Richard T. Hale. PAULA FREMONT, PhD, Faculty Mentor and Chair. BRIGIT OLSEN, PhD, Committee Member. THOMAS VAIL, PhD ... Richard Todd Hale, 2016 ...
TOWARDS A BETTER UNDERSTANDING OF EMPLOYEE ENGAGEMENT: FACTORS THAT EXPLAIN EMPLOYEE ENGAGEMENT by Richard T. Hale
PAULA FREMONT, PhD, Faculty Mentor and Chair BRIGIT OLSEN, PhD, Committee Member THOMAS VAIL, PhD, Committee Member
Andrea Miller, PhD, Dean of Psychology Harold Abel School of Social and Behavioral Sciences
A Dissertation Presented in Partial Fulfillment Of the Requirements for the Degree Doctor of Philosophy
Capella University June 2016
ProQuest Number: 10133018
All rights reserved INFORMATION TO ALL USERS The quality of this reproduction is dependent upon the quality of the copy submitted. In the unlikely event that the author did not send a complete manuscript and there are missing pages, these will be noted. Also, if material had to be removed, a note will indicate the deletion.
ProQuest 10133018 Published by ProQuest LLC (2016). Copyright of the Dissertation is held by the Author. All rights reserved. This work is protected against unauthorized copying under Title 17, United States Code Microform Edition © ProQuest LLC. ProQuest LLC. 789 East Eisenhower Parkway P.O. Box 1346 Ann Arbor, MI 48106 - 1346
© Richard Todd Hale, 2016
Abstract Although researchers have discovered many of the beneficial and positive consequences of employee engagement, little is known about the multitude of antecedent factors that lead to employee engagement. Previous research has demonstrated that an individual’s gender is a factor in engagement, and that an employee’s racioethnic similarity with a supervisor, job characteristics, and perceived organizational support, are all antecedents of engagement. The present study focused on individual personality, the perceived quality of employees’ working relationships with their supervisors, and their work roles as either managers or subordinates, to identify whether those variables contribute to employee engagement. This study’s survey, administered to 96 respondents in the work force, using hierarchical multiple regression analysis, found that the personality subdomain of conscientiousness, based on the Big Five model of personality, and the perceived quality of relationship with one’s supervisor, based on Leader-Member Exchange (LMX) theory, were positively related to, and predicted employee engagement.
Dedication This dissertation is dedicated first and foremost to my loving wife Kendallyn who has graciously supported my educational efforts. In addition, she was willing to allow significant sacrifices of our hours together that were used instead to write papers until 3:00 a.m. on many mornings. Were it not for her love, devotion, dedication to my goals, and willingness to sacrifice time used in the achievement of those goals, I would never have made it this far in my educational journey. Mere words, no matter how eloquently arranged, will not convey to her how thankful I am for her support, love, dedication, and devotion. In addition, I dedicate this dissertation to my children, who also made sacrifices. I love each and every one you, Cody, Chris, Carson, and Cassidy.
iii
Acknowledgments Additionally, I want to take this opportunity to thank and acknowledge several significant mentors in my life who have influenced and encouraged me along this doctoral educational journey. First I would want to say thank you to Dr. Fremont, my dissertation adviser, who was also the lead instructor for several of the courses in my doctoral program. Without her support, advice, constructive criticism, and positive reinforcement, I would not be where I am today. Additionally, I would like to thank my other dissertation committee members, Drs. Brigit Olsen and Thomas Vail. I would also like to thank Drs. Sina Haeri, Kim DeStefano, and Emil Fadel who continually inspired me to achieve my educational goals and who supported my efforts in doing so. Finally, I would like to acknowledge the many other professors and support staff at Capella University without whose support my completion of this dissertation would not have been possible.
iv
Table of Contents Acknowledgments
iv
List of Tables
viii
List of Figures
ix
CHAPTER 1. INTRODUCTION
1
Background of the Problem
2
Statement of the Problem
5
Purpose of the Study
8
Significance of the Study
8
Research Design
9
Research Questions and Hypotheses
10
Assumptions and Limitations
11
Definition of Terms
12
Expected Findings
16
Organization of the Remainder of the Study
16
CHAPTER 2. LITERATURE REVIEW
18
Introduction to the Literature Review
18
Theoretical Orientation for the Study
36
Review of Research Literature
40
Synthesis of the Research Findings
41
Critique of the Previous Research
42
Summary
44
v
CHAPTER 3. METHODOLOGY
45
Purpose of the Study
45
Research Design
46
Target Population and Participant Selection
47
Procedures
49
Instruments and Measures
51
Research Question and Hypotheses
56
Data Analysis
57
Ethical Considerations
58
Expected Findings
59
CHAPTER 4. DATA COLLECTION AND ANALYSIS
60
Introduction
60
Population and Sample
60
Statistical Assumptions and Diagnostics
62
Summary of the Results
65
Details of the Analysis and the Results
74
Conclusion
79
CHAPTER 5. DISCUSSION, IMPLICATIONS, RECOMMENDATIONS
80
Introduction
80
Discussion of the Results
80
Discussion of the Conclusions
84
Recommendations
86
Limitations
91
vi
Conclusion
93
REFERENCES
95
vii
List of Tables Table 1. Residuals Test of Normality
65
Table 2. Descriptive Statistics
67
Table 3. Multiple Regression Model
68
Table 4. Correlation Matrix
76
Table 5. Correlation Matrix
77
Table 6. Multiple Regression Coefficients
78
viii
List of Figures Figure 1. Theoretical Model
38
ix
CHAPTER 1. INTRODUCTION Employee engagement is a phenomenon that has garnered significant attention from scientists and researchers in just the last 25 years because of, in large part, the many desired positive consequences of employee engagement that benefit the firms that employ workers. Beginning with Kahn's (1990) qualitative study using a grounded theory method, researchers have learned a tremendous amount of information that is now beginning to be utilized in practical application by human resource (HR) managers and practitioners (B. Shuck & Wollard, 2010). This dissertation addressed the need for continued research into the conditions and antecedents that lead to employee engagement. Early research into employee engagement has resulted in several desirable outcomes, including positive relationships between employee engagement and profitability, job productivity, job tenure, and organizational citizenship behavior (Catteeuw, Flynn, & Vonderhorst, 2007; Cross, Gray, Gerbasi, & Assimakopoulos, 2012; Macey & Schneider, 2008a; Pugh & Dietz, 2008; B. Shuck & Wollard, 2010; M. Shuck, Rocco, & Albornoz, 2011). This dissertation study investigated the perceived quality of working relationships between subordinates and supervisors from the viewpoint of the subordinate, based on the vertical dyad linkage originally researched by Dansereau, Graen, and Haga (1975), which eventually culminated in the leadership theory of Leader Member Exchange (LMX), (Dansereau, 1972; Dansereau et al., 1975; Graen & Uhl-Bien, 1995; Liden & Graen, 1980). LMX, which is based on mathematical transitive law 1
theory, can be used to measure whether or not the perceived quality of an employee’s working relationship with his or her supervisor is significant in predicting that employee’s level of engagement. Because increases in the quality of working relationships between dyadic pairs of supervisors and subordinates have been demonstrated to be positively related to higher job satisfaction (Dansereau et al., 1975), and because similarly, job satisfaction has also been shown to be positively related to employee engagement (Harter, Schmidt, & Hayes, 2002), it is probable that the perceived quality of the dyadic working relationship between supervisors and employees is related to the subordinate's level of employee engagement.
Background of the Problem One of the earliest examples of the application of scientific methods and principles to the problem of inefficiency, and that inefficiency's relationship to decreased profitability, would be the work done by 19th century industrialist Frederick Taylor, who studied worker productivity (Taylor, 1911). Taylor desired to apply a scientific approach to address industry-related worker issues that can impact a firm's profitability. Those issues included productivity, high turnover, and worker-related conflict, all of which are factors that are also related to employee engagement (Macey & Schneider, 2008b; Saks & Gruman, 2011; B. Shuck & Wollard, 2010; M. Shuck et al., 2011; Taylor, 1911; Tuckey, Bakker, & Dollard, 2012). Taylor's work was rather detailed, and through his time and motion studies, he was able to calculate, for example, the ideal amount of dirt by weight and volume that should be scooped by one worker at one time in order to maximize not only the quantity of dirt that can be loaded onto a single shovel, but also to 2
maximize the worker's total amount of dirt shoveled over a period of time (Taylor, 1911). Taylor found that maximizing each shovelful with the most dirt that could physically be scooped resulted in tiring the worker out over time until eventually the number of shovels per hour decreased, which resulted in a total decrease in dirt capacity shoveled during that period. Taylor's scientific management philosophies included systematic analysis of each individual industrial operation, the design of compensation plans designed to incentivize the work desired, the creation of formal work training activities, and the provision of clear and concise instruction to employees (Taylor, 1911). Among the many constructs that Taylor investigated, one was known as soldiering, whereby Taylor observed workers intentionally operating below their best potential capacity based on a philosophy that if all workers banded together and intentionally worked sub-optimally, management could be convinced that the sub-optimal level was the most productive feasible workplace in which the workers could engage (Taylor, 1911). The concept of soldiering could be related to employee engagement because both constructs involve a worker’s physical and psychological engagement at work. Employee engagement has been demonstrated to be positively related to productivity. The more highly engaged a worker is, the more productive the worker tends to be (Harter et al., 2002). As the Industrial Revolution in America continued in the early 1900s, various practitioners and authors continued to apply science in an effort to address the problem of unmotivated workers, and the ever-important need to increase worker productivity and a firm's profitability (Landy & Conte, 2010; Viteles, 1953). Firms, such as IBM, Xerox, and Western Electric, in which the researchers discovered the Hawthorne Effect (Chiesa 3
& Hobbs, 2008; O’ Sullivan, Orbell, Rakow, & Parker, 2004; Parsons, 1982), commissioned various scientific studies designed to gain knowledge and insight into various methods by which to increase a firm's profitability by maximizing worker output (Landy & Conte, 2010). Firms quickly realized, during the early industrial time period when the vast majority of work output was attributable to humans and constrained by the technological limitations of the time, that the greatest factor in increasing a firm's profitability was the increase in worker output. This increase required great investments in time, money, and resources by firms that wished to better understand how to motivate workers (Taylor, 1911). With the passage of time, and with technological breakthroughs that included disruptive technologies that forever changed the methods, procedures, and equipment by which firms produced their respective goods and services, the stakeholders and corporate executives of firms began to shift the focus away from the industrial side of the equation. Instead, industrialists’ efforts to increase productivity and profitability shifted to the human relationships side of the equation in order to address what researchers and scholars classify as organizational behavior (Landy & Conte, 2010). Employee engagement, a relatively new phenomenon that only appeared in the literature after the term was coined by Kahn (1990) in his original qualitative study on the construct, was the focus of the present study because of its known consequences and relevance to the problem of increasing profitability (Harter et al., 2002). However, researchers still know very little about the antecedents that lead to engagement (Rana, Ardichvili, & Tkachenko, 2014; Rich, Lepine, & Crawford, 2010; Saks, 2006; Wollard & Shuck, 2011). As Church (2011) suggested, leveraging the 4
practical benefits of engagement is currently impossible due to our incomplete knowledge of its underlying constructs.
