UNDERSTANDING RESILIENCY—THE RELATIONSHIP BETWEEN USAFA CADET GRIT-S SCORES and CADET DEVELOPMENT. by JUSTIN R. STODDARD

UNDERSTANDING RESILIENCY—THE RELATIONSHIP BETWEEN

USAFA CADET GRIT-S SCORES AND CADET DEVELOPMENT.

JUSTIN R. STODDARD

B.A., University of Colorado—Boulder, 2001

M.A., University of Texas—El Paso, 2011

A dissertation submitted to the Graduate Faculty of the

University of Colorado Colorado Springs

in partial fulfillment of the

requirements for the degree of

Doctor of Philosophy

Department of Education Leadership, Research, and Foundations

2019

Ó 2019

JUSTIN R. STODDARD

This dissertation for the Doctor of Philosophy degree by

Justin R. Stoddard

has been approved for the

Department of Education Leadership, Research, and Foundations

Andrea Bingham, Chair

Joseph Taylor

Patricia Witkowsky

Phillip Morris

Christopher Luedtke

____July 29, 2019___ Date

Stoddard, Justin R. (Ph.D., Education, Leadership, and Policy)

Understanding Resiliency—The Relationship Between USAFA Cadet Grit-S Scores and Cadet Development.

Dissertation directed by Assistant Professor Andrea Bingham.

ABSTRACT

Life is filled with adversity and no one is excluded from experiencing setbacks, failures, and even catastrophes. But within all people is the potential to learn and bounce back from these experiences. Resiliency theory describes the promotive and protective factors individuals can use to process learn from setbacks and failures and to grow and bounce back from the disruptive adversities that may otherwise break us down. Using the Grit-S survey, which consists of a passion subscore and a resilience subscore, this study focuses on understanding the nature of resilience by analyzing survey responses from over 5,400 cadets from the United States Air Force Academy spanning over nine years. The quantitative analysis of the associations between Grit-S score, passion subscore, and resilience subscore and over 30 different cadet variables including background, current performance, and participation in USAFA teams and programs provided greater understanding into the cadet characteristics that were associated with their grit variable scores. Significantly associated variables unique to the military learning environment at USAFA were discovered and included performance in leadership development, athletics, and participation in a variety of activities and programs which build real-world life skills and present cadets with realistic challenges. The results confirm the statistically and practically significant associations between the grit variables and both academic and non-academic performance in several areas and also suggest there is considerable value in experiencing challenging scenarios outside the classroom.

iii

DEDICATION

This work is dedicated to my father, who is one of the best examples of resilience

I have ever known. Despite setbacks, failures, and shortcomings, he simply never quits.

He has pushed forward his entire life and has experienced many successes, including his children.

ACKNOWLEDGEMENTS

Several individuals had a significant impact on this research study and guided me to complete this work, and I simply could not have done this without them. I am sincerely grateful to the following individuals for their participation and assistance.

Dr. Andrea Bingham served as my committee chair and asked the hard questions to give this study the correct focus. Dr. Joe Taylor provided essential assistance as my methodologist and helped me navigate SPSS. Dr. Christopher Luedtke from CCLD,

USAFA was my primary connection to USAFA and served as a committee member and

USAFA expert. Dr. Patricia Witkowsky and Dr. Phillip Morris served as committee members and provided important feedback that greatly added to the study. Gene Hagan and Mark Briody both worked to get exactly the data I needed from USAFA. Laura Neal guided me through the USAFA IRB process and made sure all the documentation was correct. LTC Jessica Sullivan and Kathy McHugh provided me with the essential cadet

Grit-S survey data. Dr. Terry McFarlane took the time to discuss cadet resiliency with me and attend a book signing with Dr. Angela Duckworth. Dr. Amanda Lords was a good friend and mentor and helped me structure my research project. Dr. John

Abbatiello and Richard Ramsey provided me with the details of the USAFA Officer

Development System and insight into the various aspects of CCLD programming. Karen

Howard and Paul Torchinsky from the VA Vocational Rehab program who made this degree possible. I especially thank the USAFA cadets for being the excellent individuals they are and for completing the surveys that informed this study. Finally, I am grateful to my wife Miriam who supported me through this study and to my Heavenly Father who gave me the inspiration to build and strengthen others.

TABLE OF CONTENTS

CHAPTER

I. INTRODUCTION ...... 1

Background of Resiliency Theory ...... 7

Purpose of the Study ...... 8

Research Questions ...... 9

Research Significance ...... 10

Definitions and Terms ...... 12

Dissertation Structure ...... 15

II. LITERATURE REVIEW ...... 17

Resiliency Theory ...... 17

Origins and Development ...... 18

Resiliency Framework ...... 22

Definitions of Resiliency ...... 28

Resiliency-Building Models ...... 30

Compensatory Model ...... 31

Protective Model ...... 32

Challenge Model ...... 33

Commonalities Between Models ...... 34

Strengths and Weaknesses of Resiliency Theory ...... 35

Review of Relevant Literature ...... 39

Leadership ...... 39

Resiliency in Higher Education ...... 42

The Challenges to Students in Higher Education ...... 42

Efforts to Help Students Overcome Challenges in Higher Education ...... 43

Results of Efforts to Help Students ...... 44

Resiliency in the Military ...... 45

The Challenges to Service Members in the Military ...... 46

Efforts to Help Service Members Overcome Challenges in the Military . 47

Results of Efforts to Help Service Members ...... 49

Resiliency at USAFA ...... 51

The Challenges to Cadets at USAFA ...... 51

Efforts to Help Cadets Overcome Challenges at USAFA ...... 53

Results of Efforts to Help Cadets ...... 58

Emerging Themes and Research Gaps ...... 60

III. METHOD ...... 63

Research Setting and Participants ...... 64

Measures ...... 66

Grit-S Survey Description ...... 66

Reliability and Validity ...... 67

Research Design ...... 69

Procedures ...... 69

Data Collection ...... 69

Variables ...... 73

Missing Values ...... 81

Data Analysis ...... 83

vii

RQ1 – Grit-S score vs Cadet Predictors ...... 83

RQ2 – Performance vs Grit-S Score ...... 86

RQ3 – Attrition vs Grit-S Score ...... 89

RQ4 – Changes in Grit-S Score Over Time vs Cadet Predictors ...... 92

RQ5 – Recommended Future Research, Policies, and Practices ...... 94

Limitations ...... 95

Ethics ...... 97

IV. RESULTS ...... 99

Review of Descriptives and Research Question Results ...... 100

Descriptives ...... 100

RQ1 – Grit-S Score vs Cadet Predictors ...... 103

Grit-S Score ...... 103

Passion Subscore ...... 110

Resilience Subscore ...... 116

RQ2 – Performance vs Grit-S Score ...... 122

Grade Point Average ...... 122

Military Performance Average ...... 127

Physical Education Average ...... 132

Overall Performance Average ...... 137

RQ3 – Attrition vs Grit-S Score ...... 142

Enrollment Status ...... 142

RQ4 – Changes in Grit-S Score Over Time vs Cadet Predictors ...... 148

Average Changes in Grit-S Score Variables ...... 148

viii

Grit-S Gain Score ...... 154

Passion Gain Score ...... 159

Resilience Gain Score ...... 164

Summary of Research Question Results ...... 169

V. DISCUSSION AND CONCLUSION ...... 174

Discussion ...... 175

Discussion of Results ...... 175

RQ1 – Grit-S Score vs Cadet Predictors ...... 175

RQ2 – Performance vs Grit-S Score ...... 184

RQ3 – Attrition vs Grit-S Score ...... 186

RQ4 – Changes in Grit-S Score Over Time vs Cadet Predictors ...... 187

Addressing Earlier Criticisms ...... 194

Implications ...... 196

Implication 1 – Academic and Non-Academic Variables are Associated with Grit ...... 197

Implication 2 – Attrition is not Associated with Grit ...... 201

Implication 3 – Resilience is Critical to Leadership Development ...... 203

Recommendations ...... 205

RQ5 – Recommendations for Future Research, Policy, and Practice ...... 205

Changes to Policies and Practices ...... 206

Future Research ...... 209

Conclusion ...... 211

Final Thoughts ...... 212

REFERENCES ...... 215

APPENDICES

A. USAFA INSTITUTIONAL REVIEW BOARD APPROVAL ...... 234

B. USAFA VOLUNTEER SERVICE AGREEMENT (VSA) ...... 236

C. GRIT-S SURVEY ...... 237

D. USAFA INTERCOLLEGIATE SPORTS ...... 240

E. USAFA INTRAMURAL SPORTS ...... 241

F. USAFA COMPETITIVE PROGRAMS ...... 242

G. USAFA MISSION SUPPORT PROGRAMS ...... 243

H. USAFA PROFESSIONAL PROGRAMS ...... 244

I. USAFA CLUB PROGRAMS ...... 245

J. USAFA RECREATIONAL PROGRAMS ...... 246

K. SPSS LINEAR REGRESSION OUTPUTS ...... 247

LIST OF TABLES

TABLE

1. Description of USAFA Grit-S Survey Collected from USAFA-A9 ...... 71

2. Dependent and Independent Variables Used in Regression Analysis ...... 73

3. Statistics of Variables with Imputed Values ...... 82

4. RQ1 Variables Used in Linear Regression Analysis ...... 85

5. RQ2 Variables Used in Linear Regression Analysis ...... 88

6. RQ3 Variables Used in Logistic Regression Analysis ...... 91

7. RQ4 Variables Used in Linear Regression Analysis ...... 93

8. Descriptive Statistics Comparing Grit Variables with Cadet Predictors ...... 101

9. Model 1.1a Regression Results Comparing Grit-S Scores with Cadet Predictors ...... 104

10. Model 1.1b Regression Results Comparing Grit-S Scores with Cadet Predictors ...... 108

11. Model 1.2a Regression Results Comparing Passion Subscores with Cadet Predictors ...... 111

12. Model 1.2b Regression Results Comparing Passion Subscores with Cadet Predictors ...... 114

13. Model 1.3a Regression Results Comparing Resilience Subscores with Cadet Predictors ...... 117

14. Model 1.3b Regression Results Comparing Resilience Subscores with Cadet Predictors ...... 120

15. Model 2.1a Regression Results Comparing GPA with Cadet Predictors ...... 124

16. Model 2.1b Regression Results Comparing GPA with Cadet Predictors ...... 126

17. Model 2.2a Regression Results Comparing MPA with Cadet Predictors ...... 129

18. Model 2.2b Regression Results Comparing MPA with Cadet Predictors ...... 131

19. Model 2.3a Regression Results Comparing PEA with Cadet Predictors ...... 134

20. Model 2.3b Regression Results Comparing PEA with Cadet Predictors ...... 136

21. Model 2.4a Regression Results Comparing OPA with Cadet Predictors ...... 139

22. Model 2.4b Regression Results Comparing OPA with Cadet Predictors ...... 141

23. Model 3.1a Logistic Regression Coefficients Comparing Enrollment Status with Cadet Predictors ...... 144

24. Model 3.1b Logistic Regression Coefficients Comparing Enrollment Status with Cadet Predictors ...... 146

25. Descriptive Statistics Comparing Grit-S, Passion, and Resilience Gain Scores with Cadet Predictors ...... 153

26. Model 4.1a Regression Results Comparing Grit-S Gain Scores with Cadet Predictors ...... 156

27. Model 4.1b Regression Results Comparing Grit-S Gain Scores with Cadet Predictors ...... 158

28. Model 4.2a Regression Results Comparing Passion Gain Scores with Cadet Predictors ...... 161

29. Model 4.2b Regression Results Comparing Passion Gain Scores with Cadet Predictors ...... 163

30. Model 4.3a Regression Results Comparing Resilience Gain Scores with Cadet Predictors ...... 166

31. Model 4.3b Regression Results Comparing Resilience Gain Scores with Cadet Predictors ...... 168

32. Statistically Significant Predictors Associated with Grit Variables ...... 177

33. Statistically Significant Predictors Associated with Grit Gain Score Variables . 188

xii

LIST OF FIGURES

FIGURE

1. Richardson Resiliency Model ...... 26

2. A Comparison of Resiliency Definitions ...... 29

3. Grit-S, Passion, and Resilience Score Trends for Group One ...... 149

4. Grit-S, Passion, and Resilience Score Trends for Group Two ...... 150

5. Grit-S, Passion, and Resilience Score Trends for Group Three ...... 151

xiii

LIST OF ABBREVIATIONS

ABBREVIATIONS

ACES Academy character enrichment seminar AOC Air officer commanding AFSC Air Force Specialty Code AMT Academy military trainer ARSOF Army Special Operations Forces BCT Basic cadet training (at USAFA) CAF Comprehensive airmen fitness CBT Cadet basic training (at USMA) CCLD Center for Character and Leadership Development CSF Comprehensive soldier fitness CSF2 Comprehensive soldier and family fitness DAC Department of the Army civilian DCoE Defense Centers of Excellence of Psychological and Traumatic Brain Injury DRS Dispositional resilience scale DSAT Designated survey and assessment time GAT Global assessment tool GPA Grade point average IRB Institutional review board LIFT Leaders in flight today MPA Military performance average ODS Officer development system OML Order of merit list OPA Overall performance average PEA Physical education average PRP Penn Resiliency Project PTSD Post-traumatic stress disorder R&R Respect and responsibility SERE Survive escape resist evade SES Socioeconomic status SRI Support and resilience inventory STEM Science, Technology, Engineering, Mathematics USAFA United States Air Force Academy USAF United States Air Force USMA United States Military Academy at West Point VECTOR Vital effective character through observation and reflection

xiv

CHAPTER I

INTRODUCTION

Everyone experiences adversity. Adversity is defined as “hard times; a difficult situation or condition; a state or instance of serious or continued difficulty or misfortune”

(Adversity, 2018), and it affects virtually every aspect of life. Adversity often challenges individuals by placing difficult circumstances between them and the goals they want to achieve. As a result, people either learn to confront and grow from adversity or are overwhelmed and overcome by it. Adversity comes from a variety of sources both natural and man-made, and some of the greatest adversities are those presented when countries are at war. The United States has been in a constant state of war since the events of 9/11, and from that time until 2015, 2.77 million service members have answered the call to fight and serve their country on over 5.4 million deployments

(Wenger, O’Connell, & Cottrell, 2018). Contemporary warfighting in environments like

Iraq, Afghanistan, and Syria is difficult and dangerous, and under these arduous conditions, even the most well-prepared, best-trained, and highly experienced individuals are subject to mental fatigue and failure. Being resilient in these environments “is distinct from mere survival, and more than mere endurance. Resilience is often endurance with direction” (Greitens, 2015, p. 25). While it is difficult to imagine the challenges and adversities faced by military service members without actually walking in their boots, it is not difficult to understand the need to support, train, and prepare service members to effectively confront adversity before they head into harm’s way.

Considering the many deployments and austere conditions military men and women serve in, understanding the nature and development of individual resilience is

both essential and critical to mission success since it is their resilience that will enable them to overcome their adversities. However, exactly how to develop individual resilience, and how to measure it is a renewed field of study garnering much attention and research. An important aspect of this resiliency research is understanding the relationship between resiliency theory and practices to build resiliency so that military academies tasked with developing young military leaders can have the information necessary to develop the policies and programs designed to develop individual resiliency.

At the United States Air Force Academy (USAFA), the development of resiliency in cadets is a high priority. Exhibiting resiliency in connection with developing grit is a specific proficiency that cadets are expected to improve as they learn to cultivate “. . . a hardiness of spirit and resistance to accept failure despite physical and mental hardships”

(The United States Air Force Academy, 2016c, p. 1). Dr. Duckworth, a prominent researcher on the subject of grit has stated that resiliency is one of the two key components of grit and is also called perseverance of effort (Perkins-Gough, 2013). At

USAFA, transforming theories into operational policies and programs is essential to the goal of developing leaders of character who live honorably and embody the United States

Air Force (USAF) values of “Integrity first, Service Before Self, Excellence in All We

Do” (The United States Air Force Academy, 2016a). Training USAFA cadets to develop their individual resiliency and effectively face and overcome their own adversities contributes to their warrior ethos and is of paramount importance, since warrior leaders must not only face adversities themselves, but also effectively lead their units through those adversities.

Both by nature and design, the atmosphere and environment faced by cadets attending USAFA exposes them to a variety of adverse conditions on a regular basis.

USAFA cadets, among other military service cadets, represent future USAF leaders and will provide for the future safety and security of American lives and prosperity. From their first days of entering USAFA to the final days of graduation with the Air Force

Thunderbirds screaming overhead, USAFA cadets experience a unique environment of adversity which challenges them and attempts to develop their leadership skills and abilities. While this adversity-rich environment provides an exciting and challenging developmental adventure for many, it also represents an inhospitable and crushing environment for some who fall beneath acceptable standards, succumb to the academic pressures, and leave the academy either voluntarily or involuntarily.

Despite access to a variety of resources designed to support their successful graduation from USAFA, overall cadet attrition rates have averaged 24.4% over the last 5 years (Zubeck, 2018), compared to an average 35% for other universities in Colorado

(CollegeFactual, 2018a). This is lower than the 28% attrition rate at the United States

Military Academy at West Point (USMA) (CollegeFactual, 2018b) and higher than the

14% attrition rate at the U.S. Naval Academy (Prudente, 2014), but there is more at stake than mere comparisons. A recent report of 1,669 colleges and universities across the nation reported lost revenues close to $16.5 billion with the average school losing almost

$10 million (Raisman, 2013) due to attrition. While military academies are paid for with government budgets and tax dollars instead of receiving revenue from students, the financial loss resulting from attrition is no less a concern. A desire to decrease attrition rates while maintaining standards is of particular interest to USAFA since the Air Force

cost per graduate in 2015 was over $534K and financial losses due to attrition prior to graduation are expensive (Zubeck, 2018). However, USAFA has an even larger and more pressing interest in lowering attrition rates to help fill the large shortage of pilots currently faced by the United States Air Force (Losey, 2018a, 2018b). These challenges of minimizing financial losses and producing able and ready officers to fill much-needed pilot slots highlight the importance of reducing attrition rates at USAFA while maintaining high standards. The hope is that as cadet resiliency increases, so will their levels of grit along with their ability to rebuff failure and their ability to continue working to overcome physical and mental challenges. Efforts to find effective methods to help cadets become more successful have included the exploration of cadet resiliency and overall grit to increase cadet performance and decrease attrition despite the adversity inherent to USAFA.

This in part has been the mission and focus for USAFA’s Center for Character and Leadership Development (CCLD), to develop and graduate officers who “Live honorably consistently practicing the virtues embodied in the Air Force Core Values, Lift others to be their best possible selves, and Elevate performance toward a common and noble purpose” (The United States Air Force Academy, 2016a). CCLD is organized into four main divisions consisting of Honor, Cadet Development, Scholarship and

Innovation, and Events, and the joint focus is to improve the overall character and leadership development of USAFA cadets and produce leaders of character. The scholarship and innovation department does this is by conducting scholarly research on the development of leadership skills and strong moral character to inform the

development of policies and programs to improve cadet development. This study aids in this effort by researching the nature of cadet resiliency and its connection to performance.

The necessity for this study stemmed from two primary needs: the need to expand and clarify the relationship between resiliency and individual performance factors and the need to understand how USAFA cadet resiliency changes during their educational careers at USAFA and what influences those changes. First, a meta-analysis of 73 studies on grit and its subcomponents of passion and resilience found the relationship between grit and performance was modest at best (Credé, Tynan, & Harms, 2017). The sample sizes of the studies specifically comparing grit with academic performance ranged from 21 to

1,218 and averaged only 351 participants. Conversely, research studies at The United

States Military Academy at West Point (USMA) to understand the connection between grit and performance used one-time grit surveys to take snapshots of two groups of cadets

(N = 1,218 and 1,308) participating in Cadet Basic Training (CBT) (Duckworth,

Peterson, Matthews, & Kelly, 2007) and did find a close relationship between grit and performance. A similar study was also done to understand the hardiness, synonymous with resilience, of 2,383 USMA cadets just beginning their academic careers (Bartone,

Kelly, & Matthews, 2013) and also found close associations between grit and resilience and performance factors. This presents the need to expand and clarify the relationship between grit, passion, and resilience with individual performance factors using a larger sample size and to examine the changes in grit over time instead of a one-time snapshot.

This study addresses these needs by analyzing over 6,900 survey results completed by over 5,400 USAFA cadets over a nine-year period.

Second, among the desired outcomes established by USAFA to guide the development of cadets is an objective to instill in cadets a proficiency to “Exhibit grit: a hardiness of spirit and resistance to accept failure despite physical and mental hardships” as one of the ways to help build Warrior Ethos (The United States Air Force Academy,

2016c). The academic experience at USAFA and its programs are designed to present cadets with challenges to navigate through, with the intent of increasing their resiliency and ability to adapt and overcome. This presents the need to evaluate and understand the extent to which these programs are influencing cadet resiliency and how cadet resiliency is changing over time so policies and programs can be modified to continually improve resiliency development. This research study addresses this need by providing a quantitative analysis of cadet characteristics, including demographics, performance measures, attrition data, and participation in various clubs and programs to understand which characteristics and programs were most associated with resiliency development and how they affected cadet resiliency over time.

This study completed a thorough quantitative analysis of the data using several regression models designed to explore the data and identify the primary associations with

Grit-S score and its subscores. This study also includes recommendations for policy and program development, practices to develop resiliency, and future research opportunities to continue the exploration of resiliency theory. The results from this study serve as a baseline of information from which USAFA can conduct future studies to explore the changes and trends in the resiliency development of cadets. It also explores changes in

USAFA cadet resiliency using a larger sample population than previously available to understand the resiliency of cadets more fully at one of the five military academies in the

United States. This builds upon prior resiliency research, including studies conducted at

USMA, and expands the understanding of individual resiliency adding to the body of knowledge currently available about resiliency theory and overcoming adversity.

Background of Resiliency Theory

Researchers commonly describe resiliency as the ability of an individual to bounce back and recover after facing adversity (Dyer & McGuinness, 1996). It has been referred to as perseverance of effort (Duckworth et al., 2007; Perkins-Gough, 2013), hardiness (Bartone, 2006; Bartone, Hystad, Eid, & Brevik, 2012), and it is understood to be more accurately described as a process instead of a static characteristic or trait

(Richardson, 2002; Rutter, 1987). In Grit: The Power of Passion and Perseverance, Dr.

Duckworth (2007) defines Grit as a combination of two key subcomponents: consistency of interest, which is also called passion, and perseverance of effort (Duckworth, 2016), which Duckworth later explains as being synonymous with resiliency (Duckworth, 2016;

Perkins-Gough, 2013). Duckworth’s research included the development of a research tool called the Grit Survey, which measured both passion and resiliency to understand the grit levels of individuals including USMA cadets and participants in a national spelling bee (Duckworth, 2016; Duckworth et al., 2007). Later, Dr. Duckworth’s team developed a shorter, more reliable, and more accurate version of the Grit Survey called the Grit-S

Survey (Duckworth & Quinn, 2009). Since 2009, USAFA has used the Grit-S survey to collect data from cadets on their individual resiliency and grit levels.

While research exists exploring the need and value of resiliency in areas such as higher education and doctoral programs (Allan, McKenna, & Dominey, 2014; Cross,

2014; Garza, Bain, & Kupczynski, 2014; Morales, 2014), military leaders experiencing

deployments (Meredith et al., 2011; Reivich, Seligman, & McBride, 2011; VanBreda,

2001), and the grit of military cadets experiencing adversity at USMA (Duckworth et al.,

2007), research directly connecting the resiliency subscale of grit and resiliency theory to character and leadership development programs is limited. Military academies provide a unique research environment that combines higher education with military training.

USAFA has gathered survey data over the past several years using a variety of resiliency- related survey tools. However, this data remained unexamined or only initially explored.

This presented a unique, data-rich research environment to understand resiliency theory as it relates directly to the cadets at USAFA.

Purpose of the Study

The purpose of this study was to understand the relationship between USAFA cadet resiliency and a variety of individual characteristics using quantitative regression analyses to determine the extent to which individual characteristics were associated with individual resiliency scores. The intent was to expand the work of previous studies by analyzing a sample population much larger than typically used in resiliency research in order to explore a greater number of individual characteristics and variables possibly associated with individual resiliency. This research adds to the body of knowledge on resiliency theory and provides clarity to prior research which overall has reported mixed results as to the relationship between resiliency and performance. Understanding this enables USAFA leadership and researchers to connect resiliency theory to practice and show which characteristics are the most highly associated with cadet resiliency and deserve additional future research priority. It may also enable decision-makers to develop or modify policies and programs to strengthen or bolster resiliency-weakening

factors, capitalize and improve on resiliency-building factors, and bring awareness to the factors affecting cadet resiliency overall. This increased understanding may provide

USAFA leadership the knowledge upon which to improve the overall resiliency of cadets and the units they serve in with the desired outcome of producing a more resilient and grittier fighting force to protect this nation. Regardless of the nature of results, this research serves as a baseline of information and a precursor to future research more closely examining the programs USAFA uses to build cadet resiliency.

Research Questions

The overall research question for this study asked whether there is a relationship between USAFA Cadet Grit-S scores and cadet development. The intent was to answer this question by proposing five specific research questions:

1. To what extent are cadet characteristics, including demographics and

participation in USAFA clubs and programs, associated with Grit-S score?

2. To what extent is Grit-S score associated with cadet overall performance

average (OPA)?

3. To what extent is Grit-S score associated with cadet attrition?

4. What is the change in cadet Grit-S score over time?

5. Based on this research study, what are some recommendations for future

research, policy development, and practice to build cadet resiliency at

USAFA?

The intent of these research questions was to explore the development of USAFA cadet resiliency and to understand the extent to which individual factors affected, were associated with, and predicted individual resiliency. This study analyzed these questions

through a quantitative analysis of cadet surveys, and the findings will serve as the basis for identifying program modifications and/or development, as well as determining opportunities for future research. Exploring these questions also provides a deeper understanding of how cadet resiliency is built and developed at USAFA and identifies possible methods of improving programs designed to develop leaders of character.

Research Significance

The significance of this research is at least threefold: it adds to the body of knowledge on the value of non-cognitive factors by clarifying the relationship between resiliency and academic performance, it identifies principles that may increase the effectiveness of military leaders and their development, and it takes advantage of nine years of survey results obtained by USAFA.

First, this study analyzes the relationship between resiliency and individual characteristics and performance factors using a sample population that is both larger than previous studies on similar topics and that examines the changes to resiliency over time instead of a one-time snapshot. This unique research study adds to the global debate discussing resiliency and the development of individual abilities and behaviors that may lead to improved performance and increased success and achievement. It also serves as a firm starting point for future studies delving deeper into specific programs and activities that increase an individual’s ability to persevere through challenging obstacles.

Second, this study seeks to identify and confirm the key principles central to resiliency theory and necessary in the development of cadet resilience at USAFA. The intent is to identify and develop outcomes that military leaders, professionals, educators, and even parents may use to help both individuals and organizations better understand

how to overcome adversity regardless of the environment and how to adapt to the ever- changing global and social challenges people face.

