbal persuasion (both of which can be considered as forms of feedback) are also important antecedents of self-efficacy (Bandura, 1997). Teams that have reached critical midpoints have typically received some type of performance feedback, be it associated with the objective quality of task completion or verbal feedback from an external manager. To the extent that the leader identifies with his/her team and feels
38
responsible for the successes and failures of the team during the first half of task completion, the leader is likely to view these successes and failures as reflections of the effectiveness of his/her leadership. Moreover, any feedback that is received at the critical midpoint is likely to be particularly linked to a leader's assessment of his/her effectiveness. Often verbal feedback received at the midpoint is specifically directed toward team leaders, as they may be the liaison with external managers. Additionally, as Gersick (1988) noted, feedback received at the midpoint may be a particularly salient indication of whether the team's trajectory is on track. Teams—and their leaders—tend to be in an evaluative mode during this phase of their existence and, as such, may be more influenced by the feedback they receive. Leaders of teams that receive very positive feedback at the midpoint, then, should experience both reinforcement and an increase in their existing levels of selfefficacy, while leaders of teams that receive more negative feedback at the midpoint should experience a decrease in their existing levels of efficacy. To conclude, a team leader's efficacy shortly after his/her team receives midpoint feedback on a complex task is likely to be related to two main antecedents: his/her own core self-evaluations and the feedback the team has received on its performance. Hypothesis 1: Team leader core self-evaluations and positive midpoint team feedback are positively related to leadership efficacy.
Leader Efficacy
Transformational Leadership
Existing theory provides support for the idea that the team leader's efficacy may be related to behaviors indicative of transformational leadership. Though there is little (if
39
any) research on the link between leadership efficacy and transformational leadership, support for this relationship follows from social cognitive theory. Research has shown that high levels of efficacy seem to be associated with higher levels of performance on all types of tasks in many different realms (cf., Bandura, 1997). Part of this effect is associated with the influence that self-efficacy has on personal choice. Individuals who feel highly efficacious regarding a particular task will be more likely to choose to perform that particular task, set high performance goals, and in turn exhibit higher performance (Bandura, 1997; Locke & Latham, 1990). Transformational leadership is traditionally associated with challenging the status quo and instilling confidence in followers that they can achieve higher levels of performance (Eden, 1990; House & Howell, 1992; Kirkpatrick & Locke, 1991). It follows that the leader's own efficacy may be an important antecedent of transformational leadership, as individuals with low levels of efficacy are not likely to take the initiative in challenging situations and persuade others to do the same. More support for this relationship can be found in examining Bandura's (1997) discussion of the important role of efficacy in social processes. He wrote: ?those who have a firm belief in their efficacy, through ingenuity and perseverance, figure out ways to exercise some measure of control over social systems containing limited opportunities and many constraints. Given a social environment with surmountable barriers, people who have a high sense of efficacy will be able to exercise more control over it, and will view it as more changeable, than will self-doubters who give up in the face of difficulty. (p. 483)
40
Though Bandura was not writing in reference to transformational leadership specifically, certain elements of the statement above point to a potential connection between leadership efficacy and transformational leadership. At its core, leadership is a social influence process (e.g., Yukl & Van Fleet, 1992), and transformational leadership has been identified as being particularly effective in times of crisis and change in organizations. Therefore, though there has been little literature linking self-efficacy to transformational leadership, one can draw the conclusion that higher levels of efficacy may lead to higher levels of leadership performance. Therefore, though there is no existing literature to my knowledge that draws the specific connection between leadership efficacy and transformational leadership, I hypothesize that there should be a strong, positive relationship between the two. Hypothesis 2a: The team leader's efficacy is positively related to the extent to which the leader exhibits transformational leadership.
The Mediating Role of Leader Efficacy Though research on transformational leadership has considered personality traits closely related to core self-evaluations—such as self-confidence—as antecedents of transformational behaviors (e.g., Bass, 1990; House, 1977) and research on trait theories of leadership share conceptual links with research on transformational leadership (e.g., Lord, De Vader, & Alliger, 1986), more process-oriented models of how traits and behaviors are linked are rare in either body of work. Judge, Bono, Ilies, and Gerhardt (2002) noted that trait theories of leadership would greatly benefit from "process models that illuminate the dispositional source of leadership" (p. 36). Judge and Colbert (2002)
41
also noted a "lack of research on the processes by which personality traits affect leadership" (p.4). Though Judge and Colbert (2002) focused primarily on the mediating role of process behaviors such as idea generation, integration of others' ideas, amount of talking, and expressive communication style, a leader's efficacy at leading the group task in question may also be an important mediator between his/her traits and the expression of transformational leadership. Hypothesis 2b: The team leader's efficacy mediates the relationship between core self-evaluations and the extent to which the leader exhibits transformational leadership.
Transformational Leadership
Team Efficacy
As noted in Chapter 2, team efficacy can be defined as a "group's shared ability in its conjoint capabilities to organize and execute the courses of action required to produce given levels of attainment" (Bandura, 1997, p. 477). Though Chen and Bliese (2002) and Pescosolido (2001) have both noted the lack of research on the antecedents of team efficacy, in previous research leadership has been cited as a potential source of team efficacy. Zaccaro, Blair, Peterson, and Zazanis (1995, p. 315), for example, wrote that leadership functions "are directed at fusing a capable team from disparate individuals, and? building perceptions among individual members of their combined and collective abilities." In addition, Chen and Bliese (2002), Dusig (2000), and Sosik, Avolio, and Kahai (1997) found that leadership is, in fact, an important antecedent of team efficacy. As Lindsley et al. (1995) noted, team efficacy typically arises from common exposures of members to objective stimuli (such as the objective constraints of the situation) and the processes of social influence and comparison. It is this latter category that leader
42
behaviors come into play, as leaders play a special part in the social influence process in groups. By definition, leaders are individuals who exert influence over other group members, in addition to exerting influence over task objectives and strategies (Yukl & Van Fleet, 1992). Transformational leadership in particular may be an especially relevant source of team efficacy. Transformational leadership theorists have highlighted how transformational leaders develop, intellectually stimulate, and inspire followers to achieve their purpose, mission, or vision (Bass, 1985; Howell & Avolio, 1992). One of the ways in which these activities are accomplished is through coaching followers, which often involves verbal persuasion to convince them that they are capable of accomplishing goals. Indeed, Pescosolido (2001) suggested that verbal persuasion (an activity often associated with transformational and charismatic leaders) may be the key way leaders with high efficacy may raise team efficacy. Empowering leadership (often considered to be part of transformational leadership—Bass, 1985) may also be related to higher levels of team efficacy, as Kumpfer, Turner, Hopkins, and Librett (1993) found. Some research has begun to examine the connection between transformational leadership and team potency, a term closely related to team efficacy. Though there is a difference between the two terms (potency is more a general sense of confidence that the group can accomplish its goals, while efficacy is more related to the specific task at hand; Guzzo, Yost, Campbell, & Shea, 1993), the findings with respect to potency can easily be applied in an efficacy context. Guzzo et al. (1993) argued that transformational leadership directly influences group potency by boosting the confidence of team members and developing their belief in the ability of the team to succeed. In addition,
43
Sivasubramaniam, Murry, Avolio, and Jung (2002) suggested that transformational leadership ?may enhance the potency of groups or teams by making participation in a group's efforts more meaningful and tied to the collective identity of the group (Shamir, House, & Arthur, 1993)?to the degree that transformational leadership builds personal identification with a group, along with a sense of confidence, a group's level of potency and performance is expected to be higher. (p. 72)
In an empirical test, Sivasubramaniam et al. (2002) indeed found transformational leadership to be significantly related to group potency. Therefore, there is a fair amount of evidence in support of the idea that transformational leadership would assist in the development of team efficacy and that it would be significantly and positively related to the concept. Hypothesis 3a: Transformational leadership is positively related to team efficacy.