Statement of the Problem Researchers know very little about the antecedents of employee engagement (Saks, 2006). Yet the existing research demonstrates that employee engagement has many desired consequential outcomes that are important to businesses and the stakeholders that invest in them (Harter et al., 2002). Thus, for practitioners including CEOs and managers of firms to be able to practically achieve the desired outcome of employee engagement, they must first understand the factors that lead to engagement in the workplace. More specifically, Kim, Shin, and Swanger (2009) pointed to a gap in the literature with regard to relationships between personality and employee engagement. Kim et al. (2009) argued that researching the possible relationships between personality and employee engagement may lead to a better understanding of the underlying employee engagement construct. In addition, Andreassen, Ursin, Eriksen, and Pallesen (2012) found relationships between professional position, which can be described as managerial or subordinate, and outcome variables including work enjoyment, dedication, and narcissism. However, that particular study was conducted in Norway and may not be applicable to workers in the U.S. due to cultural work norm differences. In addition, the authors did not fully define the terms manager and subordinate. Indeed, Andreassen et al. (2012) did not define what management duties and responsibilities consist of, as opposed to subordinate duties and responsibilities, thus detracting from the duplicability of the study. Additionally, little else can be found in the 5
literature that investigates whether an employee’s work role is indeed a factor in his or her level of employee engagement as Andreassen et al. (2012) suggest. This gap in the literature pertaining to work role and its relationship with engagement was demonstrated by Rothbard (2001), who argued that future research on employee engagement should concentrate on exploring engagement from various roles within the organization. In addition, although Kim et al. (2009), as well as Langelaan, Bakker, Van Doornen, and Schaufeli (2006), have identified various relationships between subdomains of personality based on the Big Five model and employee engagement, significant gaps in the literature continue to exist as to the relationships between personality and employee engagement and whether personality is a direct antecedent of engagement in the workplace, or whether it mediates or moderates the relationship between other variables and employee engagement. In addition, Langelaan et al. (2006) pointed to considerable contradiction in the existing literature with regard to the relationships between sub-domains of the Big Five personality theory and employee engagement. Langelaan et al. (2006) pointed out that relationships have been found between the Big Five sub-domain of Conscientiousness and work-related burnout, and also between the latter and Neuroticism. Because burnout is the polar opposite of employee engagement, the authors expected to find a reversed pattern for Extraversion as it relates to employee engagement, but did not. Furthermore, nothing could be found in the literature that sought to consider the two constructs of personality and job role simultaneously, whether in an effort to consider the potential relationship between them, or to consider whether they might be antecedents of employee engagement. In light of the points alluded to above, the need to study the antecedents of employee engagement 6
becomes quite apparent, and is corroborated by the findings of B. Shuck and Reio (2014), who discussed the practical relationships between employee engagement, productivity, employee wellbeing, and profitability. However, an abundance of research does point to the positive consequences of engagement, which include increases in productivity, profitability, customer satisfaction, safety, and decreases in turnover, as well as decreases in absenteeism. Nevertheless, it is conservatively estimated that less than 30% of the workforce is engaged in the workplace (Harter et al., 2002; B. Shuck & Reio, 2014; B. Shuck & Wollard, 2010). Thus, better understanding the antecedents of employee engagement in an effort to utilize its desired positive consequences is of the utmost importance for practitioners, and this importance is beginning to be appreciated by scholars as well (Fairlie, 2011; Harter et al., 2002; Masson, Royal, Agnew, & Fine, 2008; Welch, 2011; Xu & Thomas, 2011). In addition, Rivera and Flinck (2011) found that unengaged workers cost the firms that employ them 35% of the firms’ total payroll. The same authors also pointed out the fact that the negative effects of unengaged workers cost U.S. corporations 343 billion dollars annually, which includes 65 billion tax-payer dollars due to lost productivity of unengaged federal employees. The staggering costs to both consumer taxpayers and corporations illuminate the problem and the need to study employee engagement in an effort to intervene when employees are disengaged in the workplace (Rivera & Flinck, 2011; Saks, 2006).
7
Purpose of the Study The purpose of this study was to gain a better understanding of the antecedent factors that lead to employee engagement in the workplace. Currently, little is known about the antecedents of employee engagement (Saks, 2006). By gaining a better understanding of the antecedents of engagement, those managers in senior leadership positions within companies will be better positioned to make data-driven decisions using the many known positive consequences of engagement to increase productivity and profitability within the firms that they manage. This study sought to understand the antecedents of employee engagement, such as personality's role in engagement, about which little can be found in the existing literature (Kim et al., 2009; Langelaan et al., 2006; Saks, 2006; Wollard & Shuck, 2011). In addition to gaining additional insights into personality's role in engagement, this study also investigated other possible antecedents of employee engagement including (a) gender similarity in the supervisor-subordinate dyad, (b) racioethnic similarity in the supervisor-subordinate dyad, (c) tenure, (d) work role in terms of whether an employee manages other people or not, (e) employment status as either full-time or part-time, and (f) the quality of the working relationship an employee has with his or her supervisor as understood on the basis of LMX leadership theory.
Significance of the Study This study is relevant to the industrial/organizational (I/O) psychology specialization because one of the fundamental notions of I/O psychology involves the 8
study of LMX leadership theory and employee engagement (Bakar, Mustaffa, & Mohamad, 2009; Dansereau, 1972; Dansereau et al., 1975; Graen & Schiemann, 2013; Kelloway & Day, 2005; Liden & Graen, 1980; Rothmann & Welsh, 2013; B. Shuck & Wollard, 2010). Today, due to the many desired positive consequences of employee engagement, the field of I/O psychology is clearly more focused on the organizational side of the I/O equation and frequently involves studying employee engagement (Harter et al., 2002; Saks, 2006; Church, 2011). Positive consequences of employee engagement include organizational commitment, job satisfaction, job involvement (Macey & Schneider, 2008b), organizational citizenship behavior, and disinclination to quit (Saks, 2006). These positive consequences of engagement are outcomes that organizations desire, and this fact makes the study of employee engagement extremely relevant for I/O psychology because the field involves the application of psychology in the workplace and attempts to bridge the scholar-practitioner divide. In addition, firms including the Gallup organization, the Corporate Leadership Council, the American Society for Training and Development, and the Society for Human Resource Management, are all investing resources into the study of employee engagement, and their investment gives credence to the relevance of employee engagement for the field of I/O psychology (B. Shuck & Wollard, 2010).
Research Design This study used a quantitative ex post facto design (Leedy & Ormrod, 2013). This type of design is appropriate for use when studying independent variables that are already 9
in existence and are difficult to manipulate in a laboratory setting. The independent variables in this study included the participants’ personality, race, gender, and perceived relationship with supervisors, based on LMX theory (Graen & Uhl-Bien, 1995). Personality, gender, and race are variables that cannot be easily or feasibly manipulated. Thus the lack of manipulability of human personality, gender, race, and the relationships of participants to their supervisors, would lead to significant feasibility issues if the study protocol was oriented toward an experimental design. Thus, the ex post facto quantitative design was decided upon for the study. In conjunction with the ex post facto design, and commensurate with the quantitative approach, the multiple regression statistical analysis method was used to analyze the data (Field, 2009; Warner, 2013).
Research Questions and Hypotheses This study's research question is: Do work role, individual personality, LMX working relationship, gender similarity within the supervisor-subordinate dyad, racioethnic similarity within the supervisor-subordinate dyad, full-time or part-time employment status, and/or tenure predict employee engagement? Null Hypothesis H0: The independent variables including work role, individual personality, LMX working relationship, gender similarity of the supervisor and the subordinate, racioethnic similarity of the supervisor and the subordinate, employment status as a full-time worker or part-time worker, and tenure are not statistically significant in predicting employee engagement.
10
Alternate Hypothesis H1: The independent variables of work role, individual personality, LMX working relationship, gender within the supervisor-subordinate dyad, racioethnic similarity within the supervisor-subordinate dyad, employment status as a full-time worker or part-time worker, and tenure are statistically significant in predicting employee engagement.
Assumptions and Limitations This study was conducted in a field setting as opposed to a laboratory setting. The independent variables of interest are non-manipulable, which eliminated the researcher’s ability to conduct the study using an experimental design. Thus, because of the study's ex post facto design, and because of the fact that the study was conducted in a real-world field scenario that captured data from participants working in their respective jobs, several limitations and assumptions were present in the study. Limitations Limitations include the fact that results should not be generalized to other populations of workers based on the limits of the sampling procedures, according to which samples were taken from U.S. workers between the ages of 18 to 55. In addition, the study was limited by the data collection and sampling methods, which involved purposive sampling. Because neither random selection nor random assignment methods were used in the study, one cannot eliminate the possibility of a systematic validity threat in that the participants recruited into the study may introduce into it unknown confounding variables that are both unmeasured in the study and also correlated with the outcome variable of interest, employee engagement (Cook & Campbell, 1979; Shadish, 11
Cook, & Campbell, 2002). Because it is likely that this may occur, one must consider the impact on the validity of the interpretation of the results. Assumptions As the theoretical construct of employee engagement is the foundational quantity upon which this study was built, it was assumed that the construct of employee engagement exists as a theoretical entity that is different from those posited by similar but related theories, including person-environment (P-E) fit theory, and its sub-related theories of person-job (P-J) fit and person-organization (P-O) fit theories (Landy & Conte, 2010, p. 469), role sending, role receiving, and role taking theories (Kahn, 1990), organizational citizenship behavior theory (Allen, 2006; Rioux & Penner, 2001), state engagement theory (Macey & Schneider, 2008b), and many others. This assumption is particularly important to address in light of concerns from some researchers and authors that the underlying construct of employee engagement may merely be scientific remarketing and re-purposing of constructs already in existence, as is evident with one group of authors who have questioned the construct of engagement by asking, “Old wine in new bottles?” (Macey & Schneider, 2008b, p. 6).
Definition of Terms Big-Five Personality or Big-Five The Big Five is a human personality theory based on convergent validity from several researchers that posits that the construct of human personality can be explained in five sub-domains based on a person’s language lexicon. These five subdomains are thought to be sufficient constitute a person’s overall personality profile. Although various 12
authors have used various terms to name the five sub-domains of personality, the various terms all describe the same underlying construct. The five sub-domains of the Big Five personality model are: (1) Surgency or Extraversion, (2) Agreeableness, (3) Conscientiousness or Dependability, (4) Emotional Stability vs. Neuroticism, and (5) Culture, Intellect, or Openness (Digman, 1990, 1997, 2002; Fiske, 1949; Goldberg, 1981, 1990, 1992). The five subdomains (usually capitalized hereafter) are defined as follows: Openness. “…describes the breadth, depth, originality, and complexity of an individual’s mental and experiential life” (Pervin & John, 1999, p. 121). Conscientiousness. Conscientiousness is a Big Five personality sub-domain embodied by the degree to which an individual is responsible vs. undependable, persevering vs. fickle, fussy vs. careless, and scrupulous vs. unscrupulous (Digman & Inouye, 1986, p. 117). Extraversion. Within Big Five theory, this term refers to “...the disposition towards cheerfulness, sociability, and high activity” Costa and McCrae in Langelaan et al. (2006, p. 523). Agreeableness. This term refers to a Big Five personality sub-domain embodied by the degree to which an individual is “conciliatory, cooperative, and agreeable” (Goldberg, 1990, p. 1218). Neuroticism. Within the Big Five system, neuroticism is “...the general tendency to experience distressing emotions such as fear, depression, and frustration...” (Langelaan et al., 2006, p. 523).