Finally, this study takes advantage of the time and resources spent by USAFA over at least nine years conducting cadet surveys at various times throughout the cadets’ academic careers. The time and resources applied by USAFA to understand cadet resiliency are put to work by analyzing the data, uncovering the valuable information it contains, and presenting the results in a way that benefits both USAFA, the military community, and higher education at large. The goal is to identify outcomes that increase the ability of USAFA and other higher education institutions to develop resilient leaders of character regardless of their future career paths.

Direct beneficiaries of this research include the USAFA Superintendent and

Director of CCLD along with their leadership teams as they continually develop and refine the programs and activities designed to develop cadets into leaders of character.

The list of beneficiaries also includes the leadership teams of the Dean of Faculty,

Commandant of Cadets, and Athletic Director, the professors and coaches at USAFA, the cadre of active duty officers and enlisted leaders who provide direct leadership development for USAFA cadets, and finally the cadets themselves. This research also has the potential to provide fundamental substantive principles to guide programs and policies to bring about the development of more adaptive, capable, and resilient cadets ready to face the global military challenges.

Definitions and Terms

The following definitions provide clarity for the key terms related to this study as several of the terms have a variety of connotations, interpretations, and implications depending on the audience and specialty field. The intent of these definitions is to clarify their meaning and operationalize each term for use in this study.

Adversity refers to “hard times; a difficult situation or condition; a state or instance of serious or continued difficulty or misfortune” (Adversity, 2018) and includes the various risk factors and obstacles making it difficult to achieve one’s goals.

Grit is defined as the passion and perseverance to achieve long-term goals (Duckworth,

2016).

Grit-S Survey refers to the 8-item survey developed by Dr. Angela Duckworth containing two subscales measuring passion and perseverance.

Resilience is defined as “The ability to withstand, recover and grow in the face of stressors and changing demands” (U.S. Department of the Air Force, 2014, p. 14).

Resilience and resiliency are synonymous with other terms like hardiness and perseverance but are not synonymous with the term “grit.”

Resiliency theory refers to the theoretical process an individual goes through as they face adversity, are influenced by both stabilizing and destabilizing factors, and reintegrate with either a new or unchanged level of resiliency.

Self-efficacy is “the belief in one’s capabilities to organize and execute courses of action required to produce given attainments” (Bandura, 1997, p. 3).

Mindset refers to the individual tendency to adopt either the fixed viewpoint, where value is placed on task completion, or the growth viewpoint, where value is placed on individual effort and the learning process (Dweck, 2006).

Promotive factors consisting of internal traits and abilities and external resources or the elements to which the individual can turn for assistance (Fergus & Zimmerman, 2005;

Kiswarday, 2012; Zimmerman, 2013; Zimmerman et al., 2013).

Protective factors include supportive family networks, socioeconomic status, school experiences, supportive communities, and cultural resources (Fleming & Ledogar, 2008) and serve to counteract or ameliorate the effects of adversity and guard the individual from adversity (Braverman, 2001).

Vulnerability factors represent the qualities, experiences, or lack of protective factors that both make individuals more susceptible to adversity and intensify the risk effects

(Braverman, 2001; Fleming & Ledogar, 2008; Luthar, Cicchetti, & Becker, 2000).

Bio-psycho-spiritual-homeostasis is a relatively stable emotional state an individual is in prior to experiencing a difficult situation or an adversity (Richardson, 2002).

Disruption refers to the moment when adversity or adverse conditions impact the homeostasis of an individual and cause the resiliency process to begin.

Reintegration refers to the process by which an individual attempts to regain homeostasis after experiencing disruption.

Compensatory model refers to the outcome-focused resiliency-building model which seeks to strengthen or augment an individual’s promotive factors and ability to employ them in order to better overcome the effects of adversity.

Protective model refers to the risk-focused resiliency-building model that focuses on mitigating the risks leading to negative outcomes in order to prevent or avoid the adversity altogether.

Challenge model refers to the adversity-focused resiliency-building model wherein individuals are intentionally exposed to moderate amounts of adversity in controlled environments to build an immunity to the risk (Garmezy, Masten, & Tellegen, 1984).

Attrition refers to the loss of students from an academic institution for any reason.

Warrior Ethos is defined as “the embodiment of the warrior spirit: tough mindedness, tireless motivation, an unceasing vigilance, a willingness to sacrifice one’s life for the country, if necessary, and a commitment to be the world’s premier air, space and cyberspace force” (The United States Air Force Academy, 2016c, p. 1).

Comprehensive soldier and family fitness (CSF2) program is a program initially launched in 2008 and relaunched in 2014 “to increase the resilience and enhance the performance of Soldiers, Families, and DACs,” referring to Department of the Army

Civilians (U.S. Department of the Army, 2014, para 1-5a). The CSF2 program identifies five dimensions of strength to create its primary conceptual pillars of physical, emotional, social, spiritual, and family resiliency to build the overall resiliency of soldiers (U.S.

Department of the Army, 2014).

Bonferroni adjustment is an adjustment made during multiple regression analysis in order to adjust the alpha level for multiple comparisons.

Dissertation Structure

I organized the five chapters of this dissertation to provide a thorough review of the purpose, background, and literature related to the concepts inherent to resiliency theory and the connection with leadership development. The methodology provides the analytical underpinnings of this study and describes the framework used to analyze the data. The findings inform the research questions and provide the primary discoveries of this research study, and a discussion follows to highlight the importance of the research and its contribution to the reviewed literature. In Chapter 1, I introduce the focus and intent of the research study, give a brief background of resiliency, and discuss my plan to complete the quantitative analysis to understand the relationship between Grit-S survey scores and cadet characteristics. I also provide a list of definitions and terms to guide the reader, along with an explanation of how I organized this dissertation.

In Chapter 2, I review the literature relevant to this study beginning with the origin, framework, definitions, and debated strengths and weaknesses of resiliency theory followed by a discussion of the key characteristics of leadership. Following this is a discussion of the challenges and research related to individual resiliency in both higher education and the military along with the efforts in each to build resiliency. Then I provide a review of the resiliency-developing elements of the character and leadership development programs used by USAFA, concluding with a discussion of the underlying themes and literature gaps used to develop the research questions for this study.

In Chapter 3, I provide the methodology for the study and a review of the nature of USAFA cadets and the resiliency surveys they have historically completed. I then set the stage for the quantitative analysis of survey scores and their association with

individual cadet factors. Following this, I present the data analysis of several groups of

USAFA cadets who took the Grit-S survey from 2009 to 2017 and highlight the several models of regression analyses I utilized to identify how cadet demographic predictors, performance measures, attrition rates, and participation in USAFA clubs and programs are associated with Grit-S survey scores.

In Chapter 4, I present the research findings by discussing each of the research questions and their related findings. I also present and review additional findings discovered during the analysis phase of the study along with a discussion of their relative importance and impact on the overall research study.

In Chapter 5, I present a broader discussion of the research study results in the context of known resiliency and resiliency development literature along with a discussion of the implications and recommendations for policy and/or program modifications. I also present a review of recommended future research strategies before concluding the research study with some personal anecdotal reflections.

CHAPTER II

LITERATURE REVIEW

The purpose of the literature review is to provide a detailed description of resiliency theory and the models typically used to try to build resiliency on an individual level. The review begins with a thorough examination of the origin, historical development, essential components, and current definitions of resiliency theory. An examination of the relevant literature follows by first reviewing the key characteristics of leadership and providing a foundation upon which to discuss the application of resiliency theory in higher education and the military. This leads to a discussion of cadet resiliency development at USAFA and the efforts taken to build resilient leadership within the cadet population. A review of the existing literature on these concepts includes discussions of the relevant strengths and weaknesses in each area and justifies the need for this study. I conclude by reviewing and discussing the gaps identified in the literature and presenting them as the foundation for my research questions for this study.

Resiliency Theory

This study uses resiliency theory to analyze and understand the processes individuals go through to operationalize their own resiliency and overcome obstacles when faced with adversity. Resiliency theory has changed over the years since its initial development in 1974 as “stress-resistance” in research conducted by Dr. Norman

Garmezy and Dr. Ann Masten (Garmezy, 1974). Definitions have shifted from focusing on static trait-based theories to more dynamic process-based theories, and researchers continue efforts to measure resiliency, predict it, and explain its effect on other aspects of behavior and performance. The focus of most research studies on resiliency is to identify

ways to maximize and predict human performance. This study looks at resiliency challenges in higher education, the military, and more specifically in the leadership development programs at USAFA.

Origins and Development

Research on individual resilience began in the early 1970s with the work of

Norman Garmezy and his investigation into children at risk for severe psychopathology

(Garmezy, 1974). His later work with Ann Masten focused on children raised in severely adverse circumstances and why some of them grew up with serious behavioral challenges while others seemed unaffected or even strengthened by their experiences, demonstrating a kind of stress competence and stress-resistance later termed “resilience” (Garmezy &

Masten, 1986). Early research, including that of Garmezy and Masten (1986) initially defined resilient behavior in terms of traits and characteristics that helped individuals overcome stressors (Fletcher & Sarkar, 2013). This trait-based theory remained the prevailing philosophy until it was later replaced by a concept of dynamic processes of positive adaptation whereby an individual demonstrates resilient behavior and overcomes adversity (Fleming & Ledogar, 2008; Luthar et al., 2000). This new understanding led to describing resiliency as a dynamic process of reacting to and engaging adversity in order to regain a homeostatic state that may be better, worse, or the same as when the adversity began (Richardson, 2002). This new understanding of resiliency theory, and the movement from a static trait-based theory to a process-based theory, led researchers to explore the factors that might lead individuals to develop resiliency, how resiliency influenced performance, how to measure resiliency, and whether it was possible or not to develop resiliency in different populations.

Contemporary researchers continue to explore the origins and precursors to resiliency, ways to accurately measure resiliency, and methods to develop resiliency.

Two prominent and contemporary resiliency researchers are Dr. Paul Bartone and Dr.

Angela Duckworth. Dr. Bartone is a retired Army Colonel currently working as a senior research fellow at the National Defense University and has written extensively about concepts surrounding the human ability to overcome adversity, particularly in the military

(Bartone, 2006; Bartone, Eid, Johnsen, Laberg, & Snook, 2009; Bartone et al., 2013;

Bartone, Roland, Picano, & Williams, 2008). Dr. Bartone has also dedicated a great deal of research to the study of resiliency and hardiness by creating, evaluating, and redesigning surveys and studies to measure the levels of hardiness of military personnel with results demonstrating the value of resiliency both in leaders and in those they lead

(Bartone, 2006; Bartone et al., 2013; Bartone et al., 2008).

Dr. Bartone developed a 15-item hardiness scale called the Dispositional

Resilience Scale (DRS), which has been demonstrated a reliable and valid measurement tool to evaluate the hardiness or resilience of a variety of individuals, especially soldiers and military leaders (Bartone, 1995, 2007). The DRS has been used to measure and understand the resilience of military personnel and is one of the surveys offered to cadets at USAFA (Bartone, 1995, 2007). Together, these studies highlight the importance of resiliency, especially in the military, to deal with the stresses of deployments, rigors of military life, and the risk factors associated with combat. It is notable that the terms

“hardiness” and “resilience” are often used interchangeably, showing the close and, in most cases, synonymous use of these terms.

Resiliency is one of the two key components of grit as explained by Dr. Angela

Duckworth, a professor at the University of Pennsylvania and author of the popular book

Grit: The power of passion and perseverance. Grit consists of the combination of two key subcomponents: consistency of interest referred to as passion and perseverance of effort simply referred to as perseverance (Duckworth, 2016), which Dr. Duckworth describes as being synonymous with resiliency (Perkins-Gough, 2013). It is important to note that while resiliency is synonymous with perseverance, it is not synonymous with grit. As grit is a combination of both passion and perseverance or resiliency, resiliency is a subcomponent of grit. This important distinction explains why the research study targeted both the passion of an individual to maintain an interest in a subject and their resiliency to persevere through obstacles and setbacks to achieve goals. Studies at the

United States Military Academy at West Point (USMA) show how individual grit—i.e., the passion and perseverance or resiliency of West Point cadets—enabled them to overcome extreme challenges and effectively predicted their successful graduation from

USMA (Duckworth et al., 2007).

Focusing on the resiliency subcomponent of grit, one USMA cadet described the arduous conditions imposed during their initial cadet basic training by stating, “Within two weeks I was tired, lonely, frustrated, and ready to quit—as were all of my classmates” (Duckworth, 2016, p. 25). However, as the author goes on to explain, while some quit, others did not, and the key difference seemed to be, “… a ‘never give up’ attitude” (Duckworth, 2016, p. 26). Resiliency represents the determination to keep trying, keep pushing forward, and never give up despite the many obstacles or adversities that may stand in the way. This quality is particularly valuable in the military where

individuals face constant adversities on a routine basis. Despite this, little is known about the accuracy or effectiveness of known resiliency measures, how to develop individual resiliency, or how to accurately predict whether an individual possesses resiliency or not.

Dr. Duckworth created a survey tool called the Grit Survey containing two subscales to measure the elements of passion and resiliency. This survey tool was used to predict student success, including that of USMA cadets attending basic training and students participating in the National Spelling Bee (Duckworth et al., 2007; Duckworth &

Quinn, 2009). Dr. Duckworth developed several versions of the Grit Survey, including the latest eight-item Short Grit survey, known as the Grit-S survey, which demonstrated increased improved psychometric properties, test-retest reliability, and predictive validity

(Duckworth & Quinn, 2009). This test proved accurate in a variety of settings and began to reveal how the combined effects of passion and resilience influenced the outcomes of challenges experienced by the individual. These and other related studies led to the claim that “To be gritty is to keep putting one foot in front of the other . . . to hold fast to an interesting and purposeful goal . . . to invest, day after week after year, in challenging practice . . . to fall down seven times, and rise eight” (Duckworth, 2016, p. 275).

Understanding resiliency is a critical first step to understanding how to change and improve individual resilience, especially in the adversity-filled profession of serving in the military. However, before moving to the application of resiliency theory, it is essential to understand the key elements that make up the framework of this theory and how these elements operate in relation to each other.

Resiliency Framework

Resiliency theory is a strengths-based theory that, instead of focusing on deficits, focuses on individual strengths that lead to healthy development and positive outcomes regardless of the level of risk exposure (Fergus & Zimmerman, 2005; Zimmerman,

2013). Resiliency theory involves the concepts of adversity, the promotive factors that include internal assets and external resources, and protective and vulnerability factors which all influence resiliency in various ways. Resiliency theory posits that individuals experience risks or adversities in life and possess promotive factors of varying types and degrees that may support their ability to overcome the adversities (Fergus & Zimmerman,

2005; Zimmerman, 2013). Promotive factors consist of the internal assets and external resources an individual has the knowledge and ability to employ when attempting to overcome adversity. Internal assets include traits inherent to the individual, such as positive identity, competence, hope, self-esteem, self-efficacy, coping skills (Fergus &

Zimmerman, 2005), mindset (Dweck, 2008), and mindfulness (Duckworth, 2016). These internal assets focus on the inherent positive strengths of the individual that “promote” successful resilience when disrupted by risk and adversity. External assets are the resources available to the individual to which they can turn for assistance and include parental support, youth programs, and adult mentors (Fergus & Zimmerman, 2005;

Kiswarday, 2012; Zimmerman, 2013; Zimmerman et al., 2013). They are those elements external to the individual but that the individual can choose to pursue if they decide to.

A keen sense of self-efficacy, fostering a growth mindset, and developing mindfulness are also important internal assets that contribute to resilient behavior. Initial research defined self-efficacy as “the belief in one’s capabilities to organize and execute

courses of action required to produce given attainments” (Bandura, 1997, p. 3).

Researchers working with military personnel define self-efficacy as “the belief that you can master your environment and effectively solve problems as they arise” and note its essential role in developing resilience (Reivich & Shatté, 2003, p. 19). Studies have shown that self-efficacy independently predicts resilience (Collishaw et al., 2016), strengthens resilience to challenging experiences (Bandura, 1997), and is closely related to academic performance (Chemers, Hu, & Garcia, 2001; Zajacova, Lynch, &

Espenshade, 2005). Self-efficacy plays a key role in developing resiliency because

“resilience requires some degree of belief in the ability to exert control over the social environment” (Bender & Ingram, 2018, p. 20), and persevering through challenging obstacles requires resilient self-efficacy to manage failure (Bandura, 2012). Overcoming obstacles starts with the internal belief that a person can make decisions, take action, and bring about effects resulting in overcoming obstacles and challenges in a way that brings positive growth. However, without the proper mindset, even self-efficacy can be a weak contributor to a resilient attitude.

Mindset can also be a determining factor in the development of resiliency. The idea behind mindset is that people typically adopt either a fixed viewpoint, wherein intelligence, personality, and character are considered static traits with value placed on the individual achievement of tasks, or a growth viewpoint, wherein intelligence, personality, and character is malleable, gained through both success and failure in completing tasks, with value placed on individual effort and the learning process (Dweck,

2006; Hochanadel & Finamore, 2015; O’Brien & Lomas, 2017). Research demonstrates that individuals’ mindsets can change and develop and that as they do, their resilience

increases (Yeager & Dweck, 2012). Researchers studying different groups in an outdoor personal development program found “only the group who engaged with structured processing specific to overcoming challenge with support, application of effort, and highlighting transferable strategies for overcoming setbacks showed a significant change in their resilience levels” (O’Brien & Lomas, 2017, p. 142). Their research supported the conclusion that “interweaving a structured and focused intervention . . . can have a significant impact on participants’ mindset as well as increasing their resilience”

(O’Brien & Lomas, 2017, p. 144). This suggests mindset has a direct influence on an individual’s ability to experience self-efficacy and engage in the types of behaviors that would enable self-improvement and the development of resilient behaviors.

Mindfulness can be an important element in developing resilience as it involves an awareness of the present as a way to understand, evaluate, and deliberately move forward. Researchers describe mindfulness as “a process of regulating attention in order to bring a quality of nonelaborative awareness to current experience and a quality of relating to one’s experience within an orientation of curiosity, experiential openness, and acceptance” (Bishop et al., 2004, p. 234) and an “awareness that emerges through paying attention on purpose, in the present moment, and nonjudgmentally to the unfolding of experience moment by moment” (Kabat‐Zinn, 2003, p. 145). One researcher described the experiences of an accomplished doctor who began developing mindfulness through meditation as a young man and who was later able to pass along what he had learned to patients with serious health problems (Duckworth, 2016, pp. 155-157). This ability to focus attention on each moment with openness and acceptance can empower individuals to experience hardship in a way that enables them to identify, acknowledge, and accept

their own abilities to resolve challenges and provide a path to obtaining new skills to handle the challenges ahead. Studies related to Mindfulness-Based Stress Reduction

(MBSR) strategies have found that using MBSR as an intervention helped individuals experiencing stress in a broad range of situations ranging from everyday living to extremely challenging circumstances involving severe disorders (Grossman, Niemann,

Schmidt, & Walach, 2004). Several mindfulness programs have been proposed to build resiliency among military servicemembers (Thomas & Taylor, 2015), and research shows that mindfulness training programs can help individuals learn to mitigate the negative effects of stress (Johnson et al., 2014). Thus, the research shows that self-efficacy, mindset, and mindfulness can be significant internal asset promotive factors which influence both the development and demonstration of resiliency as they affect how an individual values growth and believes that they can, in fact, bounce back from setbacks.

Protective and vulnerability factors are elements that may influence the effects of adversity depending on the individual and their background (Braverman, 2001; Fleming

& Ledogar, 2008; Luthar & Cicchetti, 2000; Luthar et al., 2000). Protective factors are sometimes closely related to promotive factors, but they extend past the individual level to include supportive family networks, socioeconomic status, school experiences, supportive communities, and cultural resources (Fleming & Ledogar, 2008). Protective factors serve to counteract or ameliorate the effects of adversity and thus guard the individual from adversity itself (Braverman, 2001). Vulnerability factors represent the negative qualities, experiences, or lack of protective factors that make individuals more susceptible to adversity and intensify the risk effects. Both factors may be present as a result of the presence or absence of family support and strong community identity and

access to or the lack of assistive resources (Braverman, 2001; Fleming & Ledogar, 2008;

Luthar & Cicchetti, 2000; Luthar et al., 2000). Thus, based on their unique life experiences, everyone will have unique combinations of protective factors and vulnerability factors and will likewise handle adversity very differently.

Dr. Glenn E. Richardson, director of graduate studies in the department of health promotion and education at the University of Utah, developed the resiliency model in figure one below; it depicts the process an individual goes through when facing adversity and the possible results.

Figure 1. Richardson Resiliency Model (Richardson, 2002).

As Richardson (2002) explains in his resiliency model, all individuals, regardless of demographic differences or prior experiences, begin the resiliency process in a state of bio-psycho-spiritual homeostasis (Richardson, 2002). This comparatively stable state is the starting point for an individual prior to experiencing a difficult situation or an adversity. At this point, the individual has already arrived at the situation with both their internal promotive factors and a measure of both protective factors and vulnerability factors developed from their environment that influence how the individual will deal with adversity. Thus, as adversity disrupts their state of homeostasis, the protective factors serve to strengthen the individual against adversity while vulnerability factors serve to weaken the individual and may intensify the effects of adversity. Of course, every individual has a unique array of experiences and influences, and therefore, each has varying amounts of both protective and vulnerability factors when faced with adversity.

According to Richardson’s model, the individual is then “disrupted” by adversity, which comes from a variety of stressors and life events to which the individual must react. Following this period of disruption begins the reintegrative process whereby the individual processes or reacts to the adversity and attempts to regain a state of stable homeostasis. The end result of this reintegrative process can range from dysfunctional reintegration, where the individual is weaker or more susceptible to adversity than they were before, to resilient reintegration, where the individual is stronger or less susceptible to adversity than when they first encountered it (Richardson, 2002). This process is affected by a variety of factors, including the protective factors, vulnerability factors, and other influences such as the individual’s environment, the promotive factors that build internal problem-solving processes, and the disruptive effects caused by both simple and

complex adversities (Zimmerman, 2013). Research has shown that individuals in general are surprisingly good at demonstrating resilient behaviors following loss and trauma

(Bonanno, 2004). However, individual results of this process depend on so many factors, including individual upbringing, that it has historically been difficult to predict individual success or failure given a particular set of adversities.

Definitions of Resiliency

It is good to note there are many definitions of resiliency from a variety of sources. A quick internet search for “resiliency definitions” returns over 4.2 million results, including a vast array of definitions from different fields like business, healthcare, and psychology. Many of these definitions include the concept of bouncing back from adversity (Dyer & McGuinness, 1996; Ledesma, 2014; VanBreda, 2001), returning to a former shape after being bent or pulled (Resilience, 2018), or rising every time you fall

(Duckworth, 2016). There are also several other words often deemed synonymous with resiliency, including “hardiness,” “perseverance,” and even “grit.” However, with so many definitions available, it is necessary for this study to operationalize a definition applicable to military personnel and higher education students.

To quantify, predict, and standardize an understanding of resilience, researchers have developed a variety of definitions and terms that have changed over time as shown in figure two below. A myriad of scales have also been developed and tested to try to capture individual resilience (Windle, Bennett, & Noyes, 2011). Some researchers have argued for definitions of resiliency applicable to specific populations, topical challenges, and research applications instead of global definitions (Ahern, 2006), and this has led to a

wide variety of definitions that may appear confusing and disjointed when considering the overall concept of resilience.

Source Population Definition Garmezy (1991) Global The capacity to recover and maintain adaptive behaviors after insult. Greenspan (1982) Infants and children The capacity to successfully undertake the work of each successive developmental stage. Hunter and Chandler (1999) Inner city, vocational high Process of defense using such tactics as insulation, isolation, school adolescents disconnecting, denial, aggression, as a process of survival. Mandleco and Peery (2000) Children and adolescents Capacity to respond, endure, and/or develop and master in spite of experienced life stressors. Markstrom et al. (2000) Rural, low income, Adaptive, stress resistant personal quality that allows the individual Appalachian adolescents to thrive despite unfortunate life experiences. Rew et al. (2001) Homeless adolescents Beliefs in one's personal competence and acceptance of self and life that enhance individual adaptation. Rouse and Ingersoll (1998) Adolescents in high school The ability to succeed, amature, and gain competence in a context of adverse circumstances or obstacles. Wagnild and Young (1993) Adult women The ability to successfully cope with change and misfortune.

Figure 2. Comparison of Resiliency Definitions (Ahern, 2006).

Researchers have cited the strong connection between resiliency and hardiness

(Ahern, 2006; Ledesma, 2014), and many have developed theories surrounding other concepts synonymous with resilience using other terms such as “hardiness” (Bartone,

2006; Bartone et al., 2008). However, reviewing this table and other common definitions from available resilience literature reveals a common thematic cycle wherein an individual possesses and demonstrates the ability to face an adversity, to process and react to that adversity, and to emerge with a new homeostatic state after regaining stability or normalization. While similar, these definitions raise questions of how resilient various populations are, how they became that way and why, and then how to compare the varying severity of adversities these populations each face. The answers vary widely along with the terms used in each case to explain resilience.

As resiliency research has progressed and definitions for resiliency have changed, researchers have moved through several “waves” of theory from identifying resiliency in terms of individual traits and qualities, to protective factors designed to help cope with

stressors, and then later combining the two into an internal motivational drive toward self-actualization (Allan et al., 2014; Fletcher & Sarkar, 2013; Masten, Herbers, Cutuli, &

Lafavor, 2008). The latest research has proposed that as resiliency theory moves past the limited understanding of individual character traits or assets (Pashak, Hagen, Allen, &

Selley, 2014; Rutter, 1985, 1987), that resiliency is actually the result of the development of dynamic processes of positive adaptation whereby an individual learns to overcome adversity (Fleming & Ledogar, 2008; Luthar et al., 2000). Thus, resiliency is not merely a static trait or state of being an individual does or does not possess and it is not just the presence or awareness of internal resources, external resources, or protective factors extant in the surrounding environment. So, what exactly is resiliency?

Since this study focuses on the resiliency of USAFA cadets, the operational definition used in this study will be the definition developed by the Defense Centers of

Excellence of Psychological and Traumatic Brain Injury (DCoE) and adopted by the Air

Force to better understand resiliency. They define resiliency as “The ability to withstand, recover and grow in the face of stressors and changing demands” (U.S. Department of the

Air Force, 2014, p. 14). This definition uses the verbs “withstand, recover, and grow” indicating resiliency is a dynamic action instead of a static trait or characteristic. This action-based definition captures the focus embodied by USAFA CCLD and provides a clear reference point from which to study the leadership and resiliency development process of USAFA cadets.

Resiliency-Building Models

In 1984, researchers proposed the compensatory model, the immunity vs. vulnerability or protective model, and the challenge model to aid in evaluating the nature

of individual resiliency and to help develop resiliency (Garmezy et al., 1984). These models continued to guide researchers as they applied the principles of resiliency theory to curricular strategies and operational practices to develop and enhance individual resiliency (Fleming & Ledogar, 2008; Ledesma, 2014; Richardson, 2002). All three of these models pursue a systematic approach to evaluating and building resiliency to stress regardless of individual background or experiences and “suggest themselves for the impact of describing stress and personal attributes on quality of adaptation” (Garmezy et al., 1984, p. 102). Resilient adaptation is the goal, and the question is how the promotive factors, protective factors, vulnerability factors, and adversities influence an individual’s ability to adapt.

