Copyright © Gregory Stephen Brown, 2016, All Rights Reserved ii Copyright © Gregory Stephen Brown, 2016, All Rights Reserved iii Acknowledgements This dissertation is the culmination of many years of self-discovery and determination. It would not have been possible without the love, support, guidance and encouragement of so many. Thank you all. My wife, Jennifer, whose sacrifice, love and encouragement made this all possible. You are my rock. My children, Owen, Abby, and Ryan whose sense of awe and wonder about the world around them inspire me to learn more. My sister, Alicia, who spent hours reading my work, and pushing me to do better. I could not have done it without you. My parents, Stephen and Joann who have always encouraged me to follow my dreams no matter where they lead. Cohort 4, we may have all gone in our own directions, but I could not have made it through Thursday nights without each of you. My committee members: Dr. Caldas, for pushing my thinking to places I did not know existed and for believing in me every step of the way. Dr. Wan, for giving sound advice, guidance, and encouragement as I navigated this journey. Dr. Fister, for being a source of encouragement. Your experience and legacy serves as an example to many. iv Dedication I dedicate this dissertation to Owen, Abigail, and Ryan Brown. I pray that as you grow, you continue to approach the world with the same sense of curiosity and wonder that you do now, with your eyes and hearts wide-open. v Table of Contents Notice of Copyright ............................................................................................................ ii Acknowledgements ............................................................................................................ iii Dedication .......................................................................................................................... iv List of Tables ................................................................................................................... viii List of Figures .................................................................................................................... ix Appendices………………………………………………………………………………...x Abstract ............................................................................................................................... 1 Chapter One: Introduction ................................................................................................. 3 Overview of the Problem ........................................................................................ 3 Theoretical Framework ........................................................................................... 4 Conceptual Framework ........................................................................................... 5 Context .................................................................................................................... 7 Summary of Literature ............................................................................................ 7 Research Question ................................................................................................ 11 Hypotheses ............................................................................................................ 11 Research Methodology ......................................................................................... 12 Significance and Limitations of Study.................................................................. 12 Dissertation Structure............................................................................................ 14 Chapter Two: Review of the Literature ............................................................................ 16 Introduction ........................................................................................................... 16 Educational Leadership ......................................................................................... 17 A Trait-Based Approach to Leadership ................................................................ 19 Leadership Skills Approach .................................................................................. 29 vi Replication Theory................................................................................................ 40 Chapter Summary ................................................................................................. 45 Chapter Three: Research Design and Methodology ......................................................... 47 Introduction ........................................................................................................... 47 Research Rationale................................................................................................ 47 Rational for Replication ........................................................................................ 47 2007 Study Statistical Findings ............................................................................ 48 Modifications to 2007 Study................................................................................. 52 Research Question and Hypothesis ....................................................................... 54 2016 Research Design and Methodology ............................................................. 56 Setting ................................................................................................................... 58 Sampling ............................................................................................................... 59 Instrumentation ..................................................................................................... 60 Dependent Variables ............................................................................................. 62 Independent Variables .......................................................................................... 66 Control Variables .................................................................................................. 66 Statistical Analysis ............................................................................................... 66 Limitations of the study ........................................................................................ 68 Chapter Summary ................................................................................................. 69 Chapter Four: Data Analysis and Findings ....................................................................... 71 Introduction ........................................................................................................... 71 Research Question ................................................................................................ 72 Hypotheses ............................................................................................................ 72 Review of The Study’s Purpose ............................................................................ 73 Summary of Setting and Data Collection Procedures........................................... 73 vii Pre-Analysis Data Screening ................................................................................ 75 Effect Sizes and Significance Levels for This Study ............................................ 83 Creation of Latent Variables ................................................................................. 84 Results / Findings .................................................................................................. 87 Themes ................................................................................................................ 116 Recapitulation ..................................................................................................... 117 Chapter Summary ............................................................................................... 120 Chapter Five: Discussion, Conclusions, and Recommendations ................................... 121 Introduction ......................................................................................................... 121 Discussion ........................................................................................................... 121 Conclusions/Implications .................................................................................... 129 Limitations .......................................................................................................... 140 Recommendations ............................................................................................... 141 Chapter Summary ............................................................................................... 142 References ....................................................................................................................... 144 Appendices ...................................................................................................................... 153 Permission Letter to Use Copyrighted Material ............................................................. 171 Biography……………………………………………………………………………….173 viii List of Tables Table Page 1. Comparison of 2007 Study and Current Study 54 2. Skewness and Kurtosis of all Leadership Skills by Group and Variable 80 3. Cohen’s (1992) Effect size Interpretations 84 4. Reliability Scores (alpha) for the 2007 and 2015 Studies 87 5. Descriptive Statistics for all Skills by Organizational Level 89 6. Pearson Correlation Matrix of Leadership Skill Requirements, Organizational Level, and Covariates 91 Fit Indices for One-Factor and Four-Factor Measurement Models Using Structural Equation Modeling and χ2 Difference Test 95 Estimated Marginal Means for Cognitive, Interpersonal, Business and Strategic Skill Requirements, Controlling for Organizational Level, Experience Level, Education Level, and District Setting 98 7. 8. 9. Partial Correlations Between Leadership Skill Requirement Categories and Organizational Level, Controlling for Years of Experience, Education Level, and District Setting 100 10. Summary of Findings and Comparison of Conclusions for the 2016 and 2007 Studies 106 Significant Univariate Effects for Organizational Level 110 11. ix List of Figures Figure Page 1. The Leadership Skills STRATAPLEX 6 2. La Sorte’s Replication Paradigm 42 3. One-factor confirmatory factor analysis structural equation measurement model of leadership skills 93 4. Multi-skill, four-factor confirmatory factory analysis structural equation measurement model 94 Estimated marginal means for leadership skill requirements across organizational levels while controlling for experience, education, and district setting 101 Estimated marginal means for Cognitive skill requirements across organizational levels while controlling for experience, education, and district setting 111 Estimated marginal means for Interpersonal skill requirements across organizational levels while controlling for experience, education, and district setting 113 Estimated marginal means for Business skill requirements across organizational levels while controlling for experience, education, and district setting 114 Estimated marginal means for Strategic skill requirements across organizational levels while controlling for experience, education, and district setting 116 Schematic of multiple Leadership Skills STRATAPLEX pyramids of varying size denoting the total amount of leadership skill required 133 11. Leadership skills interacting on a three-dimensional plane 134 12. Schematic of the reconceptualized Leadership Skills STRATAPLEX model 135 Schematic of the hypothesized, reconceptualized Leadership Skills STRATAPLEX model with extension 137 5. 6. 7. 8. 9. 10. 13. x Appendices Letter Page A. Manhattanville College IRB Approval Letter 153 B. Survey Instrument 154 C. Survey Recruitment Card 169 1 Abstract LEADERSHIP SKILL REQUIREMENTS ACROSS ORGANIZATIONAL LEVELS OF K-12 PUBLIC EDUCATION IN NEW YORK STATE: A REPLICATION AND GENERALIZATION OF THE LEADERSHIP SKILLS STRATAPLEX With a history that began with the dawn of civilization, leadership has been one of the most elusive and longest studied phenomena. Despite its long history, disagreement in the field persists over how best to define, measure, and develop it. Effective leadership has always been sought after and extremely valuable. Harvard’s Center for Public Leadership reports that the United States of America is experiencing a crisis in leadership, with this crisis felt most strongly in the realm of public leadership (Rosenthal, 2012). Educational leaders are not only public leaders but also are responsible for maintaining high standards and providing strong outcomes for our nation’s children. Waves of reform movements and ever-increasing levels of cynicism often complicate their vital work. Identifying and understanding how to develop the leadership skills required by educational leaders has never been more important. This study’s purpose was to understand more completely the leadership skills required by leaders of New York’s K-12 public schools. Through the application of the Leadership Skills STRATAPLEX model, this researcher sought to determine the degree to which this skillsbased framework could be generalized to New York’s public school leaders across multiple organizational levels. This research question was addressed by conducting a theoretical replication and extension of Mumford, Campion, & Morgeson’s 2007 study. 2 This study made use of a quantitative research design and survey data collected from 1027 educational leaders in New York State. This study’s major findings suggests that the Leadership Skills STRATAPLEX model failed to fully generalize to this new population and setting. However, the results also led this researcher to suggest a more sophisticated reconceptualization of the STRATAPLEX model that proposes a new moderating construct, the leader’s “sphere of direct influence relative to the whole organization”. This researcher’s findings and the reconceptualized Leadership Skills STRATAPLEX model have the potential to influence the ways institutions train and develop leaders while advancing the field of leadership skills-based theory. 3 Chapter One: Introduction Overview of the Problem Educational leadership is in the midst of tremendous change with the role of educational leaders, at all levels, struggling to adapt (Fullan, 2001; Malone & Caddell, 2000). The internal and external pressures of No Child Left Behind and Race to the Top have created an environment centered on accountability and high-stakes testing. This dissertation examined through the use of The Leadership Skills STRATAPLEX as a conceptual framework, the leadership skills required of New York State’s public school leaders (Mumford et al., 2007). It is important to examine leadership from this perspective as the increasing rate and complexity of educational reform demands leaders to confront complex problems that lack established solutions. Fullan (2001) states, “The more complex society gets, the more sophisticated leadership must become” (p. 1). For these reasons, this study is both timely and relevant. Nationally, Americans believe a crisis of leadership exists at all levels. The National Leadership Index 2012, conducted by The Center for Public Leadership at Harvard’s Kennedy School, found that “Sixty-nine percent of Americans still believe we have a leadership crisis” (Rosenthal, 2012, p. 1). This study also found that “educational leadership, so important to the country’s future competitive strength, continues to languish in fifth place from the bottom, among the thirteen sectors for which Americans have ‘not much’ confidence” (Rosenthal, 2012 p. 1). Kellerman and Webster (2001) lament, “the levels of our political cynicism, alienation, and now fear, are so high that they throw into question the very possibility of effective leadership in the public realms” (pp. 485-486). In the midst of this uncertain time, educational leaders face a “perfect 4 storm” which has formed as thousands of baby boomers enter retirement age and prepare to leave their leadership roles (Mascall & Leithwood, 2010). Malone and Caddell (2000) underscore the problem: “The demand for effective school principals is at an all-time high, the result of reform efforts, constant public criticism, and demographic realities” (p. 162). For this reason, the time was right to apply the theories and conceptual frames of Mumford et al., (2007) to the field of education through a theoretical replication and extension. This researcher sought to further the development of leaders of learning organizations, practitioner scholars, and leaders who develop both themselves and others through this work. The implications of this type of study aim to impact professional development and institutions that prepare future leaders. Theoretical Framework This researcher studied the relationships between leadership skill requirements and a leader’s organizational level while controlling for the effects of demographic variables, through a post-positivist framework. This framework allowed this researcher to examine the educational leadership phenomena in an attempt to help establish best practices for the field (Butin, 2010). Post-positivism acknowledges that research is often shaped by the values and beliefs held by the investigator. Additionally, post-positivism holds that theories cannot be proved conclusively (Teddie &Tashakkori, 2009). This researcher also believes that elements of Critical Realism exist in this study. Critical Realism purports that social interactions and structures existed on three levels (real, actual, and empirical; Bhaskar, 1998). According to the theoretical work of Bhaskar (1998), these levels help us understand what we observe (empirical), the actual events that take place (actual), and the underlying structures and laws that cause the events 5 (real). This stratified understanding of reality adds a critical lens to what is observed, in that the observable is influenced by events, and events are explained by theory. This framework is useful for the study of leadership as leadership can be explained through the observable actions of leaders (empirical), events that caused the leader to act (actual), and the theories that are applied to explain actions (skill-based theory). Acknowledging that invisible constructs such as leadership skills exist, are empirically distinguishable, and lead to social events necessitates an understanding of the specific amounts and proportions of leadership skills required by educational leaders across organizational levels. Conceptual Framework This research study was conducted through an application of Mumford, Campion, & Morgeson’s (2007) Leadership Skills STRATAPLEX framework to leaders of New York State’s K-12 public schools. This conceptual framework suggests that previous work done in the field of skills-based leadership can be conceptualized as four distinct skill requirements (Cognitive, Interpersonal, Business, and Strategic). These skill requirements help to explain the complexity inherent in any study of leadership by establishing the “plex” of the model. Additionally, this model suggests that leadership must be viewed as stratified organizational levels, or “strata” (Mumford et al. 2007). The basic tenants of this model are built around the following assumptions as tested in the original study and represented by Figure 1: Distinct leadership skills can be distinguished empirically. Leadership skills increase as you ascend organizational levels. 6 Leadership skills become important in different proportions at each organizational level. Figure 1. The leadership skill requirements strataplex represents the complex relationship hypothesized to exist between the four leadership skill requirements (Cognitive, Interpersonal, Business, and Strategic) and a leader’s organizational level. Adapted from “The Leadership Skills Strataplex: Leadership Skill Requirements Across Organizational Levels,” by T. V. Mumford, M. A. Campion, and F. P. Morgeson, 2007, The Leadership Quarterly, 18, p. 156. Copyright 2007 Elsevier Inc. Reproduced by permission of copyright holder, http://www.elsevier.com. Essentially, The Leadership Skills Requirements STRATAPLEX asserts that leadership can be developed and that different but increasing amounts of the four leadership skills are required at each successively higher organizational level. This developmental view of leadership acknowledges that leaders must sharpen their existing skills while preparing for future positions (Mumford et al., 2007). 7 Context This research study was designed for several specific audiences including; established and prospective educational leaders, educational policymakers, and other contexts in which stratified leadership is present at different organizational levels. This study and its findings may be of interest to these groups as they seek to hire, promote, and establish programs for the development of leaders in the field of education. According to Mascall and Leithwood (2010), the recent change in the job description of educational leaders has made the job less attractive at the same time that there is a reduced supply of people to fill the jobs that exists. This only adds to the urgency of this type of work. Summary of Literature The study and definitions of leadership date back to the beginnings of written history. Stogdill (1974), and later Bass (1990) stated, “The study of leadership rivals in age the emergence of civilization, which shaped its leaders as much as it was shaped by them. From its infancy, the study of history has been the study of leaders - what they did and why they did it” (p.3). Bass (1990) also acknowledged, “Leadership research faces a dilemma. A definition that identifies something for the factory manager or agency head is not necessarily the most useful for the development of a broad theory” (p. 19). Leadership is universal and therefore worthy of study, scrutiny, and evaluation. “No societies are known that do not have leadership in some aspects of their social life, although many may lack a single overall leader to make and enforce decisions” (Bass, 1990, p 5). According to Northouse (2010), leadership is also a “highly sought-after and highly valued commodity” (p.1). Leadership is also often misunderstood. Bass (1990) 8 claims, “there are almost as many different definitions of leadership as there are persons who have attempted to define the concept” (p.11). Bass’s (1990) “Bass and Stogdill’s Handbook of Leadership: Theory, Research and Managerial Applications”, found no fewer than thirteen different ways of conceptualizing leadership. Jago (1982) claims: Leadership is both a process and a property. The process of leadership is the use of non- coercive influence to direct and coordinate the activities of the members of an organized group toward the accomplishment of group objectives. As a property leadership is the set of qualities or characteristics attributed to those who are perceived to successfully employ such influence… It is important to recognize what this definition includes and what it excludes. Leadership is not only some quality or characteristic that one possesses or is perceived to possess it can be something that one does. It therefore can describe an act as well as a person. (pp. 315-316) By adopting a broad definition of leadership that views it as a process and recognizes common goals, we can better understand the combination of individual traits and developable skills often associated with effective leadership. The skills approach to leadership offers a unique lens through which to view the leadership phenomena. Unlike its precursor, the trait-based approach, skills approaches have been able to look at what the leader does rather than who the leader is. This more developmental or behavioral view of leadership is optimistic by nature. Some have even suggests that leadership skills are like athletic skills and as such can be improved by all, through practice, regardless of natural abilities (Northouse, 2010). 9 Arguably, the most influential work related to skills-based leadership is Katz’s 1955 article entitled “Skills of an Effective Administrator.” Contrary to dominant leadership studies of the time, this approach attempted to describe leadership as something that could be developed. For Katz (1955) “this quest for the executive stereotype has become so intense that many companies, in concentrating on certain specific traits or qualities, stand in danger of losing sight of their real concern: what a man can accomplish” (p. 33). He recognized that while traits exist, they are not the only things that matter. Katz’s three-skill approach was a new way of observing the leadership phenomenon. Katz (1955) concluded: [Skill] transcends the need to identify specific traits in an effort to provide a more useful way of looking at the administrative process. By helping to identify the skills most needed at various levels of responsibility, it may prove useful in the selection, training and promotion of executives. (p. 42) This approach inspired later studies in the leadership skills field to be conducted by others. The leadership skills STRATAPLEX model (Mumford et al., 2007), like Katz’s model, was created to assist in the examination of the leadership skills required by leaders at different organizational levels (Mumford et al., 2007). Mumford et al.’s (2007) model extended previous studies in that it focuses on the leader’s job requirements rather than the person doing the job. Mumford et al. (2007) seek to “further our understanding of leadership skill requirements across organizational levels by identifying four distinct categories of leadership skill requirements that emerge differentially across organizational levels” (p. 155). 10 The foundation of this model was built on four identifiable skills: Cognitive, Interpersonal, Business, and Strategic. The STRATAPLEX model builds on previous works by concurrently dividing leadership into tiers (strata), from junior to senior management, and into four skill categories. Mumford et al. (2007) make several hypotheses concerning the STRATAPLEX model: first, that the four leadership skill categories will be empirically distinguishable; next, that all skills will be needed at each level in a hierarchy (Cognitive, Interpersonal, Business, and Strategic), and last that the proportions of each will differ based on the organizational level. These skills were measured through language created by the Occupational Information Network (O*NET), developed by the U.S. Department of Labor (Mumford et al, 2007). Mumford et al.’s (2007) findings suggests that leadership skills can be grouped into four categories (Cognitive, Interpersonal, Business and Strategic) and that all four leadership skills are required in greater amounts as you ascend the organizational structure. Leadership skill requirement amounts vary based on the specific skill category and organizational level. Strategic and Business skills were more strongly correlated to organizational level than Interpersonal and Cognitive skills, indicating that their development may be more critical to leadership develop over time (Mumford et al, 2007). Notable are the limitations of this study. Most importantly, the sample of 1,023 government workers may not be representative of workers in other fields. Mumford et al (2007) encouraged future researchers to determine if this model of leadership skill requirements could be generalized to other settings and contexts, which is what this researcher attempted to do. 11 Research Question To what extent is Mumford et al.’s Leadership Skills STRATAPLEX model generalizable to leaders of New York State’s K-12 public schools? Hypotheses In an attempt to answer this question and describe the leadership skills required by leaders of New York’s K-12 public schools, a theoretical replication was conducted which tested the following hypotheses as found in Mumford, Campion, and Morgeson’s (2007) original study (pp. 151-159). The original hypotheses are presented as found in Mumford et al.’s (2007) original study (pp. 151-159): (H1). The Cognitive, Interpersonal, Business, and Strategic skill categories will be empirically distinguishable. (H2). Leadership skill requirements will vary by skill category such that Cognitive skills will be needed the greatest amount, followed by Interpersonal, Business, and Strategic skills, respectively. (H3). Cognitive, Interpersonal, Business, and Strategic skill requirements will be positively related to the job’s level in the organization. (H4). Leadership skill requirements will interact with organizational level such that: a. Strategic skill requirements will be more strongly related to organizational level than Cognitive, Interpersonal, and Business skill requirements. b. Business skill requirements will be more strongly related to organizational level than Cognitive or Interpersonal skill requirements. 12 c. Interpersonal skill requirements will be more strongly related to organizational level than Cognitive skill requirements. Research Methodology In order to answer this research question, the researcher conducted a theoretical replication and extension as described by La Sorte (1972), utilizing a descriptive, nonexperimental, quantitative, cross-sectionals design. The use of descriptive, bivariate, multivariate, and correlational statistics, as-well-as confirmatory factor analysis and model testing/comparison using Structural Equation Modeling (SEM), allowed for the testing of all of the aforementioned hypotheses. First, this researcher examined whether a four-factor model of leadership skills was empirically distinguishable and if so, what was the relationship between these leadership requirements and the leaders’ organizational level, controlling for other factors. Next, this researcher examined the amounts and proportion of leadership skills required by leaders at different organizational levels regardless of their experience, educational level, and district setting. Finally, this researcher concluded by conceptualizing the Leadership Skills STRATAPLEX model for educational leaders. Significance and Limitations of Study The 21st century has seen a multitude of crises on many fronts, including national security, the financial markets, and education. These uncertain times demand bold and effective leadership, and the emergence of leaders equipped with the skills to confront complex problems that lack established solutions. 13 The question many communities in New York State are asking is who will lead our schools and children through these uncertain and monumental times of change? Heifetz (1994) stated, “The common personalistic orientation to the term leadership, with its assumption that ‘leaders are born and not made’, is quite dangerous. It fosters both self-delusion and irresponsibility” (p. 20). Therefore, in the face of current challenges, the creation and validation of a skills-based model of leadership in K-12 public education has never been more necessary. The implications of such a model would allow leaders to assess and develop their skills in order to become more effective in their current role and as they prepare for future roles. In addition, such developable skill sets will help guide professional development at a time when budgets are stretched under a 2% tax cap limit in New York State. Lastly, as more educational leaders leave the field and fewer qualified replacements enter, the implications of this study will help inform providers of professional development and leadership preparation programs (Markow, Macia, & Lee, 2013). Despite the significance of this study, limitations exist. First, cross-sectional designs are restricted to measuring one moment in time. This data collection technique was limited in that it did not allow for the current mood of the person taking the survey or changes that may take place over time. Next, this study only controlled for leaders “level of education”, “level of experience”, and “district’s setting”. While the addition of these controls allowed for a clearer picture of the constructs being measured by eliminating error, more covariates could exist which were not considered or accounted for in this study. Finally, descriptive research by definition has the ability to describe a phenomenon yet lacks the ability to predict. 