The Mediating Role of Transformational Leadership Although Pescosolido (2001) found that leadership efficacy directly influenced team efficacy, the potential mediating effects of transformational leadership were not considered. While there is good reason to believe that a leader's efficacy would be positively related to team efficacy, leaders are likely to convey their efficacy to their fellow team members through their behaviors. To the extent that these behaviors help motivate followers and are indicative of the confidence the leader has, they may be the
44
way in which the level of a leader's efficacy is made clear to the rest of the team—which then is likely to result in higher levels of team efficacy. Hypothesis 3b: Transformational leadership mediates the relationship between team leader efficacy and team efficacy.
Team Efficacy
Team Goals
The above few sections have dealt with a chain of antecedents leading to team efficacy, beginning with the team leader's core self-evaluations. In the next few sections, I discuss how higher levels of team efficacy may translate into performance. First, I discuss the influence team efficacy may have on goals; I will discuss other team processes that may be influenced by team efficacy in the following section. I will close the chapter with a discussion of hypothesized connections between team performance, team goals, and team processes. A number of research teams have found team efficacy to be related to one major team process: goal-setting (e.g., Durham, Knight, & Locke, 1997; Silver & Bufiano, 1996; Weldon & Weingart, 1993). Gist (1987) argued that group perceptions of team efficacy should be related to group performance, and Prussia and Kinicki (1996) and Whitney (1994) both found support for this relationship. Scholars have found this efficacy-performance relationship to be mediated by goals, both at the individual and group levels (Mulvey & Klein, 1998; Prussia and Kinicki, 1996). Additionally, there is good theoretical reason to believe that team efficacy is related to team goal-setting; both Bandura's social cognitive theory (1986, 1998) and Locke and Latham's (1991) goalsetting theory support this connection. Bandura (1982) argued that "perceived collective
45
efficacy will influence what people choose to do as a group, how much effort they put into it, and their staying power when the group efforts fail to produce results" (p. 143). As Mulvey and Klein (1998) noted, the logical conclusion of this statement is that team efficacy should influence team goals and the team's commitment to those goals. Therefore, Hypothesis 4a: Team efficacy is positively related to team goals.
The Mediating Role of Team Efficacy in the Transformational Leadership Relationship
Team Goals
Although a critical characteristic of the transformational leader is having a vision and effectively communicating that vision to his/her followers, directive leadership (wherein leaders are involved on a more micro level with their employees) is usually the type of leadership that is associated with the setting of specific team goals (Yukl, 1998). In the case of transformational leaders, it is more likely that team members be inspired by the vision, motivation, and ideas put forth by the transformational leader, and as such take more of an active role in setting their team's own immediate performance goals. Because team members are the individuals setting those team goals, it is likely that goal levels will be dependent on team efficacy, and that transformational leadership will only indirectly influence team goals through influencing team efficacy. Hypothesis 4b. The relationship between transformational leadership and team goals is mediated by team efficacy.
46
Team Efficacy Team Processes As reviewed in Chapter 2, there is an emerging literature regarding both the antecedents and consequences of team efficacy. I postulate here that higher levels of team efficacy will be positively related to both action and transition team processes (Marks et al., 2001). Theoretically, as members feel more confident in their team's capabilities, "they are more motivated to work hard for the team, persist in the face of? obstacles, and are willing to accept more difficult challenges" (Zaccaro et al., 2001, p. 467). Despite the potential to interweave literature on team efficacy and group processes that Zaccaro and colleagues' words suggest, Marks (1999) noted that the majority of the literature on team efficacy has focused on establishing links to performance, rather than considering team efficacy as a potential determinant of team processes. Connections between processes other than goal-setting that a team might use to productively approach task work (e.g., information collection, strategy formulation, etc.) are notably absent in the literature linking team efficacy to performance. Marks (1999) did find that highly efficacious teams seemed to change their strategies more than less efficacious teams when faced with a challenging situation. Other team processes, however, may also be important to consider. Marks et al. (2001) suggested that teams progress through transition phases, wherein "teams focus primarily on evaluation and/or planning activities to guide their accomplishment of a team goal or objective" (p. 364). During this time, processes such as mission analysis, goal specification, and strategy formulation and planning typically occur. Marks et al. (2001) categorize a second important set of processes as action processes, which consist of activities leading directly to team goal accomplishment. These activities typically include monitoring progress
47
toward goals, systems monitoring, team monitoring and backup responses, and coordination activities. Though no study to my knowledge has ever directly examined the relationship between team efficacy and action or transition processes, Edmondson (1999) found that team efficacy was related to team learning behavior, which consisted of processes such as feedback seeking, experimentation, and discussion of errors. It may be that teams with higher levels of efficacy are more likely to scrutinize their own actions in an attempt to improve in the future. Similarly, teams with higher levels of efficacy may be more likely to formulate strategy and plan for the future; additionally, they may be more likely to carefully examine their progress towards their goals and monitor themselves. Given the empirical evidence that does exist on the relationship between team efficacy and team processes, and given the logical reasoning that teams that are more confident in their own ability to achieve certain performance levels will be more likely to take action to achieve those performance levels, it is likely that team efficacy is positively related to important team processes. Therefore, Hypothesis 5a: Team efficacy is positively related to team action and transition processes that are geared towards helping the team reach its goals.