Compensatory model. The compensatory model is outcome-focused and evaluates an individual’s ability to employ the internal promotive factors to reduce the effects of negative outcomes. This model is based on the idea that with personal attributes remaining the same, either the promotive factors will overpower the effects of the negative outcomes to the benefit of the individual or that the negative outcomes will overcome the promotive factors to the detriment of the individual (Garmezy et al., 1984).

The focus is on the presence, absence, and relative strength of the promotive factors and their ability to directly counteract or “compensate” for the effects of adversities or negative outcomes (Fergus & Zimmerman, 2005; Ledesma, 2014). This is represented in

Richardson’s model by the promotive arrows pushing upwards against the downward risk arrows (Richardson, 2002). The compensatory model proposes ideally that an individual’s promotive factors and ability to employ them can be developed and improved to overcome the effects of adversity more effectively.

The applications of this model could include adults mentoring children living in poverty to prevent youth violence where the adult mentorship compensates for the effects of poverty on children who become less inclined to engage in violence (Fergus &

Zimmerman, 2005). In effect, the adult mentorship to the children has a compensatory effect on the negative effects of poverty. The intended result is the children engage less in violence because of the mentorship. The level to which an individual emerges more resilient from the experience depends both upon the presence, strength, and ability to employ their promotive factors and the magnitude of the adversity and its effects

(Zimmerman et al., 2013). Thus, results of using the compensatory model will be unique to every individual.

Protective model. The protective model changes the focus from the outcomes to the risks leading to negative outcomes in an effort to prevent or avoid the adversity altogether. This model focuses on building resiliency by using the same promotive factors to moderate the effects of risks in an attempt to reduce or avoid negative outcomes (Fergus & Zimmerman, 2005; Ledesma, 2014; Zimmerman, 2013). In other words, the individual promotive factors interact with the adversity factors to change the resulting effects and outcomes. Ideally, the protective model capitalizes on the promotive factors in a way that either avoids the risk factors and negative outcomes altogether or changes them in such a way as to avoid their negative effects (Fleming & Ledogar,

2008). Thus, the goal is to change the outcome by employing the promotive factors to minimize the risk factors leading to the negative outcomes.

Application of this model may include drug use (a risk factor) and its connection to sexually risky behavior (negative outcome) that may be counteracted by

comprehensive training and education in both the dangers of drug use and sexually risky behavior (Fergus & Zimmerman, 2005). The training and education can modify the individual’s reaction and response to the risk factors and is a type of external resource within the promotive factors toolkit that an individual can call upon to change the negative effects of adversity (Kiswarday, 2012). Thus, success in the model depends on reducing both the effects of risk factors and/or the risk factors themselves.

Challenge model. The challenge model is different from the compensatory or protective models in that instead of seeking to minimize contact with adversity, counteract its effects, or avoid negative outcomes altogether, individuals are intentionally exposed to moderate amounts of adversity in controlled environments to build a kind of immunity to the risk (Garmezy et al., 1984). In this model, individuals can face adversity and wrestle with it under the supervision of others who are ready to help, provide guidance, reinforcement, and aid if necessary, in overcoming the adversity. This type of model serves as an inoculation against adversity that prepares the individual for the next adversity (Ledesma, 2014; Zimmerman, 2013). Researchers note that exposure to levels of adversity too high for the individual to adapt to are counterproductive while exposure to levels of risk too low do not pose a sufficient amount of stress, resulting in little to no positive effect (Fergus & Zimmerman, 2005; Fleming & Ledogar, 2008). After building experience through guided practice and development, individuals may enhance their abilities to process adversity as they develop an understanding of what to do, how to perform, and how to process the adversity (Kiswarday, 2012). This inoculation or steeling process familiarizes the individual with the adversities and prepares them for

facing those adversities in real life when restarts are not a possibility and the consequences are real (Fergus & Zimmerman, 2005).

Applications of this model are common especially in the context of athletics or completing physically demanding tasks like football practice, military drills, dress rehearsals, or learning to talk through a fight or argument wherein the stress of similar adversities are present with the expectation of performing a task (Zimmerman, 2013). In this practice environment, individuals can engage in a process whereby they struggle with adversity, experience both success and failure, make mistakes, discuss other strategies, evaluate viable solutions, and are then able to “restart” and face the adversities again.

For coaches, this model is commonly referred to as “doing reps” and is as simple as running athletes repetitively through difficult drills in practice to prepare them for the challenging demands of a real game. For service members in the military, conducting drills and exercises under stressful conditions has the same of effect of preparing them for the adversities they will face while deployed where the demands are extreme, and the outcomes can be lethal.

Commonalities between models. Common to all the models is the concept of repetition and practice. With enough practice, everyone can develop resiliency, as retired

Navy Seal Eric Greitens explains, simply because “We become what we do if we do it often enough” (Greitens, 2015, p. 29). In his discussion of Aristotle’s Ethics, author Will

Durant further discusses the concept of habituation by quoting the ancient philosopher who claims that excellence is a habit:

Excellence is an art won by training and habituation: we do not act rightly because we have virtue or excellence, but we rather have these because we have acted rightly; “these virtues are formed in man by his doing the actions”; we are what we repeatedly do. Excellence, then, is not an act

but a habit: “the good of man is a working of the soul in the way of excellence in a complete life . . . for as it is not one swallow or one fine day that makes a spring, so it is not one day or a short time that makes a man blessed and happy.” (Durant, 1961, pp. 61, emphasis added)

Thus, while reviewing the following models, it is important to remember that the key to success in any of them is repetition and re-evaluation until excellence, and resilience becomes a habit. This process of repetitive effort followed by introspective re-evaluation creates a steeling effect on the individual who becomes better able to process adversity.

While these models utilize promotive factors in slightly different ways, have slightly different emphases, and are appropriate in different environments, all have the goal of increasing the individual resiliency necessary to overcome actual adversity, and all try to result in a positive adaptation (Fletcher & Sarkar, 2013). While the models use different strategies, their common goal is to aid individuals in overcoming the adversities they encounter in such a way as to enable them to learn and grow more resilient as they encounter adversities, deal with them, and reintegrate into one of several new states of homeostasis (Fleming & Ledogar, 2008; Richardson, 2002). Thus, whereas the definition of resiliency refers to the ability to withstand, recover, and grow in the face of stressors and changing demands, the resiliency models seek to build and develop this ability.

Strengths and Weaknesses of Resiliency Theory

Researchers have repeatedly reported the merits of resiliency theory and resiliency-related concepts such as grit and hardiness, and several career fields have looked to resiliency theory to improve their various approaches to improve the ability to overcome adversity. Researchers have used resiliency theory to help explain the successful development of adolescents despite being at risk for substance abuse, teen pregnancy, and family dysfunction and the use of successful interventions to build and

promote resilient behaviors (Braverman, 2001; Carlos & Enfield, 1998; Fergus &

Zimmerman, 2005; Garmezy, 1993). The medical field has turned to resiliency theory and its elements of vulnerability and protective factors to assist in addressing issues faced by mental health patients and their families (Dyer & McGuinness, 1996). The education field has lauded the effects of teachers applying resiliency theory to assist struggling students (Kiswarday, 2012) and the protective factors including family and institutional support as important factors in the success of low-income, first-generation students

(Mbindyo, 2011). In several studies, Dr. Duckworth demonstrated the effectiveness grit has on predicting the successful graduation of USMA cadets and superior performance of participants in the National Spelling Bee (Duckworth et al., 2007). Grit was also a significant predictor of completing the Army Special Operations Forces (ARSOF) selection course, staying in high school, remaining faithful in marriage, and retaining employment (Eskreis-Winkler, Duckworth, Shulman, & Beal, 2014). Hardiness was determined to effectively predict success of U.S. Army Special Forces candidates

(Bartone et al., 2008) and was shown to increase leader effectiveness and unit cohesion in the military (Bartone, Barry, & Armstrong, 2009).

These examples highlight several aspects of resiliency theory proven effective in understanding human behavior in different settings and in predicting successful behavior and performance of individuals when promotive and protective factors are increased and enhanced to counter the effects of adversity. This concept of using personal strengths to overcome adversity (Perez, Espinoza, Ramos, Coronado, & Cortes, 2009) and working to turn personal weaknesses into strengths resonates with students, teachers, medical professional and their patients, and those serving in the armed forces worldwide.

Additionally, the concept of turning weaknesses into strengths to overcome adversity speaks to people of all ages and is a practice anyone can engage in regardless of prior experience or background.

However, while the benefits and strengths of resiliency theory are well documented, some critics of resiliency and grit have challenged the concept and written articles pointing out the weaknesses inherent to the concept of resiliency. One perspective is that resiliency-type theories imply more resilience is always better without recognizing the dangers of being over-resilient. Authors of this argument cite examples of workers attempting to push through unhealthy circumstances they should actually leave, overly resilient leaders who appear unemotional and lack the self-awareness to know when they should change course, and those who push their organizations to complete tasks to the detriment of their employees (Chamorro-Premuzic & Lusk, 2017).

Scholars have also argued that if resiliency is built through experiencing difficulty or hard times, many will assume the more difficulty they experience or the harder the time they put themselves through, the more resilient they will become, leading many to remain in difficult and even dangerous circumstances much longer than they should (Kashdan,

2017). These scholars question whether or not an individual can become too resilient or so resilient that it becomes a detriment to themselves or others.

Another challenge to the concepts closely related to resiliency theory are the emergence of research studies claiming grit and its subscales of passion and perseverance or resiliency have a much smaller impact on performance in a variety of settings than previously claimed. One research study conducted a meta-analysis of over 73 independent research studies on grit using 88 unique samples and 66,807 individuals and

found that grit was actually a weak predictor of overall performance, that the grit subscale of passion was not significantly predictive of overall performance, and that the subscale of perseverance or resiliency was not as reliably predictive of academic performance as other study skills and habits, adjustment to college, and class attendance

(Credé et al., 2017). This questions research findings claiming resilience predicts performance.

Another critic asserts building resilience and grit is only available to those who live in privileged communities and have the opportunities and resources available to improve their promotive and protective factors (Schreiner, 2017). The author argues the only familial or societal support that can effectively assist in building grit are those where wealth is a factor and resources are abundant. Where resources are slim, particularly in low-income areas, the family and community support necessary to help build the protective factors and increase resiliency are simply not available. Schreiner also explains how grit theory focuses on a deficit ideology that “identifies the personal shortcomings of people who are struggling, focusing on individual attitudes, behaviors, mindsets, and characteristics that impede their success” (Schreiner, 2017, p. 14). This results in a type of “victim-blaming” where students may be blamed for their own weaknesses. This standpoint supports the position that grit theory and/or resiliency theory unfairly places the responsibility of building grit solely on the shoulders of less gritty individuals without accounting for the nuances and unique circumstances contributing to their deficits in grit. This perspective suggests individuals can do relatively little to change their own grit without addressing other pre-existing and/or

current factors in their environment, and that to do so places blame primarily upon the individual for not being or not becoming more resilient.

This study engages these claims with quantitative research questions designed to add clarity to the relationship between resiliency and performance by exploring both positive and negative associations between individual factors and resiliency and the magnitude of those associations. Coupled with an examination of the extent to which changes in resiliency occur over time, this study provides additional insight into the influence of resiliency-building elements built into the character and leadership development programs cadets experience at USAFA. The following review of relevant literature examines leadership and its relationship with resiliency in higher education, the military, and in the leadership development programs at USAFA.

Review of Relevant Literature

Leadership

Leadership is inherent to the military organization, and the process of developing high-quality leaders is a priority, especially to military service academies like USAFA that are tasked with developing future military leaders. The subject of leadership has been discussed for centuries and was of great concern to those leading the earliest civilizations, including the pharaohs in Egypt and the leaders in ancient Greece and Rome

(Avery, 2004). The works of Plato discuss various aspects of leadership; the value of the best leaders and rulers over society; and the attributes, characteristics, and traits that the best leaders should have (Jowett, 1989). After several centuries, leadership theories began to emerge with the Great Man Theory in the 1840s, followed by trait theories, behavioral theories, and theories focusing on the interactions between leaders and those

whom they lead (Avery, 2004). Today, the discussion of leadership continues, although the focus has changed slightly with current philosophies focusing on leadership styles and which styles are most effective (Goleman, 2000), how to adopt different leadership styles, and whether leadership is something you learn or are born with (Avery, 2004).

As the discussions over leadership have continued, a multitude of leadership styles have been identified, beginning with the classic “Great Man Theory” leadership style (Avery, 2004; Zaccaro, 2007) and continuing with transactional leadership (Avery,

2004; Northouse, 2016), transformational or visionary leadership (Avery, 2004;

Deichmann & Stam, 2015; Hewitt, Davis, & Lashley, 2014), servant leadership

(Allender, 2006; Beck, 2014; Duncan & Pinegar, 2002; Northouse, 2016), authentic leadership (Avolio, 2007; George, Sims, McLean, & Mayer, 2007; Walumbwa, Avolio,

Gardner, Wernsing, & Peterson, 2007), level-three leadership (Clawson, 2009), principle- centered leadership (Covey, 1992), and the leaderless shared or organic leadership style

(Ancona, Bresman, & Caldwell, 2009; Avery, 2004; Pearce & Conger, 2002). Some leadership styles are regarded as superior to others due to their ability to motivate others, develop close relationships between leaders and subordinates, or result in maximum productivity and mission completion. However, scholars have highlighted research findings demonstrating the benefit and necessity of mastering all leadership styles and knowing how and when to use them in the way a professional golfer would use a variety of golf clubs to put the ball in the cup (Goleman, 2000), suggesting there is no one-size- fits-all style of leadership (Zaccaro, 2007). Leadership, then, is much more than a static or well-defined trait or style; it is an understanding of how to engage a variety of people

applying different leadership styles in a process of mutual progression. This is both the art and science of leadership in perfect balance.

Definitions of leadership abound, and most include the idea that some entity influences some other entity to act in a particular way, often resulting in the achievement of a common goal or desired endstate (Avery, 2004). However, current theorists agree that leadership is more than a characteristic, trait, or style. Leadership engages in a process of influencing others to complete a common goal (Northouse, 2016), and effective leadership occurs when leaders work with successful teams to accomplish their missions (Willink & Babin, 2015). Crisis situations add an element of stress, requiring even the greatest leaders to make complex decisions under extreme pressure, chaos, and shifting conditions. These situations pose the greatest test of leadership and test the moral and ethical character under extreme circumstances (C. E. Johnson, 2017). With these chaotic conditions in place, the stage is set for one of the most challenging environments an individual could ever face: that of leading others amidst the hellish chaos of war while influencing them to work together to achieve a common goal. In this environment, the principles of both leadership and resiliency come together in a way to help individuals process the adversity around them and lead their organizations through the adversities they face.

To understand how to apply leadership and resiliency theories in different situations, it is useful to discuss environments where some level of both leadership and resiliency is necessary to overcome the inherent adversities. A discussion of both higher education institutions and the military follows to better understand the variety of adversities and risks present, the policies, training, and programs designed to address the

adversities, and the attempts made to measure the success of these programs and evaluate the results. This is followed by a discussion of USAFA which is a uniquely challenging environment in that it combines a higher education institution with leadership development in a military atmosphere. Before looking at USAFA, it is helpful to understand resiliency in the context of both higher education and the military and review efforts to alleviate the challenges posed to university students and military service members alike.

Resiliency in Higher Education

In addition to being a military service academy, USAFA is an institution of higher education that offers 31 academic majors and four minors in a variety of subjects, including physics, computer engineering, aeronautical engineering, mathematics, and military and strategic studies (The United States Air Force Academy, 2018a).

Additionally, over half of the 31 majors offered are in the field of science, technology, engineering, and mathematics (STEM), and in 2016, USAFA was rated by Forbes as the number five STEM university in the nation (Forbes, 2018). This places USAFA in the unique position of developing future military leaders for USAF and also future leaders in

STEM fields and career paths. Being a cadet at a military academy and being a student in a STEM university are each daunting challenges by themselves due to the strenuous nature of the training programs. Combining them into a singular institution poses a uniquely demanding setting for even the best students and provides an excellent research environment to explore the relationship between resiliency and performance.

The challenges to students in higher education. The world of higher education can be challenging for students for a variety of reasons. Students enter the realm of

higher education and face a myriad of adversities, termed “risk factors” in resiliency theory, each with the potential of posing significant challenges to their educational progress. Some of these risk factors may include coming from lower socioeconomic status (SES) or low-income families, ethnic minority families with limited access to education, and single-parent homes with parents focused on income instead of academic achievement (Gutman, Sameroff, & Eccles, 2002). In addition to the typical risk factors faced by every individual, students entering higher education experience a completely new world full of adversity ranging from the challenges of admissions, dealing with roommates, professors, financial aid, classes, and ultimately the challenges of graduating

(Allan et al., 2014). Students experiencing significant challenges from this new environment often have low grade point averages and struggle with poor academic performance (Gutman et al., 2002). This can lead to an increase in attrition rates as some students get discouraged and leave the institution prior to graduation because they are unable to adapt to the new environment or maintain academic standards (Garza et al.,

2014). Additionally, STEM programs in particular have experienced high attrition rates with one study reporting 48% of bachelor’s degree students and 69% of associate degree students leaving STEM programs for non-STEM programs between 2003 and 2009

(Chen, 2013). Together, these risk factors make it difficult for many students to succeed, much less thrive in the challenging environment of higher education.

Efforts to help students overcome challenges in higher education. To address these challenges, some academic professionals have turned to the principles of resiliency theory, specifically focusing on developing the promotive and protective factors to build students’ abilities to experience adversity and learn to process and overcome the risk

factors (Perez et al., 2009). In higher education, resiliency sometimes refers to the

“interactive and accumulating process of developing different skills, abilities, knowledge and insight that a person needs for successful adaptation or to overcome adversities and meet challenges” (Kiswarday, 2012, p. 94). Being resilient can assist in overcoming these challenges and assist students in experiencing improved performance (Allan et al.,

2014). Researchers have explored ways to increase graduation rates and decrease attrition using resiliency theory to focus on compensatory models using promotive and protective influences to determine what faculty could do in their classrooms to affect and increase the resiliency of their students (Masten et al., 2008). A qualitative research study found that students appreciated teachers who built their individual self-efficacy, helped them effectively self-appraise their own strengths and weaknesses, encouraged help-seeking tendencies, and explained the clear links between academic success and future security (Morales, 2014). Multiple methods, including utilizing a list of developmental assets, have also been identified to understand the factors leading to increased performance in college students (Pashak et al., 2014). These findings explored the adverse risk factors causing students to underperform or drop out, the promotive factors that help them succeed, and the findings suggested actions to staff, faculty, and institutions to build student resilience, therefore reducing attrition and increasing performance.

Results of efforts to help students. However, despite several examples of resiliency theory in action to the benefit of students, the research evaluating resiliency through grit surveys in relation to student performance has mixed results. Researchers have reported findings indicating a close relationship between grit and academic

performance and retention (Eskreis-Winkler et al., 2014), including research by Dr.

Duckworth claiming grit is closely related to student performance with grittier students consistently performing better (Duckworth et al., 2007). However, while more recent studies also found a close correlation between performance and the resiliency component of grit, they did not find a close correlation with the consistency of interest or passion aspect of grit (Bowman, Hill, Denson, & Bronkema, 2015; Credé et al., 2017; Wolters &

Hussain, 2015). Other studies have reported no significant correlation between grit and performance, including in STEM programs (Bazelais, Lemay, & Doleck, 2016; Chang,

2014). These conflicting studies highlight the need for further research to understand how grit and its subcomponents of passion and resiliency are related to student performance and to identify and test exploratory models and methods designed to build resiliency. This research study revisits the association between resiliency and student performance and attrition to clarify the mixed results of past research studies and determine the differences and similarities which exist at USAFA.

Resiliency in the Military

Particularly in the military, personal hardiness and resiliency can reduce the effects of stressors commonly found in contemporary military operations, and leaders who improve their own hardiness are better able to influence subordinates in developing their own personal hardiness (Bartone, 2006). Researchers have determined personal hardiness, synonymous with resiliency, to be the largest predictor of military leadership performance for both men and women across different contexts (Bartone, 2006; Bartone,

Eid, et al., 2009; Bartone et al., 2013) and an important indicator of mental health (Eid,

Johnsen, Bartone, & Nissestad, 2008; Ramanaiah, Sharpe, & Byravan, 1999). Leadership

in the military is critical for mission success and is the element of combat power that unifies all other elements of combat power (U.S. Department of the Army, 2012). In an attempt to identify styles of leadership most effective in the military, studies have found that some leadership styles foster resilience in subordinates (Eid et al., 2008; Gaddy,

Gonzalez, Lathan, & Graham, 2017). However, despite this understanding, the ever- changing nature of global conflict continues to present new challenges to military service members, and no one was ready for the changes brought about by the events that occurred on the morning of September 11, 2001.

The challenges to service members in the military. Not long after the terrorist attacks on 9/11 and the ensuing war on terror in the Middle East, military leadership increased their attention on the overall psychological resilience of soldiers returning from combat tours. Some soldiers returned from combat with severe physical injuries, others with severe psychological injuries, but all returned permanently changed as a result of the ravages of war. Military leaders took increasing notice and began prioritizing ways to build resilience in soldiers to strengthen them against the adversities of combat. Around the same time in 2003, Dr. Karen Reivich from the Penn Resiliency Project (PRP) at the

University of Pennsylvania and Dr. Andrew Shattee from the Department of Family and

Community Medicine at the University of Arizona published The Resilience Factor: 7

Keys to Finding Your Inner Strengths and Overcoming Life’s Hurdles, which detailed different ways to overcome adversity (Reivich & Shatté, 2003). Military leaders took notice of Dr. Reivich’s research and began working to develop a program to help soldiers and their families.

Efforts to help service members overcome challenges in the military.

Understanding the need to engage in the research and development of resilient leaders, both the United States Army and the United States Air Force initiated research into the development of resilient leaders to better understand and develop resiliency within their ranks. The Army stresses the need for every soldier to be resilient and respond positively after facing adversity and argues that leaders should train their units to be resilient now and in preparation for future adversity (U.S. Department of the Army, 2012, 2014). The

Army defines resilience as “The mental, physical, emotional, and behavioral ability to face and cope with adversity, adapt to change, recover, learn, and grow from setbacks”

(U.S. Department of the Army, 2014, para 1-5a). This highlights the elements of facing adversity and being able to withstand, recover, learn, and grow from it.

In 2008, the Army established Army Regulation 350-53: The Comprehensive

Soldier Fitness (CSF) program in order to build soldier resiliency and address the sharply increasing number of soldiers returning from war with post-traumatic stress disorder

(PTSD), alcoholism, drug abuse, and suicidal tendencies (U.S. Department of the Army,

2014). The program was updated in 2014 and was renamed the Comprehensive Soldier and Family Fitness (CSF2) Program, adding the element of family resiliency to the program along with opportunities for families and family members to receive training that would help build family resiliency as they dealt with the difficulties of multiple deployments, soldiers who came home with injuries, soldiers who came home somehow mentally different from when they left, or soldiers who didn't come back at all. As stated in the current regulation, the purpose of the CSF2 program is “to increase the resilience

and enhance the performance of Soldiers, Families, and DACs,” referring to Department of the Army Civilians (U.S. Department of the Army, 2014, para 1-5a).

The CSF2 program identifies five dimensions of strength which serve as the primary conceptual pillars of physical, emotional, social, spiritual, and family resiliency to build the overall resiliency of soldiers (U.S. Department of the Army, 2014). The regulation briefly discusses these five dimensions of strength and focuses on the factors of physical, emotional, social, spiritual, and family as the primary factors contributing to individual resiliency.

• Physical resiliency refers to the overall physical health of the individual and

includes aspects of aerobic fitness, strength, body composition, sleep, nutrition,

and training, and their relation to psychological health.

• Emotional resiliency focuses on the ability of individuals to “approach life’s

challenges in a positive, optimistic way and to demonstrate self-control, stamina,

and good character in choices and actions” (U.S. Department of the Army, 2014,

para 2-3).

• Social resiliency refers to the relationship’s individuals have with a variety of

personal and professional contacts and the communication and friendships

developed between them.

• Spiritual resiliency includes “identifying one’s purpose, core values, beliefs,

identity, and life vision” (U.S. Department of the Army, 2014, para 2-5) and

focuses on developing the inner strength to make ethically sound choices and

persevere through adversity. This dimension does not focus on religion so much

as it focuses on drawing upon an individual’s core beliefs from a variety of

sources to strengthen his or her basic character.

• Family resiliency refers to building and strengthening the support networks

individuals have at home in order to promote a safe and secure environment and

avoid the tensions and problems associated with domestic difficulties. With this

dimension, resources and training are available to family members as well as

soldiers to aid in the improvement of family resiliency.

When combined and effectively embodied in a soldier, the Army believes these dimensions comprise the primary elements of individual resilience resulting in an

“individual [who] is better able to leverage intellectual and emotional skills and behaviors that promote enhanced performance and optimize their long-term health” (U.S.

Department of the Army, 2014, para 1-5a). This resilience enables leaders at all levels to provide better leadership to their units and strengthen both individual units and the Army as a whole.

Results of efforts to help service members. To measure and assess the resiliency techniques and skills of its members and conduct training to improve and sustain the overall force, the CSF2 program contains three assessment components consisting of an online assessment and self-development programs, specific training for both trainers and individuals in each unit, and a system of metrics and evaluation used to track and report the results of online assessments and training conducted (Reivich &

Shatté, 2003). The Global Assessment Tool (GAT) is the online training and self- assessment tool used to test the individual’s ability in each resiliency dimension and it contains a series of modules individuals can go through to learn more about each

dimension and how they can improve their capacity to embody that dimension. As of

2016, the GAT had been taken over 5.2 million times by soldiers, families, and DACs and has proved to be a measuring tool with high reliability (Vie, Scheier, Lester, &

Seligman, 2016). The CSF2 program and GAT remain in use, and research continues to evaluate its overall effectiveness and value to service members and their families in building both individual and family resiliency.

In 2011, the United States Air Force began developing its own program to improve individual resilience in a “Total Force Fitness” approach that included eight pillars of fitness and was infused with the concepts of resilience and how to increase individual resilience (Meadows, Miller, & Robson, 2016). In 2014, the Air Force established the Comprehensive Airman Fitness (CAF) program designed to “enhance the resilience of individuals, families, and communities” (U.S. Department of the Air Force,

2014, p. 1) using four domains consisting of mental, physical, social, and spiritual fitness.

This is a pared-down version of the Army CSF2 program but establishes no other measurement tool to assess the effectiveness of the program besides tools already in place, although it allows for future assessments to be completed on an as-needed basis depending on the needs of the unit (U.S. Department of the Air Force, 2014, para 4.1).

Research focused on the GAT has produced mixed results, with some lauding the program’s effectiveness at decreasing negative behaviors (Lester, Harms, Herian,

Krasikova, & Beal, 2011) while others express concerns about the overall effectiveness of the program to actually build individual resilience (Brown, 2015; Timmons, 2013).

This research study revisited the association between resiliency and military performance

to clarify the mixed results of past military studies and determine the differences and similarities which may exist at USAFA.

Resiliency at USAFA

With a review of existing challenges in higher education and military environments and the benefits of individual resiliency in the face of these adversities, the stage is set to explore the challenges and application of resiliency theory at USAFA.