13
Employee Engagement Employee engagement is the “harnessing of organization members’ selves to their work roles; in engagement, people employ and express themselves physically, cognitively, and emotionally during role performances” (Kahn, 1990, p. 694). Full-time Working a minimum of 40 hours per week in any given seven-day work week. Gender Similarity (Dichotomous) — whether a subordinate's gender is the same, or different, as compared to his or her supervisor’s. Individual Antecedents “...constructs, strategies, and conditions that were applied directly to or by individual employees and that were believed to be foundational to the development of employee engagement” (Wollard & Shuck, 2011, p. 433). Job Burnout Job burnout is a “prolonged response to chronic emotional and interpersonal stressors on the job, and is defined by the three dimensions of exhaustion, cynicism, and inefficacy” (Maslach, Schaufeli, & Leiter, 2001, p. 397). Leader-member-exchange (LMX) LMX is a leadership theory first developed by Dr. Alfred (Fred) Dansereau that posits that leaders and supervisors develop varying qualities of relationships with their subordinates resulting in some subordinates classified as the “in-group” that receive more
14
attention, more resources, and more leadership treatment than those subordinates that are placed in the “out-group” (Dansereau, 1972, p. 54). Organizational Antecedents “...constructs, strategies, and conditions that were applied across an organization as foundational to the development of employee engagement and the structural or systematic level” (Wollard & Shuck, 2011, p. 433). Part-time Working less than 40 hours per week in any given seven-day work week. Racioethnic Similarity (Dichotomous) — whether a subordinate's race is the same, or different, as compared to his or her supervisor. Subordinate A subordinate is an employee whose work role does not involve the supervision of other employees, and who both reports to a supervisor and creates output that is the responsibility of that supervisor. Supervisor A supervisor is an: employee who is given authority and responsibility for planning and controlling the work of a group through close contact. A supervisor is the first level of management in an organization. The subordinates he or she controls are usually at a nonmanagerial level, and the supervisor is wholly responsible for their work (Supervisor, 2011, para 1).
15
Tenure “The time when a position is held” (Tenure, 2006, para. 1). Work Role The term work role in this study is a term that queries whether an employee in his or her work role is responsible for the management of other workers.
Expected Findings Based on previous studies that have correlated the race and gender similarities of subordinates and their supervisors with the subordinates’ measures of employee engagement (Avery, McKay, & Wilson, 2007; Avery, Volpone, Mckay, King, & Wilson, 2012), and studies that have found relationships between the sub-domains of extraversion and neuroticism within the Big Five model of personality with employee engagement (Bakker, Tims, & Derks, 2012; Handa & Gulati, 2014; Inceoglu & Warr, 2011; Kim et al., 2009; Langelaan et al., 2006), it was expected that the results of the present study would corroborate the findings from prior research. The intriguing question was whether work role (supervisor or subordinate) and LMX relationship quality would be found to be significant predictors of employee engagement based on the results of the multiple regression analysis.
Organization of the Remainder of the Study The remainder of this study is organized as follows: Chapter 2 covers a literature review of the three primary theories on which this dissertation research is based: LMX (Dansereau, 1972; Dansereau et al., 1975; Liden & Graen, 1980), personality (Digman, 16
1997, 2002; Goldberg, 1990, 1992; Goldberg et al., 2006), and employee engagement (Bakker, Schaufeli, Leiter, & Taris, 2008; Kahn, 1990; Schaufeli, Salanova, Gonzálezromá, & Bakker, 2002). Chapter 2 also reviews the theoretical orientation and connections between the three theories of employee engagement, personality, and LMX. Chapter 3 discusses the research methodology, protocol development, and statistical analysis used in this study. The chapter begins with an explanation of the fundamental purpose of the study, an effort to better understand the antecedents that lead to employee engagement. Chapter 3 discusses the research protocol, using a quantitative, ex post facto design (Leedy & Ormrod, 2013). Furthermore, Chapter 3 addresses the target population for the study, participant selection, and sampling methods, including the use of purposive sampling (Creswell, 2007; Leedy & Ormrod, 2013; Patton, 2002). Finally, Chapter 3 concludes with a discussion of ethical considerations taken into account in the study and a discussion of expected findings. Chapter 4 includes a detailed description of the sampling plan, as well as inclusion and exclusion criteria. It also explains the use of a third party firm, Qualtrics (Qualtrics, 2014) which provided not only the population of workers from which samples were taken, but also the administration of the survey instruments over the Internet. Chapter 4 includes a report of the entire results of the study. Finally, the chapter ends with a conclusion centered on the outcome of the analysis. The dissertation closes with Chapter 5, which includes further, and more detailed, discussion of the results of the study, discussions regarding the conclusions drawn from the analysis and results, and recommendations for future research.
17
CHAPTER 2. LITERATURE REVIEW Introduction to the Literature Review The literature review for this study began with a search of the Capella online library, and consisted of multiple searchers using databases including Academic Search Premier, Business Source Complete, Credo Reference Library, Dissertations @ Capella, eBooks on EBSCOhost, ERIC, Google Scholar, Health and Psychosocial Instruments, LexisNexis Academic, Mental Measures Yearbook, ProQuest Psychological Journals, PsychARTICLES, PsychBOOKS, PsychINFO, PsychTESTS, Sage Journals Online, Sage Research Methods, ScienceDirect, Summon, and Ulrichsweb. Various search terms were used including; Workplace Engagement, Employee Engagement, and Engagement in the Workplace. Further refinement of the research terms was accomplished by searching for (antecedents AND Employee Engagement). Because the additional theories of LMX and personality are hypothesized in this research to be associated with employee engagement, other research terms were used to search the various databases including (Leader-Member Exchange AND Employee Engagement) and (Personality AND Employee Engagement). Various combinations of words taken from the three theories' nomenclatures were used in an effort to retrieve relevant sources. All searchers were limited to peer reviewed sources and the Ulrichsweb tool was used to verify that sources are peer reviewed. Employee Engagement Theory Employee engagement is a relatively new phenomenon being studied by practitioners and (I/O) psychologists (Church, 2011). The construct of employee 18
engagement began with the seminal qualitative study completed by Kahn (1990). Prior to that, little, if anything, can be found in the literature addressing the topic of employee engagement. In fact, while conducting this literature review for the dissertation, a search of the Capella online library using the search term Employee Engagement and limiting the results of the search to the period 1900 to 1989 resulted in zero search results. Conversely, after conducting the same search but changing the time frame to 1990 to 2015, the Capella online library returned 7,331 results. Of those 7,331 results, 4,095 were results from peer reviewed sources. This demonstrates the relative newness of the construct under study. Qualitative grounded theory method. In Kahn's (1990) original study on employee engagement, he utilized a qualitative, grounded theory method (Creswell, 2007; Glaser & Strauss, 1967; Khan, 2014; Patton, 2002) to study employee engagement, and more specifically, to develop a theory that explains the underlying construct. Kahn (1990) stated that his explicit goal within the study was to generate a theory demonstrating how the psychological experiences employees have at work shape employees’ avoidance of work or their attraction to it. Kahn (1990) further discussed his starting point for his research, which was based on previous work, including the work of Goffman (1961), that concerned the construct of organizational behavior. Goffman suggested that people become attached to their work and then detach from their work, and this attachment-detachment process is varied and cyclical. Kahn also considered other organizational theories that are in some way related to how an employee is engaged in his or her work including role sending and receiving (Kahn, 1990), person-role relationship theory including; person-environment fit, person-job fit, and person19
organization fit (Landy & Conte, 2010). Kahn (1990) argued that workers use varying degrees of effort in three different forms—physical effort, emotional effort, and cognitive effort—when they work. The fundamental construct that Kahn desired to study was the notion that mechanisms are in place that allow for employees to somehow choose varying amounts of discretionary physical, cognitive, and emotional effort in their job roles. Kahn (1990) continued to explain his primary goal for the study, which was to create a theoretical framework for understanding an employee’s “self-in-role” process at work (p. 692). Kahn (1990) further argued that people apply varying amounts of cognitive, physical and emotional effort while doing their jobs. Kahn’s work settings used to study employee engagement. Kahn (1990) selected two unique and divergent settings with which to apply the grounded theory qualitative method of inquiry and to collect data to generate his theory. The first location was a summer camp for boys and girls in the West Indies that he called “Camp Carrib” (p. 696). Kahn also noted that he intentionally changed the real names of the locations so as to protect the privacy of the firms, employees, and customers involved. Camp Carrib was a unique setting in which to study employee engagement because the firm operated a recreational day camp catering to 12- to 17-year-olds from mainly affluent backgrounds in America and Western Europe (Kahn, 1990). The participants included in the study from Camp Carrib included 16 camp counselors, of which 9 were men and 7 women. The average age range of those included in the study from Camp Carrib was 20 to 35 years old with an average age of 25.5 (Kahn, 1990, p. 696). Kahn's (1990) other work-related setting used to study employee engagement was a private sector for-profit architecture firm operating in the Northeastern United States. 20
Kahn (1990) pointed out that the architectural firm was well established and highly regarded in the industry when the author wrote, speaking to the demand for the firm's services, that it had “more projects than it could comfortably handle” (p. 697). Participants in the study from the architectural firm included 16 employees, of which 10 were men and six were women with an average participant age of 34.3 years (Kahn, 1990, p. 697). Kahn pointed out that those participants included in the study represented a wide range of job classifications and levels from within the organization. Kahn's interpretation of the results. Kahn's (1990) results suggested that workers vary in their level of engagement in their work, sometimes being drawn to their work with a process Kahn refers to as “personal engagement,” and at other times becoming disengaged with their work, a process that Kahn refers to as “personal disengagement.” The author compared personal engagement and disengagement to magnets that attract one another or repel one another depending on their orientation (pp. 700-701). Kahn also uses the term psychological meaningfulness, which he defines as “feeling that one is receiving a return on investments of one's self in a currency of physical, cognitive, or emotional energy” (Kahn, 1990, pp. 703-704). The author continues by explaining that workers who feel valuable and valued by their workplace, and who feel useful, have the worthwhile experience of psychological meaningfulness, and this psychological fulfillment is associated with engagement in the workplace. Conversely, employees who do not feel valued by the organization experience less psychological meaningfulness at work, and this lack of psychological meaningfulness is associated with becoming disengaged at work. This ebb and flow of increasing and decreasing levels of psychological meaningfulness, and the associated increased degrees 21
of employee engagement, together form the theoretical construct of engagement as a fluid theory in which employees can move between varying levels of engagement (Kahn, 1990). Physical, emotional, and cognitive engagement. Kahn (1990) interpreted the results of his qualitatively grounded study to mean that workers do not demonstrate engagement based on one underlying construct, but rather, the author argued, workers apply themselves in their respective work roles with varying degrees of physical, emotional, and cognitive effort. The author discussed the interplay between the three subdomains of engagement when he defined employee engagement as follows: “I defined personal engagement as the harnessing of organization members' selves to their work roles; in engagement, people employ and express themselves physically, cognitively, and emotionally during role performances” (Kahn, 1990, p. 694). Examples of physical engagement in the workplace can be demonstrated by drawing attention to an individual who chooses, for example, to train to be the fastest pizza box folder in a pizza company, as demonstrated in a rather entertaining Domino’s pizza television commercial. At the same time, another worker could demonstrate significant emotional engagement by choosing to volunteer, as opposed to being paid for their work, to accompany a team associated with the non-profit group Doctors Without Borders, and experience psychological meaningfulness while doing so. Kahn points out, however, that it is common for individuals to experience cognitive, emotional, and physical, engagement simultaneously. Continued employee engagement research. In the last decade, I/O psychologists and HR practitioners have increased their focus on the concept of employee 22