14 Dissertation Structure This dissertation conformed to the traditional five-chapter format. Chapter One’s introduction began with the context and background of leadership theory, starting with trait-based approaches to leadership and culminating with Mumford et al.’s skills-based “Leadership STRTAPLEX” as a conceptual framework for future study. Theoretical and conceptual frameworks are discussed. Next, the research problem and questions are presented along with gaps in the current body of research, which established the significance of this study to the field. Chapter Two begins with a review of the literature describing the necessity for the study of leadership and the development of skill-based models. This leads to a historical overview of the study of leadership that establishes a context in which to specifically explore trait- and skills-based approaches. Limits of this review within the broader field of leadership are discussed. Next, an exploration of current research around leadership skill acquisition and the various combinations required at different organizational levels is presented. Finally, the merits of replication methodology are examined through the work of La Sorte (1972). Chapter Three provides an extensive explanation of the research methodology and design for this study. Additionally, the rationale of sample selection, methods of data analysis, and validity concerns are addressed. Chapter Three includes the presentation of research questions and implications of this study. Chapter Four begins with a description of the study’s setting and data collection procedures. This is followed by extensive data screening that addresses; missing data, outliers, normality, linearity, and homoscedasticity. Next, the creation of latent variables 15 was reviewed along with tests for reliability. Descriptive and correlational statistics were presented followed by the examination of each of the four Hypothesis which includes; confirmatory factor analysis, model testing, model comparison (Structural Equation Modeling), partial correlations, and univariate and multivariate analysis of variance and covariance (MANCOVA). Finally, the chapter concludes with a presentation of emerging themes. Chapter Five begins with a discussion of the findings for both the theoretical replication and extension. This is followed by an interpretation of these findings. Next, conclusions are made, and suggests implications for professional development, leadership development programs, and leadership skills theory development are presented. In light of the findings and conclusions, a reconceptualized Leadership Skills STRATAPLEX model is proposed. The chapter closes with the acknowledgment of the limitations of this study and recommendations for future research. 16 Chapter Two: Review of the Literature Introduction The purpose of this study was to examine the extent to which the Leadership Skills STRATAPLEX can be generalized and validated in a sample of New York State public school leadership. This study built on the work of leadership theorists Campion, Connelly, Katz, Mumford, Morgeson, and Zaccaro by applying their theories and conceptual frameworks to the field of K-12 education in New York State, thus contributing in an original way to the field of educational leadership. In so doing, this study has implications for the hiring, professional development, and promotional practices of educational organizations. The purpose of this literature review is to explore research surrounding the necessity for and development of leadership skills across various organizational levels. This chapter discusses the vast and varied definitions of leadership, establishing the working assumptions around leadership held for this study. A historical overview of the study of leadership establishes the context in which to specifically explore skills-based leadership theory, establishing the bounds and limits of this review within the broader leadership field. Next, this review discusses the current research around leadership skills and the various skill requirements needed at different organizational levels, through an investigation of the theories and frameworks established by Campion, Connelly, Katz, Mumford, Morgeson and Zaccaro. Finally, the original study (Mumford et al., 2007) and replication theory are discussed to establish a contextual basis for the 2016 study. The varied development of definitions and concepts concerning leadership over a vast period of time necessitated the establishment of constraints or boundaries for this 17 study. This review of the literature focused on trait- and skills-based approaches to leadership in order to establish a context that necessitated the expansion of skills-based leadership models. Torraco (2005) believes, “The best literature reviews examine the literature with a particular lens… this lens points the author to specific aspects of previous research that are critically examined and evaluated” (p. 361). Through these lenses, this literature review examined leadership from a leader-centered perspective concerning what a leader is (trait approach) and what a leader does (skills approach). While leadership can be viewed through many other lenses (behavioral, contingency, emotional, moral, and servant leadership), those lenses were beyond the scope of this study. Educational Leadership The study of educational leadership is an offshoot of the larger leadership tree and must be examined and understood in the context of the larger field, while paying deference to its specific uniqueness. With these things in mind, the following review of the literature examined leadership from a holistic view in an attempt to overlay some of the larger theoretical approaches (trait- and skill-based) and frameworks (The Leadership Skills STRATAPLEX) to the specific area of K-12 education. Defining educational leadership. The study and definitions of leadership date back to the beginning of written history. Stogdill’s (1974), and later Bass’s (1990), comprehensive reviews of leadership state, “The study of leadership rivals in age the emergence of civilization, which shaped its leaders as much as it was shaped by them. From its infancy, the study of history has been the study of leaders – what they did and why they did it” (p.3). Bass (1990) also acknowledged, “Leadership research faces a 18 dilemma. A definition that identifies something for the factory manager or agency head is not necessarily the most useful for the development of a broad theory” (p. 19). Leadership is universal and therefore worthy of study, scrutiny, and evaluation. “No societies are known that do not have leadership in some aspects of their social life, although many may lack a single overall leader to make and enforce decisions” (Bass, 1990, p 5). According to Northouse (2010), leadership is also a “highly sought-after and highly valued commodity” (p.1). Leadership is also often misunderstood. Bass (1990) claims, “there are almost as many different definitions of leadership as there are persons who have attempted to define the concept” (p.11). Bass’s (1990) Bass and Stogdill’s Handbook of Leadership: Theory, Research and Managerial Applications found no fewer than thirteen different ways of conceptualizing leadership. This same problem was recognized by Pfeffer (1977) in his article entitled “The Ambiguity of Leadership”, in which he stated, “While there have been many studies of leadership, the dimensions and definition of the concept remain unclear” (p. 105). For the purposes of this review, leadership was broadly defined as “a process whereby an individual influences a group of individuals to achieve a common goal” (Northouse, 2010, p. 3). Or as Jago in his 1982 article stated: Leadership is both a process and a property. The process of leadership is the use of non-coercive influence to direct and coordinate the activities of the members of an organized group toward the accomplishment of group objectives. As a property, leadership is the set of qualities or characteristics attributed to those who are perceived to successfully employ such influence.… It is important to recognize what this definition includes and what it excludes. Leadership is not 19 only some quality or characteristic that one possesses or is perceived to possess, it can be something that one does. It therefore can describe an act as well as a person. (pp. 315-316) By adopting a broad definition of leadership that views leadership as a process and recognizes common goals, we can better understand the combination of individual traits and developable skills associated with effective leadership. A Trait-Based Approach to Leadership To understand trait-based leadership is to understand the broader history of leadership study. For thousands of years, leaders and heroes can be found universally across culture, religion, myth and literature (Bass, 1990; Hunt, 1999; Stogdill, 1974,) Both historic and mythical leaders have been examined in an attempt to establish generalizable characteristics that led to their successes. This forms the foundation of a trait-based approach to understanding leadership. Examples of ancient writings concerning such leaders and their distinct traits include but are not limited to: The Epic of Gilgamesh, The Odyssey, The Iliad, The Bible, and in the ancient philosophies of Confucianism and Taoism (Bass, 1990). In all cases, a special quality or set of qualities are present which separate the leader from the follower. This “Trait Theory” has persisted throughout the centuries, while gaining, losing, and recently regaining popularity amongst researchers. Most theories of leadership have been built upon or developed in rejection of this trait-based approach (Bass, 1990; Kirkpatric & Locke, 1991; Stogdill 1974; and Zaccaro, 2007) The fingerprints of trait theory can be seen in Nadler and Tushman’s (1988) article for Fortune magazine entitled “What Makes for 20 Magic Leadership” and the thousands of other publications on leadership that line book shelves. Many scholars have commented on the history and study of leadership, but arguably none more comprehensively than Stogdill (1974) in the Handbook of leadership: A survey of theory and research and later Bass (1990) in the Bass and Stogdill’s Handbook of Leadership: Theory Research and Managerial Applications. These volumes, reviewing more than 8,000 writings and studies, are frequently referenced and considered one of the master works in the field of leadership. A more comprehensive review of leadership, in its many forms, theories, and constructs, beyond the scope of this review, can be found within their pages. The great man theory. The explanation of historical events and the individuals who shaped them are often attributed to great men. Bass (1990) stated, “Martin Luther King is considered to be the ‘great man’ whose leadership inspired the black civil rights movement” (p. 38). Historians and psychologists share an interest in this study, for it is through the examination of impactful social and political leaders like Gandhi, Lincoln, Hitler, and Franklin D. Roosevelt that we can better understand our world and its events (Bass 1990 & Northouse, 2010). Northouse (2010) remarked, “It was believed that people were born with these traits and that only ‘great’ people possess them” (p. 15). According to Bass (1990), “The great man theory of leadership is currently espoused by those who show how faltering business corporations are turned around by transformational leaders, such as Lee Iacocca” (p. 38). Nadler and Tushman (1988) echo this supposition: 21 In the emerging folklore of corporate leadership, larger-than-life characters transform or save major American companies single-handedly. The current literature is rife with references to such leaders as Lee Iacocca, John Sculley, Jack Welch and others whose exploits assume mythic proportions. (p. 261) These examples illustrate how foundational the concepts of innate leader separating traits have become to our understanding of the world around us. Both Bass (1990) and Zaccaro (2007) agree that much of what we term trait theory, more specifically great man theory, today traces its roots in Galton’s (1869) study of hereditary genius. In this work, Galton attempted to establish a direct link between heredity and eminent men through the study of “the British”: judges, statesmen, commanders, literary men, men of science, poets, musicians, and painters, among others. Galton (1869) intended “to show ...that a man’s natural abilities are derived by inheritance, under exactly the same limitations as are the form and physical features of the whole organic world” (p. 1). While many of his theories on race and inheritance would be rejected today, the underlying construct that an inborn characteristic or trait exists in some leaders and is absent in others would strongly influence early and contemporary trait theory. Trait theory. The study of leadership traits was greatly impacted by Ralph M. Stogdill’s 1948 review of trait leadership entitled “Personal Factors Associated with Leadership: A Survey of the Literature.” In this review, Stogdill examined more than one hundred studies concerning leadership traits and characteristics, which had taken place between 1904 and 1947. Stogdill (1948) specifies, “The present survey is concerned only with those studies in which some attempt has been made to determine the traits and 22 characteristics of leaders.... The present survey lists only those factors which were studied by three or more investigators” (p. 35). Stogdill recognized two serious flaws, which existed in the study of trait leadership. First, leadership was often left undefined, and second, the methods used by the researchers were often unrelated to the stated investigation (Stogdill, 1948). In spite of these potential flaws, Stogdill (1948) identified seven common methodologies. Many of the methods for studying leadership relied heavily on the observed interactions amongst children or college-aged students, which were later generalized to adults. Only one of the seven methodologies studied relied primarily on evidence gathered through data collected from adults. However, from Stogdill’s study emerged a series of characteristics that are discussed later in this literature review. Stogdill (1948) was able to synthesize the available body of trait literature into twenty-nine areas. From that list he purported that all characteristics could be classified into one of the following categories: intelligence, alertness, insight, responsibility, initiative, persistence, self-confidence, and sociability. While this new classification system, supported by his findings, would become the foundation of further attempts to create a universal classification system, it was not the most impactful of Stogdill’s findings. The assertion with the greatest impact on the study of leadership is clearly seen in Stogdill’s (1948) statement, “the total weight of evidence presented in this group of studies suggests that if there are general traits which characterize leaders, the patterns of such traits are likely to vary with the leadership requirements of different situations” (1948, p. 61). 23 Bass (1990) would later refer to this finding as “sounding the seeming death-knell of a pure traits approach to the study of leadership” (p. 78). He would later recognize, as does Northouse (2010) and Jago (1982), that Stogdill’s 1948 review would be used “to support the view that leadership was entirely situational in origin and that particular personal characteristics could not accurately predict leadership” (p. 78). Thus, in reaction to these findings, the study of situational leadership began, and pure trait-based approaches were abandoned (Northouse, 2010). Highly influenced by Stogdill before him, Richard Mann’s 1959 work, “A Review of the Relationships between Personality Traits and Performance in Small Groups”, would consider traits, behaviors, and situational factors of leadership. In an attempt to “summarize the present state of knowledge about the relationship of an individual’s personality to his behavior or status in groups” (p. 241), Mann (1959) began the process of reconciling trait and situational theories. In a more sophisticated examination of the literature, Mann (1959) examined published and non-published studies done between 1900 through 1957. His identification of more than 500 measures and labels for personality traits illustrated the vastness and lack of consensus found in the field (Mann, 1959). Mann’s review of the literature classified personality traits into seven categories and behaviors into six. He then compared these personality traits and behaviors in light of situational considerations. Similarities are evident between Mann’s and Stogdill’s classifications of personality traits, and later leadership skills. Mann’s (1959) list includes intelligence, adjustment, extroversion-introversion, dominance, masculinity-femininity, conservatism, and interpersonal sensitivity. While the terms differed, many of the categories describe 24 the same qualities observed by Stogdill and Mann (Bass, 1990; Northouse 2010). Lastly, Mann’s study would call for research focused on personality traits, while controlling for situations. Like Stogdill before him, Mann’s work had a chilling effect on purely-trait based research and facilitated the growth in situational leadership theory that followed (Antonakis, Cianciolo, & Sternberg, 2004). Stogdill’s (1974) second review of the literature from 1948-1970 again examined the validity of personality traits and situations on effective leadership. This comprehensive volume of more than 600 pages defines, outlines, and reviews leadership theories and literature (163 studies). Unlike the study done in 1948, this work additionally examines leader stability, emergence, leaderfollower interactions, and group performance. This handbook would be reproduced several times, its third addition representing a collaborative effort between Stogdill and Bernard Bass. Like his previous review, traits were categorized (leadership, physical, social, intelligence and ability, and personality; Stogdill, 1974). The findings were explored through contemporary literature and compared to the 1948 review. According to Northouse (2010), what emerged “was more balanced in its description of the role of traits and leadership (p. 17). When responding to those promoting pure situational leadership theories, Stogdill (1974) stated, “This view seems to overemphasize the situational, and underemphasize the personal, nature of leadership. Strong evidence indicates that different leadership skills and traits are required in different situations” (p. 73). This repositioning would be seen throughout the 1974 review and would be echoed in its findings. 25 Unlike the simple list of eight leadership categories found in the 1948 review, a more descriptive set of categories emerged. Stogdill (1974) concluded: The leader is characterized by a strong drive for responsibility and task completion, vigor and persistence in pursuit of goals, venturesomeness and originality in problem solving, drive to exercise initiative in social situations, self-confidence and sense of personal identity, willingness to accept consequences of decision and action, readiness to absorb interpersonal stress, willingness to tolerate frustration and delay, ability to influence other persons’ behavior, and capacity to structure social interaction systems to the purpose at hand. (p. 81) This more tempered approach marked the reassertion of personality traits as an important factor of effective leadership. It also shared many similarities to future skill-based models. Stogdill (1974) recognized, “The conclusion that personality is a factor in leadership differentiation does not represent a return to the trait approach. It does however; represent a sensible modification of the extreme situationist point of view” (p. 82). Lord, De Vader, and Alliger (1986) took a second look at the works of Stogdill and Mann, and acknowledged, “Trait theories have not been seriously considered by leadership researchers since Mann (1959) and Stogdill (1948) reported that no traits consistently differentiated leaders from non-leaders across a variety of situations” (p. 402). This reexamination of past findings was consistent with the popular shift in position observed between Stogdill’s 1948 and 1974 works. Through the use of more sophisticated methods of analysis (validity generalization procedures), Lord et al. (1986) 26 proposed, “…first, these reviews [Stogdill and Mann] have often been misinterpreted, and second, there are both theoretical and methodological reasons for reconsidering the relations between the traits of potential leaders and their tendency to be perceived as leadership by others” (p. 402). The significance of Lord et al.’s (1986) study lies in the assertion that stronger relationships exist between personality traits and leadership perceptions than reported by Mann, and as a result were subsequently transmitted in other studies. Lord, et al. (1986) stated “personality traits are associated with leadership perceptions to a higher degree and more consistently than the popular literature” (p. 407). Their findings would reinvigorate the investigation of personality traits as a predictor of leadership perceptions and performance. Much like Lord, et al., Kirkpatric and Locke (1991) also advanced the once outmoded and now resurrected notion that leadership traits do matter. Through a brief comparison between “Great Man Theories” and “Trait Theories,” Kirkpatric and Locke (1991) drew an important dissimilarity, stating, “Trait theories did not make assumptions about whether leadership traits were inherited or acquired. They simply asserted that leaders’ characteristics are different from non-leaders” (p. 48). This observation is important in that it once again provided a venue for the discussion of traits or characteristics within modern leadership theory. By recognizing the difference between static traits and plastic characteristics, Kirkpatrick and Locke helped modernize trait leadership theory, thus laying the foundation for the emergence of the skills-based approaches that followed. 27 Like theorists before, Kirkpatric and Locke (1991) grappled with the overwhelming list of characteristics and their overlapping descriptors, which had been discovered and observed over the previous century. They attempted to classify these traits into broad categories, and subsequently a framework was attempted with six classifications emerging (drive, leadership motivation, honesty/integrity, self-confidence, cognitive ability, knowledge of the business, and other traits) (Kirkpatric & Locke, 1991). However, unlike other strict trait theorists, Kirkpatric and Locke (1991) commented: Traits alone, however, are not sufficient for successful business leadership-they are only a precondition. Leaders who possess the requisite traits must take certain actions to be successful … Possessing the appropriate traits only makes it more likely that such actions will be taken and be successful. (p. 49) The theory that traits are prerequisite to leadership (difference between leaders and nonleaders), specific traits increase the probability of leaders taking action (forming a vision, setting goals…), and successful actions determine success, was a major contribution to the field and again points to leadership skill-development. As a corollary to the six trait classifications, Kirkpatric and Locke (1991) point to “the rest of the story,” where they recommend that leadership theories require more than a simple examination of leader traits. Deference is paid to the necessity of requisite traits. However, other factors exist to maximize leadership potential (Kirkpatric & Locke, 1991). Kirkpatric and Locke (1991) stated, “Skills are narrower in meaning than traits and involve specific capacities for action such as decision making, problem solving and performance appraisal” (p. 56). Taken together, the six traits (similar to the previous works of Galton, 1869; Stogdill, 1948 and 1974; Mann, 1959; and Lord et al. 1986) and 28 the addition of skills (with their ability to be developed) marked a change in the way trait theory would be examined by others. The implications of this change would be felt in hiring, training, and promotional practices. Section summary. Between the early twentieth century and the early 1990s, trait theory underwent a metamorphosis. Purely trait-based works like, Galton (1869) were discredited and all but abandoned in light of critical reviews done by Stogdill (1948) and Mann (1959). What followed was a paradigm shift in the way leadership was conceptualized. The situation in which leadership was being exercised became the dominant point of study. Situational leadership theorists like Carter (1953) and Gibb (1954) espoused the belief that leadership was based solely on the situation and that traits could not predict leadership (Bass, 1990; Zaccaro, 2007; Zaccaro, Kemp, & Bader, 2004). However, the concept of leadership traits, characteristics, and later personality would remain, in that they help distinguish leaders from non-leaders. During the 1970s and 1980s, the pendulum would swing back, and a more balanced approach between trait and situational leadership would emerge. This was due in part to a more tempered review of the literature done by Stogdill (1974), and later a complete reevaluation of the works of Stogdill and Mann, by Lord et al. (1986). Leadership traits were once again a factor in the leadership equation. Finally, Kirkpatric and Locke (1991) blended traits and actions (skills) when stating: Since Stogdill’s early review, trait theory has made a comeback, though in altered form.… The evidence indicates that there are certain core traits which significantly contribute to business leaders’ successes. Traits alone, however, are 29 not sufficient for successful leadership- they are only a precondition. Possessing the appropriate traits only makes it more likely that such actions will be taken and be successful. (p. 49) Trait-based leadership, while markedly different in form, remained. Trait-based approaches to leadership still conclude that certain characteristics separate leaders from non-leaders. Situations play a role in the emergence of leadership, and requisite traits can cause leaders to take action which may lead to success. While these conclusions had broad implications for the field of leadership, with many models and new theories developed, this study now focuses on the skills-based approach to leadership that grew out of the traditional trait-based approach. Leadership Skills Approach A skills-based approach to leadership traces its roots to the 1950s, and theory has developed significantly in the years since. Though originally conceived during a period of leadership history marked by the search for individual leader traits, and thus similar in its leader-centered approach, the skills-based approach differs in its foundational belief that leadership skills are developed (Bass, 1990). This fundamental difference holds a more optimistic approach, one that could be seen as more compliant with the American narrative. Katz (1955) stated: This approach is based not on what good executives are (their innate traits and characteristics), but rather on what they do (the kinds of skills which they exhibit in carrying out their jobs effectively) … skill implies an ability which can be developed, not necessarily inborn, and which is manifested in performance, not merely in potential. (p. 34) 30 This section first considers two distinct skills-based approaches (the three-skill approach and the skills model) based on the works of Katz (1955 and 1974) and several studies conducted by Mumford et al. (2000a, 2000b, 2000c, 2000d). Then, it will discuss The Leadership Skills STRATAPLEX as a model for understanding leadership through four reconfigured leadership skill categories (Cognitive skill, Interpersonal skill, Business skill, and Strategic skill) and their required amounts and proportions across organizational levels. This literature review reflects the natural progression of leadership study, from trait- to skills-based approaches. Additionally, it suggests that, like trait theories before, skill-based approaches can consolidate skills into broad categories (Katz, 1974; Mumford et al., 2007; Mumford et al., 2000b; Zaccaro et al., 2004). The three-skill approach. Arguably, the most influential work related to skillsbased leadership is Katz’s 1955 article entitled “Skills of an Effective Administrator.” His framework for understanding skills-based leadership laid the foundation for all other works that followed (Bass, 1990; Mumford et al. 2000a; Mumford et al. 2007). Contrary to dominant leadership studies of the time, this approach attempted to examine leadership, developmentally, within the context of organizations. For Katz (1955), “this quest for the executive stereotype has become so intense that many companies, in concentrating on certain specific traits or qualities, stand in danger of losing sight of their real concern: what a man can accomplish” (p. 33). He recognized that while traits exist, they are not the only things that matter. Katz’s (1955) statement, “At the root of this difference is industry’s search for the traits or attributes which will objectively identify the ‘ideal executive’ who is equipped to cope effectively with any problem in any organization,” (p.1), reflects the shift away from trait-based towards skill-based 31 leadership, post Stogdill and Mann. Katz’s work created the framework from which to study leadership, a framework that acknowledges traits and situations but attempts to transcend them through the exploration of developable skills (Northouse, 2010). Katz’s (1955) model for studying leadership is built around three distinct skills: technical, human, and conceptual. Katz purports, “this approach suggests that effective administration rests on three basic developable skills which obviate the need for identifying specific traits and which may provide a useful way of looking at and understanding the administrative process” (p. 34). The three skills are seen to hold differing degrees of importance at different organizational levels. Technical skills are seen as most important at lower organizational levels. These skills are necessary to perform specific operations. In reference to technical skills, Katz (1955) suggests “technical skill implies an understanding of, and proficiency in, a specific kind of activity, particularly one involving methods, processes, procedures, or techniques” (p. 34). He also suggests that this type of skill is needed to a lesser degree as you move up organizational levels. For example, technical skills would still be a necessity at the middle level, but almost non-existent at the top levels (Katz, 1955). Human skill is defined by Katz (1955) as “the executive’s ability to work effectively as a group member and to build cooperation with the team he leads … human skill is primarily concerned with working with people” (p. 34). He later recognizes that human skill not only means cooperation but also requires the leader to understand the perceptions and beliefs of others and to anticipate reactions. This skill type dictates the way a leader reacts to others and how they approach situations. Katz (1955) warns, “human skill cannot be a sometime thing… It must become an integral part of his whole 32 being” (pp. 34-35). This skill must be present at all organizational levels but is most important to middle and top management. Conceptual skill is the ability to see the big picture (Katz, 1955). Here, the leader must take into account all the moving parts. The ability to decide between conflicting solutions and make the difficult decisions is required. Katz (1955) states, “it includes recognizing how the various functions of the organization depend on one another and how changes in any one part affect all the others” (p. 35). This skill is most important for top managers charged with setting the vision and strategic plan for the company (Northouse, 2010). While conceptual skill is predominantly needed for executive management, middle management requires a balance of human and conceptual skills (Katz, 1955). Taken together, the three-skill approach is a new way of observing the leadership phenomenon. Katz (1955) concludes: [Skill] transcends the need to identify specific traits in an effort to provide a more useful way of looking at the administrative process. By helping to identify the skills most needed at various levels of responsibility, it may prove useful in the selection, training and promotion of executives. (p. 42) This approach will greatly influence later works by Mumford, Zaccaro, and others. Twenty years after its original publication, Katz’s work was reprinted to include a retrospective. True to its original appeal, this second printing was welcomed. It reached the public at a time when the study of leadership based on traits was almost non-existent (Zaccaro, 2007). Unlike his previous statements, several positional shifts were present in the retrospective. 