Being a student at one of the five military academies in the United States, cadets at

USAFA experience a unique set of demands that combine the difficulties of higher education with the challenges of preparation for active-duty military service. In such a taxing environment, the ability of cadets to survive and thrive depends on their ability to adapt to their surroundings and overcome a variety of difficult conditions. Put simply, cadets who succeed and prosper are those who learn to demonstrate resiliency and thrive despite the adversities. Thus, efforts to build cadet resiliency have the potential to decrease attrition rates and improve student performance while still maintaining ambitious standards, resulting in more competent and resilient officers prepared for the rigors of active duty.

The challenges to cadets at USAFA. USAFA is a challenging environment for cadets for several reasons. First, it is the intent of the USAFA leadership to create and foster an environment that challenges cadets by deliberate design to build leaders of character. One of the eight key components integral to the essence of USAFA entitled

“Developing Character and Leadership” asserts:

The Academy’s unique opportunities allow cadets to practice leadership theory and learn from their experiences. Daily leadership challenges and opportunities abound to learn, apply, and refine leadership principles. The intentional and integrative nature of this officer development catalyzed by

the Center for Character and Leadership Development, but implemented throughout, is pervasive at USAFA and not available anywhere else. The Honor Code guides this leadership development to set cadets on a path of living honorably. (The United States Air Force Academy, 2017a)

This demonstrates the Academy’s commitment to setting high standards and firm expectations on a routine basis to develop the qualities, behaviors, and traits expected from future Air Force officers. This daily “inoculation” of leadership challenges and high standards reminds one of the repetitively habitual actions of excellence previously mentioned by Aristotle as a key to developing good moral character and habits of excellence.

Another challenge inherent to the environment at USAFA is the nature of the cadets themselves. Admissions rates are highly competitive, with an overall admission rate for the 2016 class of 13% with an average high school GPA of 3.85, where 75% of those admitted had a high school GPA of at least a 3.75 (CollegeData, 2016). Average cadet scores in 2016 for the SAT math and critical reading scores were 663 and 633 respectively, with a 1296 composite score, placing USAFA enrollees in the 85th percentile (Staffaroni, 2018), and the average ACT score was 30, which was in the 93rd percentile (Zhang, 2016). Ninety-eight percent of enrollees represented the top half of their class, with 56% in the top 10% of their class, including nine percent valedictorians and eight percent national merit scholars (CollegeData, 2016). The highly selective admissions process results in a group of individuals considered top performers in the country and from around the world. This creates a student body of high-caliber USAFA cadets who have demonstrated academic and athletic accomplishment, provided volunteer service, exhibited strong personal character, and are therefore already highly resilient individuals. While this results in a group of highly capable cadets, it also creates

a highly competitive environment where top students compete for top marks to improve their chances at obtaining an Air Force Specialty Code (AFSC) or career field of their choice. Cadets are ranked using order of merit lists (OMLs) according to their performance in academics, military officership, and physical fitness and OML placement has a considerable influence on which AFSC cadets receive thus deciding their careers in the Air Force. This highly competitive environment, in addition to the stresses of a military academy and a STEM institution of higher education, combines a variety of stressors and adverse conditions that cadets often find difficult to manage.

Efforts to help cadets overcome challenges at USAFA. Recognizing the need to continually develop and support cadets through their challenging academic careers,

USAFA established the Officer Development System (ODS) to develop cadets as part of the Air Force’s force development process (The United States Air Force Academy,

2014a). The purpose of the ODS is to provide the framework by which the nine institutional outcomes are accomplished in order to “1) develop each cadet’s appreciation that being an officer is a noble way of life, 2) foster a commitment to character-based officership, and 3) develop competencies essential to this identity as a character-based officer/leader” (The United States Air Force Academy, 2014a, p. 4). The ODS framework includes three main elements, including the foundation (the why), the goal

(the what), and the process (the how) (The United States Air Force Academy, 2017c).

The foundation sets forth the principles and values underlying the educational process at

USAFA; the goals sets forth USAFA’s nine desired institutional outcomes; and the process, describes the means by which cadets will learn to embody the foundational principles and achieve the institutional outcomes.

The process to achieve the goals and outcomes include three key components consisting of the Personal, Interpersonal, Team, and Organizational (PITO) model which describes the competencies cadets should develop; the Leadership Growth Model (LGM), which describes the process leaders and followers use to develop the competencies; and the Guiding Principles, which guide the development and implementation of the entire

ODS process (The United States Air Force Academy, 2017c, p. 11). The LGM model itself details four stages that include setting expectations and giving inspiration, providing instruction, providing feedback, and reflection. This growth model enables leaders and followers to work collaboratively using the guiding principles to develop the competencies in the PITO model, resulting in the accomplishment of the institutional outcomes established by USAFA to develop professional leaders of character.

The nine institutional outcomes developed by USAFA, and based on the Air

Force Institutional Competencies, focus on developing cadets into “professional Air

Force officers who think critically, lead with character, and serve the nation” (The United

States Air Force Academy, 2014b, p. 1). USAFA’s nine institutional outcomes are to develop cadets in 1) critical thinking, 2) application of engineering fundamentals, 3) scientific reasoning and the principles of science, 4) the human condition, cultures, and societies, 5) leadership, teamwork, and organizational management, 6) clear communication, 7) ethics and respect for human dignity, 8) national security of the

American republic, and 9) warrior ethos as airmen and citizens (The United States Air

Force Academy, 2014b, p. 1). Each of the outcomes has a list of proficiencies designed to achieve the intent of each outcome, and the proficiencies relate directly to the courses and programs cadets participate in. Outcome number nine, Warrior Ethos as Airmen and

Citizens, has eight proficiencies that support the accomplishment of four character traits.

The third character trait, Demonstrate Service before Self as Related to Physical Courage has two proficiencies including “Proficiency 5: Exhibit grit: a hardiness of spirit and resistance to accept failure despite physical and mental hardships” (The United States Air

Force Academy, 2016c, p. 1). This proficiency acknowledges the need for cadets to develop grit and is designed to identify, develop, and evaluate the programs and activities established to achieve that goal.

USAFA has developed and implemented courses, programs, and training opportunities to build cadet character and leadership, challenge decision-making ability, and increase mental and physical toughness. This curriculum is designed to span all four years of the cadet experience with each class of cadets experiencing unique and specific programs. While the full details of specific course and program descriptions are available online (The United States Air Force Academy, 2016a), a brief review of some of the past and current programs is presented below. It is important to note than USAFA constantly strives to capitalize on the latest research and information regarding character and leadership development. It therefore comes as no surprise that USAFA continues to review and update its programs to apply the most current research theories and maintain relevant and effective programs to develop the best officers possible.

USAFA conducts a variety of character and leadership seminars focused on developing the specific needs of each class of cadets from incoming or fourth-year cadets through seniors or first-year cadets, often referred to as “firsties”. Incoming first-year enrollees, called “basic cadets”, experience the physical and mental rigors of basic cadet training (BCT) prior to beginning the academic semester. This six-week introduction to

Air Force life consists of garrison and field training designed to challenge individuals physically, mentally, and emotionally. Prior to 2017, the Center for Character and

Leadership Development (CCLD) at USAFA was also conducting several character and leadership seminars focusing on the skills and development of different class years of cadets. The fourth-year cadets participated in a 12-hour, two-phase program called

VECTOR (Vital Effective Character Through Observation and Reflection) where they examined their own values, purpose, vision, and influence, and then committed to a career of honor and integrity. Third-year cadets (sophomores) attended the Respect and

Responsibility (R&R) seminar consisting of an eight-hour seminar/outdoor adventure program designed to expose them to differences and similarities in their own leadership styles, develop communication skills, foster cooperation and trust, and challenge negative views and biases. Second-year cadets (juniors) participated in an eight-hour Leaders in

Flight Today (LIFT) seminar focused on team-building, character, and leadership development and culminating in an exercise with their squadron. The “firsties” attended a culminating eight-hour Academy Character Enrichment Seminar (ACES) program designed to focus on the character and ethical demands placed on Air Force officers, moral and ethical decision-making skills, and their leadership impact on underclassmen.

In 2017, CCLD augmented and consolidated the program into the current ODS program

(The United States Air Force Academy, 2017c). This new system takes elements from all prior seminars and reorganizes, augments, and delivers the material to cadets in a way that better suits the leadership and development needs of each class of cadets. Cadets participating in this study experienced seminars both prior to and after the changes brought about in 2017.

Since 1993, CCLD has also hosted the annual National Character and Leadership

Symposium (NCLS) held at USAFA each Spring. NCLS typically has a variety of speakers, including distinguished scholars, military and corporate leaders, athletes, and others who are experienced in the challenges of leadership. While the focus and themes of NCLS changes every year, the goal remains to provide “. . . an opportunity for all

Academy personnel, visiting university students and faculty, and community members to experience dynamic speakers and take part in group discussions to enhance their own understanding of the importance of sound moral character and good leadership” (The

United States Air Force Academy, 2018b). While attendance of NCLS is obligatory to cadets of all classes, the freedom cadets have to select which speakers to listen to makes the event a popular respite from the typically grinding schedule of coursework.

In addition to the courses and programs offered by CCLD, USAFA includes physically challenging programs like boxing, combatives, and a variety of intramural sports. While cadets can choose which sports they participate in, participation in intramural sports is obligatory for all cadets. USAFA also provides longer-term training opportunities in the summers between semesters and these include in part, jump school, glider school, and Survive Escape Resist and Evade (SERE) school. These programs engage the cadets outside the classroom in real-world “live” scenarios where the principles and concepts of leadership and character development learned in the classroom are put into action.

The collective goal of these programs is stated in the USAFA 2015-2016 course catalog under the Center for Character and Leadership Development,

In sum, character development will be a crucial, all-encompassing part of each cadet’s Academy experience. From the time each cadet enters the

Academy until graduation, they can expect to see various character and leadership development programs in every aspect of their life with the ultimate goal of inspiring in them an intense inner drive to put integrity first, place service before self, and strive for excellence in all they do. (The United States Air Force Academy, 2016a)

The goals of these programs focus on character development, leadership skill development, and increasing the professionalism and ability of cadets. Thus, resiliency theory, as an essential aspect of character development, is a key element of the development and training of each cadet and demands attention as it appears within the overall framework of character development at USAFA.

Results of efforts to help cadets. USAFA administers a battery of surveys throughout a cadet’s academic career to better understand the effectiveness of its courses and programs and to further understand the nature and development of a variety of cadet traits and characteristics, including resiliency. Several of these surveys, including the

Grit-S survey, the Global Assessment Tool (GAT) used currently by the military, the

Dispositional Resilience Scale (DRS), and the Support and Resilience Inventory (SRI) self-assessment tool sponsored by the North Carolina Department of Health and Human

Services, measure a variety of behaviors and individual characteristics relating to individual resiliency. While these surveys vary in scale and focus, all utilize a method of using self-report survey questions and statements to measure the presence or tendency to demonstrate resiliency-related characteristics.

Initial results from data obtained from USAFA/A9 Department of Testing and

Assessment revealed that cadets from the class of 2016 and 2017 demonstrated average scores of 3.7 out of 5 for the grit survey, 8.2 out of 10 for the SRI, and 1.4 out of 3.0 for the DRS demonstrating average to above average resiliency scores for each of the

surveys. However, no comprehensive analysis has been completed to understand why cadets are resilient, how their resilience is changing at USAFA, or which programs demonstrate an impact on survey scores. While a larger meta-analysis of these, and several other surveys would be informative to review their history and development and to conduct a full comparative analysis, initial results from these surveys demonstrate

USAFA cadets display average to higher than average levels of resilience and possess and display resiliency-related qualities.

Though the results of these surveys contain myriad data describing the characteristics, traits, and behavioral development of cadets, they have yet to result in findings upon which to establish clear fundamental principles to guide the development of resiliency-building programs at USAFA. Additionally, while many of the programs build cadet character and leadership, few of them are based upon the principles of resiliency theory or explicitly address individual resiliency development as key aspects of their core frameworks. While these programs are generally considered effective, one recent study analyzing USAFA freshman cadets’ ability to identify with USAFA’s core virtues after completing the VECTOR seminar training produced only mixed results. The qualitative results of interviews with focus groups produced findings that cadets found

VECTOR training positive and impactful, but quantitative survey analysis could find no statistically significant connection (Tate, 2016). To better understand how cadets develop resiliency, this study analyzed cadet survey data to understand how demographics, performance indicators, attrition rates, and USAFA clubs and programs were associated with cadet resiliency.

Emerging Themes and Research Gaps

This review began with a discussion of the elements of resiliency theory; the history, development, and current research; and a brief discussion of some strengths and weaknesses of this theory and of how the concepts of self-efficacy and mindfulness further inform resiliency theory. The literature review discussed the need for leadership in the military and highlighted the necessity of developing resiliency to deal with the challenges to service members and their families as a result of warfighting and lengthy deployments. This set the stage for a discussion of resiliency in higher education and the military, listing examples of each and discussing the needs for greater research in each are to clarify and add to the body of existing research exploring individual resiliency.

Following this a review of the challenges faced by service members in the military was discussed noting the significant efforts made by the United States Army and the United

States Air Force to build the resiliency of their forces. The review concluded with a review of the adversities USAFA cadets face and the substantial efforts taken to develop programs and seminars to bolster cadet resiliency and improve student performance.

Several themes emerged requiring the need for further research. First is the need to add clarity to the discussion over the fundamental benefit of resiliency theory and the closely related subject of grit. A large amount of resiliency-based research exists both in higher education and in the military with conflicting findings on both sides of the debate.

This study will clarify these findings and inform current theories by adding credible research to further uncover the relationship between resiliency and performance using

USAFA cadets as its platform.

The next apparent gap is the insufficient number of research studies focused specifically on individual resiliency. Of the 73 studies listed researching resiliency and grit (Credé et al., 2017), population sample sizes varied from 21 to 1,554 individuals, with the average number of research study participants being 351 and the vast majority of the studies containing fewer than 1,000 participants. The largest grit studies conducted by Dr. Duckworth were conducted with population sample sizes of 1,218, 1,308

(Duckworth et al., 2007), and 1,554 (Duckworth & Quinn, 2009) and compared participants’ grit scores with other factors at a singular point in time. This study analyzed data from over 5,400 USAFA cadets over a period of nine years, including several groups that completed the Grit-S survey multiple times, providing both a much larger sample size than previously available together with longitudinal data to identify emerging trends and changes in grit and each of its subscales of passion and resiliency.

Another apparent gap in the literature was the lack of research connecting the principles of resiliency and the resiliency models with programs designed to develop individual resiliency. Most of the literature discusses theoretical concepts and provides research detailing the analysis of surveys completed by individuals completing various programs, but little if any mention is made citing specific programs as more or less beneficial to developing individual resiliency. The primary focus in these studies is the resiliency of the individual, not the resiliency-building effectiveness of the programs.

Necessary to the discussion of resiliency development are examples of programs and activities found to influence individual resiliency along with a discussion of effect sizes and outcomes. This research study fills this gap by associating USAFA cadet Grit-S

score with cadet participation in the myriad clubs and programs available to cadets with the intent of identifying the types of programs that may increase resiliency.

In addition to filling these gaps, this study fulfills the need to exploit previously unexamined data collected by USAFA, thereby making use of valuable resources used to collect the data over the past several years. Significant time and effort has been spent to develop the surveys to collect the data, provide time to the cadets to complete the surveys, and to ensure the surveys are kept for future analysis. This study finally makes use of this valuable information with the intent of informing USAFA leadership of resiliency-building trends and emerging themes and providing them with the information and recommendations necessary to guide the future leadership development programs that will shape the character and leadership development of future cadets.

This discussion of the thematic elements either absent or requiring further research presents the focus for this research study and provides a justification and necessary connection to the research questions chosen to address these gaps. The quantitative research designed to provide answers to these questions is presented in the next chapter.

CHAPTER III

METHOD

The purpose of this chapter is to review the setting and participants who took part in the study, provide a description of the Grit-S survey used to assess cadet resiliency, and set forth the procedures and methods used to analyze the data. I present each research question along with the associated regression model used to examine the relevant data and then provide the data and analysis used to study the several groups of

USAFA cadets who took the Grit-S survey from 2009 to 2017. I present the regression models used to identify how cadet demographic predictors, performance measures, attrition rates, and USAFA clubs and programs were associated with Grit-S survey scores and resilience subscores. A discussion of the analysis follows, answering the proposed research questions and identifying opportunities and strategies for future research studies.

This quantitative study used multiple regression models to explore the level of resiliency and grit in USAFA cadets and explore the relationship between their individual

Grit-S survey scores and other factors, including demographic information, individual performance factors, attrition, and participation in USAFA clubs and summer training programs. It is important to note that while the focus of this study was on cadet resiliency, the overall Grit-S survey that was offered to cadets by USAFA gives information about resilience, passion, and overall grit. Thus, results will be shown for all three factors, although the subject of resiliency will be emphasized. Researchers have performed individual resilience research using surveys demonstrating both reliable and valid results (Eid et al., 2008; Ramanaiah et al., 1999). This study is an exploratory study, and while Grit-S survey results were regressed on cadet performance variables

such as grade point average, this study did not seek to replicate findings from other studies. The primary focus of this study was to understand the relationship, if any, between USAFA Cadet Grit-S scores and cadet development while at USAFA. The specific research questions to explore this relationship were:

1. To what extent are cadet characteristics, including demographics and

participation in USAFA clubs and programs, associated with Grit-S score?

2. To what extent is Grit-S score associated with cadet overall performance

average (OPA)?

3. To what extent is Grit-S score associated with cadet attrition?

4. What is the change in cadet Grit-S score over time?

5. Based on this research study, what are some recommendations for future

research, policy development, and practices to build cadet resiliency at

USAFA?

The intent of the study was to examine trends in Grit-S scores to understand how the associations between Grit-S scores and other predictors changed over time, what influenced those changes, and the extent to which individual factors were associated.

Exploring these questions provided a deeper understanding of how cadet resiliency is built and developed at USAFA and identified possible methods of improving programs designed to develop leaders of character.

Research Setting and Participants

The setting for this study was the Unites States Air Force Academy (USAFA) located in Colorado Springs, Colorado. USAFA accepts a pool of unmarried applicants averaging 20 years of age and consisting of a diverse array of students including 34.1%

ethnic minority students (CollegeData, 2016). After cadets arrive at USAFA, they become “basic cadets” until they complete the basic cadet training (BCT) held in the summer prior to their first semester. Upon completion of BCT, cadets receive the rank of fourth-year cadets, equal to first-year freshman students at other universities. Third-year cadets are equal to sophomores, second-year cadets are equal to juniors, and first-year cadets, often called “Firsties,” are equal to seniors at other universities. Cadets from all ranks enter squadrons with leadership positions filled by both the cadet ranks and active- duty Air Force leadership personnel. The smallest unit is the “Element” consisting of 10-

12 cadets. Three elements form a “Flight” of 30-35 cadets, and three flights form a

“Squadron” of approximately 100 cadets. USAFA maintains 40 squadrons organized into four “groups” that comprise the “Cadet Wing” representing the entirety of the cadet population ranging in size from 3,900 to 4,100 cadets.

An active-duty leadership team leads each squadron and consists of an Air Officer

Commanding (AOC), who is a commissioned officer, and an Academy Military Trainer

(AMT), who is a non-commissioned officer. This leadership team serves as the United

States Air Force active-duty professional leadership team tasked with developing, supervising, and mentoring every cadet within their squadron in every aspect of their careers while at USAFA (USAFA, 2016). This organizational design represents the social network, community of friends and professional colleagues, and support structure they must work with throughout the duration of their academic careers at USAFA.

From day one, USAFA cadets understand the core principles and critical outcomes of USAFA as they recite the honor code, “We will not lie, steal or cheat, nor tolerate among us anyone who does”; profess the spirit of the code to “Do the right thing

and live honorably” (The United States Air Force Academy, 2016b); and strive to fully espouse the United States Air Force core values of “Integrity First, Service before Self, and Excellence in All We Do” (Center for Character & Leadership Development, 2016).

On a daily basis, cadets review the values of the United States Air Force, receive mentorship in leadership, critical-thinking, and problem solving, and are challenged by the physical demands of fitness and extracurricular sporting events. Thus, not only have individuals selected to attend USAFA demonstrated prior ability and strong moral character, they receive constant mentorship in the principles of integrity, service, excellence, and develop a commitment to challenging standards in an environment filled with extraordinary expectations as they prepare to lead in the United States Air Force.

Measures

Grit-S survey description

The survey tool used by USAFA to measure cadet resiliency is a version of the

Grit survey developed by Dr. Duckworth that contains two subscales to test individual levels of both consistency of interest or passion and perseverance of effort or resiliency which combined are referred to as grit (Duckworth et al., 2007). In 2009, USAFA began using Dr. Duckworth’s original seventeen-item Grit survey to understand individual grit as a combination of consistency of interest, perseverance of effort, and ambition. This survey included six questions to measure passion, six questions to measure resilience, and five questions to measure ambition (Duckworth et al., 2007). Beginning in 2016,

USAFA began using the Grit-S survey (Appendix C), a shorter eight-item grit survey developed by Dr. Duckworth and Dr. Quinn that they found to be a more accurate and reliable version of the survey consisting of eight of the original seventeen questions and

producing more reliable and valid results (Duckworth & Quinn, 2009). This shorter survey, called the Grit-S survey, includes four questions measuring the passion subscale and four questions measuring the resiliency subscale. A five-point Likert scale is used to answer each question, and responses consist of “very much like me,” “mostly like me,”

“somewhat like me,” “not much like me,” and “not like me at all.” Questions 1, 3, 5, and

6 of the Grit-S survey are reverse scored where the answer “very much like me” receives one point and questions 2, 4, 7, and 8 are normally scored with the answer “very much like me” receiving five points. Overall Grit-S scores are calculated by adding up the points to all the responses and dividing by the total number of questions with a 5.0 being the top possible score. The passion and resilience subscores are calculated by adding up the points from their respective four questions and dividing each by four with a 5.0 also being the top possible score.

Researchers have used the Grit-S survey and its subscales to understand the grit of cadets at West Point, students participating in a national spelling bee, and college undergraduate students. They reported Cronbach’s Alpha values ranging from .73 to .83, indicating a strong level of internal consistency in the survey (Duckworth & Quinn,

2009). While not replicating their study, in a similar fashion I analyzed the results from

Grit-S surveys taken by USAFA cadets from 2009 to 2018 and regressed the Grit-S survey scores and subscale scores on a large number of cadet-level variables.

Reliability and validity

In prior research studies, Dr. Duckworth reported psychometric properties of the

Grit-S survey with an analysis of six separate studies that presented “evidence for the

Grit-S’s internal consistency, test-retest stability, consensual validity with informant-

report versions, and predictive validity” (Duckworth & Quinn, 2009, p. 166). Of those six, study one analyzed Grit-S survey results from two samples of cadets from the United

States Military Academy at West Point (USMA). This study found Cronbach’s Alpha values measuring internal consistencies of both the overall Grit-S survey ranging from

.73 to .83 to the sub elements of Persistence of Effort and Consistency of Interest ranging from .60 to .78 and from .73 to .79 respectively (Duckworth & Quinn, 2009, p. 167).

Study two analyzed Grit-S survey results completed by adults over age 25 who reported changing careers and found adequate internal consistencies in Perseverance of Effort

(.70), Consistency of Interest (.77), and Grit-S (.82) (Duckworth & Quinn, 2009, p. 168).

Study three analyzed survey responses from adults over age 25 along with surveys completed by informants found by survey participants. The informants grouped into family members, peers, and self- and internal consistencies were found to be .84, .83, and

.83 respectively. Correlations between the self-reports and family member or peer informants were medium to large: r = .45, p < .001 and r = .47, p = < .001. Study four analyzed survey results from high-achieving middle and high school students who took the survey one year apart with the focus of determining test-retest stability. Internal consistency measured in 2006 and 2007 was strong at .82 and .84 respectively.

Correlation between the two survey periods was good with r = .68, p < .001. Study five analyzed survey results from USMA cadets participating in the summer training program and found the Grit-S survey demonstrated an internal consistency of .77 with a higher predictive value of training completion than the Whole Candidate Score typically used by

USMA (Duckworth & Quinn, 2009).

This study of USAFA cadets measured the internal consistency of the Grit-S survey collected over a nine-year span, during ten different survey periods, with a large combined sample population (N = 5,454). Over this period, the internal consistency of the Grit-S survey was measured at 0.733 with the passion and resilience subscores measuring at 0.792 and 0.692 respectively. This demonstrates an acceptable level of internal consistency in all three scales and is very similar to the values presented in Dr.

Duckworth’s research studies.

Research Design

This study includes a quantitative analysis of secondary data previously collected by USAFA during multiple scheduled survey sessions. The historical data was collected from the USAFA-A9 and Cadet Administrative Management Information System

(CAMIS) departments and then cleaned, organized, and prepared for upload into SPSS for regression analysis. I performed several iterations of regressions to analyze the associations between the dependent and independent variables and the results were compared to reveal the best fitting regression models. Details of the analysis for each research question are provided in separate sections later in this chapter. The regression results for each of the research questions is presented in chapter four.

Procedures

Data collection. USAFA administers surveys at various times during the year, including during the biannual designated survey assessment time (DSAT), at basic cadet training (BCT), and by professors conducting research in different classes. Participation in the surveys offered during the DSAT and provided by professors is always considered voluntary, and despite the elevated expectations in both military environments and

especially inherent to USAFA, participation can range from low to quite high. Each year group at USAFA begins with slightly more than 1,100 cadets and frequently graduates between 800-900 cadets, thus a quick calculation using the figures in table one shows that participation in this survey ranged between approximately 4.3% and 100%. Participation in the survey prior to BCT is mandatory, hence the 100% participation, and that will be discussed in the limitations section.

Cadets complete the surveys by logging into a secure online system, agreeing to participate in the survey, reviewing the survey items, and clicking on the answers that best match their feelings and opinions. The survey data for this study consisted of survey data collected from cadets from 2009 to 2018 and was not collected by the researcher.

After gaining USAFA IRB approval, I collected the Grit survey data from the USAFA-

A9 survey and analysis department and discovered that 8,331 Grit-S surveys had been competed by USAFA cadets from the Fall semester 2009 to the Fall semester 2018.

Many of these cadets took the surveys multiple times, providing for the analysis of changes in Grit-S scores over time. Table 1 below shows the breakdown of all surveys completed by cadets during their respective survey periods.