engagement (Macey & Schneider, 2008b). Employee engagement definitions vary but have many commonalities. All agree that employee engagement is a desirable condition; that it connotes involvement, focused effort, passion for an organization, and enthusiasm; and that it has behavioral, cognitive, and emotional components (Saks, 2006, p. 602). Discriminant validity for employee engagement can be demonstrated in the fact that the three dimensions of engagement—energy, involvement, and efficacy—are the direct opposites of the sub-dimensions of burnout, which are exhaustion, cynicism, and inefficacy (Saks, 2006, p. 601). Positive consequences of employee engagement. Although little scientific research has investigated the antecedents of engagement (Saks, 2006), many researchers have studied the consequences of engagement (Harter et al., 2002; Kelloway & Day, 2005; Macey & Schneider, 2008b; B. Shuck & Wollard, 2010; M. Shuck et al., 2011). Several desirable consequential outcomes of employee engagement that are highly coveted by practitioners, managers, and senior executives of for-profit firms have been discovered thus far in the literature on employee engagement. In one meta-analysis that investigated business-unit relationships between employee engagement and business outcomes, the researchers found that employee engagement was positively correlated with customer satisfaction and loyalty, profitability, productivity, and employee safety (Harter et al., 2002). In addition, the same study found a negative correlation between employee engagement and employee turnover (Harter et al., 2002). Demonstrating convergent validity, several other studies have similarly found a negative correlation between employee engagement and turnover (Maslach & Leiter, 2008; Saks, 2006; B. Shuck & Reio, 2014; M. Shuck et al., 2011; Wollard & Shuck, 2011). And because 23
turnover decreases profitability through loss of the intellectual talent pool and through the increased costs associated with recruiting to fill the vacancies, decreasing turnover is a goal commonly shared among for-profit entities. Triple bottom line business management theory, the term first coined by Elkington (1997) posits that in an effort to stay competitive and sustainable, businesses have been forced to move beyond thinking of profit as the single bottom line benchmark of success. Rather, businesses have turned towards a triple bottom line that includes economic profit added to the economy by the firm, value added to society, and environmental value created by the firm (Glavas & Mish, 2015; Halpern et al., 2013; Tyrrell, Paris, & Biaett, 2013; Valentin, 2015). A firm’s social contributions may involve, for example, distributive, interactive, and procedural justice in the workplace (Dayan & Benedetto, 2008; Kilchrist & Block, 2006; Witztum, 2008). And because organizational justice has been found to be related to, and an antecedent of, employee engagement, a firm’s management may begin to apply employee engagement theory in order to increase productivity, profitability, and tenure (Harter et al., 2002). Simultaneously, a firm can increase its value as measured by its contributions to society, which may include workers who are both engaged and treated justly by the organization, one of the fundamental tenets of measurement based on triple bottom line theory (Elkington, 1997; Halpern et al., 2013; Valentin, 2015). Thus, from the perspective of a firm’s stakeholders, managers, and leaders, bridging the scholar-practitioner gap (Church, 2011) will allow the stakeholders to realize not only the financial benefits of employee engagement, including profitability (Harter et al., 2002), but also organizational justice factors (Saks, 2006), which can potentially increase the firm’s social contributions. These two benefits 24
constitute two of the three fundamental tenets that make up the triple bottom line theory of strategic management (Elkington, 1997; Halpern et al., 2013; Valentin, 2015). Antecedents of engagement. Little is known about the antecedents of engagement (Saks, 2006; Wollard & Shuck, 2011). Complicating matters, Saks (2006) noted that the majority of research that has been conducted on the construct of employee engagement has been conducted by business practitioners commonly published in business journals and found in the reports of private consulting firms rather than scientific journals. In addition to the gap in the literature on the antecedents of employee engagement, little can be found on the theory that explains the construct aside from Kahn's (1990) seminal study. Antecedent factors that lead to employee engagement can be classified in various ways, and may include both individual and organizational antecedents that lead to engagement (Wollard & Shuck, 2011). Individual antecedents of employee engagement. Certain aspects of individual personality have been studied in terms of their relationship to a worker’s degree of employee engagement. Individual personality would be considered an individual antecedent factor of employee engagement as opposed to an organizational antecedent. More specifically, one study found that workers with a proactive personality, an individual antecedent, were more likely to individually craft their jobs, create additional social job resources, and increase their job challenges, all of which resulted in increased levels of worker engagement (Bakker et al., 2012). However, little research can be found in the literature examining factors, including personality, that are based on the Big Five model, and that examine the perceived quality of an employee’s working relationship with his or her supervisor as it relates to that employee’s level of employee engagement. 25
What little research there is that has investigated the linkages between individual personality and employee engagement, has tended to establish a chain of related variables, and researchers have demonstrated that specific sub-dimensions of the Big Five personality model, such as neuroticism, are significantly, positively correlated with job burnout, the antithesis of employee engagement (Kim et al., 2009; Maslach et al., 2001; Wollard & Shuck, 2011). Organizational antecedents of engagement. Because organizations are complex systems, the organizational antecedents of employee engagement involve the employee and his or her basic human and employee needs. Wollard and Shuck (2011) were able to identify 13 of 21 organizational antecedents of employee engagement that were empirically tested. These organizational antecedent factors include: authenticity of the corporate culture, clear expectations, corporate social responsibility, job characteristics, job fit, level of task challenge, manager expectations, manager self-efficacy, employee perception of workplace safety, rewards, supportive organizational culture, and use of strengths. Big Five Theoretical Model of Personality Humans have various personality traits that have been well documented in the literature. A large portion of personality research has centered on the Big Five model of personality, which is a hierarchical organization of fundamental personality traits that includes Openness to Experience, Conscientiousness, Extraversion, Agreeableness, and Neuroticism (Digman, 1990, 1997, 2002; Digman & Inouye, 1986). From an historical standpoint, Goldberg (1990) points to Galton (1884) as the first scientist to consider the underlying constructs that make up human personality. Galton used a lexicon-based 26
approach to studying personality. In the 1930s, other researchers, such as Thurstone (1934) continued investigation of personality using factor analysis. By the mid-1980s and into the 1990s, Digman and Takemoto-Chock (1981), Digman (1990, 1997, 2002), and Goldberg (1981, 1990, 1992) demonstrated convergent validity in the Big Five model when they similarly found evidence through factor analysis that a significant number of factors that explain human personality could be modeled in only five fundamental domains. In 1981, Goldberg first coined the term “Big Five” to describe the five-factor model of personality, also known as the Five Factor Model (FFM) (Goldberg, 1981, p. 159). In fact, Butcher (2009) discusses the universal acceptance of the FFM model of human personality. Furthermore, Goldberg (1981) has posited that “it should be possible to argue the case that any model for structuring individual differences will have to encompass—at some level—something like these ‘Big Five’ dimensions” (p. 159). Openness to experience, culture, intellect. The psychological construct and definition of openness to experience, commonly referred to as simply Openness in the Big Five nomenclature, was first posited by Rogers (1961). Drawing upon the philosophy of existentialism, Rogers (1961) describes the process of openness as a process whereby an individual becomes more open to his or her personal experiences in life. The author continues to define Openness by stating that the construct is the opposite of defensiveness. Rogers (1961) continues to describe the relationship between perceived personal safety in relationship with openness by explaining that as individuals feel increased levels of safety in relationships, they become increasingly open to new experiences.
27
This same sub-domain construct is also referred to as Culture, and Intellect, by other researchers of the Big Five model (Digman & Takemoto-Chock, 1981; Goldberg, 1990). Because the theoretical construct of the Big Five model of personality was originally researched based on understanding the lexicon of the English language and the descriptive characteristics of other people while “developing an adequate descriptive or observation language for personality as an area of psychological research and theory construction” (Norman, 1963, p. 574). Continuing with the same logic, it would be pertinent to communicate the other words in language that are synonymous with the construct of openness. Openness is associated with being untraditional, liberal, and having broad interest (Digman & Inouye, 1986, p. 120). Like the other sub-domains of personality based on the Big Five model, Openness is not a mutually exclusive construct where an individual is at one polar opposite or the other in terms of their openness, but rather, Openness is a continuum whereon a person may range from between being closedminded to being open to new experiences. Conscientiousness or dependability. The sub-domain of Conscientiousness involves the “conscience-governed concern for others” (Digman & Inouye, 1986, p. 119). The same author pointed out that this factor was interpreted as Agreeableness in Norman’s (1963) work on personality. The Conscientiousness sub-domain of personality measures the degree to which one engages in friendly compliance versus hostile noncompliance (Digman & Takemoto-Chock, 1981). Individuals that score high on the Conscientiousness scale are considerate of others’ beliefs, value others’ time (meaning that individuals that score high on the Conscientiousness scale are punctual to appointments), and are highly capable of empathy for others. Other traits commonly 28
found in an individual who scores high on the Conscientiousness scale include careful planning, carefulness, and persistence (Digman & Inouye, 1986). In addition to the term Conscientiousness, some authors have used the term dependability when referring to this sub-domain of the Big Five model of personality (Goldberg, 1990). Surgency or Extraversion. In the Big Five model of personality, Extraversion, also referred to in other research on the Big Five as Surgency (Goldberg, 1990), is the term used to describe one’s interpolar position between talkative vs. silent, open vs. secretive, adventurous vs. cautious, and sociable vs. reclusive (Norman, 1963). Other characteristics associated with Extraversion include dominance, gregariousness, outspokenness, and energy (Digman & Inouye, 1986). Extraversion then is a sub-domain of the Big Five personality that measures one’s traits with regard to willingness or aversion to participating in social activities and situations. Agreeableness. The term Agreeableness is a sub-domain of the Big Five model of personality that refers to a person’s personality on the bipolar dimensions of good natured vs. irritable, cooperative vs. negativistic, gentle vs. headstrong, cold vs. warm, kind vs. unkind, cooperative vs. uncooperative, selfish vs. unselfish, rude vs. polite, trustful vs. distrustful, stingy vs. generous, flexible vs. inflexible, fair vs. unfair, and not jealous vs. jealous (Goldberg, 1981, 1992; Norman, 1963). Interestingly, as one author points out, after two decades of research on personality-trait adjectives to create a foundational construct for personality, Factor II, Agreeableness, has more adjectives associated with the sub-domain than any of the other factors (Goldberg, 1992). Another author further described the sub-domain of Agreeableness when he wrote that it is “…a dimension that appears to involve the more human aspects of humanity—characteristics 29
such as altruism, nurturance, caring, and emotional support at the end of one dimension, and hostility, indifference to others, self-centeredness, spitefulness, and jealously at the other” (Digman, 1990, pp. 423-424). Thus, arguably, the Agreeableness sub-domain is the sub-domain of personality that assesses one’s humanity and how one interacts with others. Neuroticism vs. emotional stability. Neuroticism, also referred to as Emotional Stability in other language-based adjective models of human personality (Goldberg, 1990), refers to a person’s emotions in relation to his or her personality (Digman, 1990). Eysenck (1947), initially studying personality based on a “Big Two” factor system, was one of the earliest researchers to have discovered the Neuroticism sub-domain of human personality. Demonstrating convergent validity with Eysenck, Fiske (1949), only two years later, similarly found, when conducting a factor analysis of personality using selfrating and teammate ratings of individuals involved in the study, an emotional dimension to personality that Fiske called “emotional control” (p. 335). Giving further credence to the construct of an emotional component to personality based on further convergent validity, Norman in Goldberg (1981) also found evidence of what he referred to as “emotional stability,” which included polarity-based descriptors, such as calm vs. anxious, composed vs. excitable, not hypochondriacal vs. hypochondriacal, and poised vs. nervous (p. 157). Along with Extraversion, Neuroticism has been found to be a wellestablished sub-domain across many studies using a variety of research methods (Digman, 1990).