The Mediating Role of Team Efficacy in the Transformational Leadership Processes Relationship
Team
Transformational leaders motivate team members to be inspired through effectively presenting a vision for the future; however, as noted above, transformational leaders are typically not directly involved on a micro level enforcing specific goal-setting
48
or micro-managing exactly what processes teams utilize. Moreover, transformational leaders inspire team members to identify with and care more about the team itself; transformational leaders also typically build the team's confidence through verbal persuasion and encouragement. As team members become more engaged in the activities of the team and as their team efficacy reaches new heights due to the efforts of the transformational leader, it is likely that teams will exhibit subsequent upswings in the extent to which constructive team action and transition processes are utilized. Therefore, transformational leadership is likely to indirectly influence team action and transition processes through influencing team efficacy. Hypothesis 5b. The relationship between transformational leadership and team action and transition processes is mediated by team efficacy.
Team Goals and Team Action and Transition Processes
Performance
As Durham, Knight, and Locke (1997) noted, goal-setting is a well-established motivational technique for individuals and, to a certain extent, for groups as well (Locke & Latham, 1990; O'Leary-Kelley et al., 1994; Weldon & Weingert, 1993). Both individuals and teams that set higher goals exhibit higher levels of performance attainment; additionally, goals have been found to mediate the relationship between efficacy and performance, both at the individual and group level (Durham et al., 1997; Locke & Latham, 1990). Therefore, it is likely that higher team goals will be related to higher levels of team performance. Action and transition processes may also influence performance, though there is no direct empirical evidence to date to support this contention. However, some of the
49
literature on goal-setting has alluded to the importance of other team processes in predicting performance. Durham et al. (1997, p. 206), for example, wrote that the effects of goal-setting are "most reliable when the goals are specific and difficult, when there is commitment to the goals, when feedback is available regarding the goal process, high self-efficacy, and knowledge regarding how the goals can be achieved." According to Durham et al. (1997), this knowledge really consists of the knowledge of the strategies needed to perform the task so as to attain the goal, which supports the idea that effective strategy formulation (a transition process) is a critical group process to examine. For newly formed teams that are operating in highly turbulent and unfamiliar environments, or on an unknown and complex task, the importance of developing and using a strategy to cope effectively with the demands of the situation given the team's constraints is doubly critical. The team processes of information collection, analysis, strategy development, and strategy implementation should all also be positively related to objective team performance. It is necessary for teams to have knowledge of critical information in order to be effective; to the extent that teams make an effort to collect this information, the more positive the resulting performance should be. Similarly, teams that actively analyze, develop strategies for approaching the task, and implement those strategies accordingly should experience more positive performance than teams that do not actively engage in these processes. It is theoretically possible, of course, for a group that does not exhibit any of the above processes to do well; however, it is highly unlikely, as work teams that do not engage in these types of activities on complex tasks would be relying on a great
50
deal of luck. These processes, then, are hypothesized to have a positive impact on team performance. Hypothesis 6a: Team goals and team action and transition processes are positively related to team performance.
The Mediating Roles of Team Goals and Team Process in the Team Efficacy Performance Relationship The relationship between team efficacy and objective team performance has been solidly established by a number of authors, though this relationship is not quite as strong as the relationship between individual-level self-efficacy and individual performance (Collins & Parker, 2002). Gist and Mitchell (1992), Peterson, Mitchell, Thompson, and Burr (2000), Silver and Bufiano (1996), and Zander and Medow (1964), among others, have all found evidence that suggests that higher levels of group efficacy are related to higher levels of performance. In addition, Bandura (1997) and Lindsley et al. (1995) suggested that efficacy may exhibit a direct relationship with performance. The team efficacy-performance relationship, however, may be partly an indirect one. It is likely that team efficacy impacts performance partially through its influence on team goals, as goals provide an immediate focal point on which to concentrate motivation (Locke & Latham, 1990). It is also likely that team efficacy impacts performance partially through its influence on team action and transition processes. Teams that are more confident in their ability to successfully complete the task should be more likely to focus effort on how to do it; they should be more involved in analyzing feedback results, seeking further information, and adjusting team strategy, which are all critical processes
51
necessary to achieve high levels performance when dealing with complex tasks in dynamic environments. Hypothesis 6b: The relationship between team efficacy and team performance is mediated by team goals and team action and transition processes.
Conclusion In this chapter, I presented a model that suggested that a team leader's core selfevaluations and midpoint feedback the team receives influence team efficacy, goals, processes, and performance through influencing leadership efficacy and transformational leadership. Team leader core self evaluations and midpoint feedback the team receives were hypothesized to be related to leader efficacy, which in turn was hypothesized to be related to transformational leadership. Transformational leadership, in turn, was hypothesized to be an antecedent of team efficacy. Finally, team efficacy was proposed to be an antecedent of both team action and transition processes and team goal level; the latter two variables were predicted to have a direct impact on team performance. This general conceptualization of the effects of leader core self-evaluations is consistent with the findings suggesting that core self-evaluations may have an impact on job performance through their impact on individual motivation. In Chapter 4, I discuss the research methods, including sample, experimental task, procedure, and aggregation and analytic strategy.
52
Chapter 4 Research Methods
This chapter presents data collection and analysis methods for this research. In the first section, I describe research design, including a description of the task; in the second section, I describe the sample, in addition to discussing the schedule of the collection of measures; and in the third section, I describe analytic procedures.