33 Katz (1974) begins by dividing the human skill into two parts: leadership skill within the managers own unit and skill in intergroup relationships. Katz (1974) states, “managers are obligated to choose between gaining full support from subordinates and enjoying full collaboration with peers and/or superiors. Having both is rarely possible” (p. 101). Katz’s most dramatic repositioning happens within the conceptual skills category. Making a fundamental adjustment to his original model, Katz claims: I am now far less sanguine about the degree to which this way of thinking can be developed on the job … I question how easily this way of thinking can be inculcated after a person passes adolescence. In this sense, then, conceptual skill should perhaps be viewed as an innate ability. (p 101) This shift in position runs in direct conflict with his original thesis, and once again demonstrates the fine line between traits and skills. Lastly, Katz (1974) redacts his claim that technical skill is only required at the lower and middle organizational levels. He adjusts by recognizing that top managers will need this skill in smaller or larger degrees based on the size of their organization and the skill level of the teams they lead (Katz, 1974). These adjustments to the model have had profound ramifications. The addition of multiple levels of human skill and the concession that conceptual skills may in fact be traits opens the door for other theorists to construct more complex models. Not surprisingly, this modification to Katz’s original premise occurred just prior to the reinvigorated examination of trait-based research and models that combine both traits and skills. 34 Leadership skills model. During the early and mid-1990s, Michael Mumford and associates began to study leadership skills through a contract with the Unites States Army. Mumford and co-researchers, would attempt to create and validate a comprehensives theory of leadership skills (Yammarino, 2000). This departure from other studies of leadership relied on the leader’s ability to solve novel, ill-defined, complex problems. With complex problem solving and performance as the ultimate Leadership Outcomes, this model is comprised of four central components: Individual Attributes, Competencies, Career Experience, and Environmental Influences (Connelly, Gilbert, Zaccaro, et al. 2000; Mumford, Marks, Connelly, et al. 2000d; Mumford, Zaccaro, Connelly, et al. 2000a; Mumford, Zaccaro, Harding, et al. 2000b; Yammarino, 2000). Together, these components show a blending of concepts ranging from early skills-based research to individual trait models. Mumford, Zacarro, Connelly, et al. (2000) state, “The skills-based model of leader performance proposed … does not discount the importance of traits. In fact, within this model, skills are seen as developing as a function of the interaction between traits and experience” (p. 156). Mumford and associates’ skills-based model views Competencies (problemsolving skills, social judgment skills, and knowledge), the heart of the model, combined with Individual Attributes (general cognitive ability, crystallized cognitive ability, motivation, and personality), which lead to Leadership Outcomes (effective problem solving and performance). Outside the central components of this model are Career Experiences and Environmental Influences, believed to also affect Leadership Outcomes. The skills found in the Competencies component of the model (Problem-Solving skills, Social Judgment skills, and Knowledge) were shown to be measurable through 35 constructed response measures, developable over time through experience, and present in different amounts at different organizational levels (Connelly et, al. 2000; Mumford et al. 2000a; Mumford et al. 2000c). Taken together, these findings create a new way to view leadership development and emergence across organizational levels. Mumford, Zaccaro, Connelly, et al. (2000) conclude by stating: … this skills-based approach to leader performance adds a significant new element to our conception of leadership. It postulates that leadership may sometimes be a rather indirect phenomenon where influence is exercised through cognition and performance as well as through interpersonal interaction. We believe that cognitive performance or skills performance embedded in a distinctly social context, has always been a key aspect of leadership and is likely to become progressively more important as we move into the twenty-first century. (p. 167) This series of studies, while seemingly straightforward, exposes the complexity of leadership through the interplay of individual traits, skills, environmental factors, situations, and career experiences. The major contribution of these studies was to further the notion that skills can be developed, and therefore leaders can improve (Mumford et al. 2000a). Mumford, Zaccaro, Harding, et al. (2000) importantly state, “Thus, leaders are not born, nor are they made; instead, their inherent potentials are shaped by experiences enabling them to develop the capabilities needed to solve significant social problems” (p. 24). 2007 study: Leadership skills STRATAPLEX. Mumford et al.’s original 2007 study suggests that the leadership phenomena can be understood through a purely skills- 36 based model. This skills approach consists of four specific skill requirement categories which leaders must develop (Cognitive, Interpersonal, Business and Strategic). These leadership skill requirements are needed in varying amounts at different levels of the organizational structure (junior, middle and senior management). The Leadership Skills STRATAPLEX is a theoretical framework created to assist in the examination of leadership skills required at different organizational levels (Mumford et al., 2007). This model continues the line of skill-based leadership study previously mentioned in that it focuses not on characteristics but on the skills, leaders require. By doing so, Mumford et al. (2007) sought to “further our understanding of leadership skill requirements across organizational levels by identifying four distinct categories of leadership skill requirements that emerge differentially across organizational levels” (p. 155). The foundation of this model is the reconceptualization of previously researched leadership skills, consolidated into four distinct skills categories (Cognitive, Interpersonal, Business, and Strategic skills). These skill categories are similar to the ones described by Katz (1955), and later by Mumford and colleagues (2000b). However, while similar, these skill categories are more complex than Katz’s model and represent a more consolidated version of Mumford et al.’s skills model. Additionally, the Leadership Skills STRATAPLEX uniquely focuses on the leadership skills required by the position rather than the skills of the leaders in the position. With these assumptions set, the model builds on previous works by dividing leadership into tiers (strata) based on organizational levels. Mumford et al. (2007) claim, “Three levels are chosen for illustration in the figure, but the concept could be applied to more organizational levels” (p. 115). 37 Cognitive skills as conceptualized by Mumford’s STRATAPLEX model are comprised of six previously identified skills or characteristics. These skills focus primarily on thinking, communicating, and comprehending. As such, in a broad sense, speaking and active listening encapsulate both sides of verbal communication, much as writing and reading comprehension do for written communication. Lastly, active learning and critical thinking address one’s ability to work with new information in a logical way, thus addressing the strengths and weaknesses of each approach (Mumford et al., 2007). The Interpersonal skills category is comprised of four independent skill requirements: Social Perceptiveness, Coordination, Negotiation, and Persuasion. These skills align with previous works on social dynamics and are most similar to Katz’s Human Skills. Here, the leader must be aware of the reactions of others and must develop the ability to adjust, reconcile, and persuade team members (Mumford et al., 2007). Business skills address task completion. The ability to manage people, financial and material resources is the heart of this skills category. Additionally, leaders who possess Business skills are able to analyze the needs of others and manage personnel resources (Mumford et al., 2007). Lastly, the Strategic skills category focuses on the skills most necessary at the top levels of leadership (Mumford et al., 2007). Visioning and Systems Perception are important to creating and adapting systems. This executive operation also requires the ability to evaluate systems, adapting when necessary. Most closely associated with Mumford et, al.’s (2000b) work, the final four skills (Identification of Downstream 38 Consequences, Identification of Key Causes, Problem Identification, and Solution Appraisal) all deal with the ability to implement the previous three skill categories, while applying them to complex problem solving. Much like Katz (1955) and Mumford, Zaccaro, Connelly, et al. (2000), Mumford et al. (2007) recognizes that each of the four skills categories is required at every level of leadership to a greater or lesser degree: Although it is true that jobs at higher organizational levels are likely to have additional leadership skill requirements (e.g., visioning), they are also likely to require more fundamental leadership skills (e.g., oral communication). These Cognitive skills are likely to become increasingly important at higher organizational levels because the environment in which they are used grows more complex, novel and ill-defined (Mumford & Connelly, 1991). Thus, the positive relationship between organizational level and leadership skill requirements will exist for each of the four leadership skill requirement categories. (p. 158) Mumford et al.’s (2007) study made several hypotheses: first, that the four skill categories are distinguishable; next, that each will be required at every organizational level; and lastly, that the degree of each skill category required will differ based on the organizational level but generally increase as you ascend the organizational structure. Skill categories were measured using terms derived from the Occupational Information Network (O*NET). This job inventory was developed by the U.S. Department of Labor to create a common language that could be applied across job descriptions (Mumford et al, 2007). 39 Mumford et al.’s (2007) findings suggests that leadership skill requirements can be grouped into four categories (Cognitive, Interpersonal, Business, and Strategic). Overall, leadership skills were required in greater amounts as one ascends the organizational structure. In addition, leadership skill amounts varied in importance based on the skill category and organizational level. Lastly, Strategic and Business skills were more strongly correlated to organizational level than Interpersonal and Cognitive skills, indicating that their development may be more critical (Mumford et al, 2007). The limitations of this study are notable. Most importantly, the sample of 1,023 government workers may not be generalized to other fields, such as educational leadership. Future research is encouraged to determine if this model of leadership skills can be observed in other settings and contexts. This research intended to answer that call. Section summary. The skills approach offers a unique way of viewing the leadership phenomena. Unlike its precursor, the trait-based approach, skills approaches have been able to look at what the leader does, rather than who the leader is. This more developmental view of leadership is optimistic by nature. Northouse (2010) likens skills to “playing a sport such as tennis or golf. Even without natural ability in these sports, people can improve their games with practice and instruction” (p. 54). Northouse (2010) later concludes, “Whereas the early research on skills highlighted the importance of skills and the value of skills across different management level, the later work placed learning skills at the center of effective leadership performance at all management levels” (p. 54). This recognized shift reflects an evolution in leadership study from a leader attribute to leader skill development focus. 40 Replication Theory Gall, Borg and Gall (2003) state, “Replication is the process of repeating a research study with a different group of research participants using the same or similar methods” (p. 145). Yin (2003) further explains: The replication logic is analogous to that used in multiple experiments… Some of the replications might have attempted to duplicate the exact conditions of the original experiment Other replications might have altered one or two experimental conditions considered irrelevant to the original finding, to see whether the finding could still be duplicated. Only with such replications would the original finding be considered robust and worthy of continued investigating or interpretation. (p. 47) Replication studies enable research to be verified, extended, and generalized. While often seen as important, replication studies are seldom perused or published. Nosek and Lakens (2014) stated, “The signature strength of science is that the evidence is reproducible. However, direct replications rarely appear in psychology journals because standard incentives emphasize novelty over verification” (p. 59). Similarly, La Sorte (1972) noted: Even though many pay lip service to the importance of replication, one rarely finds extended discussion on it in textbooks on research methods. When applied in survey research the purpose of the replication is often left vague, its role in verifying data and theory is either exaggerated or ignored and, on occasion misused. (p. 218) 41 However, more recently social science has called for a renewed focus on replication as a valuable tool for furthering research validity, theory building, and generalization (Duvendack & Palmer-Jones, 2013; Klein et al., 2014; Nosek & Lakens, 2014). Gall et al. (2003) claim, “The need for replication is even more critical in education and other social science disciplines because studies often have weaknesses in methodology or very limited generalizability” (p. 42). Gall et al. (2003) also conclude that replication studies should be conducted in education research for the following reasons: To check the findings of a breakthrough study. To check the validity of research findings across different populations. To check trends or change over time. To check important findings using different methodology. To develop more effective or efficient interventions (pp. 42-43). While the reasons for conducting replications are clear, the methods and terminology remain less clear. La Sorte (1972) realized the lack of “common terminology” when discussing replication studies. Additionally, he recognized the benefit of a common framework for conducting replications contingent on the stated purpose of the study. La Sorte (1972) proposes “replication refers to a conscious and systematic repeat of an original study... The specific functions vary according to the aims of the replicator” (p. 281). From this position La Sorte (1972) creates a Replication Paradigm based on a review of the literature, comprised of the types and sub-types: 42 I. II. III. IV. Retest Replication 1. Confirmation Retest 2. Validity Retest 3. Data Measurement Retest 4. Stability Retest Internal Replication 1. Multiple Independent Samples 2. Single sample Independent Replication 1. Empirical Generalization a. Validity b. Extension c. Specification Theoretical Replication 1. Theoretical Generalization a. Inter-Societal b. Intra-Societal Figure 2. La Sorte’s (1972) Replication Paradigm. Adapted from outline in text “Replication as a Verification Technique in Survey Research: A Paradigm,” by M. A. La Sorte, 1972, The Sociological Quarterly, 13(2), pp. 218-219. Copyright 1972 by John Wiley & Sons, Inc. Retest replication. Retest replication is concerned with checking the original study for reliability and ruling out procedural errors or inconsistencies. However, in general this type of replication cannot rule out errors of process, nor extend or establish bounds of generalizability. It must proceed in one of two directions; it can directly repeat the original study or add new methods (La Sorte, 1972). Confirmation retest. According to La Sorte (1972), confirmation retest is the most basic type of replication where the original investigator will repeat the study with a similar sample and very little difference in time. This type of replication adds validity to and confirms the original findings. It does not allow for generalizability of theory. 43 Validity retest. This type of replication retest allows a single concept to be measured using several different indicators. Here the replicator intends to validate the use of a specific procedure for measurement. This retest has the power to validate new research techniques by comparing the new technique against the original study’s more accepted technique (La Sorte, 1972). Data measurement retest. Data measurement retest is a replication tool used to confirm qualitative research studies through quantitative analysis. For this replication, an original qualitative study is translated into survey items and the same or similar sample is retested. This type of replication allows for cross-methodological validation (La Sorte, 1972). Stability retest. This type of replication retest looks to establish the stability of research findings over time. Here the original study is faithfully reproduced or slightly adapted and compared to the original study’s findings. This type of replication will establish stability or suggest changes to the sample over time (La Sorte, 1972). Internal replication. Internal Replication focuses on the validity of an original study by conducting replications as part of the original study’s methodology. La Sorte (1972) suggests that this can happen one of two ways: “drawing two or more independent samples or taking a single sample and later dividing it into subsamples for the purposes of analysis and comparison” (p. 221). This type of replication, like retest, can add validity to findings. Independent replication. La Sorte (1972) recognizes “A major aim of survey research is that of generalization…Independent replication is the basic procedure for verifying an empirical generalization” (p. 222). He concludes that independent 44 replication has the ability to go beyond confirming and validating research findings to generalizing the findings or theories to larger populations. This is done by testing the findings of an original study with an independent sample drawn from related populations by a different investigator. La Sorte (1972) suggests that independent replication can answer the following questions: Is the empirical generalization valid? Does further investigation extend it to other social situations or subgroups outside the scope of the original study? Is the empirical generalization limited by the conditions of particular social situations or specific subgroups? Like Yin (2003) and Gall et al. (2003), La Sorte’s independent replication model seeks to test the universal generalization of research or establish the limits of its generalizability, thus confirming or extending knowledge in that field. Theoretical replication. Theoretical replication is the process of examining the ability to apply empirical findings to a larger theoretical framework, or the ability to generalize previous empirical findings to theoretical generalizations through the use of independent replications and retests. La Sorte (1972) claims: In order to move toward a theoretical generalization, where empirical variables which have concrete anchoring points are abstracted and conceptualized to a higher theoretical plane, it is necessary to sample a variety of groups using different indicators of the same concepts. (p. 233) This process can take the viewpoint of inter-societal or intra-societal replications. 45 Inter-societal. Inter-societal theoretical replication is primarily concerned with replicating and validating theoretical generalizations between two or more societal groups. Ultimately, the purpose of this type of replication is to determine whether a theory holds true from one society (western) to another (eastern), thus being universal (La Sorte, 1972). Intra-societal. Intra-societal theoretical replication seeks to generalize theory within one society but between different subgroups. La Sorte (1972) states, “A constant problem in theory is the attempt to demonstrate that a variety of seemingly unrelated social groups have certain analytical properties in common” (p. 224). This type of replication attempts to reproduce the findings of an original study or theory in a different setting within a common society. Section summary. Replication theory is often used in experimental research to establish validity. However, while seen as important to the social sciences replications are rarely conducted. The lack of replication found in the literature and a lack of consistent procedures or standards for replication have led to the overgeneralization of research findings in the social sciences and education specifically (Duvendack & Palmer-Jones, 2013; Gall, Borg and Gall 2003; Klein et al., 2014; Nosek & Lakens, 2014; Yin, 2003). La Sorte’s (1972) Replication Paradigm can be used as a framework for classifying replications based on their purpose while providing a common language through which to promote replication studies in the social sciences. Chapter Summary The purpose of this literature review was to acknowledge the progression of research surrounding the necessity for and development of leadership skills across 46 organizational levels. By beginning with the difficult but necessary task of defining leadership for the purposes of this study, the clear lack of consensus in the field was exposed. Context was established through a historical overview of the study of leadership from early to modern and postmodern iterations. Like the lack of consensus in defining leadership, the variety of lenses through which to observe and explain the leadership phenomenon created additional challenges. By examination of anchor studies in the trait- and skills-based models, this review of the literature paid deference to the complex nature and boundless interplay of conditions that predicate the trait and skill development of effective leadership. Finally, an examination of the importance and possible use of theoretical replication in the social sciences as a means to extend theory was discussed through the framework of La Sorte’s Replication Paradigm. The work continues. 47 Chapter Three: Research Design and Methodology Introduction Chapter Three describes the methods and research design used in this study. The chapter opens with a presentation of the research questions and hypotheses. Next, a rationale for conducting a quantitative, conceptual replication study, with the goal of generalizing the findings of previous studies to a new population, is discussed. The use of a non-experimental, cross-sectional, descriptive, and multivariate correlational research design is described. Chapter Three then discuss the methodology, participant selection, data collection techniques, and analysis methods used. Finally, threats to validity and reliability, their treatment, and the limitations of this study are examined. Research Rationale In light of the current national, state, and local educational reform movements, coupled with the large number of leaders leaving the field of education, this researcher sought to examine and understand the current leadership skills required of educational leadership, across organizational levels (Malone & Caddell, 2000). Specifically, this researcher wanted to know if Mumford et al.’s (2007) Leadership Skills STRATAPLEX could be used as a model to better understand the current leadership requirements of educational leaders in New York State, and perhaps nationally. Rationale for Replication Replication theory is primarily concerned with external validity. Johnson and Christensen (2012) argue that increasing the number of times research findings are found to be true, in different context or settings, increases the ability to generalize the findings beyond the original study or participants. Smith (1970) proposed “The most defensible 48 test of the reliability of data is provided by the replication or cross-validation study” (p. 971). Nosek and Lakens (2014) stated: Replication is a central tenet of science; its purpose is to confirm the accuracy of empirical findings, clarify the conditions under which an effect can be observed, and estimate the true effect size. Successful replication of an experiment requires the recreation of the essential conditions of the initial experiment. (p. 138) However, while replication is considered a main tenant of science, it is rarely carried out (Duvendack & Palmer-Jones, 2013; Gall et al., 2003; Klein et al., 2014; Nosek & Lakens, 2014; Yin, 2003). Replication studies can take several forms other than direct replication designs, which repeat the original study. This 2016 study conducted an Intra-Societal Theoretical Replication as outlined by La Sorte (1972) by replicating the critical aspects of the 2007 study in a new setting and context within a single society. Undertaking this study in a different context (New York State) and with a different sample (Educational Leaders) adds to the generalizability of Mumford et al.’s original findings. The replication of Mumford et al.’s (2007) leadership skills STRATAPLEX may further validate this model or establish a clear boundary for its use, ultimately increasing what is known about the requirements of leaders in the field of K-12 public education in New York State. 2007 Study Statistical Findings Mumford et al. (2007) sought to measure the leadership skills required of, “1,023 “professional employees working in an international agency of the U.S. government” (p.159). These positions encompassed five different career specialties; administrative, public relations, economic analysts, political analysts and multifunctional positions 49 comprised of a combination of the other four. These leaders were located in 156 different countries. Specifically, Mumford et al. (2007) hypothesized the following: (H1). The Cognitive, Interpersonal, Business, and Strategic skill categories will be empirically distinguishable. (H2). Leadership skill requirements will vary by skill category such that Cognitive skills will be needed the greatest amount, followed by Interpersonal, Business, and Strategic skills, respectively. (H3). Cognitive, Interpersonal, Business, and Strategic skill requirements will be positively related to the job’s level in the organization. (H4). Leadership skill requirements will interact with organizational level such that: a. Strategic skill requirements will be more strongly related to organizational level than Cognitive, Interpersonal, and Business skill requirements. b. Business skill requirements will be more strongly related to organizational level than Cognitive or Interpersonal skill requirements. c. Interpersonal skill requirements will be more strongly related to organizational level than Cognitive skill requirements. (pp. 157159) Each leadership skill requirement level was measured using a 7-point Likert scale with behavioral anchors provided for low, medium, and high points on the scale. Cognitive skill requirements were measured through six items with an internal 50 consistency reliability of .90. Interpersonal skill requirements were measured by four items with an internal consistency reliability of .84. Business skill requirements were assessed through the use of four questions with and internal consistency reliability of .75. Strategic skill requirements were assessed through six items with an internal consistency reliably of .91 (Mumford et al, 2007). Hypothesis 1. “The Cognitive, Interpersonal, Business, and Strategic skill categories will be empirically distinguishable” (Mumford et al., 2007, p. 157) was measured through confirmatory factor analysis, using the maximum likelihood method of covariance structure analysis (Structural Equation Modeling). One and four factor measurement models were built, tested and compared using a Chi-squared difference test. Both the four factor and one factor model fit the data (χ2=324.1 and χ2=714.8) respectively. However, the χ2 difference test suggests that the four factor model better fit the data than the one factor model (Δ χ2=390.7, 6df, p<.01) thus, supporting hypothesis 1 (Mumford et al., 2007). Hypothesis 2. “Leadership skill requirements will vary by skill category such that Cognitive skills will be needed the greatest amount, followed by Interpersonal, Business, and Strategic skills, respectively” (Mumford et al., 2007, p. 158) was measured using multivariate analysis of covariance (MANCOVA). This analysis evaluated the mean differences in leadership skill requirements among organizational levels, while controlling for organizational specialty and location. Mumford et al. (2007) used a within-subject design to control for rater effects as leaders gave ratings for each of the skill levels (Cognitive, Interpersonal, Business, and Strategic). The Bonferroni procedure was used when comparing means. The result yielded significant differences in the 51 means, with controls, for all four skills at a p<.05 level. Adjusted means for each leadership skill indicated that Cognitive skill were needed in greater amounts followed by Interpersonal, Business and Strategic skills (x̄=5.05, x̄=5.11, x̄=5.34, x̄=5.44) respectively. However, pairwise comparisons indicated that Cognitive skills were needed in greater amounts than Business, that Cognitive and Interpersonal skills were needed in greater amounts than Strategic, and all other mean differences while in the predicted direction were not significant at the p<.05 level. Thus, hypothesis 2 was found to be partially supported (Mumford et al., 2007). Hypothesis 3. “Cognitive, Interpersonal, Business, and Strategic skill requirements will be positively related to the job’s level in the organization” (Mumford et al., 2007, p. 159) was measured through the use of partial correlations. Hypothesis 3 was fully supported as all partial correlations, when controlling for the effects of the control variables, indicated positive significant relationships between all four skills requirements and organizational level such that Cognitive, Interpersonal, Business and Strategic skills correlated at (r=.15, p<.05; r=.20, p<.05; r=.28 p<.05; and r=.25, p<.05) respectively (Mumford et al., 2007). Hypothesis 4. Leadership skill requirements will interact with organizational level such that: a. Strategic skill requirements will be more strongly related to organizational level than Cognitive, Interpersonal, and Business skill requirements. 52 b. Business skill requirements will be more strongly related to organizational level than Cognitive or Interpersonal skill requirements. c. Interpersonal skill requirements will be more strongly related to organizational level than Cognitive skill requirements.” (Mumford et al., 2007, p. 159) This hypothesis was tested through an examination of the correlation coefficients between the leadership skill requirement and organizational level while controlling for organizational specialty and location. Correlations were compared using the Bonferroni correction for multiple comparisons. Hypothesis 4a was partially supported in that Strategic skill requirements (r=.25, p<.05) were more strongly related to organizational level than Cognitive and Interpersonal skill but not Business skill (r=.15, p<.05 and r=.20, p<.05; r=.28, p<.05) respectively. Hypothesis 4b was fully supported as a stronger relationship existed for Business skill requirements and organizational level than that for Interpersonal and Cognitive (r=28, p<.05; r=.20, p<.05 and r=.15, p<.05) respectively. Finally, Hypothesis 4c was fully supported as the strength of the relationship between Interpersonal skill requirements and organizational level was stronger than that of Cognitive skill requirements (r=.20, p<.05 and r=.15, p<.05) respectively (Mumford et al., 2007). Modifications to 2007 Study Modifications were made to the original 2007 study to allow for its appropriate application to the new setting and context. Mumford et al.’s sample was drawn from 1,023 U.S. professional government employees stationed around the world. Each 53 employee was classified as belonging to one of three organizational levels (junior, midlevel, and senior) dependent on their years of service (1-5, 6-20, and 21+ years). This classification system was appropriate for the 2007 study as the U.S. government practices a promotional system where each employee must pass through each of the lower organizational level, being promoted or forced to retire. Additionally, Mumford et al. (2007) controlled for several variables such as organizational specialty and location. Finally, the survey tool was distributed in paper form. The current study drew its sample from K-12 public school and district leaders in the State of New York. Each leader was classified into one of four organizational levels (Assistant Principal, Principal, Assistant Superintendent, and Superintendent). This differs from the original study in that an additional level was added and leaders must not always pass through each of the lower organizational levels. This classification system was appropriate as it represented the four major educational leadership levels found in New York State. The current study differed from the original in that the survey was distributed electronically. Also, this researcher controlled for several different variables such as years of experience, education level, district size, and district setting. A comparison of both the 2007 and 2015 study can be seen in Table 1. 