30
Vertical Dyad Linkage Theory LMX theory is the result of the seminal work of Dansereau (1972) for his dissertation on leadership and vertical dyad linkages in the workplace. In addition to Dansereau’s work on vertical dyad linkages (VDL) and LMX theory that followed, other leadership theories included (a) contingency theory (Fiedler, 1971; Heller, 1973), which is predicated on management objectives that are contingent on the characteristics of the manager and the situational aspect of his or her work; (b) great man theory (Borgatta, Bales, & Couch, 1954), which approached leadership from a personal trait angle and assumed that certain individuals are born with humanistic traits particularly conducive to leadership roles; and (c) the neo-charismatic leadership paradigm (Fiol, Harris, & House, 1999), which assumes that effective leaders are able to maximize their natural charisma to articulate their visions of desired ideological values. It was Fiedler’s (1971) work on the contingency model of leadership that began to question the assumption that leaders interact with and behave in the same way with all subordinate followers. Dansereau (1972) interpreted the work of Fiedler (1971) to mean that certain relationships exist between a manager’s work unit’s output and the manager’s style of leadership. Dansereau’s interpretation of Fiedler’s contingency theory was that jobs and tasks managed by subordinates should be engineered to best fit the manager (Dansereau, 1972). Based on this position Dansereau (1972) then described Fiedler’s (1971) process of relationship managing whereby either the leadership style as the leader relates to the subordinate, or the task situation is modified, or both, to provide for the best possible (contingent) managerial decision as to how to best manage a situation.
31
Dansereau (1972) continued with research on leadership, challenging the assumptions that leaders interact equally with subordinates as indicated by the previous work by Fiedler on contingency theory. First, Dansereau pointed out, production-centered approaches to leadership assume, first, that a supervisor’s behavior or traits should be correlated with the production of his or her unit, and, second, that a supervisor acts in the same manner and engages in the same behavior when interacting with each of his or her employees. Dansereau (1972) challenged the assumption that managers treated all subordinates equally and demonstrated empirically in his seminal work that, on the contrary, leaders developed different relationships with different subordinates. The second assumption regarding leaders engaging in relationships and interacting with subordinates that Dansereau (1972) discusses is a significant factor contributing to his consideration of a new leadership theoretical model leading to LMX theory. Dyadic relationships, the precursor to LMX theory. Dansereau (1972), and several researchers that followed him while studying LMX theory (Bakar et al., 2009; Dansereau et al., 1975; Graen & Schiemann, 2013; Graen & Uhl-Bien, 1991; Liden & Graen, 1980) began to focus on what was hypothesized to be the root of the working relationship between subordinate workers and their respective supervisors by considering the construct of vertical dyad linkages (VDL). Dansereau et al. (1975) described the groundwork for this, at the time, new theoretical approach to studying leadership using vertical dyad linkages, which culminated in the LMX theory of leadership. Dansereau et al. (1975) argued for the need to construct a theory of leadership that would not assume that leader-managers always act the same with all subordinates. Rather, by focusing on the study of the dyadic relationships between individual subordinates and supervisors, 32
theory could focus on understanding the processes that link the two, and thus would serve as a precursor to understanding work related outcomes. Distinctions between leadership and supervision. Vertical dyad linkage and the following LMX leadership theory that followed are highly dependent on a key distinction in the difference between supervision and leadership. Based on their research, Dansereau et al. (1975) found that the fundamental tenets of supervision and leadership in an organization are not the same. Rather, as the authors pointed to a difference in the vertical exchange in the dyadic pair of superior and member. In these dyadic relationships based on supervision, the leader relies exclusively on the formal contractual nature of the working relationships. This dynamic includes all proscriptions and prescriptions within the organization. In essence, in vertical dyads where a leader employs a supervisory tactic, the relationship is more formal, and the leader expects the follower to hold up his or her end of his or her contractual obligations in exchange for continued employment and the agreed-upon compensation. In stark contrast to the supervision method of leadership, the authors found the leadership technique involves not only the employment contract, but rather, the leader seeks a different basis in an attempt to influence the behavior of the follower (Dansereau et al., 1975). Dansereau et al. (1975) continued to describe the additional basis and other outcomes of value that the leader has at his or her disposal. The authors arrived at results indicating that in exchange for increased optional organizational commitment outside the formal constructs of the work contract, the supervisor can reward the subordinate with valuable and utility-laden incentives. These might include influence in decision making, open and honest communication, support of the subordinate’s actions, and job latitude. 33
Dansereau et al. (1975) found that the follower, in dyadic relationships that were grounded in the leadership method of influence, as opposed to the supervisory method, will often reciprocate for the increased incentives offered by the leader. Reciprocation might involve offering greater than required expenditures of time (in essence going above and beyond his or her formally required number of work hours) based in an effort to fulfill organizational goals, assume greater levels of organizational responsibility, and increase commitment to the greater organizational strategy (Dansereau, 1972; Dansereau et al., 1975). Leadership vs. supervision leads to in-group and out-group. Based on the findings of Dansereau et al. (1975), whose findings were supported by other researchers (Graen & Uhl-Bien, 1995; Liden & Graen, 1980; Scandura & Graen, 1984), differentiation in the approaches through which the leader in the dyadic relationship interacts with the follower resulted in 15% of the followers in the study being grouped into mutually exclusive, dichotomous groups called the “in” group and the “out” group (Dansereau et al., 1975, p. 56). The other 85% of the members demonstrated a mixture of in and out group members. Interestingly, the study found significant consequential outcomes were associated with the in-group members as opposed to the out-group members. Namely, in-group members reported significantly higher job satisfaction using the Hoppock Job Satisfaction Bank (Dansereau et al., 1975). In addition, the authors also found that in-group members also had better interpersonal relationships with their leaders, had increased intrinsic outcomes of their work, and had a higher psychological value of job performance rewards as compared to out-group members (Dansereau et al., 1975). 34
From vertical dyad linkage to leader-member exchange theory. Klein, Dansereau, and Hall (1994) made substantial arguments for the need to study leadership at various levels of analysis within organizations, and pointed to assumptions made in prior research regarding the homogeneity of groups in various levels of an organization. Klein et al. (1994) hypothesized that there may be systematic differences in the way that workers in various levels of an organization may score on various assessments, based on mixed level theory, depending on the worker’s job status or level within the organization. To that end, researchers continue to research organizational leadership theory from both a systems- and levels-perspective. Leader-Member Exchange (LMX) leadership theory (Graen & Uhl-Bien, 1995), based significantly on the work of Dansereau (1972) on vertical dyad linkages, began to emerge as a viable alternative to traditional theories of management. At its core, the fundamental tenets of LMX, which differs philosophically from prior leadership and management theories, involve LMX’s focus as a “relationshipbased approach to leadership” (Graen & Uhl-Bien, 1995, p. 225). LMX theory is predicated on the quality of the dyadic working relationship between the leader and follower. The same authors pointed out that the evolution from the original VDL theory to LMX theory occurred in four stages. 1. Validation of differentiation within working units. 2. LMX validation of differentiated relationships for organizational outcomes. 3. Theory and exploration of dyadic relationship development. 4. Team-making competency network investigating and assembling dyads into larger collectives including aggregations of dyads (Graen & Uhl-Bien, 1995).
35
It is this last stage of LMX theory where one can begin to appreciate the practical application of the LMX leadership theory in practice in an effort to facilitate the creation of highly synergistic work teams through various interventions designed to move more subordinates from the out-group to the in-group based on the known consequences of ingroup membership. Both LMX and VDL literature use the term leader (Dansereau, 1972) referring to the dominant authority member of the dyad, due to the fact that the leadership theory is applicable to various organizational entities, including for-profit, non-profit, and other entities in which the leader may or may not be in a formal power position involving coercive power or reward power. It is also possible that the leader may be solely in a position of referent power where the term supervisor, with its connotations of use in forprofit or non-profit organizations, may not be an accurate description of the leader’s role within the organization. However, because the present study focused entirely on the application of the LMX leadership theory in the for-profit business segment, the term supervisor is used in place of leader. In the LMX literature the term follower or member is used to describe the subordinate member of the vertical dyad relationship. Thus for the purpose of this study the term subordinate will be used to reference the follower.
Theoretical Orientation for the Study Significantly, previous research has demonstrated at least a partial link between aspects of personality, including positive affectivity (Bauer & Green, 1996), and their effect on both LMX theory. The same research has also linked proactive personality and employee engagement (Bakker et al., 2012). Therefore, because prior researchers have 36
demonstrated differences in the attitudes and perceptions of supervisors vs. subordinates (Romanou et al., 2010; Rothbard, 2001), logic dictates that it is possible that human personality and work role (as well as those variables' relationships with both leadership from the LMX theoretical perspective and theoretical perspectives of employee engagement) might extend each theory separately and possibly move towards synthesis of the three theories: Big Five personality, employee engagement, and LMX leadership theory. From a theoretical perspective, adding additional predictive value to LMX theory and employee engagement by formally researching the Big Five model of personality will help to bridge the gap between the scholar and practitioner. It will also help researchers investigate whether personality, based on the Big Five model and quality of relationships under LMX theory, affects employee engagement. Figure 1 demonstrates this theoretical model as presented in this dissertation.
37
Individual Antecedents: Absorption, dedication, corporate citizenship, work/life balance, and personality sub-domains of Conscientiousness and Extraversion based on the Big Five model of personality
Organizational Antecedents: Job design characteristics (full-time vs. parttime), justice in the workplace, Work environment, and perceived organizational support
Organizational Relationships
Supervisor – Subordinate relationship
Employee Engagement
Positive consequences of Engagement: Increased job performance, reduced turnover, increased corporate profitability, increased organizational citizenship behavior
Figure 1. Theoretical model
38
This model theorizes the relationships between personality, based on the Big Five model, and its effect on employee engagement and the quality of relationships. It also theorizes the relationship between personality and the process of employee engagement. In addition, the conceptualization of this model and the results of the study could help with additional understanding of the antecedents of employee engagement and could suggest further refinements of a model of employee engagement. Continued research may lead to further development of models that could be used to allow practitioners earlier intervention into possible relational issues between supervisors and subordinates that might negatively impact employee engagement. The goal of the study would be to answer the research questions to gain a better understanding of the antecedents of employee engagement. Because aspects of personality have been found to be a factor related to subordinate employee engagement, this demonstrated relationship could result in future research involving experimental methods to further expand individually and synthesize LMX theory, the Big Five personality theory, and employee engagement. In this case, from a theoretical perspective, extension and synthesis of LMX theory and employee engagement may lead to better understanding of the theoretical constructs of each theory independently, as a difference in employee engagement associated with personality would imply that perhaps it is possible, and even plausible, that the three theories are integrated. Because this study found that Conscientiousness and LMX relationship were both statistically significant in predicting employee engagement, this would suggest, at minimum, that future research efforts should continue to investigate the relationship between human personality, LMX leadership theory, and employee engagement. 39
Review of Research Literature The Capella University library was used to search for appropriate research methods to answer the research questions in this study. Databases searched within the library included the SAGE Knowledge library, which includes a section on research methodology and the associated database, SAGE Research Methods. Because this dissertation study involves the consideration of independent variables that are nonmanipulable (Shadish et al., 2002), only methods were considered that were nonexperimental in design, and would be feasible with the limitation of the non-manipulable independent variables. The results of this search resulted in the ex post facto design (Leedy & Ormrod, 2013). In this study, the independent variables including (a) individual personality, (b) work role, and (c) gender and racioethnic similarity of an employee with his or her supervisor, all constitute naturally occurring, and non-manipulable independent variables that were ideal for study using the ex post facto design. McBride (2015) alluded to the appropriateness of the ex post facto method for research plans of this type “…because the comparison of interest is based on a grouping that already exists instead of one the researcher assigns in the study. In other words, the grouping is based on something that already happened in the past” (p. 92). In this study, all of the independent variables of interest are comparisons of interest that already exist. Thus the experimental and nonexperimental designs found in the literature review—namely quantitative, ex post facto designs—were the most appropriate for use in this study. 40
Synthesis of the Research Findings This study aimed to synthesize and extend aspects of three previously developed theories in an effort to further the theory of employee engagement. The three existing theories that were investigated were: (a) personality, based on the Big Five model (Digman, 1997), (b) employee engagement (Kahn, 1990; Maslach & Leiter, 2008; B. Shuck & Wollard, 2010), and (c) LMX leadership theory (Dansereau, 1972; Dansereau et al., 1975; Liden & Graen, 1980). Whereas all three theories were individual researched, they do share some related elements. Bauer and Green (1996) investigated personality similarities between subordinates and supervisors based on the connection between those personality similarities and their effect on trust, which in turn was posited as influencing the subordinate’s tendency to delegate. However, in the same study, Bauer and Green (1996) continued by stating “The role of personality in leader-member-exchange development also appears to be a fruitful direction to pursue. Positive affectivity was a contributor here, but other measures need to be explored. Studying the Big Five measures of personality would be a good place to start” (p. 1563). Thus, although personality appears to have been an ancillary factor based on its moderation of trust, a full understanding of personality interactions between supervisors and subordinates, and an understanding of how those interactions affect a subordinate’s quality of relationship based on LMX theory, has not been achieved. In addition, like LMX theory, there has been limited investigation into personality interactions with employee engagement. More specifically, one author wrote “For instance, within the widely applied Big Five taxonomy (Neuroticism, Extraversion, Openness to Experience, Agreeableness, and Conscientiousness) it has yet to be 41
determined which factors are or are not relevant to engagement” (Inceoglu & Warr, 2011, p. 177). Another group of authors pointed to the significant gap in the employee engagement literature when they wrote, “…the relationships among potential antecedents and consequences of engagement as well as the components of engagement have not been rigorously conceptualized, much less studied” (Macey & Schneider, 2008b, pp. 34).