Research Design In order to test the hypotheses proposed in Chapter 3, a number of methodological issues had to be addressed. It was important for the study to feature a task as its centerpiece that was complex, engaging, provided objective performance assessments, and involved both ongoing and critical midpoint feedback. Additionally, the research required access to an appropriate sample of individuals whose results would be generalizeable. This study was developed with these design issues in mind. As a broad overview of when and how the data were collected, team members' core self-evaluations were assessed prior to the start of the simulation. Objective performance data was provided by the task itself, which was a business simulation that translated teams' decisions into performance. Survey data regarding leader efficacy, transformational leadership, team efficacy, team goals, and team action and transition processes were gathered from participants immediately following the critical midpoint of the task. Examining survey data collected at this point in time was particularly useful for several reasons. First, collecting data after teams had worked together during the first half
53
of task completion allowed for development of team level constructs through shared history of interaction and task feedback. Second, central concepts and relationships in the model hypothesized in Chapter 3 involved the response to critical midpoint feedback; it was therefore important to assess variables of interest after critical midpoint feedback had been received. Third, due to the fact that leaders were emergent rather than formally designated, it was important to allow team members time to gravitate toward roles in which they felt comfortable; assessing variables after the critical midpoint allowed team members the entire first half of task completion to do exactly that. Also, for the purposes of this study, individuals were placed on teams such that team composition in terms of gender and race/ethnicity was balanced across groups (demographic information had been provided to the individual who split the participants up into teams; the project director ensured that there were at least two women and two racial or ethnic minorities—but not more than three of either—in each group of five participations). Core self-evaluations, however, were allowed to randomly vary. Individuals were required to participate as part of the University of Maryland's MBA program's orientation activities for first-year students. Additionally, participants were given one incentive for participating: a summary of the findings tailored to their specific team and implications of those findings. Description of the task. MARKETPLACE (Cadotte, 2003), the task that was used in this study, is a complex, computer-based interactive team decision-making business simulation. The simulation is based on the history of the personal computer industry. During the course of the simulation, participating teams worked to build an entrepreneurial firm, experiment with strategies, make decisions regarding the major
54
business functions (e.g., marketing, manufacturing, logistics, human resources, finance, and accounting), and compete with other teams in an extremely realistic and detailed virtual business world. The simulation required each participant to imagine being a member of the top management team of a start-up computer firm. In their role as top management team members, participants made a series of strategic decisions over a period of eight "quarters" (each quarter representing a single decision cycle). Teams received performance feedback at the end of each quarter following Quarter 3 (the first quarter that teams actually sold products). Teams were pitted against each other in four "universes" of 10 teams each; teams within universes were directly competing for the same customers, though each team could choose to capture any number of five market segments of business users. These five segments ranged from the "Cost Cutter" segment (a large segment that looked for very easy-to-use computers for basic office applications) to the "Mercedes" segment (the smallest segment overall that looked for a high-performance computer to use in sophisticated engineering and manufacturing applications). Each quarter of play in MARKETPLACE involved new information and a set of decisions that built upon each other. The simulation unfolded to follow the normal lifecycle of a start-up business, though there was a natural break-point halfway through the simulation where more performance feedback and extra funding from "venture capitalists" was received. Quarters 1 and 2 essentially involved "setting up shop." In Quarter 1, team members got organized, assigned corporate responsibilities, named the company, and purchased market research. Team members also sold stock to themselves
55
during this period as owners in order to raise capital for initial operation. In Quarter 2, market research arrived, and teams analyzed the data and decided on an overall business strategy, including corporate and performance goals, target markets, and strategic direction. Several tactical decisions also had to be made during this period, which focused on brand design, plant location, production capacity, and sales office location. Teams finished Quarter 2 by answering the first survey I distributed, which included team efficacy and team goal measures (used in my analysis to derive an ability measure as a control; specifics are discussed in the below section regarding the measures). Quarter 3 involved testing the market, as fledgling firms tried out marketing strategies (including brands, prices, ad copy, media campaigns, and sales staffing). In addition, teams decided on hiring policies and production processes for the quarter, then scheduling production for the quarter. Once again, teams purchased market research in order to discover customer reactions to marketing decisions and to find out what the competition has done. Lastly, in this quarter, teams forecasted market demand and simulated their production operations given the supply chain decisions teams had made prior to that point. At the end of this quarter, MARKETPLACE made available cash flow and income statements and balance sheets in order for teams to evaluate the potential financial impact of the firm's first quarter of total business operation. In Quarter 4, team members received the market data they purchased in Quarter 3 regarding the test market and received manufacturing and accounting data from the first quarter of sales. Once again, at the end of this quarter, MARKETPLACE made available cash flow and income statements and balance sheets in order for teams to evaluate their
56
performance. Teams could then adjust the firm's strategies and tactics if necessary after examining this data. Quarter 5 was a pivotal quarter—the halfway point—with the critical midpoint feedback occurring here. In this quarter, teams had to obtain capital from venture capitalists in order to substantially expand their position in the market (which would occur by investing in research and development, new sales outlets, factory capacity, logistics, and/or employee recruitment and retention programs). Prior to making their pitch to the venture capitalists, teams received a review of marketing, human resource, manufacturing, and accounting data from Quarter 4. Then they formulated a one-year business plan to obtain outside funding. The plan had to include an assessment of market opportunities, a review of performance to date, and a strategy for the second year of business operations. It had also to include coordinated marketing, manufacturing, human resources, and financial strategies, in addition to a tactical plan that included the sequencing and timing of events to achieve the firm's goals. Finally, the plan had to include historical financial statements from the first four quarters (year 1) in addition to pro-forma statements for the second four quarters (year 2). The simulation was designed such that at this point teams provided this information as a formal presentation during the middle of the day allocated for Quarter 5 play to a panel of venture capitalists (volunteers who were chosen from the Robert H. Smith School of Business community of alumni, executives, and faculty), who on the basis of what the teams presented offered them advice and some amount of "virtual" funding. Each team presented to one panel of venture capitalists; there were eight panels of seven venture capitalists, each of which viewed five team presentations (each set of
57
five teams was in the same universe). After each team completed its presentation, venture capitalists completed a 6-item questionnaire that reflected their impressions of each team's potential and gave these questionnaire responses to the teams. Teams then met with the venture capitalists and had a chance to receive additional funding. All the venture capitalists had the same amount to offer in terms of funding; they were making their investment decisions based on the teams that they believed were in the best position to perform well during the remainder of the simulation. Although teams and venture capitalists negotiated over the funding level and price of stock, the virtual money each team could receive from the venture capitalist panel was capped at five million dollars; the funding cap ensured that the simulation could continue into the second year of play without completely eliminating any teams. It is important to note here that the amount of capital that teams received from the venture capitalists was not a pure measure of whether the venture capitalists were willing to invest money in any given team. Teams approached their negotiations with the venture capitalists with different strategies in mind. Some teams were, indeed, interested in obtaining as much capital as possible, no matter what their stock price. Other teams were more cautious about diversifying ownership; rather than sell a lot of stock, these teams were interested in driving a hard bargain for the price of their stock. Finally, some teams were simply not as interested in selling their stock to receive funding from the venture capitalists, no matter how high their stock price. Due to the many strategies utilized by different teams, the amount of capital received by each team was, again, not a pure measure of the interest venture capitalists had in investing in that team's stock.