54 Table 1. Comparison of the 2007 Study and Current Study 2007 Study Current Study Location World-Wide New York State Subjects Professional U.S. Government Employees K-12 Educational Leaders Organizational Levels Junior Mid-level Senior Assistant Principal Building Principal Assistant Superintendent Superintendent Survey Tool Paper, Mailed Electronic, Surveymonkey Control Variables Organizational Specialty Location Years of Experience Education Level District Setting Research Question and Hypothesis Quantitative research is concerned with answering “what,” “where,” and “when” questions and uncovering the variation and complexity that exists in the natural and behavioral sciences (Butin, 2010; Pedhazur, 1997). Given that leadership is complex and that leadership skills models are rarely studied, a replication of Mumford, Campion, and Morgeson’s study was conducted in hopes of generalizing and validating the Leadership Skills STRATAPLEX framework within a new population (Educational Leaders) and context (New York Public Schools; Bass, 1990; Mumford et al., 2007; Mumford, Zaccaro, Harding, Jacobs, & Fleishman, 2000). Additionally, this researcher intended to better understand the current leadership skills required by educational leaders in New York State, across organizational levels, with the hope of directing future leadership training, professional development, and the hiring and promotional practices of school districts. In this study, the researcher explored possible answers to the question: 55 To what extent is Mumford et al.’s (2007) Leadership Skills STRATAPLEX model generalizable to leaders of New York State’s K-12 public schools? In an attempt to answer this question and describe the leadership skills required by leaders of New York’s K-12 public schools, a theoretical replication was conducted which tested the following hypotheses as found in Mumford et al’s., (2007) original study (pp. 151-159). (H1). The Cognitive, Interpersonal, Business, and Strategic skill categories will be empirically distinguishable. (H2). Leadership skill requirements will vary by skill category such that Cognitive skills will be needed the greatest amount, followed by Interpersonal, Business, and Strategic skills, respectively. (H3). Cognitive, Interpersonal, Business, and Strategic skill requirements will be positively related to the job’s level in the organization. (H4). Leadership skill requirements will interact with organizational level such that: a. Strategic skill requirements will be more strongly related to organizational level than Cognitive, Interpersonal, and Business skill requirements. b. Business skill requirements will be more strongly related to organizational level than Cognitive or Interpersonal skill requirements. c. Interpersonal skill requirements will be more strongly related to organizational level than Cognitive skill requirements. 56 2016 Research Design and Methodology This theoretical replication, descriptive, non-experimental, quantitative, crosssectional study employed the use of descriptive, bivariate, multivariate, correlational statistics as-well-as structural equation modeling to test the aforementioned hypotheses, thus, answering the research question. This study first examined whether a four-factor model of leadership skills was empirically distinguishable and if so, what is the relationship between these leadership factors and leaders’ organizational level while controlling for other factors. Finally, this study examined the proportion of leadership skill required by leaders at different organizational levels regardless of their experience, educational level, and district setting. Replication. While often neglected by the social sciences, the purpose of direct and conceptual replication studies is to confirm, validate, generalize, and extend knowledge (Burman, Reed, & Alm, 2010; Duvendack & Palmer-Jones, 2010; ReiterPalmon &Tinio, 2014; Smith, 1970). In this study, the researcher intended to apply The Leadership Skills STRATAPLEX model to a new setting and in so doing attempted to determine the generalizability of a previously validated model to determine whether its relevance extends to broader populations. Descriptive research. The goal of descriptive research is to describe existing populations and make inferences about larger populations through the establishment of trends, while providing an accurate picture of the characteristics of the phenomenon being studied (Creswell, 2003; Johnson & Christensen, 2012; Leedy & Ormrod, 2005). Descriptive research is limited in that while allowing for rich description, it does not focus on cause-and-effect relationships; instead, it describes existing variables and the 57 relationships between those variables (Johnson & Christensen, 2012; Singh, 2007). Much is unknown concerning leadership skills, and less is known about how these skills relate to the positions held by leaders (Bass, 1990; Mumford et al., 2007). Additionally, no study of K-12 public school leadership skill requirements was found in the literature, justifying both a descriptive and correlational research design. Non-experimental research. Non-experimental research is undertaken when the manipulation of the independent variable(s) is not possible or ethical (Johnson & Christensen, 2012). While often weaker than experimental and quasi-experimental research in its ability to generalize, non-experimental quantitative research is able to answer many important research questions found in the field of education (Butin, 2010; Johnson & Christensen, 2012). Manipulation of variables, namely educational leadership positions, is not possible; therefore, an empirical, non-experimental, quantitative design was used. Cross-sectional design. Consistent with Mumford et al.’s (2007) original study, a cross-sectional design was used for this study due to the large sample of educational leaders and the resources available to the researcher. According to Johnson and Christensen (2012), cross-sectional research allows for large amounts of diverse data to be collected in a short amount of time. Additionally, a cross-sectional design facilitated data collection from a large number of participants comprised of multiple groups. However, several weaknesses exist when using cross-sectional research designs, including the inability to observe changes over time or account for the specific mood of the respondent as they completed the survey (Johnson & Christensen, 2012). Despite these limitations, and due to the nature of this study and its research question, a single 58 measure in time was appropriate for gathering a large sample, with the ability to describe only the current state of leadership skills required of educational leaders in New York State. Statistical analysis. Since this study attempted to describe both the current leadership skills required by educational leaders and compare a one factor and two factor model, structural equation modeling and multivariate procedures were applied. Specifically, Multivariate Analysis of Variance and Covariance (MANCOVA), withinsubject design, was used to determine the effects of several independent variables on multiple dependent variables while controlling for the impact of covariates (George & Mallory, 2011). This researcher explored the use of Mumford et al.’s Leadership Skills STRATAPLEX as a possible model for describing the current state of public school leaders in New York State through a replication study. This study used structural equation modeling (SEM) to conduct a confirmatory factor analysis to determine and compare the model fit of a one and two-factor model. This was followed by descriptive, bivariate, correlational, and multivariate statistics. Setting Seven hundred and eighty-five New York State public school districts were the population target of this study. The criteria for inclusion was as follows: public school districts (BOCES, City, Union Free, Independent Union Free, Central, City Central, Independent Central, and Common) as listed by New York State Department of Education records for 2013. The “Big Five” city school districts (New York City, Buffalo, Rochester, Syracuse, and Yonkers) were not included as their leadership 59 structure and size often differ from the rest of the state. Parochial, private, and charter districts were not included as their leadership certification requirements vary. From the remaining districts, district-level leadership and leaders of secondary schools (middle and high) were considered. Any school district that did not have a leader at each of the four stratified leadership levels (Superintendent, Assistant Superintendent, Building Principal, and Assistant Principal) was not included. Sampling Upon receiving approval from the Manhattanville College Institutional Review Board, electronic surveys were administered, through email, to 5,450 qualifying educational leaders in New York State (Superintendents, Assistant Superintendents, Building Leaders, and Assistant Principals) for which contact information was publically available. This email described the purpose of the study, parameters for participation, and statements concerning confidentiality and consent. Electronic surveys were emailed to each of the participants containing a hyperlink to SurveyMonkey.com. Follow up emails were sent to participants who did not initially respond, one, two, and four weeks after the original distribution. Surveys are appropriate for obtaining data concerning attitudes, beliefs, values, and perceptions and can be used to collect several additional types of information from participants (Johnson & Christensen, 2012). Electronic surveys capitalize on the ease of creation, distribution, and collection (Butin, 2010). The survey was distributed during the months of October and November, 2014. This period of time generally represents a slower period for educational leaders as students have transitioned back to school and the holiday season has not started, serving as an opportunity to maximize participation. Once the survey 60 period closed, responses were uploaded to IBM’s SPSS software, a powerful statistical analysis tool (George & Mallery, 2011). The specific statistical analysis and techniques that were used are discussed in subsequent sections of Chapter Three and Four. Instrumentation This researcher developed an interest in leadership skill models as an alternative to trait-based models while reading a special issue of The Leadership Quarterly [The Leadership Quarterly An International Journal of Political, Social and Behavioral Science] dedicated to the topic. Ultimately, this led to further inquiry and the development of the overarching research question: To what extent is Mumford et al.’s Leadership Skills STRATAPLEX model generalizable to leaders of New York State’s K12 public schools? In order to address the first part of the research question, after receiving permission from the author of the original study, the researcher conducted a theoretical replication study using Mumford et al.’s (2007) validated, twenty-one question survey. The addition of demographic questions specific to educational leaders was added. Appendix B shows the survey for this study. The survey consisted of four sections, which intended to measure the four leadership skill requirements found in the STRATAPLEX (Cognitive, Interpersonal, Business, and Strategic skill requirements). Each survey question in these sections was measured using a 7-point Likert scale, where one indicated low levels required and seven indicated high levels required. Anchor statements, derived from the Occupational Information Network developed by the U.S. Department of Labor, were also provided. 61 The first section of the survey consisted of six questions designed to measure the Cognitive skill requirements of the leader’s position. Each question measured a specific aspect of Cognitive skills, including: speaking, active listening, writing, reading comprehension, active learning, and critical thinking. Taken together, these questions formed the Cognitive skill requirement index (CSR). The next section included four questions intended to measure Interpersonal skill requirements. The specific components of Interpersonal skill requirements are: social perceptiveness, coordination, negotiation, and persuasion. These four questions were combined to create the Interpersonal skill requirement index (ISR). The third section of the survey measured the Business skill requirements of the leader’s job. This section consisted of four survey questions intended to measure the specific aspects of Business skills, which included: operations analysis, management of personnel resources, and management of financial resources. These questions were used to create the Business skill requirements index (BSR). The fourth section of the survey asked leaders to identify the level of Strategic skill required by their leadership position. This section consisted of seven survey questions, each of which measures one of the following components of Strategic leadership skill: visioning, systems perception, system evaluation, identification of downstream consequences, identification of key causes, problem identification, and solution appraisal. These questions were combined to create the Strategic skill requirement index (SSR). The final section of this survey consisted of demographic questions that were used as controls for this study. These questions asked the educational leaders to identify 62 current leadership level, gender, total years as a leader, total number of graduate credits earned beyond the masters’ degree, district enrollment, district geographic setting (rural, suburban, urban small city), and socioeconomic status (SES). The intent of these questions was to provide a context in which to examine leadership and to determine the unique effects of these factors on the leadership skill requirements at each organizational level. Dependent Variables Dependent variables were organized into four leadership skill requirement groups: Cognitive, Interpersonal, Business, and Strategic (Mumford et al., 2007). Cognitive skill requirements. As conceptualized by Mumford et al. (2007), Cognitive skills were the basic skills and capacities that allowed leaders to think, communicate, learn, and adapt. These foundational skills were comprised of the following specific attributes: Speaking. Speaking skill requirements referred to the leader’s required level of oral communication skills. This skill involved the leader’s ability to communicate effectively, receiving and disseminating information (Mumford et al., 2007). Active listening. Active listening skill requirements were those skills required by leaders which allowed them to understand questions, thus achieving complete understanding of a problem or situation (Mumford et al., 2007). Writing. Writing skill requirements spoke to the leaders required skill level concerning written communication. The leader’s ability to clearly communicate ideas while keeping the message and audience in mind were a key Cognitive skill (Mumford et al., 2007). 63 Reading comprehension. Reading comprehension was the leader’s required skill level to read, comprehend, and evaluate complex texts (Mumford et al., 2007). Texts were often job or level specific. Active learning. Learning how to adapt to new situations and apply new information to these situations were examples of active learning skills (Mumford et al., 2007). Critical thinking. Critical thinking skill requirements referred to the level of skill required by the leader’s position to use logic and analyze the strengths and weaknesses of different approaches. Interpersonal skill requirements. For Mumford et al. (2007), Interpersonal skills “grows out of what previous research has referred to as social capacities, social judgment, social complexity and differentiation” (pp. 156-157). This skill enabled a leader to understand the actions and reactions of others, and allowed them to coordinate, influence, and negotiate with others. Social perceptiveness. Social perceptiveness was the level of skill required by a leader to understand others, namely their actions and reactions (Mumford et al., 2007). Coordination. Coordination represented a leader’s required skill level when adjusting to the actions of others or coordinating with others to achieve a common outcome (Mumford et al., 2007). Negotiation. Negotiation skills were best described as the level of skill required to bring others together and/or to reconcile differences between two or more parties (Mumford et al., 2007). 64 Persuasion. Persuasion described the skill required by leaders to persuade others to change their minds or behaviors to align with organizational goals (Mumford et al., 2007). Business skill requirements. Business skills were originally conceptualized as management skills in previous works of Mumford et al. For the purposes of the STRATAPLEX, they described a leader’s daily management of the “functional areas that create the context in which most leaders work,” as well as management of resources specific to the leader’s area of responsibility (Mumford et al., 2007, p. 157). Operations analysis. Operations analysis described the leaders’ specific, daily job operations and the level of skill required to complete them (Mumford et al., 2007). Management of personnel resources. Management of personnel resources referred to the leader’s required amount of skill in motivating and directing people in their daily work responsibilities (Mumford et al., 2007). Management of financial resources. Management of financial resources described the leader’s required skill level for budgeting how money will be spent to accomplish organizational goals (Mumford et al., 2007). Management of material resources. Management of material resources referred to the leader’s required ability to manage, obtain, and appropriately use materials, facilities, and equipment to accomplish one’s job (Mumford et al., 2007). Strategic skill requirements. Strategic skills were those that allowed leaders to manage and understand complexity, deal with ambiguity, and see the potential consequences of decisions from a systems level (Mumford et al., 2007). These skills 65 were most important to organizational leaders who are responsible for conceptualizing and executing the steps necessary for long-term planning. Visioning. Visioning was the required leadership skill that allowed a leader to understand and develop a picture of how the organization should function (Mumford et al., 2007). Systems perception. Systems perception was the leader’s ability to determine when key changes have happened or are most likely going to happen to the organization (Mumford et al., 2007). Systems evaluation. Systems evaluation was a skill that allowed leaders to evaluate the performance of an organization (Mumford et al., 2007). Identification of downstream consequences. Downstream consequences were the predicted consequences of organizational decisions or indecision (Mumford et al., 2007). A key Strategic skill was the ability of a leader to consider and make decisions with these potential consequences in mind. Identification of key causes. A leader’s ability to identify key causes was described as the awareness of changes in the field and/or outside policy implications on the organization (Mumford et al., 2007). Problem identification. Problem identification was the leader’s ability to recognize important organizational problems that need to be solved. These problems can be routine or novel (Mumford et al., 2007). Solution appraisal. Solution appraisal was the leader's ability to reflect on and learn from the outcomes of problems solved, and redirect efforts when necessary (Mumford et al., 2007). 66 Independent Variables There were four independent variables that were considered by this researcher: Organizational levels. For the purposes of this study, organizational level was understood to describe the stratified leadership level to which the surveyed leader belonged: (a) Assistant Principal, (b) Building Principal, (c) Assistant Superintendent, and (d) Superintendent of schools or other chief executive. Control Variables Leader’s years of experience. The leader’s years of experience was defined as the total number of years served as an educational leader at any of the four stratified organizational levels. Leader’s level of education. For the purpose of this study, the leader’s level of education was represented by the number of credit hours earned beyond the master’s degree and highest degree earned (MA +30, MA+45, MA+60, Doctorate). District setting. The district setting distinguished between rural, suburban, and small city school districts. Statistical Analysis In order to answer the research question through addressing the corresponding hypotheses, quantitative methods with statistical analysis were conducted, which included: descriptive statistics, correlation coefficients (bivariate and partial), confirmatory factor analysis, a Chi-squared difference test between two structural equation measurement models, and a one-way multivariate analysis of covariance (MANCOVA). 67 The data were first screened for missing data, outliers, normality, linearity, and homoscedasticity. Descriptive univariate statistics were then presented, including: frequencies, mean, median, mode, and standard deviations for all dependent, independent, and control variables. Latent variables were then created utilizing both traditional confirmatory factor analysis and confirmatory factor analysis utilizing structural equation modeling. According to Costello and Osborne (2005) “The aim of factor analysis is to reveal any latent variable that causes the manifest variables to covary” (p.2). As with most research, this researcher collected and analyzed the data with an a priori idea about the variables already set, and thus intended to test the hypothesis that all four of the leadership skill requirements would be distinct (Costello & Osborne, 2005; Preacher & MacCallum, 2003). Additionally, Kaiser-Mayer-Olkin’s (KMO) measure of sample adequacy was used along with Bartlett’s Test of Sphericity to determine if the data were adequately distributed and analysis could therefore, be conducted. Factors were created and reliability testing conducted in order to further test construct reliability and theory validity, via calculating Cronbach’s alpha (George & Mallory, 2011). Bivariate correlations were examined. Bivariate correlation coefficients (Person’s r) were calculated to examine the strength and direction of the relationships between all study variables. A two-tailed test of significance was used with a probability level of p<.05 being considered significant (Cohen, 1992). Partial correlations between leadership skill indices (Cognitive, Interpersonal, Business and Strategic) and organization level, while controlling for variables (years of experience, education level, and district setting) were calculated. This procedure intended to examine the correlations 68 between variables, while controlling for the effects of other variables (George & Mallory, 2011). Structural equation modeling was used to create both a one and four factor model of leadership skills required. These models were tested for general model fit using the following fit indices as suggests by Byrne (2010) and Hoyle (2012): Chi-square (CS), Comparative Fit Index (CFI), Normed Fit Index (NFI), and Root Mean Square Error of Approximation (RMSEA). The two models were then compared via a Chi-square difference test to determine which model best fit the data. A multivariate analysis of covariance (MANCOVA) is a procedure that determines the effects of several independent variables on multiple dependent variables examining whether or not differences exist among dependent variables (George & Mallery, 2009). This procedure was used to determine if different amounts of one or more of the leadership skills were required at different organizational levels, while controlling for the leader’s experience, educational level, and district setting. While multiple ANCOVAs could have been run, a MANCOVA allowed for analysis to be done through the utilization of one test thus protecting against possible type 1 errors inherent in running multiple tests. Limitations of the study One limitation of this study was that while it controlled for several factors suspected to directly or indirectly affect school leadership, other factors may exist that were not considered. Their omission and the lack of skills-based leadership research conducted in the public school setting was a major limitation of this study. 69 Cross-sectional research designs presented several weaknesses, including an inability to observe changes over time or the ability to account for the specific mood of the respondent when the survey was administered. These limitations affected the reliability of the survey data collected. As a theoretical model, The Leadership Skills STRATAPLEX was originally tested and validated in a sample of professional employees working in and international U.S. government agency that used an “up or out” promotional system (Mumford et al., 2007). Consequently, if you failed to be promoted within a specific period of time, you were forced to retire, leaving a sample of leaders who demonstrated upward mobility within the organization. Conversely, when studying leaders of public schools this dynamic does not exist. Educational leaders hold positions in the organizational structure for many reasons, both personal and professional, opting at times to remain at a particular organizational level for long periods of time. Finally, this study attempted to examine the perceived leadership skills required of educational leaders in NYS at different organizational levels. Because of the specific sample, the findings cannot easily be generalized to other states or nonpublic schools. Additionally, descriptive research methodology was used, and by its nature descriptive research seeks to describe rather than predict, eliminating the ability of this study to serve as a predictive tool. Chapter Summary One goal of this study was to determine if the Leadership Skills STRATAPLEX could be generalized to leaders of public education in New York State, thus furthering the limited work in the field of leadership skills and educational leadership. Ultimately, this 70 study intended to impact leadership training, hiring protocols, and professional development practices. While this study was only one step towards that end, further research concerning educational leaders, examined through a skills-based approach is needed. This study employed the use of quantitative, cross-sectional survey methodology to describe the current skill requirements of New York State’s educational leaders, while also attempting to validate the Leadership Skills STRATAPLEX proposed by Mumford et al (2007). Additionally, this researcher intended to examine the unique effects of several control variables that have been shown to affect how leadership is carried out, thus predicting the skill requirements necessary in specific settings and conditions. While thousands of studies of leaders and leadership have been conducted through the identification and examination of traits, with a focus on the leader, few have approached the phenomena from a skills-based approach with a focus on the job of the leader (Bass, 1990; Mumford et al., 2007; Northouse, 2010). This researcher intended to add to the existing body of literature with the hope of furthering our understanding of educational leadership during a time of dramatic change. 71 Chapter Four: Data Analysis and Findings Introduction In this chapter, the findings of this study are presented through descriptive, univariate, and multivariate statistics. Additionally, confirmatory, measurement, and structural models were explored with the purpose of determining the extent to which Mumford et al.’s four-factor Leadership Skills STRATAPLEX model was generalizable to public school leaders in New York State. Replication methodology was utilized with many of the original statistical tests performed, which allowed a comparison between the two studies and their findings. Particular attention was paid to the strength of relationship between the independent variables “Organizational Level” (Superintendent, Assistant Superintendent, Building Principal, and Assistant Principal), and dependent variables “Leadership Skills Required” (Cognitive, Interpersonal, Business, and Strategic), and the mean differences between these variables, while controlling for the effects of covariates. Structural equation modeling (SEM), utilizing confirmatory factor analysis, was used to compare the relative fit of a four-factor leadership skills STRATAPLEX model compared with a one-factor model of leadership skills. Further tests were conducted beyond the scope of the 2007 study in order to more deeply examine the mean differences that existed between the dependent variables “Leadership Skills” and the grouping variable “Organizational Level”, controlling for other factors. These tests included a full factorial, one-way multiple analysis of variance and covariance (MANCOVA) with LSD post hoc tests. All tests were completed in order to formulate conclusions to the research question and corresponding hypotheses. These questions and hypotheses are listed below to provide the context and logical progression of this chapter. 72 Research Question To what extent is Mumford et al.’s Leadership Skills STRATAPLEX model generalizable to leaders of New York State’s K-12 public schools? Hypotheses In an attempt to answer this question and describe the leadership skills required by leaders of New York’s K-12 public schools, a theoretical replication was conducted which tested the following hypotheses as found in Mumford et al. (2007) original study (pp. 151-159). (H1). The Cognitive, Interpersonal, Business, and Strategic skill categories will be empirically distinguishable. (H2). Leadership skill requirements will vary by skill category such that Cognitive skills will be needed the greatest amount, followed by Interpersonal, Business, and Strategic skills, respectively. (H3). Cognitive, Interpersonal, Business, and Strategic skill requirements will be positively related to the job’s level in the organization. (H4). Leadership skill requirements will interact with organizational level such that: a. Strategic skill requirements will be more strongly related to organizational level than Cognitive, Interpersonal, and Business skill requirements. b. Business skill requirements will be more strongly related to organizational level than Cognitive or Interpersonal skill requirements. c. Interpersonal skill requirements will be more strongly related to organizational level than Cognitive skill requirements. 