Critique of the Previous Research Research on the construct of employee engagement is relatively new, taking into account the emergence of the construct was first discovered in the seminal work of Kahn (1990). The construct of employee engagement, the antecedents and consequences of engagement, and the theoretical model that explains employee engagement, continue to evolve as researchers investigate the phenomenon. Questions on the construct and debate of those questions include nomenclature with regard to the name of the construct, and confusion with regard to the underlying foundational tenets of the construct. Continuing Nomenclature Debate on the Theory Name Researchers thus far have not agreed upon a formal name for the underlying construct, with some referring to the construct as work engagement (Schaufeli & Bakker, 2003; Schaufeli, Bakker, & Salanova, 2006; Sonnentag, 2003), and some referring to the construct as employee engagement (Harter et al., 2002; Saks, 2006; Townsend & Gebhardt, 2008; Wollard & Shuck, 2011). In this dissertation the term employee engagement, for consistency, will be used whenever possible; however occasional references to work engagement may be necessary for context. In any event, use of either 42
term must currently be accepted and addresses the hypothesized same underlying research construct. Confusion Regarding State, Trait, or Behavior In Kahn’s (1990) seminal work on employee engagement, he began his discussion of the analysis of his results by explaining “The conceptual framework presented here begins with defining and illustrating the concepts of personal engagement and disengagement that emerged from this research” (Kahn, 1990, p. 699). As Kahn further developed the framework and foundation, using the emergent theory method, he began to describe personal engagement as “the simultaneous employment and expression of a person’s ‘preferred self’ in task behaviors that promote connections to work and to others, personal presence (physical, cognitive, and emotional), and active, full role performance” (Kahn, 1990, p. 700). Kahn’s use of the word behaviors is significant in that the author clearly posited that the construct of employee engagement is founded on behaviorism even in light of the fact that Kahn also discussed in his grounded theory model various organizational and individual conditions that affect employee engagement. However, Macey and Schneider (2008b) pointed to various research articles that posit that employee engagement connotes commitment, passion, focused effort, energy, and involvement, and thus the construct has behavioral and attitudinal components. Furthermore, the term engagement has been used not only in a behavior related performance construct, in terms of observable behavior or degree of effort, but also has been used to refer to a psychological state (attachment to one’s job, mood, and psychological involvement), and even used to refer to disposition with an attitudinal component (Macey & Schneider, 2008b; Maslach & Leiter, 2008; May, Gilson, & Harter, 43
2004; Rothmann & Welsh, 2013; Vander Elst, Bosman, De Cuyper, Stouten, & De Witte, 2013). These various associations with the term engagement only lead to more questions, including the question of whether, at the foundational core of the construct, employee engagement is a trait, state, or behavior, or even, a combination of all three.
Summary Scientists and practitioners are increasingly interested in removing the ambiguity currently found in the employee engagement literature by focusing on research that will help researchers better understand the fundamental tenets of the construct, the construct antecedents, and the construct consequences. Based on the previous work on employee engagement that alluded to relationships with personality theory, and LMX leadership theory, it is possible that a synthesis between the three theories exists. An exploration of the relationships between the theories may help reveal the antecedents of employee engagement. Much remains to be learned about employee engagement before researchers and practitioners can begin to utilize the knowledge by creating intervention models meant for practical application in the workplace (Church, 2011; Rana et al., 2014; Saks, 2006; Wollard & Shuck, 2011). This dissertation was designed to continue work on employee engagement by further investigating its antecedents.
44
CHAPTER 3. METHODOLOGY Purpose of the Study The purpose of the study was to gain additional insights and knowledge into the antecedent factors that lead to employee engagement. This was done by designing a research protocol to answer this study’s research question as to whether the independent variables, including (a) work role, (b) personality, (c) LMX quality of relationship, (d) gender and race similarities between supervisor and subordinate, (e) employment status, and (f) tenure statistically contribute to the prediction of employee engagement in the workplace. Adding to the literature by learning more about the antecedents of employee engagement may help future researchers develop models to increase employee engagement. Therefore, the crux of the present study was to address the need to further understand the antecedents of employee engagement. The literature appeared to be devoid of scientific, empirically-investigated research pertaining to the antecedents of employee engagement (Saks, 2006; Wollard & Shuck, 2011). Wollard and Shuck (2011) pointed out that the gap in the literature pertaining to the antecedents of engagement is not only from a theoretical standpoint, but also that the gap exists in practice. The purpose of this study was to address the gaps in the literature and the recommendations by various authors, such as Saks (2006), for continued research into the antecedents of employee engagement in an effort to add to the knowledge base upon which future scholars and practitioners could continue to build.
45
Research Design This research study used a quantitative, ex post facto design to answer the study's research question (Leedy & Ormrod, 2013). Leedy and Ormrod (2013) defined an ex post facto research design as follows: “In an ex post facto study, a researcher identifies events that have already occurred or conditions that are already present and then collects data to investigate a possible relationship between those factors and subsequent characteristics or behaviors” (p. 242). The existing conditions, in this study, that the ex post facto research method was designed to investigate, are the existing demographic and work-related independent variables of interest. Those variables are: (a) work role, (b) gender similarity of employee and supervisor, (c) racioethnic similarity of employee and supervisor, (d) the participant's tenure, (e) the participant's personality, based on the Big Five model, (f) the participant’s employment status, either full time or part time, and (g) the participant’s perceived quality of relationship with his or her supervisor, based on LMX theory. Furthermore, in this study, the characteristics examined via the ex post facto design were the study participants’ measures of employee engagement. In theory, the ex post facto protocol design in conjunction with the multiple regression analysis method (Field, 2009; Warner, 2013) is able to determine if the independent variables of interest are statistically significant in predicting employee engagement, as measured by the UWES employee engagement scale (Schaufeli & Bakker, 2003). Furthermore, when considering independent variables and the degree to which those independent variables are manipulable or not, Shadish et al. (2002) described the concept of “manipulable” and “non-manipulable” causes (p. 7). The authors pointed out there are some variables that cannot be easily, feasibly, or ethically manipulated by a 46
researcher, such as an individual’s sex, age, or raw genetic material. Shadish et al. (2002) referred to hypothesized causes, such as those listed in the previous sentence, as “nonmanipulable causes” (p. 7). The ex post facto design was a logical choice as the protocol design for this study because the method was specifically constructed, first, to consider hypothesized cause conditions that are already present, and second, to involve nonmanipulable variables. In addition, the ex post facto design is capable of considering multiple independent variables that could potentially predict an outcome variable and thus is an appropriate design for considering the multiple variables that may be factors associated with employee engagement.
Target Population and Participant Selection Because the purpose of the present study was to better understand the antecedents of employee engagement in the workplace, the population from which samples were drawn was working adults in the United States between the ages of 18 to 55 who were employed in the same position for a minimum of six months and had been reporting to the same supervisor for a minimum of six months. Participants were recruited using a third party survey and analytics company, Qualtrics (Qualtrics, 2014). A paid membership and account was established with Qualtrics, allowing the principal investigator to engage in purposive sampling using the service's extensive database of firms and prospective study participants. The inclusion-exclusion criterion was submitted to Qualtrics for screening purposes. The firm selected samples that met the inclusionexclusion criteria.
47
Purposive Sampling Method In purposive sampling, samples are chosen based on a specific purpose (Creswell, 2007; Leedy & Ormrod, 2013; Patton, 2002). In this study, that purpose involved specifically selecting participants who were employed in the workplace and whose measures of individual personality in conjunction with their perceived working relationship with their supervisors, based on the LMX leadership theory, could help answer this study’s research question. Although in an ideal research scenario, random sampling would be the preferred sampling method due to the increased internal validity offered by random sampling (Leedy & Ormrod, 2013; McBride, 2015; Shadish et al., 2002; Wilkinson, 2000), the costs involved in utilizing a random sampling method in the field made that method infeasible for this study. Thus, the purposive sampling method was employed to solicit participants who met the study’s inclusion and exclusion criteria that came from a sampling frame provided by a third party firm, Qualtrics. Inclusion and Exclusion Criteria Participants were targeted based on the purposive sampling method (Creswell, 2007; Patton, 2002) to ensure that the purpose of the study was fulfilled, and based on the limitations and restrictions of the study that were due to the non-manipulable properties of the independent variables. This study’s inclusion and exclusion criteria were as follows:
The participants must be currently employed with a minimum of six months of continuous employment in the same position with the same company.
48
Participants must have been reporting to the same supervisor for a minimum of six months.
Participants must be between the ages of 18 and 55.
Participants must be either a supervisor with the responsibility for the direct supervision and management of at least one subordinate, or a subordinate that must report to one supervisor.
Participants must not currently have any outstanding disciplinary action pending, whether verbal or written.
Participants must not have seen any of the following or been under the care of any of the following for any mental health concerns within the last year: a clergy member; a pastor; a psychiatrist; a psychologist; or a drug, alcohol, or relationship counselor.
Participants must be able to read and write at the 8th grade level.
Procedures This dissertation study, approved by the Capella Institutional Review Board (IRB), used a third party service (Qualtrics) to generate qualified samples that met the purpose for the study, based on the purposive sampling method (Creswell, 2007; Qualtrics, 2014). Working with the Qualtrics firm, a web-based content delivery survey was created that consisted of the LMX-7 instrument (Graen & Uhl-Bien, 1995), the 50-item International Personality Item Pool (IPIP) Big Five Factor Markers (Goldberg, 2015e), and the Utrecht Work Engagement Scale (UWES) (Schaufeli et al., 2006), in addition to the other questions designed to filter for the study’s inclusion and exclusion 49
criteria. The third-party service was given a list of the study’s inclusion and exclusion criteria to identify potential participants in the workplace who served the intended purpose of the study, along with other pertinent information when establishing an account with the service. The third-party service then used its proprietary algorithm to match this study’s inclusion-exclusion criterion with appropriate panels of companies whose employees matched the criteria. Once employees who met the criteria for the study were identified, Qualtrics sent an email invitation from its database to those prospective participants to determine if they would be willing to participate in the research study. Participants who were interested in enrolling in the study did so by clicking on a link provided in the email which directed the participant to a designated Universal Resource Locator (URL) address. Those participants who opened the email solicitation and clicked on the link to be taken to the survey were first presented with the study’s informed consent document, pursuant to IRB requirements, and participants were able to opt in or out of the study after reading the electronic informed consent by clicking on a designated "yes" or "no" link. Participants who selected "no" after reading the informed consent document were opted out of the study and did not participate. Those participants who opted in after reading the electronic informed consent were presented a set of self-report questions, the complete set of which included:
How long the participant had been employed with the company in which he or she was currently employed.