58
Receiving feedback and capital from the venture capitalists was considered to be the critical midpoint of the simulation. Teams had received not just objective performance results from Quarters 1-4, but also outside assessments of past performance and expected future potential. The situation provided an opportunity for teams to reflect upon whether they were on track, based on previous performance trajectory, to meet their performance goals that had been set in Quarter 2. Teams finished Quarter 5 by allocating the newly procured funding to different investment opportunities of their choice (e.g. research and development, new sales outlets, etc.) and then responding to the second survey for the purposes of this research, which included leader efficacy, transformational leadership, team efficacy, team goal, and team action and transition processes. Quarters 6-8 were a time in which the team continuously monitored and adjusted different areas of the firm (such as marketing, manufacturing, etc.). Competition also heightened during this period, as other firms began introducing new technology, more reliable products, better prices, etc. Teams during this time also had the opportunity to explore new relationships through cross-licensing and supply outsourcing. At the end of Quarter 8, the team prepared another report to assess strategy and performance in year 2 of operations and to give an assessment of how well the firm was prepared to take on the challenges of the future. Table 2 provides a summary overview of the sequence of events in the simulation, in addition to describing what variables were collected when and what were their uses. As noted above, I considered a number of parameters when choosing the appropriate research design for this study. MARKETPLACE, as the centerpiece task, provided an excellent context in which to test my theoretical model. The simulation was
59
Table 2. Sequence of Events, Measures Collected, and Measure Uses in MARKETPLACE Simulation Quarter 1 Team Activities Measures Collected Before Quarter 1: Core SelfEvaluations (collected pre-ELM) Independent variable Measure Use
• • • •
•
2
team gets organized, assigns corporate responsibilities, sets broad goals team names the company team purchases initial market research team members sell stock to themselves during this period as owners in order to raise capital for initial operation no performance feedback yet available market research arrives team engages in data analysis team makes decisions regarding overall business strategy, including corporate goals, target markets, and strategic direction team makes tactical decisions: brand design, plant location, production capacity, and sales office location no performance feedback yet available
• • •
End of Quarter 2: Team Efficacy Used to create ability residual Used to create ability residual
Team Goal
•
•
3
•
team tests market: tries its marketing strategy (including brands, prices, ad copy, media campaigns, and sales staffing) • team decides on hiring policies and production processes for the quarter, 60
End of Quarter 3: Team Performance (profit) Used to create ability residual— and provided to teams as feedback
and then schedules production for the quarter • team purchases market research in order to discover customer reactions to marketing decisions and to find out what the competition has done • team forecasts market demand and simulates its production operation given the supply chain decisions • cash-flow and income statements available at end of quarter, in addition to a balance sheet to help team members evaluate potential financial impact of the firm's first quarter of total business operation 4
• • •
team receives market data purchased in Quarter 3 regarding the test market team receives manufacturing and accounting data from the first quarter of sales team uses data to evaluate performance and adjust the firm's strategies and tactics if necessary
End of Quarter 4: Team Performance (profit) Used to create ability residual— and provided to teams as feedback
5
•
team receives review of marketing, human resource, manufacturing, and accounting data from Quarter 4 • team must formulate a oneyear business plan to obtain outside funding. Plan includes: assessment of market opportunities, review of performance to date, strategy for the Year 2 (coordinated marketing,
End of Quarter 5: VC Rating Independent variable Independent variable
Leader Efficacy
Transformational Independent Leadership variable—and used to determine leader
61
• • • •
manufacturing, human resources, and financial strategies), tactical plan (with sequencing and timing of events to achieve the goals), historical financial statements from year 1 in addition to proforma statements for year 2 team physically presents this information to venture capitalists VC panel offers team advice and some amount of funding team may negotiate with the panel team allocates funding to different investment opportunities (e.g. research and development, new sales outlets, etc.) team monitors and adjusts different areas of the firm (such as marketing, manufacturing, etc.) competition heightens during this period as other firms begin introducing new technology, more reliable products, better prices, etc. team has the opportunity to explore new relationships through cross-licensing and supply outsourcing
Team Efficacy
Independent variable
Team Goal
Independent variable Independent variable
Team Processes
6-8
• •
End of Quarter 8: Team Performance (profit) Dependent variable
•
62
meant to introduce teams to a constantly changing environment to which they had to adapt in order to survive. Also, the critical midpoint of the simulation was very clearly defined. Variance in performance at the end of the simulation was assured due to the fact that teams were directly competing against each other. Indeed, some teams finished with extremely high performance ratings, while others drifted into the final quarter bankrupt (though they did not have to stop play).
Sample Dr. Joyce Russell, coordinator of the Smith School's full-time MBA program Educational Learning Module (ELM)1 project, granted me access to the 2002 ELM. The ELM took place during the last week of August 2002; all first year full-time MBA's were included in the project. As such, 198 individual participants competed in 38 teams of five members and two teams of four members. The sample was appropriate for a number of reasons. First, teams were made up of first-year MBA's who devoted a full, uninterrupted week to the ELM project, playing two quarters each day. Teams worked full days in the hopes of gaining an edge over the competition. In addition, teams were divided so as to ensure there were at least two women and two racial/ethnic minorities (but not more than three of either), including foreign students, on each team; it was likely that differences in performance of these teams would be a result of leadership or team process factors rather than as a result of any demographic differences.
1
The ELM is designed to be the first activity first-year MBA students complete when they arrive on campus; it is intended to give them both an opportunity to receive some hands-on business decision-making experience and to gain a valuable experience in team interaction. MARKETPLACE is the simulation program used in the ELM.
63
Measures Dependent Variable Team Performance. Table 3 provides a summary of concepts and measures used in this study. Teams received financial information regarding their operating profit at the end of every quarter following Quarter 3. Operating profit was the most critical and salient measure of whether a team's strategies and decisions were effective; it was generated by the simulation after team decisions had been inputted. Therefore, Quarter 8 operating profit (the profit with which teams ended the simulation) was used as the basis of the team performance measure. However, in order to control for team ability, a standardized residual score was created from a regression that considered Quarter 8 performance as a dependent variable and the team ability control variable, described below, as the independent variable; it was this standardized residual score that was the actual dependent variable used in all the analyses.2 More information regarding the mean and standard deviation of this variable, in addition to zero-order correlations with other study variables, can be found in Chapter 5. I also examined Quarter 8 rank (a number between 1 and 10 reflecting each team's relative final position, based on operating profit, with respect to the other teams in the same universe) as a measure of final performance, using a similar procedure to that which is described above to control for ability. Due to the strong similarity of the results attained for each measure of performance, however, I will focus exclusively on Quarter 8 profit as the team performance measure in Chapter 5.
2
All analyses were also run with a separate ability control measure and the Quarter 8 profit measure as the dependent variable; these analyses yielded virtually identical results. For purposes of parsimony, I only report the use of and results associated with the standardized residual final performance measure as the dependent variable.