73 Review of The Study’s Purpose The purpose of this research study was to examine, through confirmatory factor analysis, the relative fit of Mumford et al.’s four-factor (STRATAPLEX) model for leadership skills with the leadership skill requirements reported by educational leaders in New York State. Additionally, this researcher sought to re-conceptualize the Leadership Skills STRATAPLEX model for educational leaders in New York State, providing insight into the complex roles and unique combinations of skills required by these educational leaders. The ultimate goals of this researcher was to examine and articulate these findings in a practical way in order to maximize their potential impacts on school leadership, development, training programs, hiring, and promotional practices. Summary of Setting and Data Collection Procedures Setting. The study’s setting included a stratified sample of public school leaders (Assistant Principals, Principals, Assistant Superintendents, and Superintendents) in New York State, exclusive of the Big Five city school districts (New York City, Buffalo, Rochester, Syracuse, and Yonkers), for which contact information was publicly available. Data collection. Contact information for principals and superintendents is annually compiled by the New York State Education Department (NYSED) and was obtained from a readily available database found at the NYSED’s Department of Statistics. Since no such database exists for assistant superintendents and assistant principals, this contact information was obtained and compiled manually. A database of publically available assistant superintendents and assistant principals was created by visiting all corresponding district and school web pages for which a superintendent and principal was reported in the NYSED annual report. Publically available contact 74 information was obtained and confirmed for 5,450 school leaders and recorded in an Excel spreadsheet by name, position, title, and district. On October 25, 2014, this researcher sent those identified leaders an electronic request soliciting participation in this study. The request included a detailed description of the study, an active consent form, this researcher’s contact information, and a link to the twenty-eight question survey through SurveyMonkey’s software. Two follow-up requests for participation were made on November 1, 2014, and November 11, 2014, respectively. These follow-up requests were made only to those who had yet to respond and to those who had begun the survey but failed to complete it. Additionally, this researcher attended the School Administrators Association of New York State (SAANYS) annual conference on October 26-27, 2014, where 250 business cards containing a Quick Response Code (QR scan Code), survey information, and a request to participate were distributed to educational leaders from across New York State. The data collection window concluded on November 22, 2014, and the electronic survey collection devices were closed. Of the original 5,450 leaders who were asked to participate, 207 had previously “opted out” of SurveyMonkey surveys and were immediately eliminated by the survey collection system. During the survey collection period, 40 additional leaders “opted out” of SurveyMonkey surveys, and 37 leaders indicated that they did not wish to participate in the study by selecting the choice “no” on the active consent page. Finally, 89 email addresses were returned as invalid or changed. This researcher was unable to find valid email addresses for these leaders. Of the 5,077 leaders who received requests to participate, 1,235 (24%) consented to participate in the study. Of the 1,235 who consented to participate, 1,018 completed both sections one and two of the survey with 75 nine having completed section one but having failed to complete section two. The data from these 1,027 surveys surveyed leaders were further analyzed. Pre-Analysis Data Screening Missing data. Data analysis began with the survey data collected from 1,027 school administrators in New York State, ranging in position from assistant principal to superintendent. The use of multiple stochastic imputations found within SPSS was considered for replacing the missing demographic data for nine individuals who did not complete section two (demographics) of the survey. Ultimately, due to the relatively small number of incomplete surveys, all nine incomplete responses were removed. The remaining 1,018 complete surveys yielded a 19% return rate of all leaders surveyed. Responses from the survey were then imported into SPSS from SurveyMonkey for analysis. Outliers. According to Mertler and Vannatta (2005), outliers are “cases with unusual or extreme values at one or both ends of a sample distribution” (p. 27). Researchers must be cognizant of outliers and the ill effects they can have on one’s ability to accurately interpret their data. Outliers can be both univariate and multivariate in nature, exhibiting extreme values on one variable or combinations of variables, respectively (Mertler & Vannatta, 2005; Tabachnick & Fidell, 1996). For this reason, both univariate and multivariate outliers were examined and treated through trimming. After the nine incomplete surveys were removed, the remaining 1,018 responses were evaluated for univariate normality of distribution and the “outlier labeling rule” was applied (Hoaglin, Iglewicz, & Tukey, 1986). Much like Tukey (1977), Iglewicz and Banerjee (2001) acknowledged the importance of establishing simple and effective 76 “outlier screening procedures that can easily be applied to both large and more moderate data sets” (p. 1). This becomes increasingly important, as modern research tends to evaluate ever larger datasets, which makes the traditional visual inspections of frequency distributions or histograms difficult (Inglewicz & Banerjee, 2001; Mertler &Vannatta, 2005). As this study fell within the range of a moderately large sample size as defined by Iglewicz and Banerjee (2001), and in light of Hoaglin, & Iglewicz’s (1987) revision of Tukey’s (1977) “Boxplot outlier labeling rule,” the following equation and g value were used to identify univariate outliers in this sample ((Q1-g(Q3-Q1), Q3+g(Q3-Q1) where Q1= the 25th percentile, Q3 = the 75th percentile, and g=2.2). While originally a g-value of 1.5 and 3.0 had been suggests, these values tend to yield both overly liberal and conservative results, resulting in the elimination of, or failure to eliminate, valuable data. For this reason, Hoaglin and Iglewicz’s (1987) suggests value of 2.2 is now widely used. Next, the mean scores for “total leadership skills” were calculated and evaluated by creating a new variable in SPSS consisting of the responses to survey questions 2-22, which measured the four factors of leadership skill. The mean scores were summed and divided by 21, resulting in an average “total leadership skills” score for each participant. SPSS was then used to calculate percentile scores (25th percentile = 5.333 and 75th percentile = 6.2381), respectively. Upper and lower boundaries for the outlier labeling rule were established using the previously discussed method (6.2381+2.2(6.2381-5.333) =6.93) and (5.333-2.2(6.2381-5.333) =4.64), respectively, and responses outside these limits were identified and considered for removal. Next, the box, stem, and leaf plots for each of the 21 leadership-measuring variables were reviewed as one group. Finally, the 21 variables were evaluated again using box, stem, and leaf plots by organizational group 77 (assistant principal, building principal, assistant superintendent, and superintendent). This evaluation confirmed that outliers existed both at the univariate and multivariate levels even after the outliers for construct “total leadership” were labeled and removed. Additionally, some cases were considered outliers when separated by organizational level but not when viewed as part of the whole group. For this reason, rather than this univariate approach, a multivariate approach that took into account each variable’s interaction with all other variables, was explored and ultimately used for the labeling and removal of outliers from this data set. Tests for multivariate outliers, “those comprised of unusual combinations of scores on two or more variables” (Mertler & Vannatta, 2005, p.29), were conducted using the Mahalanobis Distance Test (Mahalanobis D²), which measures a case’s distance from the centroid, with the centroid being a composite of the means of all variables (Cohen, Cohen, West & Aiken, 2003). SPSS was used to calculate the Mahalanobis D² values for all leaders sampled, and the variable “MAH_3” was created. The analysis considered each of the 21 variables intended to measure “leadership skills” in light of the control variables, which sought to measure a leader’s, “Years of Experience”, “Education Level”, “Organization Level”, and “District Setting”, respectively. The resulting D² standardized scores were evaluated as chi-square with the degrees of freedom equal to the number of variables being evaluated. Scores were considered a significant Mahalanobis distance from the centroid at an alpha level of p<.001 (Mertler & Vannatta, 2005). SPSS was then used to transform the Mahalanobis D² scores to probability scores and 67 cases were found to be significant at the p<.001 level. These cases were again individually reviewed, identified as multivariate outliers, and trimmed from the data set. 78 Finally, the dataset was screened for “total leadership skill” scores equal to any whole number, suggesting that the leader had simply selected the same number on the Likert scale for every question. The researcher found that 29 leaders had selected choice 7 on the Likert-scale as their sole answer for all questions, two had selected choice 6 as every answer, and two had selected choice 5 in response to each question. Of the 33 leaders who repeatedly selected choice 7, 6, or 5 for all survey questions, all 33 spent fewer than three minutes to complete the survey, which is less than 50% of the average completion time range of 7-10 minutes. The repeated nature of their responses and the short amount of time spent by the leader on the survey overall, suggests to this researcher that these responses were not valid. These 33 additional cases were trimmed. The 67 cases identified through a multivariate analysis and the 33 additional cases described above were removed. The trimming of 100 of the 1,018 cases represented 9.8% of the total sample and fell below the range considered for moderate sample trimming (10%-30%) according to Jose & Winkler (2008). The remaining 918 responses represent the final sample that was utilized in all subsequent analyses using SPSS and AMOS. Normality. Normality, like linearity and homoscedasticity, is often an important assumption when conducting univariate, and especially multivariate statistical analyses. When one or more of these assumptions are violated, results may be biased (Kennedy & Bush, 1985). Both univariate and multivariate tests of normality were conducted. Univariate normality was considered through the inspection of each variable’s normal probability or Q-Q plots. SPSS was used to create this graphic, which plots the expected normal values along the y-axis and the actual values along the x-axis. When 79 normality is found, the plot creates a somewhat straight line. (Mertler & Vannatta, 2005). Observations of the Q-Q plots, while subjective, led this researcher to believe that the data were not normally distributed around the mean. This qualitative test was followed by calculating and examining the Kolmogorov-Smirnov statistic with Lilliefors’ significance correction and Shapiro-Wilk statistic for each variable. Mertler and Vannatta (2005) stated, “The Kolmogorov-Smirnov statistic tests the null hypothesis that the population is normally distributed.” (p. 30). All variables were found to have a significant Kolmogorov-Smirnov and Shapiro-Wilk statistic at a p=.000 level, which confirmed the Q-Q plot observation of non-normal distribution. Finally, skewness and kurtosis values for each variable were evaluated by Organizational Level. All skewness values for each variable by Organizational Level fell within the acceptable range of ±2 (George & Mallery, 2011), except for the variable “Business skills Required, Management of Financial Resources” within the Superintendent group (-2.205). Values of kurtosis outside the acceptable range of ±7 (Hoyle, 2012) were found for one variable, “Business skills Required, Management of Financial Resources,” again within the Superintendent group (9.491). Skewness and Kurtosis scores are reported in Table 2, with scores falling outside the accepted range in bold. 80 Table 2 Skewness and Kurtosis of all Leadership Skills by Group and Variable Skill Cognitive Skill Requirements Group Assistant Principal Variable Speaking Active Listening Writing Reading Comprehension Active Learning Critical thinking Skewness -.107 -.661 .158 .026 -.354 .400 Kurtosis .593 1.336 .411 .769 -.081 -.710 Cognitive Skill Requirements Building Principal Speaking Active Listening Writing Reading Comprehension Active Learning Critical Thinking .080 -.269 .406 .169 -.303 .006 -1.436 -1.176 -.830 -.718 -.532 -1.089 Cognitive Skill Requirements Assistant Superintendent Speaking Active Listening Writing Reading Comprehension Active Learning Critical thinking -.153 -.707 .395 -.188 -.330 .001 -.209 1.088 -.929 .201 -.312 -.954 Cognitive Skill Requirements Superintendent Speaking Active Listening Writing Reading Comprehension Active Learning Critical thinking -.135 -.458 .339 .018 -.459 -.349 -1.627 -1.245 -1.038 -1.170 -1.095 -1.510 Interpersonal Skill Requirements Assistant Principal Social Perceptiveness Coordination Negotiation Persuasion -.975 -1.080 .167 -.539 1.787 2.346 -.112 1.403 Interpersonal Skill Requirements Building Principal Social Perceptiveness Coordination Negotiation Persuasion -.589 -.487 .110 -.197 -.615 -.220 -.382 -.210 Interpersonal Skill Requirements Assistant Superintendent Social Perceptiveness Coordination Negotiation Persuasion -.262 -.787 .493 .036 -.1.104 .656 -.748 -.412 Superintendent Social Perceptiveness -.564 -.892 81 Interpersonal Skill Requirements Business Skill Requirements Business Skill Requirements Business Skill Requirements Business Skill Requirements Strategic Skill Requirements Strategic Skill Requirements Coordination Negotiation Persuasion Assistant Principal Building Principal Assistant Superintendent Superintendent Assistant Principal Building Principal -1.378 -.077 -.347 3.666 -1.008 -.778 Operations Analysis Management of Personnel Resources Management of Financial Resources Management of Material Resources -.436 1.055 -.924 .678 -.465 .240 -.281 -.137 Operations Analysis Management of Personnel Resources Management of Financial Resources Management of Material Resources -.419 .649 -1.207 2.047 -.309 -.433 .451 .648 -.601 .957 -1.216 2.311 -1.087 2.352 -.852 2.577 -.302 .617 -1.148 2.222 -2.205 9.491 -.228 .370 Visioning System Perception System Evaluation ID Downstream Consequences ID of Key Causes Problem ID Solution Appraisal -.264 -.296 -1.260 -.602 .203 1.919 -.419 -.607 -.144 -.697 -.121 .333 -.516 .108 Visioning System Perception System Evaluation -.517 -.397 -.729 -.235 .660 .847 Operations Analysis Management of Personnel Resources Management of Financial Resources Management of Material Resources Operations Analysis Management of Personnel Resources Management of Financial Resources Management of Material Resources 82 ID Downstream Consequences ID of Key Causes Problem ID Solution Appraisal Strategic Skill Requirements Strategic Skill Requirements Assistant Superintendent Superintendent -.907 -.547 -.155 -.834 1.504 -.010 -1.055 .949 Visioning System Perception System Evaluation ID Downstream Consequences ID of Key Causes Problem ID Solution Appraisal -.756 -.654 -1.451 .343 1.248 4.185 -1.170 -.796 -.793 -1.076 2.169 1.004 .588 2.192 Visioning System Perception System Evaluation ID Downstream Consequences ID of Key Causes Problem ID Solution Appraisal -1.022 -1.053 -.721 1.562 2.673 .165 -1.039 -.607 -.907 -1.478 1.368 .257 .760 5.440 Note. Scores in bold fall outside the generally accepted range of ±2 for Skewness and ±7 for Kurtosis. Linearity and homoscedasticity. Linearity is an important assumption in multivariate analysis of the kind conducted in this study and assumes that the relationship between variables occurs in a straight line (Cohen, Cohen, & Stephen, 2003; Mertler & Vannatta, 2005). This relationship can occur between two or more variables or subscales in the case of this research study. A review of residual plots, the amount of the variation not accounted for by the analysis (error), was conducted by plotting the standardized predicted values along the x-axis and the standardized residuals along the y-axis. According to Mertler and Vannatta (2005), “when the assumptions of linearity, normality, and homoscedasticity are met, residuals will create an approximate rectangular distribution with a concentration of scores along the center” (p.55). As extreme 83 groupings were not present and the residual plot was rectangular in shape, general linearity and homoscedasticity were assumed for variables “Cognitive, Interpersonal, Business, and Strategic skill”. As is explored in a future section, multivariate Homoscedasticity was not assumed as a significant Box’s M statistic was reported. The Box’s M statistic has been widely criticized as being overly sensitive (Mertler & Vannatta, 2005). A significant value of p<001 suggests the rejection of the null hypothesis, that covariance matrices are equal (George & Mallery, 2011; Mertler & Vannatta, 2005). Mertler and Vannatta (2005) noted, “one may reject the assumption that covariance matrices are equal due to a lack of multivariate normality,” (p. 34). Due to the lack of multivariate normality found for these data and the overly sensitive nature of the Box’s M statistic, further analysis was conducted despite the data’s failure of this assumption. Effect Sizes and Significance Levels for This Study Cohen’s effect size. Cohen (1990) examined effect size as one of three components of statistical power. Cohen (1990) went on to report that effect sizes are an acceptable way to report the magnitude and direction of influence in quantitative studies. While only an estimate, suggests values of population effect size coefficients have become an accepted way of reporting effect sizes, such that .1 is small, .3 is middling, and .5 is large (Cohen 1962, 1988, & 1990). Despite their questioning of the use of Cohen’s (1988 & 1990) guidelines for reporting effect sizes, Bosco, Aguinis, Singh, Field and Pierce (2015) acknowledged that these conventions for understanding the magnitude of effect are present in most quantitative studies. Cohen (1988) recognized that his standards of effect size may be 84 overstated as he reported, “…if what is here defined as large is too small to meet what his area of behavioral science would consider appropriate standards he is urged to make more suitable operational definitions” (p.79). While generally accepted, Cohen’s (1988) guidelines should not be assumed to apply equally to all fields of study (Bosco et al., 2015). In spite of these recognized limitations, Cohen’s guidelines were used to report effect sizes for the purposes of this study and can be found in Table 3. Table 3 Cohen’s (1992) Effect Size Interpretations Small Medium Large r R2 F2 .10 .30 .50 .02 .13 .26 .02 .15 .35 Creation of Latent Variables Confirmatory factor analysis using the Maximum Likelihood (ML) method was performed on all 21 variables measuring the four Leadership skills categories (Cognitive, Interpersonal, Business and Strategic). This allowed for a comparison of factor loadings in an attempt to determine if the variables grouped as determined a priori. Kaiser-Mayer-Olkin’s (KMO) measure of sampling adequacy was used to examine the distribution of values and whether they were adequate for factor analysis using Kaiser’s scale (.9 > Marvelous, .8 > Meritorious, .7 > Middling, .6 > Mediocre, .5 > Miserable and .5 < Unacceptable). The KMO for this sample was .942 and fell in the “marvelous” range, thus indicating that the sample’s distribution was sufficient for conducting factor analysis (George & Mallery, 2011). 85 Bartlett’s Test of Sphericity (x²=8150.44, df =210) was used to measure the multivariate normality of the data and was significant at the p<.001 level, which indicated that the multivariate distribution of the data was not sufficiently normal and thus unacceptable for factor analysis. It is important to note that it was hypothesized, a priori, that mean differences between groups would exist and that all skill levels would increase as organizational level increased, thus suggesting a de facto non-normally multivariate distribution. Four factors with Eigen values greater than 1.00 were identified (factor 1 = 8.17, factor 2 = 1.88, factor 3 = 1.09 and factor 4 = 1.01) with 57.86% of the variance explained. The screed plots revealed three distinct factors, with the fourth being marginal. Initial review of the factor loadings suggests that the four factors with Eigen values greater than one were consistent with the four leadership skills being measured. This is important as it could be used to lend support to hypothesis 1 by confirming that each of the four specified leadership skills are empirically distinguishable. Reliability testing. In light of the four factors identified by the confirmatory factor analysis (CFA), additional reliability tests were conducted for all variables that sought to measure each of the four leadership skills. Reliability analysis in SPSS was utilized for the four identified factors. Cronbach alpha scores with deletion were calculated for each of the indices created. The removal of items did not raise the overall alpha score, and all 21 items were retained in the creation of the four indices. Cognitive skills. Six items on the survey instrument were designed to measure the perceived level of Cognitive skill required by the educational leader to perform his or her current job. Cognitive skills were measured in the areas of speaking, active listening, 86 writing, reading comprehension, active learning, and critical thinking. An internal consistency reliability score of α=.847 was determined with factor loadings ranging from .590 to .670. Interpersonal skills. Four items on the survey instrument were designed to measure the perceived level of Interpersonal skill required by the educational leader to perform his or her current job. Interpersonal skills were measured in the areas of social perceptiveness, coordination, negotiation, and persuasion, with an internal consistency reliability score of α=.722 and factor loadings ranging from .505 to .613. Business skills. Four items on the survey instrument were designed to measure the perceived level of Business skill required by the educational leader to perform his or her current job. Business skills were measured as operations analysis, management of personnel resources, management of financial resources, and management of material resources. An initial Cronbach’s alpha of .690 was noted, with factor loadings ranging from .461 to .599. Strategic skills. Seven items on the survey instrument were designed to measure the perceived level of Strategic skill required by the educational leader to perform his or her current job. Strategic skills were measured as visioning, systems perception, system evaluation, identification of downstream consequences, identification of key causes, problem identification, and solution appraisal. A Cronbach’s alpha score of .878 was noted, with factor loadings ranging from .641 to .695. All indices were found to possess adequate internal validity with Alpha scores approaching or exceeding .700 and Strategic skills required approaching the clinical level of .900 (Johnson & Christensen, 2012). This suggests that each item in the index(s) was 87 measuring the same latent variable adequately. When compared to the original study, all of the coefficients found in this study were slightly smaller. Table 4 compares the reliability statistics of the current study compared to the original 2007 study. Table 4 Reliability Scores (alpha) for the 2007 and 2015 Studies Skill requirements Cognitive Interpersonal Business Strategic 2007 Study .90 .84 .75 .91 2015 Study .85 .72 .69 .88 Results / Findings Descriptive statistics. An analysis of the data using SPSS version 22 and AMOS version 22 (structural equation modeling) was conducted. Table 5 presents univariate statistics for all four leadership skill indices and demographic/control variables. In addition, the univariate statistics for each of the four leadership skills by “organization level” are presented. The leadership skill indices had mean scores ranging from = 5.46 for Business skills to = 5.77 for Strategic skills, with wider ranges found between lower and higher organizational levels. Standard deviations ranged from 0.70 to 0.78, again with wider ranges found at the individual organizational levels. These statistics represented a fairly narrow range and variability of responses. Range and variability differed between this study and the 2007 study, where mean scores of 4.59 to 5.84 and standard deviations ranging from .75 to 1.10 were considered to show “good range and variation” (Mumford et al., 2007, p.161). This indicated that a wider variation of responses existed in the population examined by the 2007 study than the sample of 88 educational leaders analyzed by this researcher. These differences suggests that educational leaders differ from the population originally studied in that higher amounts of leadership skills are perceived to be required overall, while variation between leadership levels are less pronounced. Skewness and Kurtosis values were examined for each index at the group level. All variables except one showed moderate to low negative skew. Strategic skills had the greatest negative skew (-0.70) yet remained within the acceptable range. This suggests that survey responses tended to be grouped slightly more towards the higher end of the Strategic skill requirement range. The data were therefore slightly skewed to the left with fewer numbers of responses found in that tail. Additionally, values of kurtosis were examined. This revealed that leadership skills tended to be slightly more peaked than normally distributed data. Cognitive skill requirements were the exception with the data being slightly more platykurtic. No comparison was made between the 2007 study and the 2015 study as indications of normality of distribution were not reported in the original study. 89 Table 5 Descriptive Statistics for all Skills by Organizational Level 1. Cognitive Skills Index Assistant Principal Building Principal Assistant Superintendent Superintendent N 918 241 359 161 157 Mean 5.74 5.54 5.76 5.72 6.00 Med. 5.67 5.50 5.67 5.50 6.00 Min. 2.33 3.0 4.0 2.33 4.50 Max. 7.00 7.00 7.00 7.00 7.00 SD 0.72 0.72 0.70 0.77 0.65 2. Interpersonal Skills Index Assistant Principal Building Principal Assistant Superintendent Superintendent 918 241 359 161 157 5.76 5.60 5.77 5.73 6.04 5.75 5.50 5.75 5.75 6.00 3. Business Skills Index Assistant Principal Building Principal Assistant Superintendent Superintendent 918 241 359 161 157 5.46 5.15 5.38 5.65 5.89 4. Strategic Skills Index Assistant Principal Building Principal Assistant Superintendent Superintendent 918 241 359 161 157 5. Organizational Level Assistant Principal Building Principal Assistant Superintendent Superintendent 241 359 161 157 6. Experience Level 0-10 years 11-20 years 21-30 years 30+ years 365 348 150 88 7. Educational Level Master’s Degree 1-30 Beyond Master’s 31-60 Beyond Master’s Terminal Degree Skewness 0.06 0.28 0.12 -0.23 -0.05 Kurtosis -0.16 0.20 -0.78 1.21 -1.08 2.75 2.75 3.75 4.0 4.00 7.00 7.00 7.00 7.00 7.00 0.70 0.74 0.70 0.64 0.62 -0.23 -0.32 -0.13 0.04 -0.31 0.15 0.68 -0.19 -0.26 -0.78 5.50 5.25 5.50 5.75 6.00 2.25 2.75 2.25 3.50 3.50 7.00 7.00 7.00 7.00 7.00 0.78 0.84 0.74 0.70 0.59 -0.56 -0.36 -0.50 -0.49 -0.69 0.49 -0.03 0.73 0.46 2.04 5.77 5.48 5.76 5.86 6.13 5.86 5.57 5.86 5.86 6.14 2.43 2.43 2.86 3.14 3.71 7.00 7.00 7.00 7.00 7.00 0.75 0.81 0.73 0.70 0.58 -0.70 -0.53 -0.68 -0.73 -0.73 0.94 0.58 1.08 1.15 1.15 -- -- -- -- -- -- -- -- -- -- -- -- -- -- 18 146 549 205 -- -- -- -- -- -- -- 8. District Setting Rural Suburban Small City 250 565 103 -- -- -- -- -- -- -- 9. % Free and Reduced Lunch 0-50% 51-100% 671 247 -- -- -- -- -- -- -- Correlation coefficients. Table 6 presents a bivariate correlation matrix (Pearson’s r) with all leadership skill indices and demographic/control variables included. 90 There were strong correlations between the four leadership skill requirements, with the strongest positive relationship found between variables “Business Skill Requirements” and “Strategic skill Requirements” (r = .658, p < .001). The average correlation found between the four leadership skill requirements was r = .572, which accounted for 42.8% of the total shared variance explained. Therefore, while strongly correlated, 57.2% of the total variance was unexplained and a result of other outside factors. Similar findings were reported by the authors of the original 2007 study, with an average correlation between the leadership skill requirements reported as r = .62 or 38% of the shared variance (Mumford et al., 2007). Similar to the original study, strong positive relationships between the leadership skill categories were established and expected “given that we expect all skills to be required of all leaders” (Mumford et al., 2007, p.161). While both direction and magnitude of correlation were similar between the 2007 and 2015 samples, comparisons of Pearson’s product moment (r) using Fisher’s ztransformation revealed that while generally slight, the difference between all but two correlations (“Strategic skill” and “Cognitive skill” and “Strategic skill and Organizational Level”) were found to be statistically significantly different, at a p<.05 level, between the two studies. This suggests that the correlation between Cognitive skill and Strategic skill, and Strategic skill and organizational level, are the same in both populations. While all other correlations may share direction and magnitude of correlation, the two populations were different in a meaningful way. The most notable difference was that educational leaders tended to have a weaker positive correlation between “Interpersonal skill Requirements” and “Organizational Level” than those in the 91 original study (r= .184, p<.000; r= .32, p<.000), respectively. This suggests that “Interpersonal skill Requirements” are less closely associated with a school leader’s “Organizational Level” than among those in the original population, and, when combined with higher overall mean scores, may suggest that “Interpersonal skill Requirements” are more important for educational leader’s at all organizational levels. Table 6 Pearson Correlation Matrix of Leadership Skill Requirements, Organizational Level, and Covariates Variables 1. Cognitive skill requirements 2. Interpersonal skill requirements 3. Business skill requirements 4. Strategic skill requirements 2 .598** (.71**) z=-4.33** --- 3 .472** (.23**) z=6.12** .545** (.41**) z=3.86** 4 .565** (.62**) z=-1.86 .594** (.70**) z=-4.03** 5 .190** (.28**) z=-2.09** .184** (.32**) z=-3.20** 6 .138** 7 .130* 8 .047 .148** .096** .061 --- .658** (.53**) z=-4.37** --- .329** (.24**) z=2.13** .278** (.34**) z=-1.51 --- .168** .100** -.006 .160** .161** .067* 5. Organizational .430** .214** -.123** Level 6. Experience Level --.239** .099** 7. Educational Level --.066* 8. District Setting --Note. N= 918. Numbers in parentheses represent correlation coefficients from 2007 study. * p<.05 and **p<.001 two-tailed. Hypothesis 1: The Cognitive, Interpersonal, Business, and Strategic skill categories will be empirically distinguishable. Structural equation modeling (SEM) was utilized for the purposes of testing whether the four leadership skill categories were empirically distinguishable. While the 2016 study previously used traditional confirmatory factor analysis (CFA) utilizing the maximum likelihood method (ML) and reliability coefficients (Cronbach’s alpha), which supported a four-factor model of 92 leadership and the creation of four leadership skill indices, CFA with ML utilizing SEM (with AMOS) was also conducted for the purposes of comparing the data’s relative fit to a one-factor and four-factor leadership skills model as in the original study. The following model fit indices were reported: Maximum Likelihood (ML) Chi-Square (X2), the Comparative Fit Index (CFI), the Normed Fit Index (NFI), and the Root Mean Square Error of Approximation (RMSEA) with 90 percent confidence intervals (CI). Statistical analyses of the data were conducted utilizing first-order CFA with the ML estimation procedure in SEM utilizing AMOS version 22. Both one-factor and fourfactor measurement models were created. Both CFA Models were examined for model fit. Fit indices, as outlined by Byrne (2010), suggests that a one-factor model fit the data very poorly (X2 =1879.44/df=189 p<.001), CFI=.789, NFI=.771, RMSEA=.099 (90% CI=.095-.103) and a four-factor model fit the data marginally well (X2 =926.65/df=183 p<.001), CFI=.907, NFI=.887, RMSEA=.067 (90% CI=.062-.071). Additionally, a X2 difference test revealed a statistically significant difference between the one-factor and four-factor models (X2 difference = -952.79, p <.001). While neither hypothesized model fit the data exceptionally well, the four-factor model fit the data significantly better. These findings lend support to Hypothesis One, in that a four-factor model of leadership skills is empirically identifiable and preferable to a single-factor model. Both measurement models are presented in Figures 3 and 4, and a comparison of the two measurement models, fit indices, and chi-square difference test are reported in table 7. 