Whether the participant had been under the care of a psychologist, psychiatrist, or counselor during the past year.
The participant’s age. 50
Whether the participant had worked under his or her current supervisor for a minimum of six months.
Whether the participant had been under any form of disciplinary action at his or her place of employment.
The participant’s gender.
Whether the participant’s gender was the same as his or her supervisor’s.
The participant’s race.
The participant’s racioethnic similarity with his or her supervisor.
Whether the participant worked full-time or part-time.
Whether or not the participant’s work role involved managing other people or not.
After the self-report questions, the participants were then presented with the remainder of the battery of assessment questions. Upon completing the assessments, the data were collected by the Qualtrics web-based service and stored electronically on the Qualtrics server. After collection of the data from all participants to meet the minimum number of 95 participants necessary to sufficiently power the study, an Excel spread sheet was emailed to the study’s principal investigator from Qualtrics containing all the raw data.
Instruments and Measures In order to conduct this research study, three psychological instruments were administered to participants. Those three instruments were (a) the UWES, (b) 50-Item International Personality Item Pool--Big Five Factor Markers, and (c) the LMX-7 instrument. The three instruments were administered electronically over the Internet to 51
participants after participants accepted the terms of the informed consent. The results of the LMX-7 and the 50-Item International Personality Item Pool—Big Five Factor Markers where included in the multiple regression analysis as predictor variables. Conversely, the result of the results of the UWES instrument was entered into the multiple regression analysis as the dependent variable. Utrecht Work Engagement Scale To investigate the dependent variable, employee engagement, the UWES instrument was used (Schaufeli & Bakker, 2003, p. 48). The constructs measured by the instrument included three dimensions of work engagement: vigor, dedication, and absorption (Schaufeli et al., 2006, p. 702). The instrument created by Schaufeli et al. (2006, p. 702) included 17 total items, comprised of six items measuring aspects of vigor, five items measuring dedication, and six items measuring absorption. Norming data for the instrument came from 10 countries included in a study in which a total of 14,521 participants (N = 15,521) were studied (Schaufeli et al., 2006, p. 704). Furthermore, the participants in the Schaufeli et al. (2006, p. 705) study in which the instrument was created and validated were categorized by occupational group as follows: social work, blue collar, health care, white collar for-profit, white collar not-for-profit, teachers, police officers, management, and a final category labeled information missing. In the Schaufeli et al. (2006, p. 705) study, of the 15,521 participants, 7,621 were male [53.3%], while 6,684 where female [46.7%] (Schaufeli et al., 2006, p. 705). Ages of participants in the Schaufeli et al., (2006, p. 705) study ranged from 16 to 68 years (M = 40.3, SD = 11.7). The UWES instrument utilized a Likert scale, resulting in a continuous variable from 0 to 102. A confirmatory factor analysis confirmed that the hypothesized three52
factor version of the UWES is superior to the instrument's one-factor model and fits well to various samples (Schaufeli & Bakker, 2003, p. 7). A number of validity studies comparing the construct of employee engagement using the UWES scale as the polar opposite of burnout using the Maslach Burnout Inventory (MBI) demonstrated that, as expected, the three scales of the UWES are negatively correlated with aspects of the MBI (Schaufeli & Bakker, 2003, p. 8). This is relevant due to the fact that in many of the previous studies conducted on employee engagement, the MBI was administered to participants as a measure of employee engagement. 50-Item International Personality Item Pool--Big Five Factor Markers To investigate the participant's personality, the 50-item IPIP Big Five Factor Markers instrument (Goldberg, 1999, 2015e; Goldberg et al., 2006) was administered to all participants. The items in the instrument were created by Goldberg (2015e) and are based on his previous work on the Big Five model of personality (Goldberg, 1981, 1990, 1992). The International Personality Item Pool Goldberg (2015b) is a repository for public domain psychometric assessments and scales created by Dr. Lewis Goldberg, in collaboration with other scientists who have a mutual interest in the philosophical goal of creating a public domain repository of psychiatric and psychological assessments. Goldberg (2015d) clearly provided anyone permission to use the collection of items and instruments in any way that one would want to use the items or instruments where the author wrote, “Asking permission...Please don't! One neat thing about the world of public domain is that NOTHING is a problem. You are free to use the IPIP items and/or scales in any way you want. You don't have to ask permission. Cool, huh” (p. 1). The 50 item IPIP Big Five Markers measures the five sub-dimensions of 53
personality based on the Big Five model: Openness to Experience, Conscientiousness, Extraversion, Agreeableness, and Neuroticism (Goldberg, 1992, 1999, 2015c; Goldberg et al., 2006). Goldberg (2015a) created and provide scoring instructions for the instrument. Convergent validity has been demonstrated for the 50-item IPIP Five Factor scale by comparing its results with several well-researched commercially available instruments designed to measure the same hypothesized underlying five constructs of personality. The instruments included the Big Five Inventory (Zheng, et. al, 2008), and the Revised NEO Personality Inventory (Maples, Guan, Carter, & Miller, 2014). The 50 item IPIP Five Factor scale has also been demonstrated to have high correlations with the 16PF instrument, resulting in an average correlation among the five domains of .51 (Goldberg, in press). In addition, the 50-item IPIP Five Factor Markers has been demonstrated to have above average correlations with the commercially available NEO five factor inventory (Costa, 1996; McCrae & Costa, 2004, 2007) with correlations ranging from .46 to .48 (Goldberg, in press). Finally, one group of researchers conducted a comparison of the 50 item IPIP Five Factor Markers and the Big Five Inventory (John, Donahue, Kentle, & Benet-Martinez, 1998) finding convergent validity between the two instruments with an average correlation of .65 on all five sub-domains (Zheng et al., 2008). Leader-Member Exchange To investigate the variable relationship quality based on LMX leadership theory, the LMX-7 questionnaire was used (Graen & Uhl-Bien, 1995, p. 237). The LMX-7 questionnaire is a Likert-style instrument with 7 questions. Each question can be answered with a number ranging from 1 to 5. Thus, the instrument results in a composite, 54
continuous score from 7 to 35. Development of a scale to measure the theoretical leadership theory of LMX began with a two-item survey (Dansereau et al., 1975). Further refinement of both the theory and instrumentation to measure the underlying construct of LMX resulted in a four-item instrument (Liden & Graen, 1980). Continued debate over whether the LMX construct is unidimensional or multidimensional resulted in the development of other instruments, including a 7-item instrument (Seers & Graen, 1984), and even a 14-item instrument (Uhl-Bien, Tierney, Graen, & Wakabayashi, 1990). However, Graen and Uhl-Bien (1995) noted that although the additional items in various forms of the instrument were added to test the dimensionality of LMX, the other instruments are highly correlated with the more precise seven-item instrument and produced the same effects. Thus, for the present research study, the seven-item version of the instrument was used. Convergent validity has been established by correlating the results of the LMX-7 with an existing instrument, the LMX-MDM, which measures the same hypothesized underlying constructs (Liden & Maslyn, 1998). The correlation between the LMX-MDM and the LMX-7 questionnaire resulted in coefficients between .33 and .71 on the subscales of loyalty, contribution, and professional respect (Liden & Maslyn, 1998, p. 61). A study conducted on 302 working students and 251 organizational employees resulted in internal consistency reliabilities with coefficient alphas ranging from .56 to .92 (Liden & Maslyn, 1998, p. 59). The LMX-7 questionnaire measures three underlying constructs that constitute the theory: the leader, the follower, and the relationship (Graen & UhlBien, 1995, p. 223).
55
Research Question and Hypotheses This study’s research question was formulated to address gaps in the literature pertaining to the antecedents of engagement in conjunction with the philosophies inherent with quantitative, ex post facto research methods (Leedy & Ormrod, 2013) and falsification theory (Popper, 2002). The study’s research question posits whether potential work-related and person-related variables are statistically significant in predicting employee engagement using multiple regression analysis. Research Question This study’s research question is as follows: Do work role, individual personality, LMX working relationship, gender similarity between members of the supervisorsubordinate dyad, racioethnic similarity between members of the supervisor-subordinate dyad, employment status as a full-time worker or part-time worker, and/or tenure predict employee engagement? This research question is written in an attempt to address gaps in the employee engagement literature with regard to the antecedents of engagement, the primary goal of this research study. Null hypothesis. The null hypothesis, which is consistent with the philosophical tenets of falsification in quantitative research (Popper, 2002), is H0: The independent variables, work role, individual personality, LMX working relationship, gender similarity between the members of supervisor-subordinate dyad, racioethnic similarity between the members of the supervisor-subordinate dyad, full- or part-time employment status, and/or tenure are not statistically significant in predicting employee engagement. Alternative hypothesis. Conversely, the alternative hypothesis will be H1: The independent variables of work role, individual personality, LMX working relationship, 56
gender similarity between the members of the supervisor-subordinate dyad, racioethnic similarity between the members of the supervisor-subordinate dyad, full- or part-time employment status, and/or tenure are statistically significant in predicting employee engagement.
Data Analysis To answer this study's research question, multiple regression analysis was used to fit a regression model to the data (Field, 2009; Warner, 2013). This study included 12 potential independent predictor variables: (a) work role, (b) Big Five personality omnibus total score, (c) the raw scores for the five sub-domains that make up the Big-Five model of personality (Extraversion, Agreeableness, Conscientiousness, Emotional Stability, and Intellect), (d) LMX quality of relationship as measured by the LMX-7 instrument, (e) gender similarity between supervisor and subordinate, (f) racioethnic similarity between supervisor and subordinate, (g) employment status (full time or part time), and (e) job tenure, which were inserted as the potential predictor variables into the regression equation using the stepwise hierarchical method of multiple regression (Field, 2009; Warner, 2013). The outcome variable of interest was employee engagement and was measured by the UWES instrument (Schaufeli & Bakker, 2003). Field (2009) alludes to the appropriateness of using the multiple regression method as an extension of the same statistical principles found in linear regression method when several predictor variables are under consideration. This demonstrates that the multiple regression method was an appropriate method of statistical analysis for this study based on the multiple independent, predictor variables of interest which are (a) work role, (b) LMX quality of 57
relationship, (c) gender similarity within the supervisor-subordinate dyad, (d) racioethnic within the supervisor-subordinate dyad, (e) employment status, (f) the results of the personality omnibus score based on the Big Five model, (g) each of the five sub-domains of personality based on the Big Five model of personality, and (h) tenure. The answer to this study’s research question was determined by the significance of the F statistic outcome in the omnibus multiple regression analysis. Because previous research demonstrated that race similarities between supervisors and subordinates (Avery et al., 2012), as well as two specific subdomains of personality, Extraversion and Conscientiousness, (Handa & Gulati, 2014; Kim et al., 2009; Langelaan et al., 2006) are related to employee engagement, the hierarchical blockwise entry method (Field, 2009; Warner, 2013) of multiple regression was utilized. When inserting the predictor variables in the multiple regression analyses, the first block included only the predictor variables of Conscientiousness and Extraversion, along with the racioethnic similarity variable between supervisor and subordinate into the multiple regression equation used in the analysis. After those three variables were entered into the equation, the remaining potential predictor variables were entered into the equation as the second block of variables. The alpha level used for the study was 0.05. Finally, the results of the multiple regression output were analyzed and they were interpreted to mean that a rejection of the null hypothesis was in order.