64
Table 3. List of Concepts and Measures Used
Concept Team Performance
Measures Standardized residual of regression with team profit in Quarter 8 as dependent variable and ability (i.e., standardized residual of first half performance regressed on team efficacy and goal assessed after Quarter 2) as the independent variable I am confident I get the success I deserve in life. Sometimes I feel depressed. (r) When I try, I generally succeed. Sometimes when I fail I feel worthless. (r) I complete tasks successfully. Sometimes, I do not feel in control of my life. (r) Overall, I am satisfied with myself. I am filled with doubts about my competence. (r) I determine what will happen in my life. I do not feel in control of my success in my life. (r) I am capable of coping with most of my problems. There are times when things look pretty bleak and hopeless to me. (r) This team acts like a unified team. This team seems to have strong leadership. I might invest some money in this team. This team seems to have a clear strategy for the future. This team was responsive in answering questions. This team's presentation was effective. During Quarter 5 of the simulation, I had a high degree of confidence in my ability to? ?steer this team in a successful direction. ?get the team to develop viable strategies. ?inspire others to be motivated to do well. ?build this team's sense of spirit and cohesiveness. ?get the people on this team to be excited about working together.
Core Self-Evaluations
Venture Capitalist Feedback
Leader Efficacy
Transformational Leadership
Leadership behavior ratings matrix—please see Figure 4
65
Team Efficacy
Please take a moment to discuss and agree upon what your team's confidence levels are. What is your team's confidence level (ranging from 0% to 100%) in being able to attain the following performance rankings in your universe by the end of the game? Top 9 or better: _______ Top 8 or better: _______ Top 7 or better: _______ Top 6 or better: _______ Top 5 or better: _______ Top 4 or better: _______ Top 3 or better: _______ Top 2 or better: _______ Top 1 or better: _______
Team Goal
Please take a moment to discuss and agree upon what your team's goal is in terms of end ranking within your universe. At the conclusion of this simulation, our team's goal is to be ranked ______ out of the 10 teams in our universe (1 = top performing team; 10 = lowest performing team).
Team Processes
During Quarter 5 of the simulation, the members of my team? ?set specific goals for our team to accomplish. ?collected and analyzed all information we could before making decisions. ?monitored changing environmental conditions. ?developed and implemented a comprehensive strategy. ?adjusted our strategy in keeping with what we had learned. ?coordinated activities among team members. Standardized residual of regression with first half (Quarters 3 and 4) profit as dependent variable and team efficacy and team goal in Quarter 2 as independent variables
Team Ability
66
Control Team Ability. Performance prior to the critical midpoint— an average of operating profit information provided by the simulation in Quarters 3 and 4—was used as the basis for the measure of team ability. In order to control for motivational variables that may have been affecting performance, I conducted a regression analysis, with profit prior to the critical midpoint (an average of Quarter 3 and Quarter 4 profits) as the dependent variable and Quarter 2 team efficacy and team goals as the independent variables3. Because the residuals from these regressions represented observed team performance early in the simulation, controlled for team efficacy and goals, they served as measures of actual team ability. These residuals were used as independent variables in the regression to create the Quarter 8 performance residuals—the standardized final performance residuals—to control for team ability.
Independent Variables Venture Capitalist Feedback. As noted in the description of the critical midpoint transition (Quarter 5), venture capitalists completed a six-item, Likert-type assessment of each team immediately after viewing each team's presentation of their past performance and future strategies and goals. Sample items were as follows (1 = strongly disagree; 5 = strongly agree): "This team acts like a unified team"; "I might invest some money in this team"; and "This team seems to have a clear strategy for the future." The inter-item reliability for the six-item measure as assessed by Cronbach's alpha was .82; the withingroup agreement as assessed by rwg (James, Demaree, & Wolff, 1984; 1993) was .86;
3
In creating the rank-based ability measure, I used rank prior to the critical midpoint, as determined by an average of Quarter 3 and Quarter 4 profit.
67
Lindell's (Lindell & Brandt, 1999) alternative within-group agreement statistic was .83; ICC(1) (Bartko, 1976; James, 1982; McGraw & Wong, 1996) was .45; and ICC(2) (Bartko, 1976; James, 1982) was .79. These agreement and reliability statistics were at levels that allowed me to meaningfully average venture capitalists' individual assessments to reflect an overall venture capitalist assessment for each team. Team Leader Core Self-Evaluations. Core self-evaluations was measured by selfreport using the 12-item Core Self-Evaluations Scale, developed by Judge, Erez, Bono, & Thoreson (2003). Judge et al. (2003) found substantial support for the validity of the scale; it is reliable and displays acceptable levels of internal consistency and test-retest reliability. In addition, inter-source (self-significant other) level of agreement is comparable to that of other personality measures. Sample items are as follows (assessed on a 1-5 Likert scale; 1 = strongly disagree, 5 = strongly agree): "I am confident I get the success I deserve in life"; "Sometimes I feel depressed" (r); "When I try, I generally succeed"; and "Overall, I am satisfied with myself." Cronbach's alpha for the scale was .80, suggesting adequate inter-item reliability. Leaders were emergent rather than formally designated. I first used the transformational leadership scale score (discussed below) collected after the critical midpoint transition to identify which team member was considered to be the team leader by his/her teammates during the critical midpoint period. I then utilized that individual's core self-evaluations score as the team leader's core self-evaluations score. Team Leader Efficacy. I developed a measure for this concept that assessed the extent to which individuals felt confident in assuming different aspects of the team leadership role. Just as was noted for the assessment of the leader's core self-evaluations,
68
I was only interested in the leader efficacy scale scores of the individual who was recognized by his/her teammates as being the emergent leader during the critical midpoint transition period. The 5-item, Likert-type scale was tailored to refer to the level of efficacy experienced by the leader during the critical midpoint. Sample items are as follows (1-5 Likert scale, 1 = strongly disagree, 5 = strongly agree): "During Quarter 5 of the simulation, I had a high degree of confidence in my ability to?" 1) "?steer this team in a successful direction"; 2) "?get the team to develop viable strategies"; 3) "?inspire others on this team to be motivated to do well." The scale demonstrated a high level of inter-item reliability; Cronbach's alpha was .90. Team Leader's Transformational Leadership. All participants completed a measure assessing their perceptions of their teammates' transformational leadership. For each team, I created a matrix (see Figure 1) with space for team member names horizontally across the top and three items listed vertically down the left-hand column. The three items were intended assess transformational leadership as conceptualized by Bass's (1985) four dimensions: idealized influence, inspirational motivation, intellectual stimulation, and individual consideration. Each participant rated all his/her team members on the extent to which they exhibited the transformational leadership behaviors listed. Team members assessed each other following the critical midpoint period (end of Quarter 5); the emergent leader for the team was defined as the individual who received the highest score from his/her teammates. I used that individual's score to serve as the leader's transformational leadership rating, and—as noted above—I used that individual's core self-evaluations and leadership efficacy scores. Team members were
69
Figure 4. Transformational Leadership Ratings Matrix
Directions: Please write each of your fellow team members' names on the diagonal lines across the top of the grid. Then, rate each team member on the extent to which he/she exhibited the behaviors listed below in the first two quarters of the simulation. Do not rate yourself.