93 Figure 3. One-factor confirmatory factor analysis structural equation measurement model of leadership skills. Completely standardized robust maximum likelihood parameter estimates (factor loadings). The residual variance components (error variances) indicate unexplained variance. Fit Indices: Chi-square=1879.44 / df = 189. (p<.001); CFI = .789; NFI = .771; RMSEA = .099 94 Figure 4. Multi-skill, four-factor confirmatory factory analysis structural equation measurement model. Completely standardized robust maximum likelihood parameter estimates (factor loadings). The residual variance components (error variances) indicate unexplained variance. Ovals represent the four leadership skill types (latent variables). Curved lines represent correlations between leadership skill categories. Fit Indices: Chisquare=926.65/df=183, (p<.001); CFI=.907; NFI=.887; RMSEA=.067 95 Table 7 Fit Indices for One-Factor and Four-Factor Measurement Models Using Structural Equation Modeling and χ2 Difference Test Test χ2 Four926.65 factor (324.1) model One-factor 1879.44 model (714.8) Model difference df CFI 183 (14) .907 (.991) NFI .887 (.991) 189 (20) .789 (.981) .771 (.980) RMSEA .067 (90% CI=.062-.071) (.147) .099 (90% CI=.095-.103) (.184) χ2 Difference df -952.79** (-390.7**) 6 (6) Note. CFI = comparative fit index; NFI = normed fit index; RMSEA = root mean square error of approximation. Scores in parentheses represent the findings of the 2007 study. N=951, **p=<.01 Hypothesis 2. Leadership skill requirements will vary by skill category such that Cognitive skills will be needed the greatest amount, followed by Interpersonal, Business, and Strategic skills, respectively. Multivariate analysis of covariance (MANCOVA) was utilized to evaluate the degree to which each of the four identified leadership skills were required and if those requirement levels lent support to the aforementioned hypothesis. The advantage of using MANCOVA for this type of analysis is that one or more covariates can be introduced and the effects of the covariates are then controlled for. Removing the effects of covariates decreases error and helps ensure a clearer understanding of the true effects of the independent variables on the multiple dependent variables (Mertler & Vannatta, 2005). MANCOVA is recommended over multiple t tests or multiple one-way ANOVAs conducted separately, especially when dependent variables are correlated, as was the case with this study, due to an increased chance of committing a type I error. Therefore, MANCOVA was conducted to establish the mean differences in the skill requirements. Additionally, the Bonferroni procedure 96 for multiple tests was used to protect against a type I error, which can occur when multiple tests are conducted simultaneously as in MANCOVA (Holm, 1979). Therefore, the probability statistic (α=.05) was divided by the number of tests being run (H=4), and a new threshold for significance was established at α<.013. This more conservative alpha level was used to determine the statistical significance of the marginal means for each skill requirement. Finally, the Wilks Lambda statistic was used as the multivariate test statistic, which reported the main effect and significance for each leadership skill category. The analysis was conducted by performing a one-way MANCOVA utilizing SPSS version 22. This test, as in the 2007 study, was selected in order to investigate possible categorical differences between organizational levels of educational leaders and the four leadership skills (Cognitive, Interpersonal, Business, and Strategic), while controlling for organizational level, years of experience, amount of education, and district setting. Linearity of the four dependent variables (Cognitive, Interpersonal, Business, and Strategic skills) and the covariates (experience, education, and district setting) was tested by reviewing a matrix scatterplot and Pearson correlation coefficients (Table 6). As previously discussed, results suggest a linear relationship. Correlation coefficients between the dependent variables were considered high, ranging from r=.472 to r=.658 and were statistically significant at the p<.001 level. Correlation coefficients between the dependent variables and the covariates were significant at the p<.05 level and ranged from small r=.006 to medium r=.430. The final assumption of MANCOVA, homogeneity of variance-covariance, was tested using the Box’s M test. A significant Box’s M (p=.001) indicated that equality of variance could not be assumed. Therefore, 97 rather than Wilks Lambda, the more robust Pillai’s Trace statistic was utilized as the appropriate test statistic (Mertler & Vannatta, 2005). For the purpose of comparison, Wilks’ Lambda was also reported, since Wilks Lambda was reported in the original 2007 study. The main effect of the leadership skill categories, while controlling for organizational level, experience, education, and district setting was statistically significant with a (Pillai’s Trace = .111, p<.001) and (Wilks’ Lambda = .892, p<.001). This suggests that significant differences in the mean scores of the dependent variables were found between skill groups, controlling for the covariates. An evaluation of the estimated marginal means for each of the perceived leadership skill requirements revealed that overall, Strategic skills were required the most, followed by Interpersonal skills, Cognitive skills, and Business skills ( = 5.80, = 5.78, = 5.75 and =5.52), respectively (Table 8). These results suggests that when controlling for an educational leader’s organizational level, years of experience, level of education, and the district’s geographic setting, there is a statistically significant difference among the average required amounts of the four leadership skills. However, statistically significant differences did not exist between all skill groups and the order in which skills are required in the greatest amounts differed from the original study. Therefore, Hypothesis Two was not supported, and the results suggests that educational leaders might in fact require a very different combination of leadership skills than those proposed by Mumford et. al.’s Leadership Skills STRATAPLEX model. 98 Table 8 Estimated Marginal Means for Cognitive, Interpersonal, Business, and Strategic Skill Requirements Controlling for Organizational Level, Experience Level, Education Level, and District Setting. Variable Estimated marginal mean with controls Strategic skill requirements 5.80a (5.05a) Interpersonal skill requirements 5.78ab (5.34bc) Cognitive skill requirements 5.75b (5.44c) Business skill requirements 5.52c (5.11abc) Wilks Lambda .892, F=8.874** (.982, F=6.020**) Pillai’s Trace .111 F=8.704** Note. Numbers in parentheses represent estimated marginal means from the 2007 study. Means in each column that do not share subscripts differ at p<.05 when corrected using the Bonferroni procedure. **p<.01, two tailed. Hypothesis 3. Cognitive, Interpersonal, Business, and Strategic skill requirements will be positively related to the job’s level in the organization. Partial correlations with controls were used to explore the magnitude and direction of the relationships between a leader’s organizational level and leadership skills required for the job, while controlling for the influence of years of experience, level of education, and school district setting. According to George & Mallery (2011), “Partial correlation is the process of finding the correlation between two variables after the influence of other variables has been controlled for” (p. 127). Cohen et al. (1983) stated that, “…partial correlation is the correlation of Xi with that part of Y that is independent of the other IV’s in the equation” (p. 316). Thus, partial correlations allowed a clearer view of the direction and magnitude of association between the dependent variable (Leadership Skills) and the independent variable (Organizational Level) by removing the impacts of a leader’s years of experience, level of education, and district’s setting. 99 The results of the partial correlational analysis revealed that after controlling for covariates, which were also highly correlated with the DV, statistically significant, positive, small to medium relationships were found between each of the four leadership skills and organizational level. Interpersonal and Cognitive skills had a small, positive, statistically significant relationship with organizational level (r=.140, p<.05; r=.143, p<.05) that were of about the same magnitude. Strategic and Business skills had moderate, positive, statistically significant correlations with organizational level (r=.236, p<.05; r=.283, p<.05) again, with about the same magnitude. All four of these findings suggests that as organizational level increases, so too does the required amounts of each of the four leadership skill categories. The strongest relationship was found between organizational level and Business skills followed by Strategic, Cognitive, and Interpersonal, respectively. Partial correlations with Fisher’s z-transformation comparisons between the original 2007 and current 2016 studies found that the magnitude of relationship between each of the four leadership skill categories were about the same. These results and comparisons between the two studies are found in table 9. Fisher’s z-transformations were conducted to determine if there was a statistically significant difference between each of the skill requirements. Standardized z-scores were calculated and revealed that there was a statistically significant difference between the rscores of Strategic skill requirements and Interpersonal skill requirements (z=2.13, p=.033); Strategic skill requirements and Cognitive skill requirements (z=2.07, p=.039); Business skill requirements and Interpersonal skill requirements (z=3.21, p=.001); and Business skill requirements and Cognitive skill requirements (z=3.14, p=.002) respectively. 100 Table 9 Partial Correlation Between Leadership Skill Requirement Categories and Organizational Level, Controlling for Years of Experience, Education Level, and District Setting. Variable Estimated marginal mean with controls Strategic skill requirements Interpersonal skill requirements Cognitive skill requirements .236a ** (.25a **) .140b ** (.20b **) .143b ** (.15c **) z=-.33, p=.741 z=1.36, p=.174 z=-.16, p=.873 Business skill requirements .283a ** (.28a **) z=.07, p=.944 Note. N = 918. Correlations that do not share subscripts differ at p<.05. Scores in parentheses represent the findings of the 2007 study. Fisher’s z-scores with probability coefficients are given for comparison between the 2007 and 2015 study. **p=<.05, two tailed. This statistically significant, positive relationship between organizational level and leadership skill requirement lent support to Hypothesis Three by showing that as organizational level increased so too did the amount of the leadership skill required. However, while all four skills are positively correlated to Organizational level in a statistically significant way, the estimated marginal means of variables Cognitive skill required and Interpersonal skill required, while controlling for covariates, tended to decline in a non-statistically significant way between organization level “Principal” and “Assistant Superintendent”. This suggests that no meaningful mean difference existed between Organizational levels “Principal” and “Assistant Superintendent” and Leadership skill Types “Interpersonal” and Cognitive”. This finding was inconsistent with the findings of the 2007 study (Table 10). This mostly positive relationship is also 101 illustrated in figure 5. Amount of Leadership Skills Required 6.5 6.25 6 5.75 Cognitive Skill 5.5 Interpersonal Skill Business Skill 5.25 Strategic Skill 5 Assistant Principal Principal Assistant Supt. Organization Level Supt. Figure 5. Estimated marginal means for leadership skill requirements across organizational levels while controlling for experience, education, and district setting. Additionally, Fisher’s z-transformations were performed between the partial correlations found in the original study and those reported in this 2015 study. Standardized z-scores for Cognitive, Interpersonal, Business, and Strategic skill requirements were found to be non-statistically significant (z=-.16, p=.873; z=1.36, p=.174; z=.07, p=.944; and z=-.33, p=.741), respectively. This suggests that there was no statistically significant difference in the direction or magnitude of correlation between leadership skill requirements and organizational level found between the original and current studies. In effect, both population samples, while very different in characteristics 102 and occupational fields, require overall increasing amounts of leadership skill requirements as one ascends the organizational structure. Thus, this finding validates the portion of Mumford et al.’s 2007 STRATAPLEX model that suggests that all skill levels will increase as organizational level increases such that leaders at the highest organizational level require all the skills required at each subordinate level in this new population sample. Another way of conceptualizing this phenomenon is to view leadership skills as cumulative, increasing at each organizational level as one ascends the hierarchy. For instance, as Cognitive skills increase from assistant principal ( =5.54) to superintendent ( =6.01) all the skills of an assistant principal are needed in conjunction with the additional skills required by the superintendent. Hypothesis 4. Interaction of skill requirements and organizational level. Hypothesis 4, by means of three sub hypotheses, sought to examine the magnitude of the relationship between organizational level and leadership skill requirements. It suggests that the strength of relationship with organizational level will be different for each of the four leadership skill types and that these differences would be found in a hypothesized way. Support for hypotheses 4 could be suggests if correlation coefficients, controlling for covariates (“years of experience,” “level of education,” and “district setting”), were found to be statistically significantly different and of successively greater magnitudes for Cognitive, Interpersonal, Business, and Strategic skills. Partial correlations between leadership skill requirements and organizational level, while controlling for covariates, were examined for this study. The correlations were compared to each other using Fisher’s z-transformation to determine if significant 103 differences existed between the four skill types. Statistically significant differences were found between Strategic skills and both Cognitive and Interpersonal skills and between Business skills and both Cognitive and Interpersonal skills. However, no difference was found between Strategic and Business skills or between Interpersonal and Cognitive skills. These multiple comparisons, adjusted using the Bonferroni procedure in SPSS version 22 as suggests by Cohen et al. (1983 and 2003) and Pedhazur (1997) are reported in Table 9 and Figure 5. Hypothesis 4a stated that Strategic skill requirements would be more strongly related to organizational level than Cognitive, Interpersonal, and Business skill requirements. Results from the partial correlations reported in table 9 indicated that Strategic skill requirements were indeed more strongly correlated to organizational level (r=.236, p<.001) than either Cognitive or Interpersonal skill requirements (r=.143, p<.001 and r=.140, p<.001, respectively), even when controlling for other variables. However, Business skill requirements were slightly more strongly correlated to organizational level than Strategic skill requirements (r=.283, p=.000 and r=.236, p=.000). The differences among the correlations between Business skill requirements and organizational level (r=.283, p=.000) and Strategic skill requirements and organizational level (r=.236, p=.000) were not different in a statistically significant way (z=-1.08, p=.280). This suggests that both Business skill requirements and Strategic skill requirements are about the same in regards to their strength of relationship to organizational level, while controlling for other variables. Therefore, hypothesis 4a was only partially supported. 104 These findings were similar to the findings from the 2007 study, which also reported stronger relationships between Strategic skill requirements and organizational level than Interpersonal and Cognitive skill requirements and organizational level. Again, strong similarities exist between both population samples with Business skills having the strongest statistically significant positive relationship to organizational level in both studies (Table 9). Hypothesis 4b suggests that Business skill requirements would be more strongly related to organizational level than Cognitive or Interpersonal skill requirements. Results from the partial correlations reported in Table 9 indicated that Business skill requirements were indeed more strongly associated with organizational level than both Cognitive and Interpersonal skill requirements (r=.283, p<.001; r=.143, p<.001 and r=.140, p<.001), respectively. The magnitude of the correlations was significantly different when Business skill requirements were compared to Cognitive and Interpersonal skill requirements (z=3.14, p=.002 and z=3.21, p=.001). Therefore, hypothesis 4b was fully supported. As with the previous hypothesis, the findings for hypothesis 4b were the same among both the original and present study. Additionally, the lack of statistical significance found between the magnitude of the partial correlations reported in both studies suggests that the findings were similar, failing to differ from each other in a statistically meaningful way. This finding lent further support to hypothesis 4 in this sample of school administrators. Hypothesis 4c stated that Interpersonal skill requirements would be more strongly related to organizational level than Cognitive skill requirements. An examination of the 105 partial correlations with controls reported in Table 9 demonstrated that Interpersonal skills were less strongly correlated to organizational level than Cognitive skills (r=.140, p<.001 and r=.143, p<.001). An evaluation of the partial correlations using Fisher’s ztransformation indicated that no statistically significant difference existed between the two reported correlation coefficients suggesting that they did not differ in a meaningful way. Therefore, no support was given for Hypothesis 4c. Unlike hypothesis 4a and 4b, hypothesis 4c was not in line with the findings from the 2007 study, which reported support of hypothesis 4c. This difference suggests that educational leaders in New York State required about the same amount of both Cognitive and Interpersonal skill and that these two skills were positively related to one’s organizational level. This differed from the original 2007 study, which found Cognitive skill to be required more than Interpersonal skill among that sample of diplomats. This finding would suggest that the Leadership Skills STRATAPLEX model could not be fully generalized to educational leaders in New York State. A comparison of the findings of this study to the original 2007 study is summarized in table 10. 106 Table 10 Summary of Findings and Comparison of Conclusions for the 2016 and 2007 Studies Hypothesis H1: Cognitive, Interpersonal, Business, and Strategic skill categories will be empirically distinguishable Test(s) Confirmatory Factor Analysis Structural Equation Modeling Chi Squared Difference Test Findings Variables loaded on 4 distinct factors with Engine values greater than 1. Conclusions 2007 Study: Full Support 2016 Study: Full Support The 4 Factor measurement model (SEM) fitted the data better than the 1 factor model. Leadership skills were measureable and were empirically distinguishable. H2: Leadership skill requirements will vary by skill category such that Cognitive skills will be needed the greatest amount, followed by Interpersonal, Business and Strategic H3: Cognitive, Interpersonal, Business and Strategic skill requirements will be positively related to the job's level in the organization Multivariate Analysis of Variance and Covariance (MANCOVA) Educational leaders required over all greater amounts of all four leadership skills. 2007 Study: Partial Support 2016 Study: No Support Skills varied such that Strategic skills were needed in the greatest amounts followed by Interpersonal, Cognitive and Business. Test of Partial Correlations Significant positive relationships existed between each Leadership skill and Organizational Level, while controlling for other factors. 2007 Study: Full Support 2016 Study: Full Support The magnitude of relationship between each skill and organizational level was the same between both studies. H4: Leadership skill requirements will interact with organizational level such that: All four skills were positively related to Organizational level such that Business skills had the strongest relationship followed by Strategic, Cognitive and Interpersonal. H4a: Strategic skill requirements will be more strongly related to organizational level than cognitive, Interpersonal, and Business Skills Partial Correlations H4b: Business skill requirements will be more strongly related to organizational level than Cognitive or Interpersonal skills Partial Correlations The Magnitude of relationship between Business and Strategic skills, and Organizational level were statistically the same. The Magnitude of relationship between Cognitive and Interpersonal skills, and Organizational level were statistically the same. 2007 Study: Partial Support 2016 Study: Partial Support 2007 Study: Full Support 2016 Study: Full Support H4c: Interpersonal skill requirements Partial Correlations 2007 Study: Full Support will be more strongly related to 2016 Study: No Support organizational level than Cognitive skills Note. Due to the differences between the 2015 and 2007 studies, the Leadership Skills STRATAPLEX model is not generalizable to leaders of K-12 Public Education in New York State 107 Extended multivariate analysis. In light of these findings, further analyses were conducted to examine the mean differences in the four leadership skill requirements at each of the four organizational levels, while controlling for covariates “level of education”, “years of experience” and “district setting.” The purpose of these follow-up tests was to better understand the differences between the average amounts of each of the four leadership skill types required at each of the four organizational levels examined. A MANCOVA was performed on the means to help protect against a Type 1 error that can result from performing repeated ANCOVAs and post-hoc comparisons. Pearson product moment correlations were generated between all of the independent variables and dependent variable All four skill categories were either moderately or highly correlated (Table 6). These relationships ranged from r=.472, p<.001 to r=.658, p<.001, thus satisfying one of the assumptions of MANCOVA. Additionally, a Box’s M value of 83.037 (f=2.74) had an associated p value of .000, which was interpreted as significant. Therefore, equality of variance-covariance between groups was not assumed equal for the purposes of this MANCOVA. While this was a violation of the assumptions of MANCOVA, the failure likely resulted due to differences in sample sizes among the groups (Table 5). According to Mertler and Vannatta (2005) “Violations of this assumption of homoscedasticity, similar to violations of homogeneity, will not prove fatal to an analysis” (p. 34). A one-way multivariate analysis of variance was then conducted. The purpose of the test was to examine the mean differences in leadership skills required by leaders at different levels of the organizational structure. While this study has already confirmed that these skills can be empirically tested, that they were positively correlated to 108 organizational level, and that the amount of leadership skill required was moderated by the leader’s organizational level, it was unclear how these skill requirements differed among leaders at different organizational levels. A statistically significant MANCOVA effect was obtained with Pillai’s Trace = .111, F= (12, 2730) = 8.70, p<.001, and multivariate eta squared = .037. A Multivariate effect size of .037 was estimated and indicated that 3.7% of the variance in the canonically derived dependent variable was accounted for by Organizational level. This suggests that 3.7% of the difference in the mean scores reported for each of the four leadership skill categories at each of the four organizational levels could be explained by the effects of organizational level. Before conducting follow-up ANCOVAs for the purpose of examining the individual differences that existed between the four leadership skills at the four organizational levels, the homogeneity of variance was tested for all four leadership skill subscales by conducting a series of Levene’s tests. Homogeneity of variance was assumed despite two of the four Levene’s tests being reported as statistically significant (p<.05). According to Howell (2007), ANOVAs can be considered robust even if they fail the assumption of homogeneity of variance as long as the largest standard deviations are not more than four times the smallest. None of the largest standard deviations were four times that of the smallest (Table 5) and therefore a series of one-way ANCOVA’s were conducted on each of the four dependent variables as a follow-up to MANCOVA. As seen in Table 11, Organizational level differences were significant for Cognitive skills, F(3, 911) = 9.728, p<.001, partial eta squared = .031. Differences in Interpersonal skills were significant, F(3, 911) = 9.481, p<.001, partial eta squared = .030. Differences in Business skills were also significant F(3, 911) = 26.499, p<.001, partial eta squared = 109 .080. Finally, differences in Strategic skills among the four organizational levels were also significant F(3, 911) = 19.620, p<.001, partial eta squared= .061. This indicated that while controlling for covariates, significant mean differences existed among groups for each of the four leadership skill requirements tested and the magnitude of these differences were small for Cognitive and Interpersonal skills and moderate for Business and Strategic skills (Cohen, 1988). Finally, a series of t-tests with Bonferroni post hoc analyses were calculated to examine individual mean differences across all four organizational levels and all four leadership skill categories, while controlling for the covariates (years of experience, education level, and district setting). 110 Table 11 Significant Univariate Effects for Organizational Level Dependent Variable df df error Cognitive Skill requirement 3 911 Interpersonal Skill requirement Business Skill requirement Strategic Skill requirement 3 3 3 911 911 911 F 9.73 9.48 26.50 19.62 Organizational Level Means 99.9% Confidence Interval Lower Upper Bound bound Assistant Principal 5.55a 5.46 5.65 Building Principal 5.78b 5.70 5.85 Assistant Supt. 5.69ab 5.58 5.80 Superintendent 5.97c 5.85 6.09 Assistant Principal 5.62a 5.53 5.71 Building Principal 5.78b 5.71 5.85 Assistant Supt. 5.70ab 5.59 5.81 Superintendent 6.02c 5.90 6.14 Assistant Principal 5.16a 5.06 5.26 Building Principal 5.40b 5.31 5.57 Assistant Supt. 5.63c 5.51 5.75 Superintendent 5.88d 5.76 6.00 Assistant Principal 5.49a 5.39 5.58 Building Principal 5.78b 5.70 5.85 Assistant Supt. 5.83b 5.72 5.95 Superintendent 6.12c 6.00 6.24 Note. N= 918. Means in each column that do not share subscripts differ at p<.05 using the Bonferroni procedure for multicomparison adjustment. Cognitive skills. There was a statistically significant difference in the means reported for the construct “Cognitive skills” between grouping variables “assistant principal” ( =5.55) when compared to “building principals” ( =5.78, p=.001), and “superintendents” ( =5.97, p<.001). However, there was no significant difference in the means reported for “Cognitive skills” between the groups “assistant principal” ( =5.78, p=.001) and “assistant superintendent” ( =5.69, p=.427). This suggests that as one increases in organizational level from assistant principal to building principal and from assistant principal to superintendent, the amount of Cognitive skills required for the job 111 increases in significant ways. Statistically significant differences also existed for Cognitive skills between “building principals” ( =5.78) and “superintendents” (5.97, p=.040) but not between “building principals” ( =5.78) and “assistant superintendents.” This would also suggest that as one advances from building principal to superintendent an increased amount of Cognitive skills is required. However, there is no difference between the amounts of Cognitive skills required to be a “building principal” or “assistant superintendent.” Finally, there was a significant mean difference in Cognitive skills required between assistant superintendents ( =5.69) and “superintendents” ( =5.97, p=.003) which indicated that educational leaders require higher amounts of Cognitive skills as they move from assistant superintendent to superintendent ( =5.69, p=1.00). This linear relationship is represented in Figure 6. Amount of Cognitive Skill Required 6.5 6.25 6 5.75 Assistant Principal Principal 5.5 Assistant Supt. Supt. 5.25 5 Assistant Principal Principal Assistant Supt. Organizational Level Supt. Figure 6. Estimated marginal means for Cognitive skill requirements across organizational levels while controlling for experience, education, and district setting. 112 Interpersonal skills. There was a statistically significant difference in the reported means for the construct “Interpersonal skills” among the grouping variable "organizational level” such that “assistant principals” ( =5.60) had significantly smaller means than “building principals” ( =5.80, p=.037) and “superintendents” ( =6.04, p<.001). However, “assistant principals” ( =5.60) did not differ in a statistically significant way from “assistant superintendents” ( =5.73 p=1.00). This suggests that as one increases in organizational level from assistant principal to building principal and from assistant principal to superintendent; one requires greater amounts of Interpersonal skills. Conversely, while directionally consistent, the increase in Interpersonal skills required when advancing from assistant principal to assistant superintendent was not significantly different. Additionally, “building principals” ( =5.78) had significantly smaller mean scores on Interpersonal skills than “superintendents” ( =6.02, p=.003). However, there was no significant difference in the mean scores reported among “building principals” ( =5.78) and “assistant superintendents” ( =5.70, p=1.00). This would suggest that as one increased in organizational level from building principal to assistant superintendent, there is no difference in the amount of Interpersonal skills required. In fact, the amount needed tends to slightly decrease. However, there is a significant increase in the amount of Interpersonal skills required as you move from building principal to superintendent. Next, “assistant superintendents” had significantly smaller mean scores on Interpersonal skills ( =5.70) than superintendents ( =6.02, p<.001), which suggests that 113 as one moves from assistant superintendent to superintendent, one requires additional amounts of Interpersonal skills to do the job. Finally, “superintendents” ( =6.02) reported needing significantly higher levels of Interpersonal skills than did all other organizational levels. This suggests that as one moves from any other organizational level to superintendent, one would need to acquire additional amounts of Interpersonal skills to do the job. Figure 7 reflects these relationships. Amount of Interpersonal Skills Required 6.5 6.25 6 5.75 Assistant Principal Principal 5.5 Assistant Supt. Supt. 5.25 5 Assistant Principal Principal Assistant Supt. Organizational Level Supt. Figure 7. Estimated marginal means for Interpersonal skill requirements across organizational levels while controlling for experience, education, and district setting. Business skills. A statistically significant mean difference was reported for all combinations of the dependent variable “Business skills” and the grouping variable “Organization Level” such that “assistant principals”( =5.16) reported needing 114 significantly lower levels of Business skills than “building principals”( =5.39, p=.002), “assistant superintendents”( =5.63, p<.001) and “superintendents”( =5.88, p<.001). Also, the mean for “building principals” ( =5.39) was significantly smaller than that of “assistant superintendents” ( =5.63, p=.004) and “superintendents” ( =5.88, p<.001). Finally, “assistant superintendents” ( =5.63) reported needing significantly lower levels of Business skills than did “superintendents” ( =5.88, p=.020). These findings suggest that no matter one’s current level in the organizational structure, one would require increasing amounts of Business skills as one increases in position. These relationships are shown in Figure 8. Amount of Business Skill Required 6.5 6.25 6 5.75 5.5 Assistant Principal Principal Assistant Supt. 5.25 Supt. 5 Assistant Principal Principal Assistant Supt. Organizational Level Supt. Figure 8. Estimated marginal means for Business skill requirements across organizational levels while controlling for experience, education, and district setting. 