187 256 582 765 222 469 925 154 574 1684 1482 1031 8331 Surveys Available Available 41 85 Fall Fall 2018 1142 304 (J)* 307 (Sr)* 405 (So)* 0 111 2018 1142 Summer Summer 1031 (B) 2 39 818 2018 Spring Spring urveys 253 (J)* 237 (Sr)* 287 (So)* 0 90 2017 1167 Summer Summer 1077 (B)* 2 12 363 2017 Spring Spring 78 (Sr)* 103 (J)* 168 (So)* 45 77 Fall Fall 435 2016 77 (J)* Grit-S SurveyGrit-S Groups 101 (Sr)* 135 (So)* 0 49 2016 1142 Summer Summer 1093 (B)* 2 23 Fall Fall 450 2015 43 (J)* 62 (F)* 52 (So)* 268 (Sr)* 7 40 2014 1126 Spring Spring 582 (J) 497 (So)* 22 81 Fall Fall 546 2009 187 (J) 256 (F) 2011 2013 2015 2016 2017 2018 2019 2020 2021 2022 Unknown Student Class Student Surveys Total Incomplete data Incomplete Table 1 Table From USAFA-A9 Collected Data Grit-S Survey USAFA of Description s matched groups *denotes containing Sr=Senior; So=Sophomore, F=Freshman, J=Junior, Cadets, B=Basic Note:

The USAFA CAMIS department was able to find and provide demographic, performance, attrition, and other cadet data for 5,454 of the cadets, but it found that the remaining cases had blank or erroneous identification information, making it impossible to connect cadet information in CAMIS with the results of the surveys. CAMIS personnel indicated this periodically occurs during survey periods as some cadets, apprehensive about providing their identification information, enter erroneous information to prevent any possible connection to their answers. The data was further inspected for missing and incomplete data, and cases that could not be matched with cadet data, contained erroneous data, or that were missing most, or all of the grit survey values were deleted. This reduced the number of useable surveys by 16.29%, resulting in

6,974 surveys available for analysis. The final result was a large sample population of cadets (N = 5,454) who completed a large number of Grit-S surveys (N = 6,974) from

Fall 2009 to Fall 2018.

After receiving the raw data from USAFA-A9 and the CAMIS department, it was necessary to process the data and prepare it for further regression analysis. Since data regarding cadet Grit-S survey scores, demographics, performance measures, attrition, and participation in USAFA clubs and programs came from more than a single source, the data first had to be organized and arranged so the data for all cases was displayed in a single row. Once the entire dataset was properly organized, a final dataset consisting of all cases (N = 5,454) and surveys (N = 6,974) was compiled in preparation for the analysis of missing data. The multiple-survey cases consisting of those who took the survey twice, three times, and four times (n = 1,335) were used to conduct the trend analysis in research question four.

Variables. This exploratory research study used a total of 44 variables throughout the entire analysis, and each research question used a different combination of the dependent and independent variables depending on the requirements of the research questions. Table 2 below shows all the variables used in the study, their regression code, measure, and coding used. A review of each of the variables follows the table and provides more details as to the reason the variables were included in the study.

Table 2

Dependent And Independent Variables Used In Regression Analysis

Variables Code Measure Coding Gender GEN dichotomous 1= female, 0 = male Black BLK dichotomous 1 = yes, 0 = no Caucasian (control) CAUC dichotomous 1 = yes, 0 = no Hispanic HISP dichotomous 1 = yes, 0 = no American Indian AMIN dichotomous 1 = yes, 0 = no Asian ASIA dichotomous 1 = yes, 0 = no Native Hawaiian/Pacific NHPI dichotomous 1 = yes, 0 = no Islander Unknown UNK dichotomous 1 = yes, 0 = no Family education FEDU ordinal 1=High School, 2=Some college, 3=Associates, 4=Bachelors, 5=Graduate, 6=unknown Family income FINC ordinal 1= don't know, 2= <25K, 3= 25K-74,999K, 4= 75K- 124,999K, 5= 125K-174,999K, 6= >175K Family members FMEM ordinal 1=1, 2=2, 3=3, 4=4, 5=5, 6=6, 7=7, 8=8, 9=9, 10=10 or more From single parent FSGP dichotomous 1 = yes, 0 = no First generation college FGEN dichotomous 1 = yes, 0 = no HS GPA HS_GPA% continuous scale HS Athletics HSAH dichotomous 1 = yes, 0 = no Recruited college athlete RCAT dichotomous 1 = yes, 0 = no Honors List HON dichotomous 1 = yes, 0 = no Grade Point Average GPA continuous scale Military Performance Average MPA continuous scale Physical Education Average PEA continuous scale Overall Performance Average OPA continuous scale Grit score GRIT continuous scale Passion subscore PASS continuous scale

Dependent And Independent Variables Used In Regression Analysis

Variables Code Measure Coding Resilience subscore RES continuous scale Grit survey gain score GTGN continuous scale Passion gain score PNGN continuous scale Resiliency gain score RYGN continuous scale Not disenrolled (control) NDSN dichotomous 1 = yes, 0 = no Disenrolled voluntarily VOL dichotomous 1 = yes, 0 = no Disenrolled involuntarily INVOL dichotomous 1 = yes, 0 = no Probation PROB dichotomous 1 = yes, 0 = no Probation-Academic PRAC dichotomous 1 = yes, 0 = no Probation-Aptitude PRAP dichotomous 1 = yes, 0 = no Probation-Athletic PRAT dichotomous 1 = yes, 0 = no Probation-Conduct PRCO dichotomous 1 = yes, 0 = no Probation-Honor PRHO dichotomous 1 = yes, 0 = no Intercollegiate sports ICAT dichotomous 1 = yes, 0 = no Intramural sports INTR dichotomous 1 = yes, 0 = no Competitive COMP dichotomous 1 = yes, 0 = no Mission Support MSPT dichotomous 1 = yes, 0 = no Professional PROF dichotomous 1 = yes, 0 = no Club CLUB dichotomous 1 = yes, 0 = no Recreational REC dichotomous 1 = yes, 0 = no Civil Air Patrol, Boy/Girl SRV dichotomous 1 = yes, 0 = no Scouts, Campfire Girls

The gender and race variables are self-explanatory and are designed to place cadets into gender and race groups for comparative purposes. Most questionnaires and surveys completed by USAFA cadets ask for this information, so they were included to be in agreement with common surveying practices.

The family variables consisting of family education, family income, family members, from single parent home, and first-generation college student are exploratory variables designed to identify factors present in the cadet’s lives prior to attending

USAFA. These variables were included to understand how diverse background characteristics such as socioeconomic status (SES) and other factors related to upbringing and family dynamics impact the growth and development of cadets. These variables

most closely align with the protective factors within the resiliency theory framework as they point to the family support aspects that influence cadets during their upbringing.

The intent of these variables is to understand to what degree diverse background characteristics influence the associations that may exist between these characteristics and the grit, passion, and resilience levels cadets demonstrate while attending USAFA.

The variables high school GPA percentage of max, high school athletics, and recruited college athlete are exploratory variables designed to identify factors related to the cadet’s individually diverse life experiences prior to attending USAFA. The GPA percentage of max was used instead of straight GPA because USAFA accepts a wide variety of candidates both international and domestic, some of whom complete school with GPAs measured on 3.0, 4.0, 5.0, and even 6.0 scales. To compare all these GPAs evenly, a percentage of max was calculated by dividing each GPA by its respective scale to obtain the GPA percentage each cadet had out of 100%. This made it possible to compare all the GPAs from all the cases evenly and set the groundwork to impute missing variables where applicable. These variables most closely align with the promotive factors within the resiliency theory framework as they point to the individual characteristic’s cadets have obtained and developed throughout their lives. The intent is to identify and compare the cadet’s individual characteristics developed throughout their lives with the grit and resilience levels demonstrated while at USAFA.

The performance variables consisting of honors list, grade point average (GPA), military performance average (MPA), physical education average (PEA), and overall performance average (OPA) are variables designed to understand the relationship between Grit-S score and cadet performance measures. The honors list variable

represents cadets who at some point during their career at USAFA were included in one of several different honors lists including the Dean’s list, the Commandant’s list, the

Director of Athletics list, and the Superintendent’s list. The minimum standards require an average GPA of 3.0 for the Dean’s list, an average MPA of 3.0 for the Commandant’s list, an average PEA of 3.0 for the Director of Athletics list, and all three standards must be met for inclusion on the Superintendent’s list. The OPA variable is comprised of the

GPA, MPA, and PEA variables, so to avoid multi-collinearity issues, two different regression models were used to understand the associations between OPA and the independent variables, and then again with each of the GPA, MPA, and PEA variables.

The Grit-S score, passion subscore, and resilience subscore served as the primary dependent variables for research question one and four against which all the independent variables were compared in order to identify any predictive associations. These same variables were included as independent variables in questions two and three. A total of

5,454 cases were included when comparing Grit-S score and its subscores with the other independent variables. For the cadets who completed multiple surveys, only the final survey was included in this analysis. As was mentioned previously, Dr. Duckworth developed and used the Grit-S survey to measure and understand grit levels in a variety of populations, including USMA cadets, elementary school students, and adults

(Duckworth et al., 2007). Because this study had a particular focus on resilience, all the results from the Grit-S surveys completed by cadets were included for regression and comparison. However, in order to avoid multi-collinearity issues, the Grit-S score variable was used as an independent variable in one regression model, and the passion

and resilience subscores were used similarly in another variant, and both models were compared for associations.

The Grit-S, passion, and resilience gain scores variables were derived from analyzing the cases where more than one survey was completed. Cadets who took the survey twice (n = 1,157) produced 2,314 surveys, those who took it three times (n = 171) produced 513 surveys, and those who took it four times (n = 7) produced 28 surveys, resulting in an overall multiple-survey group of 1,335 cases who produced 2,855 surveys.

These cadets completed the surveys at various times throughout their careers, at different time intervals between surveys, and were from different graduating classes, so a standardized gain score based on a common time interval between first and last survey was calculated in order to conduct a reasonably even comparison. The differences were standardized by assessing a survey time interval that captured the greatest number of cases and yet did not cover too long or too short a span of time. Analyzing the responses revealed that 73.86% of the cases had survey time intervals of between 16 and 28 months, a difference of only 12 months. Removing the cases outside the interval window

(n = 198) reduced the total number of cases by 14.8% leaving a final group of 1,137 cases for comparison. The gain score was calculated by subtracting the first Grit-S survey score from the last survey score. The same procedure was completed for the passion gain scores and resilience gain scores, and these scores were regressed on the list of independent variables to understand the relative associations. The gain scores identified any longitudinal trends to understand how cadet behavior measured by the Grit-S survey changed as they moved through their educational careers at USAFA. For gain scores, multiple regression analysis compared the gain scores to the list of independent variables,

and an additional graph was included showing the growth trend of cadet Grit-S score, passion subscore, and resilience subscore.

The exploratory attrition outcome variables of not disenrolled, voluntarily disenrolled, and involuntarily disenrolled were included to identify any association between Grit-S scores and its subscores and cadets who were voluntarily or involuntarily disenrolled from USAFA. In research questions one, two, and four, these variables were used as independent variables, but in research question three, each of these three variables was used as the dependent variable. “Not disenrolled” refers to those cadets who were not disenrolled from USAFA or those who remained enrolled through graduation, and SPSS automatically used this variable as a control for the other two attrition variables in research questions one, two, and four. Disenrollment from USAFA can either be categorized as voluntary or involuntary depending on the circumstances surrounding the disenrollment. Voluntary disenrollment occurs when cadets who have not yet received their AFSC and fully committed to the Air Force decide on their own to leave USAFA. Reasons for voluntary disenrollment include a change in career goals, medical or administrative turnback, pregnancy, medical injuries, international students who only attend for a few semesters, cadets who left to serve religious missions, psychological challenges like depression or mental distress, minor drug and alcohol infractions, or death of a family member. Involuntary disenrollment involves the mandatory removal of a cadet from USAFA for reasons such as academic deficiencies, forced medical retirement, misconduct, failing to recover from any of the six types of probation (academic, aptitude, athletic, conduct, honor, honor rehab), driving under the

influence (DUI), major drug or alcohol violations, cheating, assaulting other cadets, and violations classified under the title of honor code violation.

The exploratory, probation-related variables in this study were included to identify associations between the six types of probations at USAFA and the grit scores and subscores. The variable “probation—all types” is comprised of all probations that can occur at USAFA, namely academic probation, aptitude probation, athletic probation, conduct probation, honor probation, and honor rehab probation. In the 5,454 sample of cadets who completed the surveys, only four cadets had ever been placed on honor rehab probation and regression models excluded it from analysis. Similar to the performance variables and grit variables, regression models were completed using both the singular

“probation—all types” variable and then again with the five probation type variables in order to avoid issues with multi-collinearity.

The exploratory USAFA program variables compared seven categories of activities cadets may participate in white attending USAFA. The goal of including these categories was to identify any associations between Grit-S scores or subscores and the various categories of programs and activities. The seven categories are intercollegiate sports (which include both varsity and junior varsity level sports), intramural sports, competitive programs, mission support programs, professional programs, clubs, and recreational programs and activities. The list of specific activities in each category is long and almost all cadets participate in more than one category during their careers at

USAFA. A full list of specific activities in each category is provided in Appendices C through I, but a brief explanation of each category is instructive.

Intercollegiate sports (Appendix C) includes all sports, male and female, in which cadets compete with other Universities. Participation in intercollegiate sports is time consuming with daily practices, competitions on weekends, and extended periods of travel to compete with institutions all over the United States. These sports include football, basketball, swimming, water polo, fencing, tennis, and several others.

Intramural sports (Appendix D) include sports cadets participate in seasonally at the school, competing against other USAFA cadets, and is by far the most inclusive of any category since the majority of cadets are required to participate. The time requirement for these sports is much less than intercollegiate since there are fewer practices and no travel required. These sports include flag football, ultimate frisbee, boxing, rugby, racquetball, team handball, futsal (a form of indoor soccer), and several others.

The remaining categories include activities and programs cadets participate in on their own time or in connection with the classes they may be taking. Competitive activities (Appendix E) include both athletic and non-athletic activities like alpine skiing, performing arts, forensics, cycling, rodeo, parachute/skydiving, triathlon club, coed volleyball, and several others. Mission support programs (Appendix F) include choir, drum & bugle corp., orchestra, combat pistol, cyber warfare, the honor guard, the

Sandhurst Competition Team (which competes annually), the Arnold Air Society (which is an Air Force service organization), and several others. Professional programs

(Appendix G) include astrophysics, future business leaders, history, language club, chemistry, Chinese, tutoring, robotics, forensics, STEM (science, technology, engineering, mathematics), and several others. Club activities (Appendix H) include the

Hispanic/Latino club, Secular Cadet Alliance, Native American Heritage, car club,

Karate, the international club, steel script, and a few others. Recreational activities

(Appendix I) include Aikido, karate, archery, equestrian, model building, paintball, mountaineering, trap & skeet, wargaming, some performing arts, and several others.

Finally, the exploratory variable comprised of individuals who had participated in the Civil Air Patrol, Boy and/or Girl Scouts, or Campfire Girls prior to attending USAFA was included to see if the characteristic of service or being active in service organizations was associated with Grit-S score and its subscores. I made the decision to include this variable because from my own experiences in Boy Scouts for over 15 years both as a youth and an adult, I noticed an increase in gritty and resilient behavior from individuals who participated in these types of organizations. This variable is most closely aligned with the promotive factors within the resiliency theory framework as they point to other characteristics cadets may have obtained and developed throughout their lives. The

CAMIS department at USAFA routinely collects this information from applicants, making it possible to include it in regression analysis.

While the total number of variables included in this research study was particularly large, the large sample size (N = 5,454) made it possible to utilize all 41 independent variables while maintaining adequate statistical power. Taking into account, the general guideline of including at a minimum 10 cases per variable, this study far surpassed the necessary 410 cadets, enabling a unique opportunity to explore a wide variety of variables and identify any factors that previously had gone unresearched.

Missing values. While compiling and organizing the data, it became apparent that there were missing values in a few of the variables. An analysis of missing values revealed that 10 of the 41 (24.39%) of the variables were missing at least one value,

5,050 of the 5,454 (92.59%) of the cases were missing at least one value, and that 8,214

(3.673%) of all values were missing. Missing data occurred in cadet information because

CAMIS did not record all the information for all the cadets, but a review of the missing value patterns found that no patterns were evident. Conducting Little’s MCAR test of the variables in SPSS was statistically significant at p < .000, and multiple imputation was deemed appropriate. Due to the large sample size, the decision was made to impute values for all 10 variables missing data using 10 imputations for each variable. The imputation data for each of the variables is shown in Table 3.

Table 3

Statistics of Variables with Imputed Values

Missing Percent Total Imputed Variables Data Points Missing Values First generation college study (FGEN) 3,671 67.30% 36,710 Family member education level (FEDU) 2,225 40.80% 22,250 From a single parent home (FSGP) 1,195 21.90% 11,950 Overall performance average (OPA) 295 5.40% 2,950 Family income (FINC) 204 3.70% 2,040 Number of family members (FMEM) 203 3.70% 2,030 Physical education average (PEA) 174 3.20% 1,740 Military performance average (MPA) 110 2.00% 1,100 Grade point average (GPA) 105 1.90% 1,050 High school grade point average percentage (HS_GPA%) 32 0.60% 320 Note: N = 5,454 after imputing missing data

After reviewing the number and percentages for the missing data values, regressions were completed and the pooled regression results were compared using first the dataset without any variable imputations, a second dataset with the seven imputed variables missing less than 5.40% of the data, and a third dataset imputing values for all ten variables. An analysis of the regressions models for each of the research questions

using each of these datasets found the best fitting model using the pooled results from the dataset imputing missing data to all ten variables, and this was the model that was selected for the primary analysis in this study. While the first-generation college student variable had a high percentage of missing values (67.3%), the variable was included because doing so resulted in a better fitting model. However, it is important to take this into account when interpreting both the statistical significance and impact of this variable. While the results from the other datasets were reviewed and analyzed for comparison purposes with the other datasets, only the results from the dataset with ten imputed variables will be presented in this study.

Data analysis

Analysis of the combined Grit-S survey data and cadet information data was conducted using the Statistical Package for the Social Sciences (SPSS) version 25, release 25.0.0.1, 64-bit edition (IBM Corp., 2017). Analysis of the five research questions required using different regression models tailored specifically to each question. Each of the regression models is presented separately in relation to the research questions each answered.

Research question 1 (RQ1) – Grit-S score vs Cadet Predictors. The first research question asked which cadet predictors are associated with Grit-S score. To answer this question, I created a linear regression model to regress the continuous dependent variable of USAFA cadet Grit-S survey score, where Y is the grit score of the i-th cadet, on a set of 36 independent variables consisting of the variables shown on Table

4 at the end of this section. The regression model was:

Yi(grit) = β0 + Sβ1-36(X1-36)…+ ei

Two similar regression models also compared the relationship between the passion and resiliency subscales and the same list of independent variables. Additionally, a second variant of each of these three regression models was created using the OPA variable in place of the GPA, MPA, and PEA variables, since the three are simply added together to create OPA. Likewise, the “probation—all types” variable was used in place of the five different probation variation variables since combining these together creates the PROB variable. In both cases, this was done to avoid multi-collinearity between variables. Finally, each of these six variations were run without any imputed variables, with seven imputed variables, and with 10 imputed variables, resulting in 18 separate regressions computed and compared to identify model fit and significance. As stated previously, the models using imputed data for all 10 variables demonstrated the best fitting models and were used throughout this study.

Categorical dummy codes were created for non-continuous independent variables, and all independent variables were grand-mean centered. Dummy codes were created for specified demographic variables when more than two categories existed. For binary variables, participation or membership to a group was coded as 1, and non-participation or non-membership was coded as 0. Overall regression results for all RQ1 models are presented in chapter four.

Table 4

RQ1 Variables Used in Linear Regression Analysis

Dependent variables Independent variables Abbrev. Codes Grit-S survey score Gender GEN 1 = female, 0 = male (GRIT) Passion subscore Black BLK 1 = yes, 0 = no (PASS) Resilience subscore Caucasian CAUC 1 = yes, 0 = no (RES)

Hispanic HISP 1 = yes, 0 = no American Indian AMIN 1 = yes, 0 = no Asian ASN 1 = yes, 0 = no Native Hawaiian / Pacific NHPI 1 = yes, 0 = no Islander

Unknown UNK 1 = yes, 0 = no Family education FED 1=High School, 2=Some college, 3=Associates, 4=Bachelors, 5=Graduate, 6=unknown Family income FIN 1= don't know, 2= <25K, 3= 25K-74,999K, 4= 75K- 124,999K, 5= 125K-174,999K, 6= >175K Family members FMEM 1=1, 2=2, 3=3, 4=4, 5=5, 6=6, 7=7, 8=8, 9=9, 10=10 or more

From single parent FSP 1 = yes, 0 = no First generation college FGC 1 = yes, 0 = no High school GPA HS_GPA% scale percentage of max High school athletics HAS 1 = yes, 0 = no Recruited college athlete RCA 1 = yes, 0 = no Honors list HON 1 = yes, 0 = no Grade point average GPA scale Military performance MPA scale average Physical education PEA scale average Not disenrolled NDSN 1 = yes, 0 = no Disenrolled voluntarily VOL 1 = yes, 0 = no Disenrolled involuntarily INVOL 1 = yes, 0 = no Academic probation PRAC 1 = yes, 0 = no Aptitude probation PRAP 1 = yes, 0 = no Athletic probation PRAT 1 = yes, 0 = no Conduct probation PRCO 1 = yes, 0 = no Honor probation PRHO 1 = yes, 0 = no Intercollegiate sports ICSP 1 = yes, 0 = no

RQ1 Variables Used in Linear Regression Analysis

Dependent variables Independent variables Abbrev. Codes Intramural sports IMSP 1 = yes, 0 = no Competitive programs COM 1 = yes, 0 = no Mission support programs MSPT 1 = yes, 0 = no Professional programs FESS 1 = yes, 0 = no Club programs CLU 1 = yes, 0 = no Recreational programs REC 1 = yes, 0 = no Civil Air Patrol, Boy/Girl SRV 1 = yes, 0 = no Scouts, Campfire Girls

This regression model explored the relationship between Grit-S scores, passion, and resilience subscores, and the list of independent variables, revealing both positive and negative associations with performance scores and the magnitude and statistical significance of each association.

Research question 2 (RQ2) – Performance vs Grit-S Score. The second research question asked to what extent Grit-S score is associated with the cadet overall performance average (OPA). To answer this question, I created a linear regression model to regress the continuous dependent variables of OPA and the subfactors of grade point average (GPA), military performance average (MPA), and physical education average

(PEA), where Y is the score of the i-th cadet, on a set of 35 independent variables consisting of the variables shown on Table 5 at the end of this section. It is important to note that since GPA + MPA + PEA = OPA and since USAFA assigns a different weight to GPA (50%), MPA (40%), and PEA (10%) (The United States Air Force Academy,

2017b), I created unique regression models for each of these factors as the dependent variable. After performing the regressions, I used the Bonferroni correction (Shi, Pavey,

& Carter, 2012) to include alpha levels .05, .025, and .0125 to adjust for multiple comparisons. The resulting regression model for the dependent variables was as follows:

Yi(opa) = β0 + Sβ1-35(X1-35)…+ ei

Three similar regression models also compared the relationship between the GPA,

MPA, and PEA subscores with the same list of independent variables. Additionally, a second variant of each of these three regression models was created using the GRIT variable in place of the passion and resilience subscales variables, since both together comprise the overall Grit-S score. Similarly to the previous research question, the

“probation—all types” variable was used in place of the five different probation variation variables since combining these together creates the PROB variable. In both cases, this was done to avoid multi-collinearity between variables. Finally, each of these eight variations were run without any imputed variables, with seven imputed variables, and with 10 imputed variables, resulting in 24 separate regressions computed and compared to identify model fit and significance. As stated previously, the models using imputed data for all 10 variables demonstrated the best fitting models and were used throughout this study.

Categorical dummy codes were created for non-continuous independent variables, and all independent variables were grand-mean centered. Dummy codes were created for specified demographic variables if more than two categories existed. For binary variables, participation or membership to a group was coded as 1, non-participation or non-membership was coded as 0, and the overall regression results for all RQ2 models are presented in chapter four.

Table 5

RQ2 Variables Used in Linear Regression Analysis

Dependent variables Independent variables Abbrev. Codes Overall performance Gender GEN 1 = female, 0 = male average (OPA) Grade point average Black BLK 1 = yes, 0 = no (GPA) Military performance Caucasian CAUC 1 = yes, 0 = no average (MPA) Physical education Hispanic HISP 1 = yes, 0 = no average (PEA)

American Indian AMIN 1 = yes, 0 = no Asian ASN 1 = yes, 0 = no Native Hawaiian / NHPI 1 = yes, 0 = no Pacific Islander

Family members FMEM 1=1, 2=2, 3=3, 4=4, 5=5, 6=6, 7=7, 8=8, 9=9, 10=10 or more

From single parent FSP 1 = yes, 0 = no First generation college FGC 1 = yes, 0 = no High school GPA HS_GP% scale percentage of max High school athletics HAS 1 = yes, 0 = no Recruited college RCA 1 = yes, 0 = no athlete Honors list HON 1 = yes, 0 = no Passion subscore PASS scale Resilience subscore RES scale Not disenrolled NDSN 1 = yes, 0 = no Disenrolled voluntarily VOL 1 = yes, 0 = no Disenrolled INVOL 1 = yes, 0 = no involuntarily Academic probation PRAC 1 = yes, 0 = no Aptitude probation PRAP 1 = yes, 0 = no Athletic probation PRAT 1 = yes, 0 = no Conduct probation PRCO 1 = yes, 0 = no Honor probation PRHO 1 = yes, 0 = no Intercollegiate sports ICSP 1 = yes, 0 = no

RQ2 Variables Used in Linear Regression Analysis

Dependent variables Independent variables Abbrev. Codes Intramural sports IMSP 1 = yes, 0 = no Competitive programs COM 1 = yes, 0 = no Mission support MSPT 1 = yes, 0 = no programs Professional programs FESS 1 = yes, 0 = no Club programs CLU 1 = yes, 0 = no Recreational programs REC 1 = yes, 0 = no Civil Air Patrol, SRV 1 = yes, 0 = no Boy/Girl Scouts, Campfire Girls

This regression model explored the relationship between the performance measures OPA, GPA, MPA, and PEA and the list of independent variables, revealing both positive and negative associations with performance scores and the magnitude and statistical significance of each association.

Research question 3 (RQ3) – Attrition vs Grit-S Score. The third research question asked to what extent Grit-S score is associated with cadet attrition. This question was analyzed using one regression model with two different variants to understand the relationship between Grit-S score for those who were not disenrolled and for those who were disenrolled from USAFA either voluntarily or involuntarily. The first regression model regressed the binary dependent variable of attrition for those cadets who did not leave USAFA prior to graduation, where disenrollees are coded as 0 and non- disenrollees are coded as 1, and where i is the log odds of non-attrition of 퐿푛 the i-th cadet, on a set of 30 independent variables shown in Table 6 at the end of this section. This regression model for non-disenrollees versus independent variables is shown below:

= β0 + Sβ1-35(X1-35)…+ ei 퐿푛

A second variant of this regression model was created using the OPA variable in place of the GPA, MPA, and PEA variables, since the three are simply added together to create OPA. Likewise, the GRIT variable was used in place of the passion and resilience subscales variables since both together comprise the overall Grit-S score. In both cases, this was done to avoid multi-collinearity between variables. Finally, each of the variations were run without any imputed variables, with seven imputed variables, and with 10 imputed variables, resulting in six separate regressions computed and compared to identify model fit and significance. As stated previously, the models using imputed data for all 10 variables demonstrated the best fitting models and were used throughout this study.

For this model, I created categorical dummy codes for non-continuous independent variables and all independent variables were grand-mean centered. Dummy codes were created for specified demographic variables when more than two categories existed. For binary independent variables, participation or membership to a group was coded as 1, and non-participation or non-membership was coded as 0.