1=never; 2=very infrequently; 3=sometimes; 4=often; 5=always Inspired team members by demonstrating confidence in the team and showing optimism. Encouraged other team members to be creative problem solvers. Demonstrated sensitivity to and support of the needs and concerns of other team members.
70
instructed with the following directions to fill out the matrix: "Please write each of your fellow team members' names on the diagonal lines across the top of the grid. Then rate each team member on the extent to which he/she exhibited the behaviors listed below during Quarter 5 of the simulation." Sample items team members used to rate each other are as follows (1-5 Likert, 1 = Never, 5 = Always): "Encouraged others to be creative problem solvers" and "Inspired team members by demonstrating confidence in the team and showing optimism." Cronbach's alpha for this 3-item measure was .89, suggesting a high level of inter-item reliability. Additionally, rwg was .97, Lindell's alternative rwg was .92, ICC(1) was .52, and ICC(2) was .85, suggesting outstanding levels of within-team agreement and inter-rater reliability for this measure that justified the aggregation of this measure across raters. In order to ensure that the transformational leadership measure was stable, I also collected the same measure at the end of the simulation. 70% of the individuals who received the highest transformational leadership scores following the midpoint feedback also received the highest transformational leadership scores at the conclusion of the simulation, suggesting that the measure indeed was stable and was being consistently utilized by raters. Therefore, each individual's transformational leadership score reflected an aggregation of their peer's ratings. Team Efficacy. A scale was created to assess this measure based on the recommendations of Locke, Frederick, Lee, and Bobko (1984) and others (e.g., Gibson, Randel, & Earley, 1999; Gist & Mitchell, 1992; etc.). This measure was tied directly to the team's belief in its capabilities to achieve 10 specific performance rankings. Teams were asked to indicate their confidence (using a 100 point scale, where 0 = certainly cannot be achieved and 100 = certainly can be achieved) as to whether their team will
71
finish in the top 9 or better, top 8 or better, top 7 or better, etc. all the way up to top 1 in their respective 10-team universes, based on their assessments of the team's capabilities. Teams completed this scale by consensus—directions were as follows: "Please take a moment to discuss and agree upon what your team's confidence levels are. Please write your team's assessment of confidence below." The team efficacy score that was used to test hypotheses was assessed at the end of the midpoint transition period (end of Quarter 5). Cronbach's alpha of this scale was .94, suggesting high levels of inter-item agreement. Due to the fact that the scores were captured through group consensus, intra-team agreement and reliability measures could not be calculated. The team efficacy scale was also captured following Quarter 2 of the simulation; that initial efficacy score was used in the calculation of the team's ability measure. Cronbach's alpha for the post-Quarter 5 collection of team efficacy was .94. Cronbach's alpha for team efficacy post-Quarter 2 was .92. Team End Goal. Teams were asked to provide a response to one question regarding their team's end goal after the critical midpoint transition period. The team completed this item by consensus. Directions were as follows: "Please take a moment to discuss and agree upon what your team's goal is in terms of end ranking within your universe." The item itself read as follows: "At the conclusion of this simulation, our team's goal is to be ranked ____ out of the 10 teams within our universe (1 = top performing team; 10 = lowest performing team)." Team end goal was also assessed in the same manner at the end of Quarter 2; this initial goal measure was used as a control to calculate the team ability measure.
72
Team Processes. Team members provided individual responses to questions developed for this dissertation that were intended to reflect Marks et al.'s (2001) action and transition processes, discussed in Chapters 2 and 3. At the end of the critical midpoint transition period (end of Quarter 5), team members assessed the extent to which they believed that their team had engaged in action and transition processes reflected by a six-item, Likert-type scale. Sample items (1 = strongly disagree to 5 = strongly agree) are as follows: "The members of my team?" (1) "?set specific goals for our team to accomplish"; (2) "collected and analyzed all information we could before making decisions"; (3) "developed and implemented a comprehensive strategy"; and "coordinated activities among team members." Cronbach's alpha was .83, suggesting an acceptable level of inter-item agreement; rwg was .98; Lindell's alternative rwg measure was .88; ICC(1) was .26, and ICC(2) was .64. These statistics suggested that the items could be meaningfully aggregated into a score that reflected the team's action and transition processes.
Analytic Strategy I used path analysis through covariance structure analysis in Bentler's (1989) EQS program to test the strength of the relationships among leader core self-evaluations, venture capitalist feedback, leader efficacy, transformational leadership, team efficacy, team processes, team goals, and team performance (hypotheses 1, 2a, 3a, 4a, 5a, and 6a). Data from the 40 groups were input into EQS in the form of a correlation matrix with accompanying standard deviations. Model structural paths were evaluated for significance, and goodness of fit of the overall model was assessed by several fit indices:
73
the comparative fit index (CFI; .90 and above is considered to be indicative of acceptable model fit; Bentler, 1989), the LISREL goodness of fit index (GFI; .90 and above is considered to be indicative of acceptable model fit; Joreskog & Sorbom, 1989), the Bentler-Bonett Normed Fit Index (NFI; .90 and above is considered to be indicative of acceptable model fit; Bentler, 1989), and the Bentler-Bonett Non-Normed Fit Index (NNFI; .90 and above is considered to be indicative of acceptable model fit; Bentler, 1989). Because the model tested 16 parameters with a sample size of 40, which falls short a great deal of the rule-of-thumb requirement that there should be at least 5 cases per parameter tested in a structural model, I also tested the model using traditional path analysis in SPSS regression. This regression revealed virtually identical results, suggesting that the parameter estimates derived from structural equations modeling were stable. Klein, Conn, and Sorra (2001) advocated the use of both regression and covariance structure modeling in their study, which also considered a particularly small sample size (n = 39); they also noted that in cases of such small sample size, fit statistics for the overall model should be interpreted very cautiously. Chi-squared difference tests (James, Mulaik, & Brett, 1982) were conducted to determine the degree to which hypothesized mediators indeed acted as mediators, as proposed by hypotheses 2b, 3b, 4b, 5b, and 6b. I followed Baron and Kenny's (1986) guidelines for testing mediation in this set of analyses in conjunction with a modelcomparison approach. I compared the chi-squared statistic of my hypothesized model with chi-squared statistics of models that included the hypothesized mediated paths in addition to direct paths. If the difference in chi-squared statistics was significant, the
74
model with the direct path was a better fit with the data, indicating either partial mediation or no mediation. The results from these mediation tests also mirrored the results from traditional SPSS regression mediation tests, once again supporting the stability of the model in EQS.