115 Strategic skills. A statistically significant mean difference was reported for the construct “Strategic skills” and the grouping variable “Organizational Level”. “Assistant principals” ( =5.49) reported requiring significantly lower amounts of “Strategic skills” than “building principals” ( =5.78, p<.001), “assistant superintendents” ( =5.83, p<.001), and “superintendents” ( =6.11, p<.001). In addition, “building principals” ( =5.78, p<.001), reported requiring lower amounts of “Strategic skills” than “assistant superintendents” ( =5.83, p<.001) and “superintendents” ( =6.11, p<.001). Finally, “assistant superintendents” ( =5.83) reported requiring lower amounts of “Strategic skills” than “superintendents” ( =6.11, p<.001). This would suggest that as you increase in position from assistant principal to any other higher position in the organization, you also tend to require successively larger amounts of Strategic skill to perform the job. Additionally, “building principals” ( =5.78) had significantly smaller mean scores on “Strategic skills” than “superintendents” ( =6.11, p<.001). However, there was no significant difference in the mean scores reported among “building principals” ( =5.78) and “assistant superintendents” ( =5.83, p=1.00). This suggests that as you increase in organizational level from building principal to assistant superintendent there is no reported increase in the amount of Strategic skills required to do the job. However, there is a significant increase in the amount of Strategic skills required as you move from building principal to superintendent. Also, “assistant superintendents” ( =5.83) reported significantly lower amounts of Strategic skills required to do their job than “superintendents” ( =6.11, p=.004). Finally, “superintendents” ( =6.11) reported requiring greater amounts of Strategic skills 116 than any other leadership group. This suggests that assistant superintendents, like all other lower level leadership groups, require lower mounts of Strategic skills to do their jobs than superintendents do. These relationships are illustrated in Figure 9. 6.5 Amount of Strategic Skill Required 6.25 6 5.75 Assistant Principal Principal 5.5 Assistant Supt. Supt. 5.25 5 Assistant Principal Principal Assistant Supt. Organizational Level Supt. Figure 9. Estimated marginal means for Strategic skill requirements across organizational levels while controlling for experience, education, and district setting. Themes As anticipated, the leadership skills of leaders of K-12 public education in New York State can be conceptualized as the four empirically distinguishable skills theorized by Mumford et al (2007). Additionally, these four skills were found to be required in increasing amounts as the leader’s organizational level increased from assistant principal to superintendent. Despite these similarities, leaders in the current study differed in important ways from those of the original 2007 study. These differences were found in 117 the total amount and proportions of leadership skills required at each level of the organization. Interestingly, leaders of K-12 public schools in New York State seem to have required a higher level of overall skills than the diplomats in the original study. Also, the required amount of Interpersonal skills required by school administrators was found to be almost equally as important as Cognitive skills at all levels, rather than subordinate to Cognitive skills as found in the original 2007 study. These findings suggest that while a four-factor model of leadership skill requirements did explain the leadership skills phenomena in this study’s setting, the amount and way in which these skills are required are quite different from the original group studied. This uniqueness should be noted, as it will potentially impact the way educational leaders are formed, developed, and ultimately, perform. While the original 2007 study validated most aspects of the Leadership Skills STRATAPLEX model, this current study was unable to generalize those findings to educational leaders. The fact that these two sample groups were quite different in the amount and proportion of required leadership skills across organizational levels, suggests that a clear boundary or limit to the generalization of the STRATAPLEX model exists. Recapitulation The research question for this study was: To what extent is Mumford et al.’s Leadership Skills STRATAPLEX model generalizable to leaders of New York State’s K-12 public schools? This question was answered by examining each of several hypotheses as presented by Mumford et al. 2007 (pp. 151-159): 118 (H1). The Cognitive, Interpersonal, Business, and Strategic skill categories will be empirically distinguishable. (H2). Leadership skill requirements will vary by skill category such that Cognitive skills will be needed the greatest amount, followed by Interpersonal, Business, and Strategic skills, respectively. (H3). Cognitive, Interpersonal, Business, and Strategic skill requirements will be positively related to the job’s level in the organization. (H4). Leadership skill requirements will interact with organizational level such that: a. Strategic skill requirements will be more strongly related to organizational level than Cognitive, Interpersonal, and Business skill requirements. b. Business skill requirements will be more strongly related to organizational level than Cognitive or Interpersonal skill requirements. c. Interpersonal skill requirements will be more strongly related to organizational level than Cognitive skill requirements. Hypothesis 1 was fully supported as traditional factor analysis suggests that leadership skills were empirically measurable on four distinct factors. Additional support for hypothesis 1 was found when one-factor and four-factor confirmatory factor analysis measurement models using structural equation modeling (SEM) were compared using a chi-squared difference test. Findings suggests that the one factor model did not fit the data well, while a four-factor model fit markedly better. These results were similar to those found by Mumford et al. (2007). 119 While the main effect of the leadership skills required, controlling for covariates, was statistically significant, an evaluation of the estimated marginal means for each of the individual skill requirements revealed that Strategic Skills were required the most followed by Interpersonal skills, Cognitive skills, and Business skills. This suggests that while there is a statistically significant difference in the amount of skills required by leaders, the amounts of each individual skill category (Cognitive, Interpersonal, Business, and Strategic) were not found to be in the hypothesized order of importance. Thus unlike the original 2007 study which found partial support, the data from the 2015 study did not lend any support to Hypothesis 2. Hypothesis 3, consistent with the original study, received full support. Partial correlations between leadership skill requirements and organizational level, controlling for covariates, were found to be statistically significant, positive relationships. This would suggest that as you ascend the organizational structure from front line leadership to executive leadership, all of the leadership skill requirements increase also. Finally, Hypothesis 4, which sought to examine the interaction of leadership skills required with organizational level while controlling for other factors, received mixed results. Rather than all skill levels being statistically significantly different, such that Cognitive skills were most strongly correlated to organizational level followed by Interpersonal, Business, and Strategic, this study found Strategic skills to be most strongly correlated to organizational level, followed by Business, Interpersonal, and Cognitive, when controlling for other factors. Thus, Hypothesis 4a received only partial support, which was consistent with the original 2007 study findings. Hypothesis 4b received full support, which was also consistent with the original study’s findings. 120 However, Hypothesis 4c received no support, which was inconsistent with the original 2007 study’s findings Chapter Summary Researchers have debated for decades how best to conceptualize and measure leadership. Previous researchers conceptualized the leadership phenomena as being made up of traits, styles, personality traits, and interactions between leaders and situations, and as skills (Bass, 1990; Connelly, et al., 2000, Heifetz, 1994; Katz, 1955and 1974; Kellerman & Webster, 2001; Mumford et al. 2007; Mumford et al., 2000a, 2000b, 2000c; Stogdill, 1948 and 1974; Zaccaro, 2007; Zaccaro et al., 2004; Zaccaro et al., 2000). One thing is clear; leadership is complex and a lack of consensus about how it should be conceptualized continues to exist. This researcher chose to examine leadership through a skills-based lens, thus providing additional insight into a segment of leadership research that is still emerging. This study provides strong evidence that educational leaders’ skills, like public administration, can in fact be measured and that those required skills change depending on the position held by that leader. What this study could not prove is that Mumford et. al’s (2007) Leadership Skills STRATAPLEX model was completely generalizable to educational leaders of New York State’s K-12 public schools. In spite of this, a better understanding of the leadership skills required by school administrators in New York State was gained, and this study lays the groundwork for the establishment of a modified STRATAPLEX model, elaborated in the next chapter, that may better fit educational leaders. 121 Chapter Five: Discussion, Conclusions, and Recommendations Introduction The purpose of this study was to establish the extent to which the Leadership Skills STRATAPLEX, as a theoretical model, could be generalized from professional government embassy workers, as in the original study, to leaders of K-12 public schools in New York State. Additionally, given the implications of a stratified, skills-based model of leadership, this study intended to add to the specific field of educational leadership as well as the larger study of leadership skill theory. Both theoretical and practical implications of this study are given. This chapter begins with a discussion of this study’s results, along with its major findings and interpretations of these findings. This chapter explores the relevance of this skills-based leadership study and makes suggestions to the field. Findings from this theoretical replication and extension are discussed and interpreted. Conclusions and implications for professional development, leadership preparation programs, and for the field of leadership skills theory are presented. Finally, the chapter concludes with a reconceptualization of the Leadership Skills STRATAPLEX model followed by a presentation of recommendations and limitations for this study. Discussion This section explores the study’s major findings and interpretations of those findings. Key comparisons are made between the original 2007 and current 2015 studies. These findings lead to implications for and recommendations to the field. Purpose of this study. Like Mumford et al. (2007), this researcher intended to better understand the nature of the leadership skill requirements for leaders at different 122 levels of an organization. Unlike Mumford et al., this researcher sought to move beyond the validation of theory to the specific application of the Leadership Skills STRATAPLEX model to the context of public school leaders in New York State. By using the Leadership Skills STRATAPLEX model, which intends to capture the “layered (strata) and divided complex (plex) nature of leadership skill requirements” (Mumford et al., 2007, p.163), this researcher examined the discrete leadership skill requirements of K-12 public school leaders in New York State and compared these findings to Mumford et al.’s 2007 study. This theoretical replication and extension allowed the researcher to test the extent to which the Leadership Skills STRATAPLEX model could be generalized to a new setting and context, either extending or identifying limits to the model’s usefulness. Additionally, this study allowed for an examination of the unique leadership skills required of current leaders in the field of K-12 public education in New York State. Theoretical replication, findings, and interpretations. All aspects of the Leadership Skills STRATAPLEX, as tested in Mumford et al., were tested by this researcher within a large sample of educational leaders in New York State. Overall, the Leadership Skills STRATAPLEX model failed to fully generalize to this new context and setting. The results suggest that a possible theoretical boundary or limit may exist for this specific leadership skills-based model. The identification of a limit or boundary to this model is useful as it serves to draw important distinctions between the leaders studied in the original sample (professional U.S. government employees), and the educational leaders examined in this current study. These distinctions led this researcher to suggest a reconceptualized Leadership Skills STRATAPLEX model that better explains the unique skill requirements of educational leaders in New York. Like the original, the proposed 123 skill-based model accounts for leadership skill requirements across organizational levels through the use of a STRATAPLEX triangle. However, this new model suggests that organizational levels can be organized into sub-groups and that more than one STRATAPLEX triangle may actually exist. It also suggests that these independent STRATAPLEX triangles may exist in different sizes and proportions, representing the different amounts and relative importance of each of the four leadership skills required by each sub-group. Finally, this researcher proposes that the series of STRATAPLEX triangles can be arraigned along a third dimension that represents the sub-group’s sphere of direct influence relative to the whole organization. Major findings from this study are presented below. The first and most foundational finding of this study was that the leadership skills, as theorized by Mumford et al. (2007), were empirically measured within this large sample of New York State public school leaders. This seemingly basic finding confirmed, within this sample, the theoretical base upon which previous leadership skills theories have been built (Connelly et, al. 2000; Katz, 1955; Mumford et al. 2000a; Mumford et al., 2000d). The ability to identify individual, discrete leadership skill types is a significant finding of this study as very little empirical evidence of this phenomenon is present in the literature, and no evidence at all of this type of research was found in studies of educational leadership. Therefore, the findings of this study suggest that educational leadership can and should be further examined through the lens of a skillsbased approach. Second, like the original 2007 study, leadership skill requirements were overall, positively related to the leader’s organizational level. While controlling for the 124 differences due to the leader’s education level, experience level, and district setting, the amount of leadership skills reportedly required by educational leaders tended to increase as the educational leader’s organizational level increased. In other words, the leadership skills required by educational leaders to do their jobs may be cumulative, such that each subsequently higher leadership position requires all the skills gained at the lower organizational levels and more. This finding was consistent with those found in the original study where a sample of professional government employees tended to require overall increasing levels of leadership skills as they moved from lower to higher positions of leadership. Again, the general overall increase in required leadership skills found by this researcher further demonstrates and extends the theorized cumulative nature of leadership skills found by Mumford et al. 2007. This finding suggests that this relationship between organizational level and leadership skills can be demonstrated in New York public school leaders. Third, among the current sample of New York State educational leaders, at each of the four organizational levels, the amounts and proportions of each of the four leadership skills were found to be different from those found among U.S. diplomats by Mumford et al. in 2007. The researcher found that across organizational levels, Strategic skills were generally needed in the largest amounts, followed by Interpersonal, Cognitive, and Business skills. Interestingly, Interpersonal and Cognitive skills tended to be required in roughly the same amounts, at most organizational levels. These findings suggest that educational leaders may require a very different combination of leadership skills than those reported by the original 2007 study. In addition, educational leaders reported needing leadership skills in more balanced amounts, especially Cognitive and 125 Interpersonal. While three of the four leadership skill categories (Strategic, Cognitive, and Interpersonal skills) were required in similar amounts, it was Cognitive and Interpersonal skills that tended to be most closely related to one another. Therefore, educational leaders reportedly require a different and more balanced amount of leadership skills than previously suggests by the Leadership Skills STRATAPLEX model. Fourth, the total overall amount of leadership skills required by leaders to do their job, across and within all organizational levels, was found to be greater for educational leaders than those examined in the original study. This difference suggests that educational leaders of New York State’s K-12 public schools feel that their positions require greater amounts of each of the four leadership skills, at each of the four organizational levels, than do their counterparts, working as professionals in an international agency of the U.S. government. Fifth, similarly to the original study, this study found a significant relationship between “organization level” and each of the four “leadership skill categories,” which means that as educational leaders increase in position they also tend to need more Strategic, Cognitive, Interpersonal, and Business skills. The magnitude of the relationship was such that Strategic and Business skills were more strongly related to organizational level than Interpersonal and Cognitive skills, which means that the amount of Strategic and Business skills required by educational leaders tends to increase more rapidly from one organizational level to the next than do Interpersonal and Cognitive skills. This suggests that it would be more important for an aspiring leader to focus on the development of Strategic and Business skills over Cognitive and Interpersonal skills as this leader progresses from lower leadership positions to higher as Strategic and Business 126 skills tend to increase the most from level to level. This finding was somewhat surprising in that generally, educators and educational leaders are thought to rely heavily on soft skills, such as the ability to empathize, communicate, work with, and influence others. However, this study revealed that educational leaders may in fact be more similar to leaders in other organizations, than we tend to believe. The strong relationships between Strategic and Business management abilities and a leader’s position would suggest that our traditional assumptions about educational leaders might not be true. Educational leaders and leaders of other organizations seem to need very similar amounts of Strategic and Business skills, especially as you move towards executive levels such as superintendent. It is important to note that while these findings suggest practitioners focus more strongly on the importance of Strategic and Business skill development, continued development of all four skills is necessary. This relationship does not indicate the order of importance of each skill at each level, but rather only indicates that Strategic and Business skills increase at the highest rates from level to level. The cumulative nature of these four leadership skills, demonstrated by the tendency of educational leaders to report needing greater and greater amounts of each skill as they move from lower to higher leadership positions, necessitates a continued focus on the development of all four leadership skills by leaders at all levels. Extension, findings, and interpretations. Analysis beyond the scope of the original 2007 study found that the multivariate main effect of organizational level on all four leadership skills was significant. This means that an educational leader’s position in the organization has a meaningful impact on the combination of leadership skill amounts 127 reportedly required by those leaders. Through the use of MANCOVA, the researcher was able to eliminate the effects that experience, education, and district setting have on the required leadership skills reported by leaders. In addition, this statistical technique allowed the researcher to account for the positive relationships that existed among all four leadership skill types. This means that the use of MANCOVA allowed the researcher to account for the effects that resulted from the tendency of all four leadership skills to increase as one or more of the other skills increases. The ability to remove the impacts of experience, education level, and district setting, while also accounting for the relationships that existed between the four leadership skill types, provided a clearer picture of the unique influence an educational leader’s position has on the leadership skills required of that leader. Educational leaders of public schools in New York State reported that overall Strategic skills were needed in the greatest amounts. This was followed by Interpersonal and Cognitive skills which seemed to be needed in very similar amounts to one another at all four of the organizational levels studied. Finally, leaders reported that Business skills were needed in the smallest amounts at each of the four levels. It is important to remember that while Business skills were needed in the smallest amounts at all four levels; they also tended to increase at a much more rapid rate from one level to the next than Cognitive or Interpersonal skills. Overall, the amounts of leadership skills reported suggests that educational leaders at all levels need Strategic skills in the highest amounts, followed by Interpersonal and Cognitive and Business. Analysis at the univariate level allowed the researcher to examine the differences in the amounts of each of the four leadership skill types among the four organizational levels studied. Because MANCOVA was utilized, these results again accounted for the 128 impact of experience, education level, and district setting on the leadership skill amounts reported. Unlike the multivariate analysis, which measured the overall effect of organizational level on leadership skills required, a univariate analysis examined the differences in the amounts of leadership skills reported among leaders at each of the four organizational levels. This univariate analyses revealed important inconsistencies between this researcher’s findings and the hypothesized Leadership Skills STRATAPLEX model. It was discovered that while overall, educational leaders required Strategic skills the most, followed by Interpersonal, Cognitive, and Business skills, the order of importance and amount of each leadership skill varied from level to level. This means that unlike the original 2007 study, this study’s findings did not show meaningful increases in all four leadership skills at all four organizational levels. This was true specifically between building principals and assistant superintendents. Instead, this researcher found the following results: First, rather than reporting distinctly different required amounts of each of the four leadership skills, assistant principals tended to require equally high amounts of Cognitive and Interpersonal skills, followed by lower amounts of Strategic and Business skills (Figure 5). Second, the amount of each of the four leadership skills tended to increase as leaders moved from assistant principal to building principal and again from assistant superintendent to superintendent. However, from assistant principal to building principal, the increases in the amount of Strategic, Interpersonal, and Cognitive skills resulted in leaders reporting that they needed very similar amounts of all three skills. This suggests that building principals may require the most balanced amounts of Strategic, Interpersonal, and Cognitive skills than leaders at any of the other three organizational levels examined. 129 Third, the largest inconsistency between this studies results and the theorized Leadership Skills STRATALEX model was found between building principals and assistant superintendents. The researcher found that the required amounts of Cognitive, Interpersonal, and Strategic skills remained roughly the same for building principals and assistant superintendents. Moreover, Interpersonal and Cognitive skills decreased slightly when moving to the assistant superintendence from the principalship, though in a nonstatistically significant way. This finding, which is inconsistent with hypothesis 3, suggests that despite holding very different levels in the organizational structure, building principals and assistant superintendents tend to report very little difference in the amount of Strategic, Interpersonal, and Cognitive skill levels required to do their jobs. This anomaly demonstrates that the leadership skills of educational leaders do not always increase from organizational level to organizational level as previously hypothesized. This finding led this researcher to conclude that the Leadership Skills STRATAPLEX model, in its current form, could not be completely generalized to leaders of K-12 public education in New York State. However, due to the many similarities found between the original 2007 study and the current study, this researcher suggests a reconceptualization of the Leadership Skills STRATAPLEX that may better account for the leadership skills required by leaders of complex multi-tracked leadership structures such as are found in public education. Conclusions/Implications Implications for professional development. Like Mumford et al.’s (2007) study, this study demonstrates that leadership skills tend to increase as you increase in organizational level, as conceptualized in educational leadership. This increase intimates 130 that leadership skills are cumulative, such that all skills required at lower levels of the organization are also required at higher levels. In terms of professional development, this suggests that professional development opportunities should, to an extent, focus on supporting the development of all four skill areas (Cognitive, Interpersonal, Business, and Strategic), regardless of the leader’s current organizational level. Professional development must also be individualized, taking into account the unique amounts and proportions of each skill category at each organizational level. Therefore, professional development must consider both the requirements of the leader’s current level in the organization and those of successively higher levels, allowing the leader to sharpen the skills needed for his/her current position, while developing the leader’s skills necessary at the next level. Organizational level matters. When looking to develop leaders within an organization, having a clear understanding of the skills needed at each level of the organization is imperative. Matching appropriate experiences, instruction, and professional development to the needs of a leader’s position, rather than the individual person, provides a more efficient way of offering targeted professional development. The findings of this study suggest a clear place to start. Finally, the findings of this study would implore organizations to focus additional professional development efforts on Strategic and Business skills. The findings demonstrate that as you increase in organizational level, the need for Business and Strategic skills increase at the fastest rates. Knowing this allows professional development programs to target the skills that have the greatest ability to differentiate leaders of higher and lower organizational levels. Increasing a leader’s ability to 131 understand downstream consequences construct a vision for the organization; and manage resources, people, and equipment will prepare these leaders to assume new responsibilities. A strong focus on Strategic and Business skills will ensure that leaders are prepared for the challenges of their current position and beyond. Implications for leadership preparation programs. Unlike professional development, leadership preparation programs and certification programs are a requirement for leaders of public schools in New York State. Again, understanding that leadership skills are identifiable, cumulative, and required in different proportions at different organizational levels, has important implications for these programs. First, leaders must prioritize the acquisition and exercise of Strategic skills. Strategic leadership courses should therefore be the centerpiece of a strong leadership preparation program becoming even more important when preparing executive level leaders. Providing leaders the skills necessary to successfully navigate novel, ill-defined problems of practice is essential for leadership programs as the responsibilities of all leaders continue to change (Fullan, 2001; Heifetz, 1994; Kellerman & Webster, 2001; Mumford et al., 2000b). While the study of leadership is complex and disagreement persists in the field over how to define it, leadership development programs would do well to pay particular attention to skills based theories, with an emphasis on strategic planning. Unlike leadership traits, leadership skills are, by definition, developable. That they can be developed is suggests by the findings of this study and others (Bass, 1990; Mumford et al., 2007; Mumford et al., 2000a; Mumford et al., 2000b; Mumford et al. 2000d). Preparatory programs can use this study’s findings to further refine their curriculum to provide potential leaders with the foundational skills required for the job. In fact, the 132 specific findings of this study might be worthy of sharing with students in educational leadership programs, superintendent development programs, and practicing educational leaders alike. Implications for leadership skills theory development. In light of the study’s findings and interpretations, the Leadership Skills STRATAPLEX model was reconceptualized to offer a more highly developed and realistic model of the complex relationship between organizational level and skill level. As all leadership skill requirements were not found to increase between all organizational levels, a simple reorganization of required leadership skills within the existing model’s structure (Figure 1) was not possible. The failure of Cognitive, Interpersonal, and Strategic