Table 6

RQ3 Variables Used in Logistic Regression Analysis

Dependent variables Independent variables Abbrev. Codes Not disenrolled Gender GEN 1 = female, 0 = male (NDSN) Disenrolled (DSN) Black BLK 1 = yes, 0 = no Caucasian CAUC 1 = yes, 0 = no Hispanic HISP 1 = yes, 0 = no American Indian AMIN 1 = yes, 0 = no Asian ASN 1 = yes, 0 = no Native NHPI 1 = yes, 0 = no Hawaiian/Pacific Islander Unknown UNK 1 = yes, 0 = no Family education FED 1=High School, 2=Some college, 3=Associates, 4=Bachelors, 5=Graduate, 6=unknown Family income FIN 1= don't know, 2= <25K, 3= 25K-74,999K, 4= 75K- 124,999K, 5= 125K-174,999K, 6= >175K Family members FMEM 1=1, 2=2, 3=3, 4=4, 5=5, 6=6, 7=7, 8=8, 9=9, 10=10 or more From single parent FSP 1 = yes, 0 = no First generation FGC 1 = yes, 0 = no college High school GPA HS_GPA% scale percentage of max

High school athletics HAS 1 = yes, 0 = no Recruited college RCA 1 = yes, 0 = no athlete Honors list HON 1 = yes, 0 = no Grade point average GPA scale Military performance MPA scale average

Physical education PEA scale average

Passion subscore PASS scale Resilience subscore RES scale Intercollegiate sports ICSP 1 = yes, 0 = no Intramural sports IMSP 1 = yes, 0 = no Competitive programs COM 1 = yes, 0 = no Mission support MSPT 1 = yes, 0 = no programs

Professional programs FESS 1 = yes, 0 = no Club programs CLU 1 = yes, 0 = no

RQ3 Variables Used in Logistic Regression Analysis

Dependent variables Independent variables Abbrev. Codes Recreational programs REC 1 = yes, 0 = no Civil Air Patrol, SRV 1 = yes, 0 = no Boy/Girl Scouts, Campfire Girls

This model explored the relationship between changes in enrollment status and the list of independent variables. This revealed the odds that a member of a particular independent variable group would also be a member of the dependent variable group, which further clarifies the relationship between attrition and other cadet factors.

Research question 4 (RQ4) – Changes in Grit-S Score over time vs Cadet

Predictors. The fourth research question asked what the change in cadet Grit-S score is over time. Similarly to RQ1, I also created a linear regression model to regress the continuous dependent variable of Grit-S survey gain score, where Y is the grit gain score of the i-th cadet, on a set of 35 independent variables consisting of the variables shown in

Table 7 at the end of this section. The resulting regression model for the dependent variables was as follows:

Yi(gtgn) = β0 + Sβ1-32(X1-32)…+ ei

Two similar regression models also compared the relationship between the passion and resiliency gain scores and the same list of independent variables. As in RQ1, a second variant of each of these three regression models was created using the OPA variable in place of the GPA, MPA, and PEA variables, since the three are simply added together to create OPA. The “probation—all types” variable was used in place of the five different probation variation variables since combining these together creates the PROB variable. In both cases, this was done to avoid multi-collinearity between variables.

Finally, each of these six variations were run without any imputed variables, with seven imputed variables, and with 10 imputed variables, resulting in 18 separate regressions computed and compared to identify model fit and significance. As stated previously, the models using imputed data for all 10 variables demonstrated the best fitting models and were used throughout this study.

I created categorical dummy codes for non-continuous independent variables and all independent variables were grand-mean centered. Dummy codes applied to demographic information used 0, 1, 2, etc. to specify descriptive statistics. For binary variables, participation or membership to a group was coded as 1, and non-participation or non-membership was coded as 0.

Table 7

RQ4 Variables Used in Linear Regression Analysis

Dependent variables Independent variables Abbrev. Codes Grit-S gain score Gender GEN 1 = female, 0 = male (GTGN) Passion gain score Black BLK 1 = yes, 0 = no (PNGN) Resiliency gain Caucasian CAUC 1 = yes, 0 = no score (RYGN)

Hispanic HISP 1 = yes, 0 = no American Indian AMIN 1 = yes, 0 = no Asian ASN 1 = yes, 0 = no Native Hawaiian/Pacific NHPI 1 = yes, 0 = no Islander

Unknown UNK 1 = yes, 0 = no Family education FED 1=High School, 2=Some college, 3=Associates, 4=Bachelors, 5=Graduate, 6=unknown

Family income FIN 1= don't know, 2= <25K, 3= 25K-74,999K, 4= 75K- 124,999K, 5= 125K-174,999K, 6= >175K

Family members FMEM 1=1, 2=2, 3=3, 4=4, 5=5, 6=6, 7=7, 8=8, 9=9, 10=10 or more

RQ4 Variables Used in Linear Regression Analysis

Dependent variables Independent variables Abbrev. Codes From single parent FSP 1 = yes, 0 = no First generation college FGC 1 = yes, 0 = no High school GPA HS_GPA% scale percentage of max

High school athletics HAS 1 = yes, 0 = no Recruited college athlete RCA 1 = yes, 0 = no Honors list HON 1 = yes, 0 = no Grade point average GPA scale Military performance MPA scale average

Physical education PEA scale average Not disenrolled NDSN 1 = yes, 0 = no Disenrolled voluntarily VOL 1 = yes, 0 = no Disenrolled INVOL 1 = yes, 0 = no involuntarily Academic probation PRAC 1 = yes, 0 = no Aptitude probation PRAP 1 = yes, 0 = no Athletic probation PRAT 1 = yes, 0 = no Conduct probation PRCO 1 = yes, 0 = no Honor probation PRHO 1 = yes, 0 = no Intercollegiate sports ICSP 1 = yes, 0 = no Intramural sports IMSP 1 = yes, 0 = no Competitive programs COM 1 = yes, 0 = no Mission support MSPT 1 = yes, 0 = no programs Professional programs FESS 1 = yes, 0 = no Club programs CLU 1 = yes, 0 = no Recreational programs REC 1 = yes, 0 = no Civil Air Patrol, SRV 1 = yes, 0 = no Boy/Girl Scouts, Campfire Girls

This model explored the relationship between changes in Grit-S gain score, and passion and resilience gain scores, and the list of independent variables, revealing both the positive and negative associations with gain scores and the magnitude and statistical significance of each association.

Research question 5 (RQ5) – Recommended Future Research, Policies, and

Practices. The last research question concluded this study by using the results of the data

analysis to recommend future research opportunities and possible changes to policies and practices to build cadet resiliency at USAFA. To answer this question, I consolidated, organized, and present the themes in a comparative manner to identify both the common and unique themes and evaluate whether or not these themes are in accordance with the information on resiliency research available in the literature. This last question is essential to bring the results of the study together and identify the emergent themes as well as the opportunities and next steps for future research opportunities and to provide policy and practice recommendations.

Limitations

Limitations in this study may have included several types of self-report survey bias explained in detail below, selection and non-response/response bias from those who answered the survey, and the nature of mandatory surveys completed during BCT. In their article regarding comparative values of Grit-S scores, Dr. Duckworth and Dr.

Yeager explain how self-report questionnaires and surveys present challenges for participants in the form of reference bias, acquiescence bias, social desirability bias, and faking (Duckworth & Yeager, 2015). Reference bias refers to the way different individuals decode questions and the frame of reference each person uses to answer the survey questions. Acquiescence bias refers to the inclination of students to select answers they believe their superiors would want them to select and is of particular interest since this is prevalent in military settings. Social desirability bias refers to the fact that society views certain behaviors as desirable or undesirable, and students often want to select answers more favorable to society. Faking simply refers to the bias created when participants blatantly fake or make up answers irrelevant to their beliefs, behaviors,

or actual performance (Duckworth & Yeager, 2015). In this particular study, some cadets left the identification number blank, and some entered a false number such as “00000”, or

“12345”, or simply a different number altogether. Since the data is historical in nature, it is impossible to determine why exactly this was done. While it is difficult if not impossible to reduce reference bias and faking when conducting research using historical data, the anonymous way in which the survey data was collected and the large survey sample sizes of cadets both served to reduce the effects of acquiescence bias and social desirability bias. Conducting face-to-face interviews and discussion groups could provide a means to clarify survey results, and I include this in the recommendation for future research opportunities.

The effects of both selection bias and non-response/response bias may also have been introduced because the Grit-S survey was administered in a voluntary manner to all classes of USAFA cadets with no method to control who responded. Since the response rates for both the survey groups and class years within groups ranged from 25% to 100%, it may not be possible to ensure the sample was a true representation of each cadet class or of the entire population of USAFA cadets at the time they completed the survey.

Annually, there are two Designated Survey Assessment Times (DSAT), one in the Fall and one in the Spring. Cadets are given time off from classes and are highly encouraged to participate in the surveys as a way to provide the feedback necessary to modify and improve programs. While it is unknown what exactly impacts the response rates,

USAFA employees who have conducted other surveys mentioned that time for cadets is extremely limited due to the large numbers of surveys cadets are asked to complete during the DSAT periods. Also, the limited amount of free time available to complete

surveys, a high premium on available time to study or relax, and low cadet interest serve as possible factors affecting response rates. However, the large survey group sample sizes of cadets served to reduce these effects.

Additionally, while most of the survey respondents voluntarily completed the surveys, cadets attending BCT are required to complete a series of forms and surveys.

The Grit-S survey is one of these surveys, and so participation in these survey sessions is

100%. This may affect survey responses since participants have no choice in participation. Dropping the results from the three mandatory survey groups may serve to alleviate this effect, but the large survey group sample size collected from the other seven survey groups also served to reduce the negative effects of mandatory participation.

Ethics

Ethical considerations particularly important to this study included the protection of participants’ confidentiality, ensuring the quality of the research process itself, and avoiding bias as the researcher (Strauss & Corbin, 1998). In regard to the study participants, who are all USAFA cadets, extreme care must be given to protect the anonymity and confidentiality of the research participants (Creswell, 2003). The raw data obtained from USAFA contained partial cadet identification numbers used to correlate the cadet grit scores with demographic, performance, and program participation data. These identification numbers were essential in order to correlate and organize the data from various sources and prepare the datasets for regression analysis. However, since the primary investigator is a USAFA employee, and since no non-USAFA personnel would have access to the raw data, the USAFA IRB determined that a de- identification plan for this research study was unnecessary.

It was also requisite for the research study itself to adhere to the rigors and structure of the research design set forth, allowing for some flexibility to achieve maximum research quality but respecting the process and not allowing undisciplined or careless actions to taint the research results (Strauss & Corbin, 1998). The research addressed the issues posed in the research questions (Creswell, 2003) but did not extend past those criteria into areas not pertaining to the research questions or not previously approved by the appropriate IRB authorities (Creswell & Clark, 2011). This ensures the research design and rigor lends strength, credibility, and focus to the results of the research and avoids diluting results that might obscure outcomes.

Finally, as the researcher, I had the responsibility of presenting the findings of this study in their actual essence unaffected either by my own personal biases or by the biases of others within the research environment. It was also my responsibility to do my best to produce the most accurate and highest quality work to advance the knowledge of resiliency theory both in regard to the field of social sciences and in respect to the profession of military services (Strauss & Corbin, 1998). While this is a logical requirement, my duty to add to the body of knowledge increasing the effectiveness and performance of military officers in the United States of America, while ensuring no harm or damage to individuals, was paramount to other influences or personal desires.

CHAPTER IV

RESULTS

The purpose of this study was to understand the relationship between USAFA cadet resiliency and a variety of individual characteristics using quantitative regression analyses to determine the extent to which individual characteristics were associated with individual grit, passion, and resiliency scores. This results section presents the data collected during the analysis process for each of the research questions along with an interpretation of the data. After this introduction, a complete analysis of each research question follows, presenting the associated descriptive statistics, explanation of the results, tables showing the regression coefficients used and resulting data, and an interpretation of the data. This will set the stage for the discussion of the results, which is presented in chapter five. The specific research questions to used explore this relationship were:

1. To what extent are cadet characteristics, including demographics and

participation in USAFA clubs and programs, associated with Grit-S score?

2. To what extent is Grit-S score associated with cadet overall performance

average (OPA)?

3. To what extent is Grit-S score associated with cadet attrition?

4. What is the change in cadet Grit-S score over time?

5. Based on this research study, what are some recommendations for future

research, policy development, and practices to build cadet resiliency at

USAFA?

Review of Descriptives and Research Questions

Descriptives

USAFA cadets completed the Grit-S survey at various times as part of classes, during the biannual Designated Survey and Assessment Time (DSAT), or as a mandatory requirement before beginning basic cadet training (BCT) prior to commencing their freshman year. Participants consisted of basic cadets, freshman, sophomores, juniors, and seniors ranging in age 17 to 27 years of age who completed the Grit survey from

2009 to 2017. As shown in Table 8 below, survey participants included in the final sample consisted of both male (n = 3,981, 73.0%) and female (n = 1,473, 27.0%) cadets from a variety of ethnic backgrounds, including American Indian (n = 78, 1.4%), Asian

(n = 483, 8.9%), Black (n = 407, 7.5%), Caucasian (n = 3,667, 67.2%), Latin/Hispanic (n

= 502, 9.2%), Native Hawaiian/Pacific Islander (n = 104, 1.9%), and a group marked in the surveys as “unknown” (n = 213, 3.9%). For research questions one, two, and four,

Pearson’s correlation coefficient benchmarks (.1 = small, .3 = medium, .5 = large) were used to evaluate and compare all ß values (Cohen, 1988). The mean and standard deviation for the regression variables used in the models for research question one are presented in Table 8.

100

Table 8

Descriptive Statistics Comparing Grit Variables with Cadet Predictors

Variables Mean/Proportion SD Grit-S survey score 3.536 .536 Passion subscore 3.125 .768 Resilience subscore 3.946 .600 Gender 27.00% Black 7.00% Hispanic 9.00% American Indian 1.00% Asian 9.00% Native Hawaiian / Pacific Islander 2.00% Race unknown 4.00% Family education 3.860

High school 4.50% Some college 7.00% Associate degree 4.60% Bachelor's degree 21.60% Graduate degree 20.80% Unknown 0.60% Family income 4.210

Unknown 6.20% <25K 3.40% 25-74,999K 16.60% 75-124,999K 20.10% 125-174,999K 20.10% ≥175K 20.90% Family members 4.300

1 2.70% 2 4.80% 3 15.10% 4 35.40% 5 23.50% 6 9.90% 7 2.90% 8 1.10% 9 0.40% ≥10 0.50%

101

Descriptive Statistics Comparing Grit Variables with Cadet Predictors

Variables Mean/Proportion SD From single parent home 16.00% First generation college student 45.00% High school GPA percentage of max 95.22% .108 High school athletics 93.00% Recruited college athlete 21.00% Honors list 84.00% Grade point average (GPA) 2.998 .549 Military performance average (MPA) 3.257 .272 Physical education average (PEA) 2.760 .423 Voluntary disenrollment 7.00% Involuntary disenrollment 2.00% Academic probation 28.00% Aptitude probation 6.00% Athletic probation 7.00% Conduct probation 5.00% Honor probation 2.00% Intercollegiate sports 36.00% Intramural sports 73.00% Competitive programs 16.00% Mission support programs 19.00% Professional programs 3.00% Club programs 3.00% Recreational programs 5.00% Civil Air Patrol, Boy/Girl Scouts, Campfire Girls 30.00% Overall performance average (OPA) 3.016 .352 Probations—all types 36.00% Note. N = 5,454

The total number of survey responses collected was 8,331, and after screening for missing or erroneous data, 6,974 surveys were available for analysis. This sample of surveys completed by USAFA cadets (N = 5,454) included 4,119 cadets who took the survey one time, 1,157 cadets who took the survey twice, 171 cadets who took the survey three times, and 7 cadets who took the survey four times. For the first three research

102

questions, only the final survey from each respondent was included for analysis. For research question four, cases with multiple responses were used for the trend analysis.

RQ1 – Grit-S Score vs Cadet Predictors

The first research question explored which cadet predictors were associated with

Grit-S score. Separate models were completed using Grit-S score, passion subscore, and resiliency subscore each separately as the dependent variable, and two sub-models were completed separating out each of the performance and probation variables, and then combining those into the OPA and “probation—all types” variables.

Grit-S score. The first set of models compared Grit-S survey scores with the list of cadet predictor variables as shown in Tables 9 and 10 below. For comparison, Model

1.1a included the performance variables of a cadet’s grade point average (GPA), military performance average (MPA), and physical education average (PEA), and Model 1.1b used the overall performance average (OPA) predictor variable since the GPA, MPA, and

PEA are combined to create the cadet OPA. The same procedure was applied to the probation variables, and Model 1.1a included academic, aptitude, athletic, conduct, and honor probations while Model 1.1b combined the five probation types into a

“probation—all types” variable. The race variables were dummy coded, and Caucasian was used as the control variable.

Regression results indicated that for Model 1.1a, the overall model significantly predicted Grit-S score (R2 = .049, p = .048) and accounted for 4.9% of variance in Grit-S score. A summary of regression coefficients is presented in Table 9 below and shows that 12 of the 34 variables significantly contributed to the model, controlling for the other covariates. These variables included gender (B = .048, p = .005), Asian (B = -.053, p =

103

.038), family members (B = -.018, p = .001), first generation college student (B = -.039, p

= .052), high school GPA percentage of max (B = -.003, p < .001), honors list (B =.099, p < .001), GPA (B =.092, p < .001), PEA (B =.082, p < .001), intercollegiate sports (B

=.052, p = .018), intramural sports (B =.074, p < .001), mission support programs (B

=.077, p < .001), and recreational programs (B = -.068, p = .045).

Table 9

Model 1.1a Regression Results Comparing Grit-S Scores with Cadet Predictors

CI Lower CI Upper Variables B SE ß p bound bound Gender .048 .017 .040 .005 .015 .082

Black .037 .029 .018 .198 -.019 .093

Hispanic .027 .025 .015 .278 -.022 .077

American Indian .079 .060 .018 .191 -.039 .197

Asian -.053 .026 -.028 .038 -.104 -.003

Native Hawaiian/Pacific Islander -.030 .052 -.008 .563 -.133 .073

Race unknown -.012 .038 -.004 .758 -.085 .062

Family education -.012 .007 -.029 .106 -.026 .003

Family income -.005 .006 -.013 .428 -.017 .007

Family members -.018 .006 -.048 .001 -.029 -.007

From single parent home -.025 .024 -.017 .298 -.072 .022

First generation college student -.039 .020 -.036 .052 -.078 .000

High school GPA percentage of max -.003 .001 -.052 .000 -.004 -.001 High school athletics -.025 .028 -.012 .369 -.081 .030

Recruited college athlete .027 .026 .021 .294 -.024 .078

Honors list .099 .025 .068 .000 .051 .148

Grade point average (GPA) .092 .021 .094 .000 .051 .132

Military performance average (MPA) .069 .038 .035 .069 -.005 .143 Physical education average (PEA) .082 .022 .065 .000 .040 .125 Voluntary disenrollment .018 .029 .009 .528 -.038 .075

Involuntary disenrollment .028 .061 .006 .649 -.092 .147

Academic probation .009 .020 .008 .643 -.030 .049

Aptitude probation -.044 .052 -.020 .396 -.146 .058

Athletic probation .010 .030 .005 .740 -.049 .069

Conduct probation .009 .055 .004 .868 -.098 .117

Honor probation .024 .047 .007 .608 -.068 .117

104

Model 1.1a Regression Results Comparing Grit-S Scores with Cadet Predictors

CI Lower CI Upper Variables B SE ß p bound bound Intercollegiate sports .052 .022 .046 .018 .009 .095

Intramural sports .074 .020 .061 .000 .034 .113

Competitive programs .033 .020 .023 .108 -.007 .072

Mission support programs .077 .019 .056 .000 .040 .114

Professional programs -.038 .043 -.012 .383 -.123 .047

Club programs -.050 .045 -.015 .270 -.138 .039

Recreational programs -.068 .034 -.027 .045 -.134 -.001

Civil Air Patrol, Boy/Girl Scouts, .012 .016 .011 .442 -.019 .044 Campfire Girls

2 Note. Model 1.1a R = .049, p < .001, Confidence Interval (CI) = 95%

The results for Model 1.1a show that, on average, when controlling for the other covariates, Grit-S scores for female cadets were predicted to be higher than male cadets by .048 points (B = .048, p = .005) and lower for Asian cadets than white cadets by .053 points (B = -.053, p = .038). For each one-person increase in number of family members, cadet Grit-S scores were predicted to be lower by .018 points (B = .018, p =

.001). Grit-S scores for cadets who were first generation college students were predicted to be lower by .039 points (B = .039, p = .052) than for those who were not first- generation college students. For every one-point increase in high school GPA percentage of max, Grit-S score was predicted to be lower by .003 points (B = .003, p < .001). Grit-

S scores for cadets who were on the USAFA honors list were predicted to be higher by

.099 points (B = .099, p < .001) than those who had never been on the list. For every one-point increase in cadet GPA, Grit-S score was predicted to be higher by .092 points

(B = .092, p < .001). For every one-point increase in cadet PEA, Grit-S score was predicted to be higher by .082 points (B = .082, p < .001). Grit-S scores for cadets who participated in intercollegiate sports were predicted to be higher by .052 points (B = .052,

105

p = .005) than non-participants. Grit-S scores for cadets who participated in intramurals were predicted to be higher by .074 points (B = .074 points, p < .001) than non- participants. Grit-S scores for cadets who participated in mission support programs were predicted to be higher by .077 points (B = .077, p < .001) than non-participants. Grit-S scores for cadets who participated in recreational programs were predicted to be lower by

.068 points (B = .068, p = .045) than non-participants.

Comparing the absolute value of the standardized regression coefficient (ß) for each of the statistically significant variables revealed that the variables demonstrated little practical significance according to Cohen’s broad guidelines, a comparison I make here and elsewhere in this dissertation with caution as the applicability of these guidelines for this research needs further exploration. As such, my conclusions about the importance of these variables will draw from both these standardized coefficients and the unstandardized coefficients that are expressed in the raw metric of the outcome measure.

These guidelines were used throughout research question one to provide practical effect size comparisons and to discuss the practical significance of each variable. The practical significance of the variables, listed in order of their ß values, were as follows: GPA (ß =

.094), honors list (ß = .068), PEA (ß = .065), intramural sports (ß = .061), mission support programs (ß = .056), high school GPA percentage of max (ß = -.052), family members (ß = -.048), intercollegiate sports (ß = .046), gender (ß = .040), first generation college student (ß = -.036), Asian (ß = -.028), and recreational programs (ß = -.027).

Regression results indicated that for Model 1.1b, the overall model also significantly predicted Grit-S score (R2 = .052, p = .048) and accounted for 5.2% of variance in Grit-S score. A summary of regression coefficients is presented in Table 10

106

below and shows that 12 of the 34 variables significantly contributed to the model, controlling for the other covariates. These variables included gender (B = .038, p =

.023), Asian (B = -.053, p = .039), family members (B = -.017, p = .002), high school

GPA percentage (B = -.003, p < .001), honors list (B = .098, p < .001), intercollegiate sports (B = .066, p = .002), intramural sports (B = .066, p = .001), competitive programs

(B = .040, p = .044), mission support programs (B = .075, p < .001), recreational programs (B = -.066, p = .051), and OPA (B = .262, p < .001).

107

Table 10

Model 1.1b Regression Results Comparing Grit-S Scores with Cadet Predictors

CI Lower CI Upper Variables B SE ß p bound bound Gender .038 .017 .031 .023 .005 .071

Black .047 .028 .023 .098 -.009 .103

Hispanic .030 .025 .016 .240 -.020 .079

American Indian .088 .060 .020 .142 -.030 .206

Asian -.053 .026 -.028 .039 -.103 -.003

Native Hawaiian/Pacific Islander -.026 .052 -.007 .618 -.129 .076

Race unknown -.008 .038 -.003 .835 -.081 .066

Family education -.010 .007 -.027 .144 -.024 .004

Family income -.005 .006 -.013 .407 -.017 .007

Family members -.017 .006 -.044 .002 -.028 -.006

From single parent home -.025 .024 -.017 .306 -.072 .023

First generation college student -.027 .018 -.025 .131 -.063 .008

High school GPA percentage of max -.003 .001 -.058 .000 -.004 -.001

High school athletics -.016 .028 -.008 .561 -.072 .039

Recruited college athlete .040 .026 .031 .117 -.010 .091

Honors list .098 .024 .068 .000 .051 .145

Voluntary disenrollment .019 .029 .009 .510 -.037 .075

Involuntary disenrollment .033 .060 .007 .587 -.085 .150

Intercollegiate sports .066 .021 .059 .002 .024 .108

Intramural sports .066 .020 .055 .001 .027 .105

Competitive programs .040 .020 .028 .044 .001 .080

Mission support programs .075 .019 .054 .000 .038 .112

Professional programs -.045 .043 -.014 .301 -.129 .040

Club programs -.048 .045 -.015 .282 -.136 .040

Recreational programs -.066 .034 -.026 .051 -.132 .000

Civil Air Patrol, Boy/Girl Scouts, .011 .016 .009 .500 -.021 .042 Campfire Girls

Overall performance average (OPA) .262 .030 .172 .000 .202 .321 Probations—all types .021 .019 .019 .254 -.015 .058 Note. Model 1.1b R2 = .052, p < .001, Confidence Interval (CI) = 95%

The results of Model 1.1b show that, on average, when controlling for the other covariates, Grit-S scores for female cadets were predicted to be higher than male cadets by .038 points (B = .038, p = .023), and lower for Asian cadets than white cadets by .053

108

points (B = -.053, p = .039). For each one-person increase in number of family members,

Grit-S score was predicted to be lower by .017 points (B = -0.017, p = .002). For every one-point increase in high school GPA percentage of max, Grit-S scores were predicted to be lower by .003 points (B = -0.003, p < .001). Grit-S scores for cadets who were on the USAFA honors list were predicted to be higher by .098 points (B = .098, p < .001) than those who were never on the list. Grit-S scores for cadets who participated in intercollegiate sports were predicted to be higher by .066 points (B = .066, p = .002) than non-participants. Grit-S scores for cadets who participated in intramural sports were predicted to be higher by .066 points (B = .066, p = .001) than non-participants. Grit-S scores for cadets who participated in competitive programs were predicted to be higher by .040 points (B = .040, p = .044) than non-participants. Grit-S scores for cadets who participated in mission support programs were predicted to be higher by .075 points (B =

.075, p < .001) than non-participants. Grit-S scores for cadets who participated in recreational programs were predicted to be lower by .066 points (B = -.066, p = .051) than non-participants. For every one-point increase in OPA, cadet Grit-S scores were predicted to be higher by .262 points (B = .262, p < .001).

Comparing the practical significance or absolute value of the ß for each of the statistically significant variables revealed that OPA (ß = .172) had a small to medium effect on the model. The practical significance of the remaining variables, listed in order of their ß values were as follows: honors list (ß = .068), intercollegiate sports (ß = .059), high school GPA percentage of max (ß = -.058), intramural sports (ß = .055), mission support programs (ß = .054), number of family members (ß = -.044), gender (ß = .031),

109

Asian (ß = -.028), competitive programs (ß = .028), and recreational programs (ß = -

.026).