75
Chapter 5 Results
The following report of results begins with the results of the covariance structure analysis of the original model and mediation analyses. I then present a revised version of the model and discuss results of exploratory analyses.
Descriptive Statistics Table 4 presents the descriptive statistics, scale reliabilities, rwgs, and ICCs of the study variables; Table 5 presents the intercorrelations among study variables. In both cases, team goal is reverse scored (a score of 10 indicates the team's goal was to end in first place within its universe). Also, team performance is a standardized variable representing the residual of a regression of ability on Quarter 8 (end) profit.4 Additionally, the results reported here are associated exclusively with the use of final (Quarter 8) profit as the basis for the dependent variable. As noted in the measures section, all analyses were also run with Quarter 8 rank as the dependent variable. Results for each set of analyses were virtually identical; therefore, only results associated with Quarter 8 profit are reported here.
4
The use of this residual, as noted in the previous chapter, was intended to control for any differences in ability across teams. In the following chapter, I report only the results associated with this performance measure. Analyses were also run, however, using a separate variable for ability and the raw final profit scores for each team; due to the similarity of the two sets of results, and the relative simplicity associated with reporting the results associated with the residual performance measure, I have reported only the results associated with the residual performance measure.
76
Table 4. Descriptive Statistics, Scale Reliabilities, Rwgs, and ICCs
Number N 1. Leader CSE's 2. VC Feedback 3. Leader Efficacy 4. Trans. Leadership 5. Team Efficacy 77 6. Team Goals 7. Team Processes 8. Team Performance 40 40 40 8.60 4.07 0.00 2.16 .53 .99 1 6 --.82 --.98 -38 40 40 40 40 M Lindell 5.46 3.68 3.92 4.31 76.26 SD .67 .35 .62 .34 20.77 12 6 5 3 9 .80 .82 .86 .89 .94 rwg -.86 -.97 --
rwg ICC(1) -.83 -.92 --.88 --
ICC(2) of Items -.45 -.52 --.26 --.79 -.85 --.64 --
Notes. Team goal is reverse scored (scores of 10 indicate goals of first place). Team performance is a standardized variable representing the residual of a regression of ability on Quarter 8 (end) profit.
Table 5. All Study Variable Intercorrelations.
1 1. Leader CSE's 2. VC Feedback 3. Leader Efficacy 4. Transformational Leadership 5. Team Efficacy 6. Team Goals 7. Team Processes 78 Notes. * p < .05 ** p < .01 Team Performance --.03 .40* .18 .16 .20 .39* .29
2 -.33* .27 .38* .33* .37* .32*
3
4
5
6
7
8
-.39* .43** .23 .70** .26 -.24 .11 .28 .19 -.58** .52** .16 -.54** .20 --8. .37* --
Team goal is reverse scored (a score of 10 indicates the team's goal was to end in first place within its universe. Team performance is a standardized variable representing the residual of a regression of ability on Quarter 8 (end) profit.
Results I tested the hypothesized model presented in Chapter 3 by analyzing the covariance matrix using EQS; the results of this test are depicted in graphically in Figure 5 below. Hypothesis 1 suggested that the team leader's core self-evaluations and feedback the team receives at critical points would simultaneously have an impact on the leader's own efficacy to lead effectively. This hypothesis was supported; paths leading from the leader's core self-evaluations and team feedback to leader efficacy were both positive and significant (.38, p < .01, and .35, p < .01, respectively). The percentage of variance explained in leader efficacy was 26% (R2 = .26). Hypothesis 2a postulated that the leader's efficacy would be related to transformational leadership behaviors. This hypothesis was also supported; the path leading from leader efficacy to transformational leadership was significant (.39, p < .01). The percentage of variance explained in transformational leadership was 16%. Hypothesis 2b postulated that leader efficacy would mediate between both the leader's core self-evaluations and venture capitalist feedback and subsequent transformational leadership. In order to test for the presence of mediation, I separately tested two models with direct paths added between core self-evaluations and transformational leadership and between venture capitalist feedback and transformational leadership, respectively. Neither direct path model proved to be a better fit with the data ( of freedom, NS;
2 2
= .05 on one degree
= 1.01 on one degree of freedom, NS; respectively). Therefore,
hypothesis 2b was also supported.
79
Figure 5. Results of the Hypothesized Model.
Leader CSE's 80 VC Feedback
.38** Leader's Efficacy .35**
.39** Transformational Leadership
.24
.58*** Team Efficacy .52***
Team Goals
.00
Team Performance Team Processes
.38*
Notes. Note that the values are standardized parameter estimates *** p < .001 ** p < .01 * p < .05
Hypothesis 3a concerned the relationship between the leader's transformational leadership and team efficacy. This hypothesis was not supported (.24, NS); transformational leadership was not significantly related to team efficacy (with R2 = .06). Hypothesis 3b, which suggested that transformational leadership would mediate the relationship between leader efficacy and team efficacy, was also not supported, as there was no evidence of a direct effect between transformational leadership and team efficacy. Hypothesis 4a suggested that team efficacy would be related to team goal. This hypothesis was supported; team efficacy was significantly related to team goal (.58, p < .001, R2 = .34). Hypothesis 4b suggested that team efficacy would mediate the relationship between transformational leadership and team goal. To begin, transformational leadership had to be significantly related to team efficacy (hypothesis 3a); because hypothesis 3a was not supported, hypothesis 4b also was not supported. Hypothesis 5a suggested that team efficacy would be related to team action and transition processes. This hypothesis was supported; team efficacy was significantly related to team action and transition processes (.52, p < .001, R2 = .27). Hypothesis 5b suggested that team efficacy would mediate the relationship between transformational leadership and team action and transition processes; once again, because transformational leadership was unrelated to team efficacy (hypothesis 3a), hypothesis 5b was not supported. Hypothesis 6a suggested that both team goals and team action and transition processes would be related to higher levels of team performance. This hypothesis was partially supported; team action and transition processes were significantly related to team performance (.38, p < .05), but team goal was not (.00, NS). Team action and
81
transition processes accounted for 14% of the variance in performance (R2 = .14). Hypothesis 6b suggested that team goal and team action and transition processes would mediate the relationship between team efficacy and team performance; although team efficacy was related to both team action and transition processes and team goals (hypotheses 4a and 5a), only team action and transition processes was related to performance. When an alternative model was tested that included a direct path from team efficacy to team performance, a chi-squared difference test revealed no model improvement (
2
= .11 on one degree of freedom, NS). Therefore, it is reasonable to
conclude that team action and transition processes do, in fact, serve as a mediating mechanism between team efficacy and team performance; hypothesis 6b was partially supported. On the whole, the hypothesized model provided a poor fit with the data, with (20) = 47.34; p