skills to increase from organizational level “building principal” to “assistant superintendent” created an interesting anomaly. Visual observations of Figure 5 led this researcher to consider dividing the four organizational levels into two, with levels assistant principal and building principal forming one sub-group, and assistant superintendent and superintendent forming another. The STRATAPLEX model’s assumption that skills will increase at all organizational levels is satisfied when the data are re-examined in this way. Each new leadership subgroup then creates its own unique STRATAPLEX triangle. This triangle contains the subgroup’s unique amount and proportion of the four required leadership skills. Therefore, the STRATAPLEX triangle housing the two organizational levels “assistant principals” and “building principals” are arraigned such that the overall lower amount of total leadership skills (size of the triangle), and more balanced proportions of these skills (bands within the triangle) are reflected. The STRATAPLEX triangle housing the two organizational levels “assistant superintendents” 133 and “superintendents” reflects the overall larger amounts and less balanced proportions of the four leadership skills required (Figure 10). Figure 10. Schematic of multiple Leadership Skills STRATAPLEX pyramids of varying size denoting the total amount of leadership skill required. Internal divisions represent the amounts and proportions of the four leadership skill types (Cognitive, Interpersonal, Business, and Strategic) required at each organizational level (Assistant Principal, Building Principal, Assistant Superintendent, and Superintendent). Next, this new model suggests that each STRATAPLEX triangle can be thought to be operating in three dimensions. The horizontal x-axis represents the total amount of leadership skills required. Along this axis, the base of the equilateral triangle increases or decreases (moving out from the origin) based on the total amount of leadership skills required by the leaders represented by that triangle or sub-group. The vertical y-axis represents the organizational level within and among each STRATAPLEX triangle. The x and y-axis are related in that as the base of the equilateral triangle broadens along the xaxis, indicating greater amounts of total leadership skills required, the peak of the triangle also increases, indicating higher organizational levels. This positive relationship between “leadership skills” and “organizational level” was found by both the current study and the original 2007 study and is graphically represent here. The third dimension of this 134 reconceptualized model is represented by the addition of the z-axis. This axis graphically represents the differences found between leadership subgroups. As a leader and their sub-group gain a greater sphere of direct influence relative to the whole organization, the triangle moves out from the origin along the z-axis. This third dimension was not considered or accounted for in Mumford et al.’s original model and is an original contribution to the field of leadership skills theory (Figure 11). Figure 11. Leadership skills interacting on a three-dimensional plane. The x-axis denotes amount of total leadership. The y-axis represents organizational level and the z-axis represents the leadership sub-groups sphere of direct influence relative to the whole organization. This reconceptualization of the Leadership Skills STRATAPLEX not only accounts for the leadership skill requirements of leaders at each organizational level but adds a third dimension that accounts for the leader’s sphere of direct influence relative to the whole organization. Specifically, this model better represents the differences found 135 between educational leaders who operate at the building level (assistant principals and building principals) and those who operate at the district level (assistant superintendents and superintendents) while also representing the overlap in skill requirements found to exist between building principals and assistant superintendents. As such, this researcher proposes that educational leadership is not one STRATAPLEX triangle, but rather two or more triangles along a continuum that varies by the leader’s leadership skill amounts, organizational level, and scope of direct influence relative to the whole organization (Figure 12). Figure 12. Schematic of the reconceptualized Leadership Skills STRATAPLEX model. The smaller triangle represents the leadership sub-group “School Building Leaders” made up of Assistant Principals and Building Principals. The larger triangle denotes the leadership sub-group “School District Leaders” comprised of Assistant Superintendents and Superintendents. Note. The “School District Leaders” sub-group is larger denoting greater amounts of total leadership required and is located further from the origin along the z-axis, representing this group’s increased sphere of direct influence relative to the whole organization, as compared to sub-group “School Building Leaders”. 136 This new conceptualization of the Leadership Skills STRATAPLEX differs from previous leadership skills models such as Katz (1955) and (1974) and Mumford et al. (2007). Unlike Katz (1974), who proposed that at different organizational levels some skills are present and others are not, this model proposes that all four-skill categories are present at every organizational level. This new model, like Mumford et al. (2007), suggests that the required amount of each leadership skill increases as the leader’s position increases. However, unlike Mumford et al.’s model, this new model suggests that organizational levels may exist in bands and represents increases in overall leadership through individual, successively larger triangles. This new reconceptualization allows the ability to place these leadership sub-groups along a third axis representing that group’s sphere of direct influence relative to the whole organization. This more sophisticated model suggests that skills are cumulative, increase as organizational levels increase, and are required in different amounts and proportions at each level. This is supported by the differences found between the current and original 2007 studies such that building principals required a more balanced amount of Cognitive, Interpersonal, and Strategic skills than did assistant principals. In addition, this model suggests that organizational levels are not continuous and can be grouped based on the sub-groups sphere of direct influence relative to the whole organization. Therefore, leaders at higher organizational levels within a sub-group who exercises less influence (building principals) can share leadership skill characteristics with leaders at lower organizational levels within a higher sub-group who exercises more influence (assistant superintendents). This is supported by the lack of differentiation in three of the four skill categories between building principals and assistant superintendents. 137 The reconceptualized Leadership Skills STRATAPLEX model could be hypothesized to extend beyond the traditional school leaders examined by this study. Teacher leaders, who hold non-traditional or informal leadership roles, are hypothesized to require increasing amounts of Strategic, Interpersonal, Cognitive, and Business skills as they advance from the role of classroom teacher to department chair, grade, and team leader. The amounts and proportions of these leadership skill requirements would also be moderated by the scope of the leader’s sphere of direct influence relative to the whole organization (Figure 13). Figure 13. Schematic of the hypothesized, reconceptualized Leadership Skills STRATAPLEX model with extension. The smallest triangle represents the leadership sub-group “Teacher Leaders” and is made up of classroom teachers and teacher leaders”. The small triangle located closest to the origin along the z-axis denotes the lower overall skills required and limited sphere of direct influence relative to the whole organization. 138 The medium sized triangle represents the leadership sub-group “School Building Leaders” made up of Assistant Principals and Building Principals. This sub-groups triangle represents its’ increased skill requirements and sphere of direct influence, in relation to “teacher leaders”. The larger triangle denotes the leadership sub-group “School District Leaders” comprised of Assistant Superintendents and Superintendents. Note. The “School District Leaders” sub-group is larger denoting greater amounts of total leadership required and is located further from the origin along the z-axis, representing this group’s increased sphere of direct influence relative to the whole organization, as compared to sub-group “School Building Leaders”. Finally, this new leadership skills model may be useful for the examination of other complex organizations. Organizations like the military that require increasing levels of leadership skill requirements as you increase in organizational level and have organizational sub-groups that can be divided based on their spheres of direct influence relative to the whole organization provide a clear parallel. Like the educational leaders studied by this researcher, military leaders require increasing amounts of leadership skills as they rise in rank and also need these skills in different amounts and proportions at each level (Mumford et al., 2000b). For example, the first STRATAPLEX triangle, signifying a sub-group of leaders within a school building, is comprised of teachers and teacher leaders. A teacher who leads a class of 30 students is much like a non-commissioned staff sergeant in the United States Army who leads a squad of 6-10 soldiers. In both cases, this is the smallest organizational group who requires leadership. Next, a team leader, grade level leaders, or department chairperson is a teacher leader who leads groups of teachers and assists the building administrator, much like a lieutenant who leads a platoon made up of 3-4 squads assists the captain who leads a company. These two organizational levels, teachers and teacher leaders, sergeants and lieutenants, make up the basic unit of school and military leadership and is the first STRATAPLEX triangle in this new model. 139 At the school building level, another organizational sub-group emerges with broader direct influence relative to the whole organization. An assistant principal is often the leader of several grade levels or departments, comprised of several teacher leaders who lead groups of classroom teachers. Assistant principals are often assisted by teacher leaders and in turn support the building principal. The assistant principal and building principal make up a group of educational leaders who must hold New York State School Building Leader (SBL) certification. The building principal is charged with leading the whole school building made up of multiple grade level teams or departments and is supported by assistant principals and teacher leaders. Again, a parallel can be made between school building leaders and an army company commander or captain who is a commissioned officer and leads 3-4 platoons each of which is made up of 3-4 squads of soldiers. The captain is supported by the lieutenants and staff sergeants in much the same way the building principal is supported by the assistant principal(s) and teacher leaders. The captain and lieutenant, building principal, and assistant principal make up the second STRATAPLEX pyramid with broader overall amounts of direct influence relative to the whole organization. Despite the general differences in the overall scope of direct influence held by teachers and teacher leaders, and assistant principals and building principals, teacher leaders, and assistant principals tend to share responsibilities and in turn may require similar amounts of the four leadership skills studied. This overlap from the top of one sub-group to the bottom of the next is thought to exist much like the overlap between building principals and assistant superintendents found in this study. 140 Limitations While this study controlled for factors like “years of experience,” “level of education,” and “district setting,” suspected to directly or indirectly affect the leadership skills of public school leaders in New York State, other factors may exist. The omission of these potential factors and the overall lack of skills-based leadership research conducted in the public school setting are major limitations of this study. Cross-sectional research designs, used by this researcher, present several weaknesses, including the inability to observe changes over time or the ability to account for the specific mood of the respondent at the time the survey was completed. Collecting data at a single point in time can affect the reliability and generalizability of the survey data collected. As a theoretical model, The Leadership Skills STRATAPLEX was originally tested and validated in a sample of professional government employees. This group operated within an “up or out” promotional system. Consequently, if you failed to be promoted within a specific period of time, you were forced to retire, much as is the case in the U.S. military. This left a sample of leaders in positions who demonstrated upward mobility within the organization and an absence of leaders who had reached their highest level without the ability to progress. Conversely, this dynamic does not exist when studying leaders of public schools. Educational leaders hold positions in the organizational structure for many reasons, both personal and professional, opting at times to remain at a particular organizational level for long periods of time for a multitude of reasons. Additionally, some leaders remain at their current organizational level for many years despite attempts to move up. This may suggest in part, that there could be limits to an 141 individual’s capacity or desire to acquire the levels of skills necessary to operate at higher levels in the organization. This study attempted to examine the perceived leadership skills required of educational leaders at different organizational levels in New York State. Because of the state-specific sample, the findings cannot be generalized easily to other states, nonpublic schools, or other occupations. Finally, descriptive research methodology was used, and, by its nature, descriptive research seeks to describe rather than predict. This eliminates the ability of this study’s findings to be used as a predictive tool. Recommendations While this study was able to quantitatively examine a large sample of public school leaders in New York State, future research should focus on non-traditional leadership positions, such as teachers and teacher leaders. The addition of this subgroup of leaders would further test this new leadership skill model and potentially validate the construct “sphere of direct influence.” Additionally, future researchers are encouraged to conduct skills-based research in different states, thus establishing the generalizability, limits, or boundaries of this study’s findings. Future research should be conducted to validate this new reconceptualized Leadership Skills STRATAPLEX model, as proposed by this researcher. In order to test the three-dimensional nature of this new model, it is necessary that this research be done in settings/contexts with highly stratified organizational levels and distinct variations in the sphere of influence relative to the whole organization. It is also suggests that future research be conducted in settings like the military, where leadership is distributed along 142 multiple tracts, (non-commissioned and commissioned officers) and clear delineations exist in the sphere of direct influence between and among these groups. Future, researchers are also encouraged to conduct qualitative and mixed methods studies of leadership skills across organizational levels. These studies would better capture nuances that exist between and among leadership skills at different organizational levels, which were not captured by the purely quantitative approach of this study. Finally, leadership skills research, specifically related to public school leaders would benefit from longitudinal designs. This methodological design would enable the researchers to evaluate specific changes over time and better protect against the potential variation in the moods of individuals at the time of the survey. This could also include following a junior leader from the start of his or her career to retirement. A longitudinal study would serve to complement cross-sectional studies by protecting against the limitations inherent in their design. Chapter Summary This study tested a previously established, skills-based leadership model within a large sample of public school leaders in New York State. Results suggests that the Leadership Skills STRATAPLEX model did not fully generalize to this new context and setting. However, several aspects of the model’s core tenets did. Specifically, the four leadership skills (Cognitive, Interpersonal, Strategic, and Business) as conceptualized by Mumford et al. (2007) were empirically distinguishable. Next, leadership skills are positively related to organizational level. In addition, leadership skills tended to be cumulative such that all the skills required at lower organizational levels were still needed at higher levels. Finally, a clear violation of the models assumptions occurred; 143 the leadership skill interactions between building principals and assistant superintendents did not follow the hypothesized trends. In fact, three skill types (Cognitive, Interpersonal, and Strategic) were required in essentially the same amounts for building principals and assistant superintendents. This difference between the theorized model and this study’s findings became the basis for a reconceptualized Leadership Skills STRATAPLEX model. This new model not only reorganized the skill requirements within the existing model’s triangle, such that Strategic skills were required the most, followed by Interpersonal, Cognitive, and Business but also suggests that multiple STRATAPLEX triangles exist and can be conceptualized as differentiating along a third dimension. This third dimension, representing a leader’s “direct sphere of influence” relative to the whole organization, then becomes a potentially important additional factor in understanding leadership skill requirements across organizational levels. 144 References Antonakis, J., Cianciolo, A. T., & Sternberg, R. J. (Eds.). (2004). The nature of leadership. Thousand Oaks: CA: Sage Publications. Bass, B. M. (1990). Bass and Stogdill’s handbook of leadership: Theory, research and managerial applications (3rd ed.). New York: The Free Press. Bhaskar, R. (1998). The possibility of naturalism: A philosophical critique of the contemporary human sciences (3rd ed.). London: Routledge. Bosco, F. A., Aguinis, H., Singh, K., Field, J. G., & Pierce, C. A. (2015). Correlational effect size benchmarks. Journal of Applied Psychology, 100(2), 431-449. http://dx.doi.org/10.1037/a0038047 Burman, L. E., Reed, W. R., & Alm, J. (2010). A call for replication studies. Public Finance Review, 38(6), 787-793. doi: 10.1177/1091142110385210 Butin, D. W. (2010). The education dissertation: A guide for practitioner scholars Thousand Oaks, CA: Sage. Byrne, B. M. (2013). Structural equation modeling with AMOS: Basic concepts, applications, and programming. New York: Routledge. Carter, L. F. (1952). Leadership and small-group behavior. Army Field Forces Human Research Unit No. 2. Cohen, J. (1962). The statistical power of abnormal-social psychological research: A review. The Journal of Abnormal and Social Psychology, 65(3), 145153. http://dx.doi.org/10.1037/h0045186 145 Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.). Hillsdale, NJ: L. Erlbaum. Cohen, J. (1990). Things I have learned (so far). American Psychologist, 45(12), 13041312. doi:10.1037/0003-066X.45.12.1304 Cohen, J. (1992). A power primer. Psychological Bulletin, 112(1), 155159. http://dx.doi.org/10.1037/0033-2909.112.1.155 Cohen, J., & Cohen, P., (1983). Applied multiple regression/correlation analysis for the behavioral sciences (2nd ed.). Hillsdale, NJ: Lawrence Erlbaum Associates. Cohen, J., Cohen, P., West, S. G., & Aiken, L. S. (2003). Applied multiple regression/correlation analysis for the behavioral sciences (3rd ed.). Mahwah, NJ: L. Erlbaum Associates. Connelly, M. S., Gilbert, J. A., Zaccaro, S. J., Threlfall, K. V., Marks, & M. A., & Mumford, M. D.(2000). Exploring the relationship of leadership skills and knowledge to leader performance. The Leadership Quarterly, 11(1), 65-86. http://dx.doi.org/10.1016/S1048-9843(99)00043Costello, A. B., & Osborne, J. W. (2005). Best practices in exploratory factor analysis: Four recommendations for getting the most from your analysis. Practical Assessment, Research & Evaluation, 10(7), 1-9. Retrieved from http://pareonline.net/ Creswell, J. W. (2003). Research design: Qualitative, quantitative, and mixed methods approaches (2nd ed.). Thousand Oaks, CA: Sage. 146 Duvendack, M., & Palmer-Jones, R. (2013). Replication of quantitative work in development studies: Experiences and suggestions. Progress in Development Studies. 13(4), 307-322. doi: 10.1177/1464993413490480 Fullan, M. (2001). Leading in a culture of change. San Francisco, CA: JosseyBass Gall, M. D., Gall, J. P., & Borg, W. R. (2003). Education research: An introduction (7th ed.). New York, NY: Pearson Education Inc. Galton, F. (1869). Hereditary genius: An inquiry into its laws and consequences. London: Macmillan and CO. George, D., & Mallery, P. (2011). SPSS for windows step by step: A simple guide and reference (11th Ed.). New York, NY: Allyn & Bacon. Gibb, D. (1954). Leadership. In G. Lindzey, & E. Aronson (Eds.), The handbook of social psychology (pp. 877-918). Reading Mass: Addison-Wesley Publishing Company. Heifetz, R. A. (1994). Leadership without easy answers. Cambridge MA: Harvard University Press. Hoaglin, D. C., & Iglewicz, B. (1987). Fine-tuning some resistant rules for outlier labeling. Journal of the American Statistical Association, 82(400), 1147-1149. doi: 10.1080/01621459.1987.10478551. Hoaglin, D. C., Iglewicz, B., & Tukey, J. W. (1986). Performance of some resistant rules for outlier labeling. Journal of the American Statistical Association, 81(396), 991999. doi:10.1080/01621459.1986.10478363 147 Holm, S. (1979). A simple sequentially rejective multiple test procedure. Scandinavian Journal of Statistics, 6 (2), 65-70. Retrieved from http://onlinelibrary.wiley.com/journal/10.1111/(ISSN)1467-9469 Howell, D. C. (2007). Statistical methods for psychology (6th ed.). Belmont, CA.: Thomson Wadsworth. Hoyle, R. H. (Ed.). (2012). Handbook of structural equation modeling. New York: Guilford Press. Hunt, J. G. (1999). Transformational/charismatic leadership’s transformation of the field: An historical essay. The Leadership Quarterly, 10 (2), 129–144. doi: 10.1016/S1048-9843(99)000015-610. Iglewicz, B., & Banerjee, S. (2001, August). A simple univariate outlier identification procedure. In Proceedings of the Annual Meeting of the American Statistical Association. Retrieved from http://www.amstat.org/sections/srms/proceedings/y2001/proceed/00523.pdf Jago, A, G. (1982). Leadership: Perspectives in theory and research. Management Science, 28 (3), 315-336. http://dx.doi.org/10.1287/mnsc.28.3.315 Johnson, B., & Christensen, L. B. (2012). Educational research: Quantitative, qualitative, and mixed approaches (4th ed.). Thousand Oaks, CA: Sage. Jose, V. R. R., & Winkler, R. L. (2008). Simple robust averages of forecasts: Some empirical results. International Journal of Forecasting, 24(1), 163-169. Retrieved from http://doi:10.1016/j.ijforecast.2007.06.001 148 Katz, R. L. (1955, January-February). Skills of an effective administrator. Harvard Business Review, 33(1), 33-42. Retrieved from https://hbr.org/ Katz, R. L. (1974, September-October). Skills of an effective administrator. Harvard Business Review, 52(5)., 90-102. Retrieved from https://hbr.org/ Kellerman, B., & Webster, S. W. (2001). The recent literature on public leadership: Reviewed and considered. The Leadership Quarterly, 12(4), 485-514. http://dx.doi.org/10.1016/S1048-9843(01)00091-1 Kennedy, J. J., & Bush, A. J. (1985). An introduction to the design and analysis of experiments in behavioral research. Lanham, MD: University Press of America. Kirkpatrick, S. A., & Locke, E. A. (1991). Leadership: Do traits matter?. The Executive, 5(2), 48-60. doi:10.5465/AME.1991.4274679 Klein, R. A., Ratliff, K. A., Vianello, M., Adams, R. B., Jr., Bahnik, S., Bernstein, M.J….Noresk, B.A. (2014). Investigating variation in replicability: A “many labs” replication project. Social Psychology, 45(3), 142-152. doi:10.1027/18649335/a000178 La Sorte, M. (1972). Replication as a verification technique in survey research: A paradigm. The Sociological Quarterly, 13(2), 218-227. doi:10.1111/j.15338525.1972.tb00805.x. Leedy, P. D., & Ormrod, J. E. (2005). Practical research: Planning and design (8th ed.). Upper Saddle River, NJ: Pearson. 149 Lord, R. G., de Vader, C. L., & Alliger, G. M. (1986). A meta analysis of the relations between personality traits and leadership perceptions: An application of validity generalization procedures. Journal of Applied Psychology, 7(3), 402-410. http://dx.doi.org/10.1037/0021-9010.71.3.402 Malone, B. A., & Caddell, T. A. (2000). A crisis in leadership: Where are tomorrow’s principals? The Clearing House, 73(3), 162-164. doi:10.1080/00098650009600938 Mann, R. D. (1959). A review of the relationships between personality and performance in small groups. Psychological Bulletin, 56(4), 241-270. doi:10.1037/h0044587. Markow, D., Macia, L., & Lee, H. (2013). The MetLife survey of the American teacher: Challenges for school leadership. 1-120. Retrieved from MetLife, Inc. website: https://www.metlife.com/assets/cao/foundation/MetLife-Teacher-Survey2012.pdf Mascall, B., & Leithwood, K. (2010). Investing in leadership: The district’s role in managing principal turnover. Leadership and Policy in Schools, 9(4), 367-383. doi:10.1080/1570000763.2010.493633 Mertler, C. A., & Vannatta, R. A. (2005). Advanced and multivariate statistical methods: Practical application and interpretation. Glendale, CA: Pyrczak. Mumford, T. V, Campion, M. A., & Morgeson, F. P. (2007). The leadership skills strataplex: Leadership skill requirements across organizational levels. The Leadership Quarterly, 18(2), 154-166. doi:10.16/j.leaqua.2007.01.005 150 Mumford, M. D., Marks, M. A., Connelly, M. S., Zaccaro, S. J., & Reiter-Palmon, R. (2000a). Development of leadership skills: Experience and timing. The Leadership Quarterly, 11(1), 87-114. doi:10.1016/S1048-9843(99)00044-2 Mumford, M. D., Zaccaro, S. J., Connelly, M. S., & Marks, M. A. (2000b). Leadership skills: Conclusions and future directions. The Leadership Quarterly, 11(1), 155170. doi:10.1016/S1048-9843(99)00047-8 Mumford, M. D., Zaccaro, S. J., Harding, F. D., Jacobs, T.O., & Fleishman, E. A. (2000c). Leadership skills for a changing world: Solving complex social problems. The Leadership Quarterly, 11(1), 11-35. doi:10.1016/S10489843(99)00041-7 Mumford, M. D., Zaccaro, S. J., Johnson, J. F., Diana, M., Gilbert, J. A., & Threlfall, V.K. (2000d). Patterns of leader characteristics: Implications for performance and development. The Leadership Quarterly, 11(1), 115-133. doi:10.1016/S10489843(99)00045-4 Nadler, D, A., & Tushman, M. L. (1988, June 6). What makes for magic leadership. Fortune, 117(12), 261-262. Retrieved from http://fortune.com/ Norsek, B. A., & Lakens, D. (2014). Registered reports: A method to increase the credibility of published results. Social Psychology, 45(3), 137-141. doi:10.1027/1864-9335/a000192. Northouse, P. G. (2010). Leadership: Theory and practice (5th ed.). Thousand Oaks, CA: Sage. Pedhazur, E. J. (1997). Multiple regression in behavioral research: Explanation and prediction (3rd ed.). New York: Harcourt Brace. 151 Pfeffer, J. (1977). The ambiguity of leadership. Academy of Management Review, 2(1), 104-112. http://dx.doi.org/10.5465/AMR.1977.4409175 Preacher, K. J., & MacCallum, R. C. (2003). Repairing Tom Swift’s electric factor analysis machine. Understanding Statistics, 2(1), 13-43 doi: 10.1207/S15328031US0201_02 Reiter-Palmon, R., & Tinio, P. (2014). Pipeline revisions: a call to change. Psychology of Aesthetics, Creativity and the Arts, 8(1), 13-14. doi: 10.1037/a0035845 Rosenthal, S. A. (2012). National leadership index 2012: A national study of confidence in leadership. Center for Public Leadership, Harvard Kennedy School, Harvard University, Cambridge, Massachusetts. Retrieved from http://andresraya.com/wp-content/uploads/2012/12/cpl_nli_2012.pdf Singh, K. (2007). Quantitative social research methods. Thousand Oaks, CA: Sage. Smith, N. C. (1970). Replication studies: A neglected aspect of psychological research. American Psychologist, 25(10), 970-975. doi: 10.1037/h0029774 Stogdill, R. M., (1948). Personal factors associated with leadership: A survey of the literature. The Journal of Psychology, 25(1), 35-71. http://dx.doi.org/10.1080/00223980.1948.9917362 Stogdill, R. M. (1974). Handbook of leadership: A survey of theory and research. New York: The Free Press. Tabachnick, B. G., & Fidell, L. S. (1996). SPSS for windows workbook to accompany large sample examples of using multivariate statistics. HarperCollins College Publishers. 152 Teddie, C., & Tashakkori, A. (2009). Foundations of mixed methods research: Integrating quantitative and qualitative approaches in the social and behavioral sciences. Thousand Oaks, CA: Sage. Torraco. R. (2005). Writing integrative literature reviews: Guidelines and examples. Human Resource Development Review, 4(3), 356-366. doi:10.1177/1534484305278283. Tukey, J. W. (1977). Exploratory data analysis. Reading, MA: Addison-Wesley. Yammarino, F. J. (2000). Leader skills: Introduction and overview. The Leadership Quarterly, 11(1), 5-9. doi:10.1016/S1048-9843(99)00040-5 Yin, R. K. (2003). Case study research (3rd ed.). New York, NY: Sage Zaccaro, S, J. (2007). Trait-based perspectives of leadership. American Psychologist, 62, (1), 6-16. doi: 10.137/0003-066X.62.1.6 Zaccaro, S. J., Kemp, C., & Bader, P. (2004). Leader traits and attributes. In J. Antonakis, A. T. Cianciolo, & R. J. Sternberg.(Eds.), The nature of leadership. (pp. 101-123). Thousand Oaks, CA: Sage. Zaccaro, S. J., Mumford, M. D., Connelly, M. S., Marks, M. A., & Gilbert, J. A. (2000). Assessment of leader problem-solving capabilities. The Leadership Quarterly, 11 (1), 37-64. doi:10.1016/S1048-9843(99)00042-9 153 APPENDIX A MANHATTANVILLE COLLEGE IRB APPROVAL LETTER 154 APPENDIX B SURVEY INSTRUMENT 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 APPENDIX C SURVEY RECRUITMENT CARD 170 PERMISSION LETTERS TO USE COPYRIGHTED MATERIAL PERMISSION TO USE THE STRATAPLEX SURVEY 171 PERMISSION TO USE THE STRATAPLEX FIGURE 1 The following is an excerpt of a License Agreement between Gregory Brown and Elsevier. A full copy of the License Agreement can be obtained by contacting the author of this dissertation. 172 173 Biography Gregory S. Brown was born December 15, 1979, in Poughkeepsie, New York. He attended Arlington High School in LaGrangeville, New York, graduating in 1998. During his undergraduate program at the State University of New York, College at Oneonta, Gregory was inducted into the National Residence Hall Honorary Society before graduating in 2003 with a Bachelor of Science in Secondary Education with a concentration in Social Studies. He received initial and then permanent New York State Teacher Certification in the area of Social Studies grades 7-12. Gregory began teaching in the North Rockland Central School district in 2003 leaving in 2011. In 2006 he earned a Master of Arts in Educational Psychology from Marist College, followed by a Master of Science in Education in Educational Leadership and Administration in 2009 from Iona College. Additionally, he received New York State Teacher Certification in both School Building and School District Leadership. In 2011 Gregory became an assistant principal at Arlington High school where he supervised a cohort of more than 850 students and 250 teachers. During this time he was accepted into Manhattanville’s Doctoral program in Educational Leadership where he earned the 2016 Doctoral Award for Academic Excellence. Gregory took the position of Regional Coordinator for Social Studies and Blended Learning at Putnam Northern Westchester BOCES in 2015. Besides speaking regionally and nationally, on a variety of topics such as leadership and social studies, he plans to continue working in K-12 and Higher Educational settings after graduation.