These two models showed there were 14 variables that demonstrated statistically significant associations with Grit-S score. These variables were gender, Asian, number of family members, high school GPA percentage of max, first generation college student, honors list, GPA, PEA, OPA, intercollegiate sports, intramural sports, competitive programs, mission support programs, and recreational programs. However, only OPA (B

= .262, ß = .172, p < .001) demonstrated both statistical and practical significance.

Passion subscore. The second set of models compared passion subscores with the list of cadet predictor variables as shown in Tables 11 and 12 below. Model 1.2a included the performance variables GPA, MPA, and PEA, and Model 1.2b used OPA.

Likewise, Model 1.2a used the same probation variables academic, aptitude, athletic, conduct, and honor probations, while Model 1.2b included the “probation—all types” variable. Race variables were dummy coded, and Caucasian was the control variable.

Regression results indicated that for Model 1.2a, the overall model significantly predicted passion subscore (R2 = 0.034, p < .001) and accounted for 3.4% of variance in passion subscore. A summary of regression coefficients is presented in Table 11 below and shows that 11 of the 34 variables significantly contributed to the model, controlling for the other covariates. These variables included gender (B = .079, p = .001), family members (B = -.019, p = .019), high school GPA percentage of max (B = -.005, p <

.001), high school sports (B = -.099, p = .015), recruited college athlete (B = .071, p =

.059), honors list (B = .102, p = .004), GPA (B = .136, p < .001), PEA (B = .060, p =

110

.049), intercollegiate sports (B = .064, p = .041), intramural sports (B = .125, p < .001), and mission support programs (B = .089, p = .001).

Table 11

Model 1.2a Regression Results Comparing Passion Subscores with Cadet Predictors

CI Lower CI Upper Variables B SE ß p bound bound Gender .079 .025 .046 .001 .031 .128

Black .076 .041 .026 .063 -.004 .157

Hispanic .028 .036 .011 .446 -.044 .099

American Indian .073 .087 .011 .402 -.098 .244

Asian -.060 .037 -.022 .105 -.133 .012

Native Hawaiian/Pacific Islander -.007 .076 -.001 .923 -.156 .141

Race unknown .007 .054 .002 .897 -.099 .113

Family education -.003 .011 -.005 .811 -.023 .018

Family income -.006 .009 -.010 .517 -.022 .011

Family members -.019 .008 -.034 .019 -.035 -.003

From single parent home -.032 .035 -.015 .365 -.101 .037

First generation college student -.015 .028 -.010 .607 -.071 .042

High school GPA percentage of max -.005 .001 -.068 .000 -.007 -.003

High school athletics -.099 .041 -.033 .015 -.180 -.019

Recruited college athlete .071 .037 .037 .059 -.003 .144

Honors list .102 .036 .049 .004 .032 .172

Grade point average (GPA) .136 .029 .098 .000 .079 .193

Military performance average (MPA) .017 .054 .006 .748 -.088 .123

Physical education average (PEA) .060 .031 .033 .049 .000 .120

Voluntary disenrollment .048 .041 .016 .243 -.033 .130

Involuntary disenrollment .045 .088 .007 .607 -.127 .217

Academic probation .054 .029 .031 .065 -.003 .111

Aptitude probation -.029 .075 -.009 .698 -.176 .118

Athletic probation .009 .043 .003 .843 -.077 .094

Conduct probation -.010 .079 -.003 .899 -.165 .145

Honor probation .015 .068 .003 .827 -.119 .149

Intercollegiate sports .064 .032 .040 .041 .003 .126

Intramural sports .125 .029 .073 .000 .068 .182

Competitive programs .043 .029 .021 .138 -.014 .100

Mission support programs .089 .027 .045 .001 .035 .143

Professional programs -.027 .062 -.006 .667 -.149 .095

111

Model 1.2a Regression Results Comparing Passion Subscores with Cadet Predictors

CI Lower CI Upper Variables B SE ß p bound bound Club programs .014 .065 .003 .829 -.113 .141

Recreational programs -.080 .049 -.022 .101 -.175 .016

Civil Air Patrol, Boy/Girl Scouts, .031 .023 .018 .182 -.015 .077 Campfire Girls Note. Model 1.2a R2 = .034, p < .001, Confidence Interval (CI) = 95%

The results for Model 1.2a show that, on average, when controlling for the other covariates, passion subscores for female cadets were predicted to be higher than male cadets by .079 points (B = .079, p = .001). For each one-person increase in number of family members, passion subscores were predicted to be lower by .019 points (B = -.019, p = .019). For every one-point increase in high school GPA percentage of max, passion subscore was predicted to be lower by .005 points (B = -.005, p < .001). Passion subscores for cadets who participated in high school sports were predicted to be lower by

.099 points (B = -.099, p = .015) than non-participants. Passion subscores for cadets who were recruited to be college athletes were predicted to be higher by .071 points (B = .071, p = .059) than non-recruits. Passion subscores for cadets who were on the USAFA honors list were predicted to be higher by .102 points (B = .102, p = .004) than for those who were never on the list. For every one-point increase in GPA, passion subscore was predicted to be higher by .136 points (B = .136, p < .001). For every one-point increase in physical education average, passion subscores were predicted to be higher by .060 points (B = .060, p = .049). Passion subscores for cadets who participated in intercollegiate sports were predicted to be higher by .064 points (B = .064, p = .041) than non-participants. Passion subscores for cadets who participated intramural sports were predicted to be higher by .125 points (B = .125, p < .001) than non-participants. Passion

112

subscores for cadets who participated in mission support programs were predicted to be higher by .089 points (B = .089, p = .001) than non-participants.

Comparing the practical significance or absolute value of the ß for each of the statistically significant variables revealed that the variables demonstrated little practical effect. The practical significance of the variables, listed in order of their ß values, were as follows: GPA (ß = .098), intramural sports (ß = .073), high school GPA percentage of max (ß = -.068), honors list (ß = .049), gender (ß = .046), mission support programs (ß =

.045), intercollegiate sports (ß = .040), recruited college athlete (ß = .037), number of family members (ß = -.034), high school athlete (ß = -.033), and PEA (ß = .033).

Regression results indicated that for Model 1.2b, the overall model significantly predicted passion subscore (R2 = .034, p < .001) and similarly accounted for 3.4% of variance in passion subscore. A summary of regression coefficients is presented in Table

12 and shows 12 of the 34 variables significantly contributed to the model, controlling for the other covariates. These variables included gender (B = .067, p = .005), Black (B =

.078, p = .056), family members (B = -.018, p = .030), high school GPA percentage of max (B = -.005, p < .001), high school athletics (B = -.099, p = .021), recruited college athlete (B = .079, p = .035), honors list (B = .121, p < .001), intercollegiate sports (B =

.076, p = .014), intramural sports (B = .122, p < .001), mission support programs (B =

.089, p = .001), and OPA (B = .229, p < .001).

113

Table 12

Model 1.2b Regression Results Comparing Passion Subscore s with Cadet Predictors

CI Lower CI Upper Variables B SE ß p bound bound Gender .067 .024 .039 .005 .020 .115

Black .078 .041 .027 .056 -.002 .159

Hispanic .026 .036 .010 .468 -.045 .098

American Indian .078 .087 .012 .373 -.093 .248

Asian -.058 .037 -.022 .116 -.131 .014

Native Hawaiian/Pacific Islander -.007 .076 -.001 .928 -.155 .141

Race unknown .012 .054 .003 .824 -.094 .118

Family education -.001 .011 -.003 .888 -.023 .020

Family income -.006 .009 -.011 .483 -.023 .011

Family members -.018 .008 -.032 .030 -.033 -.002

From single parent home -.031 .035 -.014 .384 -.100 .039

First generation college student -.013 .026 -.008 .628 -.064 .038

High school GPA percentage of max -.005 .001 -.069 .000 -.007 -.003

High school athletics -.094 .041 -.031 .021 -.174 -.014

Recruited college athlete .079 .037 .042 .035 .006 .152

Honors list .121 .034 .058 .000 .054 .189

Voluntary disenrollment .055 .041 .019 .180 -.025 .136

Involuntary disenrollment .042 .087 .006 .633 -.129 .212

Intercollegiate sports .076 .031 .047 .014 .015 .136

Intramural sports .122 .029 .071 .000 .065 .178

Competitive programs .050 .029 .024 .088 -.007 .106

Mission support programs .089 .027 .045 .001 .035 .143

Professional programs -.028 .062 -.006 .654 -.150 .094

Club programs .014 .065 .003 .830 -.113 .141

Recreational programs -.077 .049 -.021 .113 -.172 .018

Civil Air Patrol, Boy/Girl Scouts, .031 .023 .018 .187 -.015 .076 Campfire Girls

Overall performance average (OPA) .229 .043 .105 .000 .144 .314

Probations—all types .041 .027 .025 .134 -.013 .094 Note. Model 1.2b R2 = .034, p < .001, Confidence Interval (CI) = 95%

The results in Model 1.2b show that, on average, when controlling for the other covariates, female cadet passion subscores were predicted to be higher by .067 points (B

= .067, p = .005) than male cadets and were also predicted to be higher for Black cadets

114

by .078 points (B = .078, p = .056) than for white cadets. For each one-person increase in family members, passion subscores were predicted to be lower by .018 points (B = -

.018, p = .030). For every one-point increase in high school GPA percentage of max, passion subscores were predicted to be lower by .005 points (B = -.005, p < .001).

Passion subscores for cadets who participated in high school sports were predicted to be lower by .094 points (B = -.094, p = .021) than non-participants. Passion subscores for cadets recruited to play intercollegiate sports were predicted to be higher by .079 points

(B = .079, p = .035) than non-recruits. Passion subscores for cadets on the USAFA honors lists were predicted to be higher by .121 (B = .121, p < .001) than those who were never on the list. Passion subscores for cadets who participated in intercollegiate sports were predicted to be higher by .076 points (B = .076, p = .014) than non-participants.

Passion subscores for cadets who participated in intramural sports were predicted to be higher by .122 points (B = .122, p < .001) than non-participants. Passion subscores for cadets who participated in mission support programs were predicted to be .089 points higher (B = .089, p = .001) than non-participants. For every one-point increase in OPA, passion subscore was predicted to be higher by .229 points (B = .229, p < .001).

Comparing the practical significance or absolute value of the ß for each of the statistically significant variables revealed that OPA (ß = .105) had a small to medium effect on the model. The practical significance of the other variables, listed in order of their ß values were as follows: intramural sports (ß = .071), high school GPA percentage of max (ß = -.069), honors list (ß = .058), intercollegiate sports (ß = .047), mission support programs (ß = .045), recruited college athlete (ß = .042), gender (ß = .039),

115

number of family members (ß = -.032), high school athlete (ß = -.031), and Black (ß =

.027).

These two models showed that there were 13 variables that demonstrated statistically significant associations with passion subscore. These variables were gender,

Black, number of family members, high school GPA percentage of max, high school athlete, recruited college athlete, honors list, GPA, PEA, OPA, intercollegiate sports, intramural sports, and mission support programs. However, only OPA (B = .229, ß =

.105, p < .001) demonstrated both statistical and practical significance.

Resilience subscore. The third set of models compared resiliency subscores with the list of cadet predictor variables as shown in Tables 13 and 14 below. In equivalent manner as the comparisons in the first two models, Model 1.3a included the performance variables GPA, MPA, and PEA and Model 1.3b used OPA. Likewise, Model 1.3a included the same probation variables academic, aptitude, athletic, conduct, and honor probations while Model 1.3b included the “probation—all types variable”. The race variables were dummy coded, and Caucasian was used as the control variable.

Regression results indicated that for Model 1.3a the overall model significantly predicted passion subscore (R2 = .039, p < .001) and accounted for 3.9% of variance in passion subscore. A summary of regression coefficients is presented in Table 13 below and shows that nine of the 34 variables significantly contributed to the model, controlling for the other covariates. These variables included family education (B = -.021, p = .019), family members (B = -.018, p = .005), first generation college student (B = -.063, p =

.007), honors list (B = .097, p = .001), GPA (B = .047, p = .048), MPA (B = .120, p =

116

.005), PEA, (B = .104, p < .001), mission support programs (B = .065, p = .002), and club programs (B = -.114, p = .025).

Table 13

Model 1.3a Regression Results Comparing Resilience Subscores with Cadet Predictors

CI Lower CI Upper Variables B SE ß p bound bound Gender .018 .019 .013 .359 -.020 .055 Black -.003 .032 -.001 .927 -.066 .060 Hispanic .027 .028 .013 .341 -.029 .083 American Indian .085 .068 .017 .212 -.048 .218 Asian -.047 .029 -.022 .106 -.103 .010 Native Hawaiian/Pacific Islander -.053 .059 -.012 .367 -.169 .062 Race unknown -.030 .042 -.010 .477 -.113 .053 Family education -.021 .009 -.047 .019 -.038 -.003 Family income -.004 .007 -.010 .562 -.018 .010 Family members -.018 .006 -.041 .005 -.030 -.005 From single parent home -.018 .026 -.011 .498 -.069 .034 First generation college student -.063 .023 -.053 .007 -.109 -.018 High school GPA percentage of max .000 .001 -.005 .719 -.002 .001 High school athletics .048 .032 .021 .129 -.014 .111 Recruited college athlete -.016 .029 -.011 .576 -.073 .041 Honors list .097 .028 .059 .001 .042 .152 Grade point average (GPA) .047 .024 .043 .048 .000 .094 Military performance average (MPA) .120 .043 .055 .005 .036 .204 Physical education average (PEA) .104 .024 .074 .000 .057 .152 Voluntary disenrollment -.012 .032 -.005 .709 -.076 .051 Involuntary disenrollment .010 .069 .002 .880 -.124 .145 Academic probation -.035 .023 -.026 .129 -.080 .010 Aptitude probation -.059 .059 -.023 .312 -.174 .056 Athletic probation .011 .034 .005 .736 -.055 .078 Conduct probation .028 .062 .010 .646 -.093 .149 Honor probation .034 .053 .009 .528 -.071 .138 Intercollegiate sports .039 .025 .031 .113 -.009 .087 Intramural sports .022 .023 .016 .328 -.022 .066 Competitive programs .022 .023 .013 .339 -.023 .066 Mission support programs .065 .021 .042 .002 .023 .107 Professional programs -.049 .049 -.014 .318 -.144 .047 Club programs -.114 .051 -.031 .025 -.213 -.014 Recreational programs -.055 .038 -.020 .147 -.130 .019 Civil Air Patrol, Boy/Girl Scouts, -.006 .018 -.005 .730 -.042 .029 Campfire Girls Note. Model 1.3a R2 = .039, p < .001, Confidence Interval (CI) = 95%

117

The results for Model 1.3a show that, on average, when controlling for the other covariates, for every one-category increase in family education, resilience subscores were predicted to be lower by .021 points (B = -.021, p = .019). For each one-person increase in number of family members, resilience subscores were predicted to be lower by .018 points (B = -.018, p = .005). Resilience subscores for cadets who were first generation college students were predicted to be lower by .063 points (B = -.063, p = .007) than those cadets who were not first-generation college students. Resilience subscores for cadets who were on the USAFA honors list were predicted to be higher by .097 points (B

= .097, p = .001) than those who were never on the list. For every one-point increase in

GPA, resilience subscores were predicted to be higher by .047 points (B = .047, p =

.048). For every one-point increase in MPA, resilience subscores were predicted to be higher by .120 points (B = .120, p = .005). For every one-point increase in PEA, resilience subscores were predicted to be higher by .104 points (B = .104, p < .001).

Resilience subscores for cadets who participated in mission support programs were predicted to be higher by .065 points (B = .065, p = .002) than non-participants.

Resilience subscores for cadets who participated in club programs were predicted to be lower by .114 points (B = -.114, p = .025) than non-participants.

Comparing the practical significance or absolute values of ß for the statistically significant variables revealed that the variables demonstrated little practical effect. The practical significance of the variables, listed in order of their ß values are as follows:

PEA (ß = .074), honors list (ß = .059), MPA (ß = .055), first generation college student (ß

= -.053), family education (ß = -.047), GPA (ß = .043), mission support programs (ß =

.042), number of family members (ß = -.041), and club programs (ß = -.031).

118

Regression results indicated that for Model 1.3b, the overall model significantly predicted passion subscore (R2 = .044, p < .001) and accounted for 4.4% of variance in passion subscore. A summary of regression coefficients is presented in Table 14 and shows nine of the 34 variables significantly contributed to the model, controlling for the other covariates. These variables included family education (B = -.019, p = .025), family members (B = -.017, p = .007), first generation college student (B = -.042, p = .050), high school athletics (B = .061, p = .053), honors list (B = .027, p = .006), intercollegiate sports (B = .057, p = .018), mission support programs (B = .060, p = .005), club programs (B = -.111, p = .029), and OPA (B = .295, p < .001).

119

Table 14

Model 1.3b Regression Results Comparing Resilience Subscores with Cadet Predictors

CI Lower CI Upper Variables B SE ß p bound bound Gender .009 .019 .006 .650 -.028 .045 Black .016 .032 .007 .626 -.047 .078 Hispanic .033 .028 .016 .247 -.023 .089 American Indian .099 .068 .020 .144 -.034 .232 Asian -.048 .029 -.023 .097 -.104 .009 Native Hawaiian/Pacific Islander -.045 .059 -.010 .441 -.161 .070 Race unknown -.028 .042 -.009 .512 -.110 .055 Family education -.019 .008 -.044 .025 -.036 -.003 Family income -.004 .007 -.009 .565 -.018 .010 Family members -.017 .006 -.039 .007 -.029 -.005 From single parent home -.019 .026 -.011 .478 -.070 .033 First generation college student -.042 .021 -.035 .050 -.084 .000 High school GPA percentage of max -.001 .001 -.014 .307 -.002 .001 High school athletics .061 .032 .026 .053 -.001 .123 Recruited college athlete .002 .029 .002 .940 -.055 .059 Honors list .075 .027 .046 .006 .022 .128 Voluntary disenrollment -.018 .032 -.008 .583 -.080 .045 Involuntary disenrollment .024 .068 .005 .725 -.109 .156 Intercollegiate sports .057 .024 .045 .018 .010 .104 Intramural sports .010 .022 .008 .647 -.034 .054 Competitive programs .031 .023 .019 .167 -.013 .076 Mission support programs .060 .021 .039 .005 .018 .102 Professional programs -.061 .049 -.017 .207 -.156 .034 Club programs -.111 .051 -.030 .029 -.210 -.011 Recreational programs -.054 .038 -.019 .150 -.129 .020 Civil Air Patrol, Boy/Girl Scouts, -.009 .018 -.007 .620 -.044 .026 Campfire Girls Overall performance average (OPA) .295 .035 .173 .000 .225 .364 Probations—all types .002 .021 .002 .916 -.040 .044 Note. Model 1.3b R2 = .044, p < .001, Confidence Interval (CI) = 95%

The results in Model 1.3b show that, on average, when controlling for the other covariates, for every one-category increase in education, resilience subscores were predicted to be lower by .019 points (B = -.019, p = .025). For each one-person increase in number of family members, resilience subscores were predicted to be lower by .017 points (B = -.017, p = .007). Resilience subscores for cadets who were first generation

120

college students were predicted to be lower by .042 points (B = -.042, p = .050) than those who were not first-generation college students. Resilience subscores for cadets who participated in high school sports were predicted to be higher by .061 points (B =

.061, p = .053) than non-participants. Resilience subscores for cadets who were on the

USAFA honors list were predicted to be higher by .075 points (B = .075, p = .006) than those who were never on the lists. Resilience subscores for cadets who participated in intercollegiate sports were predicted to be higher by .057 points (B = .057, p = .018) than non-participants. Resilience subscores for cadets who participated in mission support programs were predicted to be higher by .060 points (B = .060, p = .005) than non- participants. Resilience subscores for cadets who participated in club programs were predicted to be lower by .111 points (B = -.111, p = .029) than non-participants. For every one-point increase in OPA, resilience subscores were predicted to be higher by

.295 points (B = .295, p < .001).

Comparing the practical significance or absolute value of the ß for each of the statistically significant variables revealed that OPA (ß = .173) had a small to medium effect on the model. The remaining statistically significant variables did not produce practically significant effect sizes and are listed in order of their ß values: honors list (ß =

.046), intercollegiate sports (ß = .045), family education (ß = -.044), mission support programs (ß = .039), number of family members (ß = -.039), first generation college student (ß = -.035), club programs (ß = -.030), and high school athlete (ß = .026).

These two models showed that there were 12 variables that demonstrated statistically significant associations with resilience subscore. These variables were family education, number of family members, first generation college student, high

121

school athlete, honors list, GPA, MPA, PEA, OPA, intercollegiate sports, mission support programs, and club programs. However, only OPA (B = .295, ß = .173, p < .001) demonstrated both statistical and practical significance.

RQ2 – Performance vs Grit-S Score

The second research question explored the extent to which Grit-S score was associated with cadet performance measures such as overall performance average (OPA), and included analysis of the GPA, MPA, and PEA performance measures that combine to form the OPA. Separate models using the GPA, MPA, PEA, and OPA as dependent variables were completed, and as in research question one, two sub-models were completed separating out each of the Grit-S and probation statistics, and then combining those into the Grit-S score and “probations—all types.” This research question focuses on the association between the dependent variables and Grit-S score, passion subscore, and resilience subscore; thus, while the results show other statistically significant variables only interpretations for the three grit variables will be discussed.

Grade point average. The first set of models compared GPA with Grit-S score and subscores and other cadet predictor variables as shown in Tables 15 and 16. The

Bonferroni Method was applied to the results and includes an indication for statistical significance post Bonferroni adjustment (p < .025, p < .0125) for multiple tests within dependent samples and/or correlated outcomes (Shi et al., 2012). For comparison, Model

2.1a included the Grit-S survey passion and resilience subscores, and Model 2.1b used the combined Grit-S score. The same procedure was applied to the probation variables, and Model 2.1a included academic, aptitude, athletic, conduct, and honor probations while Model 2.1b combined the five probation types into a “probation—all types”

122

variable. The race variables were dummy coded, and Caucasian was used as the control variable.

Regression results indicated that for Model 2.1a the overall model significantly predicted GPA (R2 = 0.542, p < .001) and accounted for 54.2% of variance in GPA. A summary of regression coefficients is presented in Table 15 below and shows that 19 of the 33 variables significantly contributed to the model, controlling for the other covariates. These variables included Black (B = -.143, p < .001), Hispanic (B = -.072, p

< .001), American Indian (B = -.110, p < .012), family education (B = .022, p = .019), family members (B = .008, p = .051), high school GPA percentage of max (B = .004, p <

.001), recruited college athlete (B = -.095, p < .001), honors list (B = .528, p < .001), passion subscore (B = .032, p < .001), resilience subscore (B = .023, p = .014), voluntary disenrollment (B = .059, p = .022), involuntary disenrollment (B = -.302, p < .001), academic probation (B = -.533, p < .001), athletic probation (B = -.076, p < .001), intercollegiate sports (B = -.040, p = .009), intramural sports (B = .037, p = .014), mission support programs (B = .026, p = .054), professional programs (B = .115, p <

.001), and recreational programs (B = .049, p = .040).

123

Table 15

Model 2.1a Regression Results Comparing GPA with Cadet Predictors

CI Lower CI Upper Variables B SE ß p bound bound Gender -.021 .012 -.017 .087 -.045 .003 Black -.143 .021 -.069 .000*** -.185 -.102 Hispanic -.072 .018 -.038 .000*** -.109 -.036 American Indian -.110 .044 -.024 .012 -.196 -.024 Asian .019 .019 .010 .313 -.018 .057 Native Hawaiian/Pacific Islander -.039 .038 -.010 .309 -.114 .036 Race unknown .018 .027 .006 .495 -.034 .071 Family education .022 .008 .053 .019** .004 .039 Family income .001 .005 .002 .866 -.009 .010 Family members .008 .004 .021 .051* .000 .016 From single parent home .020 .021 .014 .345 -.022 .063 First generation college student -.016 .023 -.014 .506 -.065 .033 High school GPA percentage of max .004 .001 .087 .000*** .003 .005 High school athletics -.015 .020 -.007 .471 -.055 .025 Recruited college athlete -.095 .019 -.070 .000*** -.132 -.059 Honors list .528 .017 .354 .000*** .495 .560 Passion subscore .032 .007 .045 .000*** .019 .046 Resilience subscore .023 .009 .025 .014** .005 .042 Voluntary disenrollment .059 .025 .028 .022** .009 .110 Involuntary disenrollment -.302 .043 -.067 .000*** -.387 -.218 Academic probation -.533 .013 -.436 .000*** -.559 -.508 Aptitude probation .034 .036 .015 .353 -.038 .105 Athletic probation -.076 .020 -.036 .000*** -.115 -.036 Conduct probation -.064 .039 -.026 .102 -.140 .013 Honor probation -.062 .033 -.017 .065 -.127 .004 Intercollegiate sports -.040 .015 -.035 .009 -.070 -.010 Intramural sports .037 .015 .030 .014** .007 .066 Competitive programs -.019 .014 -.013 .193 -.047 .010 Mission support programs .026 .014 .018 .054* .000 .053 Professional programs .115 .031 .035 .000*** .054 .175 Club programs .031 .032 .009 .338 -.032 .094 Recreational programs .049 .024 .019 .040* .002 .096 Civil Air Patrol, Boy/Girl Scouts, .002 .012 .002 .879 -.022 .026 Campfire Girls Note. Model 2.1a R2 = .542, p < .001, Confidence interval (CI) = 95%, *p < .05, **p < .025, ***p < .0125

The regression results for Model 2.1a show that, on average, when controlling for the other covariates, for every one-point increase in passion subscore, GPA was higher by

.032 points (B = .032, p < .001) and that for every one-point increase in resilience score,

124

GPA was higher by .023 points (B = .023, p = .014). Comparing the practical significance and absolute value of the ß for all of the statistically significant variables in the model revealed that the highest effect sizes were academic probation (ß = -.436, p <

.001) with a medium to large effect on the GPA followed by honors list (ß = .354, p <

.001) with a medium effect. However, while statistically significant, neither the passion subscore (ß = .045) nor resilience subscore (ß = .025) demonstrated practical significance using Cohen’s broad guidelines. These guidelines were used throughout research question two to provide practical effect size comparisons.

For Model 2.1b, regression results indicated that the overall model significantly predicted GPA (R2 = 0.509, p < .001) and accounted for 50.9% of variance in GPA. A summary of regression coefficients is presented in Table 16 below and shows that 18 of the 28 variables significantly contributed to the model, controlling for the other covariates. These variables included, gender (B = -.025, p = .050), Black (B = -.168, p <

.001), Hispanic (B = -.075, p < .001), American Indian (B = -.147, p < .001), family education (B = .023, p = .016), family members (B = .009, p = .046), high school GPA percentage of max (B = .005, p < .001), recruited college athlete (B = -.101, p < .001), honors list (B = .558, p < .001), voluntary disenrollment (B = .065, p = .014), involuntary disenrollment (B = -.306, p < .001), intercollegiate sports (B = -.043, p = .007), intramural sports (B = .039, p = .012), competitive programs (B = -.030, p = .048), professional programs (B = .119, p < .001), recreational programs (B = .049, p = .047),

Grit-S survey score (B = .057, p < .001), and “probations—all types” (B = -.455, p <

.001).