Ethical Considerations This research proposal received approval from the Capella University IRB before solicitation of participants or data collection began. The participants in this study are not 58
considered a vulnerable population. In addition, because there was no treatment being applied to an experimental group, there was no ethical concern regarding withholding potentially beneficial treatment from a control group. Ethical concerns that were considered involved ensuring methods were in place to protect the privacy of participants and their personal information. All participant responses were de-identified using a unique identifier for each participant, as explained in the Qualtrics frequently-asked questions section of the company's website (Qualtrics, 2015). Thus, the study author did not have any identifying information for any of the study participants.
Expected Findings This study may corroborate prior research, including one study that found that race and gender similarities between subordinates and supervisors was related to the subordinates’ measure of employee engagement (Avery et al., 2007, 2012). This study may also corroborate studies that found that relationships between specific sub-domains of the Big Five personality model, namely Extraversion and Conscientiousness (Handa & Gulati, 2014; Inceoglu & Warr, 2011; Kim et al., 2009; Langelaan et al., 2006), were factors in predicting employee engagement. The intriguing questions were whether work role (supervisor or subordinate) and LMX relationship quality would be found to be significant predictors of employee engagement based on the multiple regression analysis.
59
CHAPTER 4. DATA COLLECTION AND ANALYSIS
Introduction Data were collected for this research study using the purposive sampling method (Creswell, 2007; Patton, 2002; Shadish et al., 2002). The purpose of the study was to determine if the following hypothesized predictor variables significantly predict employee engagement: (a) work role, (b) the result of the omnibus individual personality score based on the Big Five model, (c) the result of the five sub-domains of personality based on the Big five model, (d) LMX working relationship, (e) gender similarity within the supervisor-subordinate dyad, (f) racioethnic similarity within the supervisorsubordinate dyad, (g) employment status as a full-time worker or part-time worker, and (h) tenure. In an effort to answer this study’s research question and pursuant to the purposive sampling method, ideal subject profiles were identified and the profile information was converted into a series of inclusion and exclusion criteria. Ideal participants were solicited for participation in the study via email using a third party analytics firm, Qualtrics (Qualtrics, 2014).
Population and Sample I was provided the samples which came from a proprietary list of participant panels owned by the third party analytics company Qualtrics (Qualtrics, 2014). The sampling frame was created by including persons who had opted in to Qualtrics surveying service. Potential participants were contacted in the United States of America and were actively employed at the time of solicitation. Participants came from a wide 60
variety of industries and professions. Participants within that sampling frame who met the inclusion and exclusion criteria for the study were solicited electronically via email. The email solicitation included a link and Universal Resource Locator (URL) that directed the prospective participants to take an online survey that included the complete battery of assessments, UWES, LMX-7, and the 50-Item International Personality Item Pool—Big Five Factor Markers. Those prospective participants who clicked on the link in the solicitation email were first presented with an electronic informed consent pursuant to the requirements of the Capella University IRB. After reading the electronic informed consent, prospective participants were required to accept the remainder of the assessments and questions by first electronically acknowledging the informed consent document. Participants who declined were not allowed to participate in the survey. Respondents came from a variety of industries and professions. All respondents lived in the United States of America and met the inclusion and exclusion criteria established for the study. In an effort to ensure that respondents would fit the purpose of the study pursuant to the purposive sampling method, all respondents were asked selfreport questions involving length of employment with their current employer, length of time reporting to the same supervisor, age, and other demographic questions, and whether or not they had been under the care of any of the following: clergy members; pastors; psychiatrists; psychologists; or drug, alcohol, or relationship counselors. Any potential participants who did not meet the inclusion and exclusion criteria based on the self-report questions were eliminated.
61
After eliminating those respondents who did not consent to participate in the study, and any participants eliminated based on the self-report demographic questions, 96 respondents remained who continued by answering the survey questions (N = 96).
Statistical Assumptions and Diagnostics In an effort to reduce statistical validity threats (Shadish et al., 2002), I addressed common assumptions inherent with the multiple regression statistical method used in this study. In addition, I ran and interpreted multicollinearity, homoscedasticity, and normally distributed errors diagnostics using SPSS (version 20). The results of these interpretations suggest that no assumptions inherent with multiple regression where violated. Finally, the diagnostics support the notion that results of the study are statistically valid. Independence of Observations Independence of observation of the outcome variable data is an assumption inherent in multiple regression analysis (Field, 2009; Warner, 2013). All data collected in this study were collected independently. No data input was influenced by other data. Respondents entered their answers into the web-based survey one question at a time and no data input influenced other data inputs. Thus the assumption of independence of observation of the outcome variable was not violated in this study. Outcome Variable The outcome variable, the result of the UWES assessment, is a quantitative, continuous, ratio-level variable, as is required to meet assumptions for use in a parametric statistical method, such as this study’s multiple regression analysis. Thus, the outcome variable type and measurement level met the assumptions necessary for outcome 62
variables used in multiple regression analysis (Field, 2009; Warner, 2013). A visual inspection of the histogram for the outcome variable UWES was performed in an effort to make a determination regarding the assumption of normality of the variable’s distribution (Field, 2009; Warner, 2013). The visual inspection of the histogram for the UWES variable suggests that the variable meets the assumption of normality required when using parametric statistical methods. To further ensure that the outcome variable met the assumption of normality, a Shapiro-Wilk test of normality was run on the UWES variable, which demonstrated a non-significant result, W(96) = .978, p < .098. The result of the Shapiro-Wilk test, along with the visual inspection of the outcome variable’s histogram, suggested that the assumption of normality was not violated. Predictor Variables All predictor variables are either quantitative or categorical in terms of variable level. In addition, all predictor variables that are categorical have at least two categories (as in cases of race and gender similarity between employee and supervisor). Finally, predictor variables that are not categorical are quantitative, continuous, interval, or ratio level variables (in the cases of the five sub-domain measures of personality, along with the omnibus personality score, and the result of the LMX-7 questionnaire). Multicollinearity Multicollinearity is a concern in multiple regression analysis. To meet the multicollinearity assumption, no two or more variables should have a perfectly linear correlation (Field, 2009; Warner, 2013). One method used to test for multicollinearity concerns, according to Field (2009), is investigation of the correlation matrix for all predictor variables to see if any correlations above .80 exist. As can be seen in this 63
study’s correlation matrix in Tables 4 and 5, no two predictor variables have a correlation greater than .80, which suggests that a threat to the results of the statistical analysis due to multicollinearity is not a concern. To further investigate multicollinearity concerns, the variance inflation factor (VIF) was calculated in SPSS (version 20) for all predictor variables (Cook & Campbell, 1979; Warner, 2013). One predictor variable, the personality omnibus score, was found to have a statistically significant VIF, VIF = 4.95, p < .001, and thus was eliminated from inclusion in the multiple regression models. Homoscedasticity Homoscedasticity is an important assumption to consider when using a multiple regression statistical analysis (Field, 2009). To check for the homoscedasticity assumption, visual inspections of the scatter plots of the residuals for each predictor variable were compared against one another. A visual inspection of the scatter plots of the residuals for each predictor variable suggested that the assumption had not been violated. Normally Distributed Errors To diagnose potential violation of the normally distributed errors assumption, three new variables were created when the multiple regression analysis was run in SPSS (version 20) to save both the standardized and un-standardized residuals, as well as the unstandardized predicted value, for each predictor variable. Next, the distributions of the newly created variables were tested for normality with a Shapiro-Wilk test of normality. No significant results were found in any of the three distributions, demonstrating that the residual distributions for each predictor variable met the assumption of normally distributed errors. The visual inspection suggested that the assumption of 64
homoscedasticity had not been violated. The results of the Shapiro-Wilk test of the distribution of errors can be viewed in Table 1. Table 1 Residuals Test of Normality Shapiro-Wilk Result Residual Distribution
Statistic
df
Significance
Unstandardized Predicted Value
0.979
96
0.128
Unstandardized Residual
0.989
96
0.581
Standardized Residual
0.989
96
0.581
Summary of the Results Ninety-six respondents completed the online assessment and submitted their data to be included in the study. After analyzing the raw data, it was determined that there was no missing data. Demographic and descriptive statistics were analyzed for the data set. Finally, the multiple regression analysis using the hierarchical blockwise entry method was run using SPSS (version 20). Demographics In terms of race, 76 participants (79.17%) identified as Caucasian, nine participants (9.38%) identified as African-American, seven participants (7.29%) identified as Hispanic, one participant (1.04%) identified as Native American, and one participant (1.04%) identified as mixed-race. Furthermore, two participants (2.08%) identified as other with respect to race. Sixty-four of the respondents (66.66%) worked a 65
minimum of 40 hours per week and thus were considered full-time employees. Thirtytwo of the respondents (33.33%) worked fewer than 40 hours per week and are considered part-time employees. The mean statistic for tenure was 6.01 years of employment. Of the 96 respondents to the study, 29 were male (30.20%) and 67 were female (69.79%). Thirty-eight of the 96 respondents (39.58%) reported that they were responsible for the management of other workers as part of their job duties and thus were considered supervisors for the purposes of this study. Conversely, 58 of the 96 respondents (60.41%) reported that they were in non-managerial roles and were classified as non-managers in this study. The mean age statistic for the sample was 35.96 and the median age was 36.00. Descriptive statistics A case processing summary run using SPSS (version 20) demonstrated that there was no missing data from the 96 respondents. Sixty-six of the respondents (68.75%) reported that their race was the same as their supervisor’s race. Conversely, 30 respondents (31.25%) reported that their race was different than their supervisor’s race. In addition, 63 respondents (65.62%) reported that their gender was the same their respective supervisor’s gender. Conversely, 33 respondents (34.38%) reported that their gender was opposite of their respective supervisor’s gender. Descriptive statistics for the continuous variables entered into the multiple regression equation can be seen in Table 2.
66
Table 2 Descriptive Statistics Variable UWES LMX-7 Factor I - Extraversion Factor II - Agreeableness Factor III - Conscientiousness Factor IV - Emotional Stability Factor V - Intellect Personality Totals Age Tenure (in years)
Mean SD 81.46 18.18 27.03 5.21 31.03 9.11 41.05 6.39 40.60 5.61 34.28 8.92 38.85 5.47 185.82 25.17 35.96 10.55 6.01 5.24
Multiple Regression Results The multiple regression analysis yielded a significant result in both models. The first model included the hierarchical block wherein the predictor variables were Extraversion and Conscientiousness along with the racioethnic similarity of employee and supervisor. These three predictor variables were loaded into the equation based on the results of prior research that demonstrates relationships between the aforementioned predictor variables and employee engagement (Avery et al., 2012; Handa & Gulati, 2014; Inceoglu & Warr, 2011; Kim et al., 2009). This block resulted in a significant finding of F(3, 92) = 12.595, p < .001 as confirmed by both the multiple regression model summary, as can be seen in Table 3, and the ANOVA result included in the omnibus SPSS output. The first model resulted in a multiple correlation coefficient of .540 (R = .540). The second model with the hierarchical block that included all predictor variables (with the exception of the personality omnibus score due to multicollinearity concerns) also 67
demonstrated a significant result of F(11,84) = 9.014, p < .001 according to the ANOVA result. The second model of the equation resulted in a multiple regression coefficient of .736 (R = .736). The remaining multiple regression model data for both hierarchical blocks can be seen in Table 3.
Table 3 Multiple Regression Model Statistics
Model Summary
Adjusted R R Square Square
Std. Error of the Estimate
R Square Change
F Change
df1
df2
Sig. F Change
Model
R
1
.540
.291
.268
15.555
.291
12.595
3
92