125

Table 16

Model 2.1b Regression Results Comparing GPA with Cadet Predictors

CI Lower CI Upper Variables B SE ß p bound bound Gender -.025 .013 -.020 .050 -.050 .000 Black -.168 .022 -.080 .000*** -.211 -.125 Hispanic -.075 .019 -.039 .000*** -.112 -.037 American Indian -.147 .045 -.032 .001*** -.236 -.059 Asian .024 .020 .012 .223 -.015 .063 Native Hawaiian/Pacific Islander -.021 .039 -.005 .592 -.099 .056 Race unknown .022 .028 .008 .439 -.033 .076 Family education .023 .008 .055 .016** .005 .040 Family income .002 .005 .006 .625 -.007 .012 Family members .009 .004 .022 .046* .000 .017 From single parent home .023 .022 .015 .312 -.022 .067 First generation college student -.026 .024 -.023 .308 -.077 .026 High school GPA percentage of max .005 .001 .097 .000*** .004 .006 High school athletics -.026 .021 -.012 .222 -.067 .016 Recruited college athlete -.101 .019 -.075 .000*** -.139 -.063 Honors list .558 .017 .374 .000*** .525 .592 Voluntary disenrollment .065 .026 .031 .014** .013 .117 Involuntary disenrollment -.306 .044 -.068 .000*** -.393 -.220 Intercollegiate sports -.043 .016 -.037 .007 -.074 -.011 Intramural sports .039 .015 .031 .012*** .008 .069 Competitive programs -.030 .015 -.020 .048* -.059 .000 Mission support programs .019 .014 .014 .178 -.009 .046 Professional programs .119 .032 .036 .000*** .056 .181 Club programs .031 .033 .009 .346 -.034 .097 Recreational programs .049 .025 .019 .047** .001 .098 Civil Air Patrol, Boy/Girl Scouts, -.001 .012 .000 .964 -.025 .024 Campfire Girls Grit-S survey score .057 .010 .056 .000*** .037 .077 Probations—all types -.455 .012 -.397 .000*** -.479 -.431 Note. Model 2.1b R2 = .509, p < .001, Confidence Interval (CI) = 95%, *p < .05, **p < .025, ***p < .0125

The regression results for Model 2.1b show that, on average, when controlling for the other covariates, for every one-point increase in Grit-S score, GPA was higher by

.057 points (B = .057, p < .001). Comparing the practical significance or absolute value of the ß for all the statistically significant variables in the model revealed that the highest effect sizes were “probations—all types” (ß = -.397) with a medium effect on the GPA

126

followed by honors list (ß = .374) also with a medium effect. However, while statistically significant, Grit-S score (ß = .056) did not demonstrate practical significance.

These two models showed there were several variables that demonstrated statistically significant associations with GPA including in part academic probation,

“probation—all types”, honors list, Grit-S score, passion subscore, and resilience subscore. However, neither Grit-S score, passion subscore, nor resilience subscore demonstrated significant effect sizes on GPA.

Military performance average. The second set of models compared MPA with

Grit-S score and subscores and other cadet predictor variables as shown in Tables 17 and

18. The Bonferroni Method was applied to the results and includes an indication for statistical significance post Bonferroni adjustment (p < .025, p < .0125) for multiple tests within dependent samples and/or correlated outcomes (Shi et al., 2012). For comparison,

Model 2.2a included the Grit-S survey passion and resilience subscores and Model 2.2b used the combined Grit-S score. The same procedure applied to the probation variables and Model 2.2a included academic, aptitude, athletic, conduct, and honor probations while Model 2.2b combined the five probation types into a “probations—all types” variable. The race variables were dummy coded, and Caucasian was used as the control variable.

Regression results indicated that for Model 2.2a, the overall model significantly predicted MPA (R2 = .425, p < .001) and accounted for 42.5% of variance in MPA. A summary of regression coefficients is presented in Table 17 below and shows that 19 of the 33 variables significantly contributed to the model, controlling for the other covariates. These variables included gender (B = .066, p < .001), family education (B =

127

.025, p < .001), first generation college student (B = .168, p < .001), high school GPA percentage of max (B = .001, p = .008), recruited college athlete (B = -.052, p < .001), honors list (B = .121, p < .001), resilience subscore (B = .020, p < .001), voluntary disenrollment (B = -.084, p < .001), involuntary disenrollment (B = -.197, p < .001), academic probation (B = -.114, p < .001), aptitude probation (B = -.233, p < .001), athletic probation (B = -.061, p < .001), conduct probation (B = -.065, p = .004), honor probation (B = -.176, p < .001), intercollegiate sports (B = -.066, p < .001), competitive programs (B = -.020, p = .017), mission support programs, (B = -.018, p = .023), professional programs (B = .057, p = .001), and club programs (B = .063, p = .001).

Passion subscore (B = .004, p = .273) was not a statistically significant contributor.

128

Table 17

Model 2.2a Regression Results Comparing MPA with Cadet Predictors

CI Lower CI Upper Variables B SE ß p bound bound Gender .066 .007 .108 .000*** .052 .081 Black -.015 .013 -.014 .236 -.040 .010 Hispanic -.015 .011 -.016 .149 -.036 .006 American Indian -.026 .024 -.012 .290 -.074 .022 Asian -.014 .010 -.014 .195 -.034 .007 Native Hawaiian/Pacific Islander .001 .022 .000 .972 -.042 .044 Race unknown -.026 .016 -.019 .110 -.058 .006 Family education .025 .004 .124 .000*** .016 .035 Family income .006 .004 .032 .127 -.002 .015 Family members -.003 .003 -.016 .280 -.009 .003 From single parent home -.006 .012 -.009 .602 -.031 .018 First generation college student .168 .015 .307 .000*** .137 .199 High school GPA percentage of max .001 .000 .037 .008*** .000 .002 High school athletics .020 .011 .019 .080 -.002 .042 Recruited college athlete -.052 .010 -.078 .000*** -.073 -.032 Honors list .121 .010 .163 .000*** .101 .141 Passion subscore .004 .004 .012 .273 -.003 .012 Resilience subscore .020 .005 .044 .000*** .010 .030 Voluntary disenrollment -.084 .015 -.081 .000*** -.114 -.053 Involuntary disenrollment -.197 .024 -.088 .000*** -.244 -.149 Academic probation -.114 .007 -.188 .000*** -.129 -.100 Aptitude probation -.233 .021 -.203 .000*** -.274 -.192 Athletic probation -.061 .012 -.059 .000*** -.085 -.038 Conduct probation -.065 .023 -.052 .004*** -.109 -.020 Honor probation -.176 .019 -.100 .000*** -.215 -.138 Intercollegiate sports -.066 .009 -.116 .000*** -.084 -.049 Intramural sports -.014 .009 -.023 .101 -.031 .003 Competitive programs -.020 .008 -.027 .017** -.036 -.003 Mission support programs -.018 .008 -.026 .023** -.033 -.002 Professional programs .057 .018 .035 .001*** .023 .092 Club programs .063 .019 .037 .001*** .026 .100 Recreational programs .016 .015 .013 .282 -.013 .045 Civil Air Patrol, Boy/Girl Scouts, -.007 .007 -.012 .263 -.020 .006 Campfire Girls Note. Model 2.2a R2 = .425, p < .001, Confidence Interval (CI) = 95%, *p < .05, **p < .025, ***p < .0125

The regression results for Model 2.2a show that, on average, when controlling for the covariates, for every one-point increase in resilience subscore, MPA was higher by

.020 points (B = .020, p < .001), but passion subscore was not a statistically significant

129

contributor. Comparing the practical significance or absolute value of the ß for all of the statistically significant variables in the model revealed that the highest effect sizes were first generation college student (ß = .307) with a medium effect on MPA followed by aptitude probation (ß = -.203) with a small to medium effect. However, while statistically significant, resilience subscore (ß = .044) did not demonstrate a practically significant effect size.

For Model 2.2b, regression results indicated the overall model significantly predicted MPA (R2 = .383, p < .001) and accounted for 38.3% of variance in MPA with

16 of 28 significant variables. A summary of regression coefficients is presented in

Table 18 below and shows that 15 of the 33 variables significantly contributed to the model, controlling for the other covariates. These variables included gender (B = .064, p

< .001), family education (B = .025, p < .001), first generation college student (B = .170, p < .001), high school GPA percentage of max (B = .001, p = .002), recruited college athlete (B = -.052, p < .001), honors list (B = .120, p < .001), voluntary disenrollment (B

= -.095, p < .001), involuntary disenrollment (B = -.256, p < .001), intercollegiate sports

(B = -.073, p = .013), intramural sports (B = -.021, p = .049), competitive programs (B =

-.021, p = .005), mission support programs (B = .052, p < .001), recreational programs (B

= .022, p < .001), Civil Air Patrol, Boy/Girl Scouts, Campfire Girls (B = -.003, p < .001),

Grit-S score (B = .022, p < .001), and “probations—all types” (B = -.172, p = .016).

130

Table 18

Model 2.2b Regression Results Comparing MPA with Cadet Predictors

CI Lower CI Upper Variables B SE ß p bound bound Gender .064 .007 .105 .000*** .050 .079 Black -.020 .013 -.019 .132 -.046 .006 Hispanic -.015 .011 -.016 .172 -.037 .007 American Indian -.021 .025 -.009 .402 -.071 .028 Asian -.011 .011 -.011 .313 -.032 .010 Native Hawaiian/Pacific Islander -.009 .023 -.005 .675 -.054 .035 Race unknown -.021 .017 -.015 .200 -.054 .011 Family education .025 .005 .122 .000*** .015 .034 Family income .007 .004 .037 .089 -.001 .016 Family members -.002 .003 -.013 .413 -.008 .004 From single parent home -.006 .013 -.007 .680 -.033 .022 First generation college student .170 .015 .310 .000*** .138 .201 High school GPA percentage of max .001 .000 .045 .002*** .000 .002 High school athletics .016 .012 .015 .178 -.007 .039 Recruited college athlete -.052 .011 -.078 .000*** -.073 -.031 Honors list .120 .010 .163 .000*** .100 .140 Voluntary disenrollment -.095 .016 -.091 .000*** -.126 -.063 Involuntary disenrollment -.256 .025 -.114 .000*** -.304 -.207 Intercollegiate sports -.073 .009 -.128 .013** -.091 -.054 Intramural sports -.021 .009 -.035 .049 -.039 -.004 Competitive programs -.021 .009 -.029 .005*** -.038 -.005 Mission support programs -.016 .008 -.023 .000*** -.032 .000 Professional programs .052 .018 .032 .144 .016 .088 Club programs .070 .019 .042 .709 .032 .108 Recreational programs .022 .015 .018 .000*** -.008 .052 Civil Air Patrol, Boy/Girl Scouts, -.003 .007 -.004 .000*** -.016 .011 Campfire Girls Grit-S survey score .022 .006 .044 .000*** .011 .034 Probations—all types -.172 .007 -.301 .016** -.186 -.157 Note. Model 2.2b R2 = .383, p < .001, Confidence Interval (CI) = 95%, *p < .05, **p < .025, ***p < .0125

The regression results for Model 2.2b show that, on average, when controlling for the other covariates, for every one-point increase in Grit-S score, MPA was higher by

.022 points (B = .022, p < .001). Comparing the practical significance or absolute value of the ß for all of the statistically significant variables in the model revealed that the highest effect sizes were first generation college student (ß = .310) with a small to

131

medium effect on MPA followed by “probations—all types” (ß = -.301), also with a small to medium effect. However, while statistically significant, Grit-S score (ß = .044) did not demonstrate a practically significant effect size.

These two models showed there were several variables that demonstrated statistically significant associations with MPA including first generation college student, aptitude probation, “probations—all types”, Grit-S score, and resilience subscore.

However, neither Grit-S score nor resilience subscore demonstrated significant effect sizes on MPA.

Physical education average. The third set of models compared PEA with the

Grit-S score and subscores and other cadet predictor variables as shown in Tables 19 and

20 below. The Bonferroni Method was applied to the results and includes an indication for statistical significance post Bonferroni adjustment (p < .025, p < .0125) for multiple tests within dependent samples and/or correlated outcomes (Shi et al., 2012). For comparison, Model 2.3a included the Grit-S survey passion and resilience subscores and

Model 2.3b used the combined Grit-S score. The same procedure was applied to the probation variables and Model 2.3a included academic, aptitude, athletic, conduct, and honor probations while Model 2.3b combined the five probation types into a

“probations—all types” variable. The race variables were dummy coded, and Caucasian was used as the control variable.

Regression results indicated that for Model 2.3a the overall model significantly predicted PEA (R2 = .320, p < .001) and accounted for 32.0% of variance in PEA. A summary of regression coefficients is presented in Table 19 below and shows that 15 of the 33 variables significantly contributed to the model, controlling for the other variables.

132

These variables included gender (B = -.072, p < .001), family education (B = .014, p =

.012), family members (B = .016, p < .001), first generation college student (B = .036, p

= .056), high school athletics (B = .108, p < .001), honors list (B = .273, p < .001), resilience subscore (B = .040, p < .001), academic probation (B = -.091, p < .001), athletic probation (B = -.514, p < .001), intercollegiate sports (B = .170, p < .001), intramural sports (B = -.079, p < .001), competitive programs (B = .063, p < .001), mission support programs (B = .042, p = .001), recreational programs (B = .063, p =

.005), and Civil Air Patrol, Boy/Girl Scouts, Campfire Girls (B = -.035, p = .001).

Passion subscore (B = .011, p = .085) was not a statistically significant contributor.

133

Table 19

Model 2.3a Regression Results Comparing PEA with Cadet Predictors

CI Lower CI Upper Variables B SE ß p bound bound Gender -.072 .011 -.076 .000*** -.094 -.050 Black .028 .020 .017 .170 -.012 .067 Hispanic -.018 .017 -.012 .293 -.052 .016 American Indian -.042 .042 -.012 .315 -.124 .040 Asian -.031 .018 -.021 .081 -.065 .004 Native Hawaiian/Pacific Islander .044 .035 .014 .213 -.025 .114 Race unknown .011 .026 .005 .660 -.040 .062 Family education .014 .005 .045 .012*** .003 .025 Family income .005 .004 .015 .300 -.004 .013 Family members .016 .004 .052 .000*** .008 .023 From single parent home .015 .017 .013 .381 -.019 .049 First generation college student .036 .018 .043 .056* -.001 .074 High school GPA percentage of max .000 .000 -.007 .560 -.001 .001 High school athletics .108 .019 .065 .000*** .070 .145 Recruited college athlete .020 .018 .020 .257 -.015 .055 Honors list .273 .015 .237 .000*** .243 .302 Passion subscore .011 .007 .020 .085 -.002 .024 Resilience subscore .040 .009 .057 .000*** .023 .057 Voluntary disenrollment .031 .021 .019 .130 -.009 .071 Involuntary disenrollment -.032 .041 -.009 .434 -.112 .048 Academic probation -.091 .012 -.097 .000*** -.115 -.068 Aptitude probation -.001 .034 -.001 .971 -.069 .066 Athletic probation -.514 .019 -.318 .000*** -.551 -.477 Conduct probation -.037 .037 -.019 .311 -.109 .035 Honor probation -.005 .031 -.002 .874 -.066 .057 Intercollegiate sports .170 .014 .193 .000*** .142 .199 Intramural sports -.079 .014 -.084 .000*** -.107 -.052 Competitive programs .063 .014 .055 .000*** .036 .090 Mission support programs .042 .013 .038 .001*** .017 .067 Professional programs -.044 .029 -.017 .130 -.101 .013 Club programs .048 .030 .018 .116 -.012 .107 Recreational programs .063 .023 .032 .005*** .018 .107 Civil Air Patrol, Boy/Girl Scouts, -.035 .011 -.038 .001*** -.057 -.013 Campfire Girls Note. Model 2.3a R2 = .320, p < .001, Confidence Interval (CI) = 95%, *p < .05, **p < .025, ***p < .0125

The regression results for Model 2.3a show that, on average, when controlling for the other covariates, for every one-point increase in resilience subscore PEA was higher by .040 points (B = .040, p < .001) while passion subscore was not statistically

134

significant. Comparing the practical significance or absolute value of the ß for all of the statistically significant variables in the model revealed that the highest effect sizes were athletic probation (ß = -.318) with a medium effect on PEA followed by honors list (ß =

.237) with a small to medium effect. However, while statistically significant, resilience subscore (ß = .057) did not demonstrate a practically significant effect size.

For Model 2.3b, the overall model significantly predicted PEA (R2 = .251, p <

.001) and accounted for 25.1% of variance in PEA. A summary of regression coefficients is presented in Table 20 below and shows that 16 of the 28 variables significantly contributed to the model, controlling for the other covariates. These variables included gender (B = -.089, p < .001), Black (B = .042, p = .047), family education (B = .016, p = .005), family members (B = .018, p < .001), first generation college student (B = .039, p = .043), high school athletics (B = .128, p < .001), honors list

(B = .294, p < .001), intercollegiate sports (B = .188, p < .001), intramural sports (B = -

.094, p < .001), competitive programs (B = .073, p < .001), mission support programs (B

= .030, p = .026), professional programs (B = -.059, p = .053), recreational programs (B =

.076, p = .001), Civil Air Patrol, Boy/Girl Scouts, Campfire Girls (B = -.042, p < .001),

Grit-S score (B = .047, p < .001), and “probations—all types” (B = -.192, p < .001).

135

Table 20

Model 2.3b Regression Results Comparing PEA with Cadet Predictors

CI Lower CI Upper Variables B SE ß p bound bound Gender -.089 .012 -.093 .000*** -.112 -.066 Black .042 .021 .026 .047* .001 .083 Hispanic -.015 .018 -.010 .417 -.050 .021 American Indian -.056 .044 -.016 .195 -.142 .029 Asian -.028 .018 -.019 .130 -.064 .008 Native Hawaiian/Pacific Islander .041 .037 .013 .271 -.032 .114 Race unknown -.001 .027 -.001 .967 -.054 .052 Family education .016 .006 .052 .005*** .005 .028 Family income .004 .005 .012 .425 -.006 .013 Family members .018 .004 .059 .000*** .010 .026 From single parent home .007 .018 .007 .684 -.029 .044 First generation college student .039 .018 .046 .043* .001 .077 High school GPA percentage of max -.001 .001 -.013 .320 -.001 .000 High school athletics .128 .020 .077 .000*** .089 .168 Recruited college athlete .027 .019 .026 .153 -.010 .063 Honors list .294 .016 .256 .000*** .263 .325 Voluntary disenrollment .016 .021 .010 .449 -.026 .058 Involuntary disenrollment -.057 .043 -.016 .176 -.141 .026 Intercollegiate sports .188 .015 .213 .000*** .158 .217 Intramural sports -.094 .014 -.099 .000*** -.122 -.065 Competitive programs .073 .014 .064 .000*** .045 .100 Mission support programs .030 .013 .028 .026* .004 .056 Professional programs -.059 .030 -.023 .053* -.118 .001 Club programs .043 .032 .017 .180 -.020 .105 Recreational programs .076 .024 .038 .001*** .030 .123 Civil Air Patrol, Boy/Girl Scouts, -.042 .012 -.046 .000*** -.065 -.020 Campfire Girls Grit-S survey score .047 .010 .060 .000*** .028 .067 Probations—all types -.192 .011 -.218 .000*** -.215 -.170 Note. Model 2.3b R2 = .251, p < .001, Confidence Interval (CI) = 95%, *p < .05, **p < .025, ***p < .0125

For Model 2.3b, the regression results also showed that, on average, when controlling for the other covariates, for every one-point increase in Grit-S score, PEA was higher by .047 points (B = .047, p < .001). Comparing the practical significance or absolute value of the ß for all of the statistically significant variables in the model revealed that the highest effect sizes were honors list (ß = .256) with a small to medium

136

effect on PEA followed by “probations—all types” (ß = -.218), also with a small to medium effect. However, while statistically significant, Grit-S score (ß = .060) did not demonstrate a practically significant effect size.

These two models showed there were several variables that demonstrated statistically significant associations with PEA including in part athletic probation, honors list, “probations—all types”, Grit-S score, and resilience subscore. However, neither

Grit-S score nor resilience subscore demonstrated significant effect sizes on PEA.

Overall performance average. The fourth set of models compared OPA with the Grit-S score and subscores and other cadet predictor variables as shown in Tables 21 and 22 below. The Bonferroni Method was applied to the results and includes an indication for statistical significance post Bonferroni adjustment (p < .025, p < .0125) for multiple tests within dependent samples and/or correlated outcomes (Shi et al., 2012).

For comparison, Model 2.4a included the Grit-S survey passion and resilience subscores, and Model 2.4b used the combined Grit-S score. The same procedure was applied to the probation variables, and Model 2.4a included academic, aptitude, athletic, conduct, and honor probations while Model 2.4b combined the five probation types into a

“probations—all types” variable. The race variables were dummy coded, and Caucasian was used as the control variable.

Regression results from Model 2.4a indicated the overall model significantly predicted OPA (R2 = .565, p < .001) and accounted for 56.5% of variance in OPA. A summary of regression coefficients is shown in Table 21 and shows that 21 of the 33 variables significantly contributed to the model, controlling for the other covariates. The variables included gender (B = .024, p = .002), Black (B = -.075, p < .001), Hispanic (B =

137

-.044, p < .001), American Indian (B = -.085, p = .002), family education (B = .014, p =

.002), high school GPA percentage of max (B = .003, p < .001), recruited college athlete

(B = -.086, p < .001), honors list (B = .300, p < .001), passion subscore (B = .015, p <

.001), resilience subscore (B = .043, p < .001), involuntary disenrollment (B = -.198, p <

.001), academic probation (B = -.322, p < .001), aptitude probation (B = -.136, p < .001), athletic probation (B = -.118, p < .001), conduct probation (B = -.063, p = .010), honor probation (B = -.147, p < .001), intercollegiate sports (B = -.029, p = .003), competitive programs (B = -.018, p = .049), mission support programs (B = .024, p = .004), professional programs (B = .064, p = .001), and recreational programs (B = .036, p =

.016).

138

Table 21

Model 2.4a Regression Results Comparing OPA with Cadet Predictors

CI Lower CI Upper Variables B SE ß p bound bound Gender .024 .008 .030 .002*** .009 .039 Black -.075 .014 -.056 .000*** -.102 -.048 Hispanic -.044 .012 -.036 .000*** -.066 -.021 American Indian -.085 .027 -.029 .002*** -.139 -.031 Asian -.007 .012 -.006 .539 -.030 .016 Native Hawaiian/Pacific Islander -.017 .024 -.006 .491 -.064 .031 Race unknown -.014 .017 -.008 .399 -.047 .019 Family education .014 .004 .056 .002*** .006 .023 Family income .003 .003 .011 .371 -.003 .009 Family members .003 .003 .012 .241 -.002 .008 From single parent home .008 .013 .008 .550 -.019 .035 First generation college student .009 .013 .013 .504 -.018 .036 High school GPA percentage of max .003 .000 .079 .000*** .002 .003 High school athletics .001 .013 .001 .916 -.024 .026 Recruited college athlete -.086 .012 -.099 .000*** -.109 -.063 Honors list .300 .011 .313 .000*** .279 .321 Passion subscore .015 .004 .034 .000*** .007 .024 Resilience subscore .043 .006 .074 .000*** .032 .055 Voluntary disenrollment .007 .017 .005 .703 -.028 .041 Involuntary disenrollment -.198 .027 -.068 .000*** -.251 -.145 Academic probation -.322 .008 -.410 .000*** -.338 -.306 Aptitude probation -.136 .023 -.091 .000*** -.181 -.091 Athletic probation -.118 .013 -.088 .000*** -.143 -.093 Conduct probation -.063 .024 -.039 .010*** -.110 -.015 Honor probation -.147 .021 -.065 .000*** -.188 -.107 Intercollegiate sports -.029 .010 -.040 .003*** -.048 -.010 Intramural sports .013 .010 .017 .175 -.006 .032 Competitive programs -.018 .009 -.019 .049* -.035 .000 Mission support programs .024 .008 .027 .004*** .008 .041 Professional programs .064 .019 .030 .001*** .027 .102 Club programs .037 .020 .017 .065 -.002 .076 Recreational programs .036 .015 .022 .016** .007 .066 Civil Air Patrol, Boy/Girl Scouts, -.007 .008 -.009 .367 -.022 .008 Campfire Girls Note. Model 2.4a R2 = .565, p < .001, Confidence Interval (CI) = 95%, *p < .05, **p < .025, ***p < .0125

The regression results for Model 2.4a showed that, on average, when controlling for the other covariates, for every one-point increase in passion subscore, OPA was higher by .015 points (B = .015, p < .001) and that for every one-point increase in

139

resilience subscore, OPA was higher by .043 points (B = .043, p < .001). Comparing the practical significance or absolute value of the ß for all the statistically significant variables in the model revealed that the highest effect sizes were academic probation (ß =

-.410) with a medium to large effect on OPA followed by honors list (ß = .313) with a medium effect. However, while both statistically significant, neither resilience subscore

(ß = .074) nor passion subscore (ß = .034) demonstrated practically significant effect sizes.

For Model 2.4b, the overall model significantly predicted OPA (R2 = .542, p <

.001), and accounted for 54.2% of variance in OPA. A summary of regression coefficients is presented in Table 22 below and shows that 17 of the 28 variables significantly contributed to the model, controlling for the other variables. These variables included gender (B = .019, p = .015), Black (B = -.089, p < .001), Hispanic (B = -.044, p

< .001), American Indian (B = -.100, p < .001), family education (B = .015, p < .001), high school GPA percentage of max (B = .003, p < .001), recruited college athlete (B = -

.090, p < .001), honors list (B = .317, p < .001), involuntary disenrollment (B = -.240, p <

.001), intercollegiate sports (B = -.033, p = .001), competitive programs (B = -.023, p =

.014), mission support programs (B = .021, p = .014), professional programs (B = .061, p

= .002), club programs (B = .040, p = .053), recreational programs (B = .042, p = .007),

Grit-S score (B = .054, p < .001), and “probations—all types” (B = -.327, p < .001).

140

Table 22

Model 2.4b Regression Results Comparing OPA with Cadet Predictors

CI Lower CI Upper Variables B SE ß p bound bound Gender .019 .008 .024 .015** .004 .034 Black -.089 .014 -.067 .000*** -.116 -.062 Hispanic -.044 .012 -.036 .000*** -.067 -.021 American Indian -.100 .028 -.034 .000*** -.155 -.046 Asian -.003 .012 -.003 .778 -.027 .020 Native Hawaiian/Pacific Islander -.016 .025 -.006 .512 -.065 .032 Race unknown -.012 .017 -.006 .500 -.045 .022 Family education .015 .004 .056 .002*** .006 .023 Family income .004 .003 .016 .197 -.002 .010 Family members .004 .003 .014 .159 -.001 .009 From single parent home .009 .014 .009 .515 -.019 .037 First generation college student .005 .013 .008 .684 -.022 .033 High school GPA percentage of max .003 .000 .092 .000*** .002 .004 High school athletics -.003 .013 -.002 .846 -.028 .023 Recruited college athlete -.090 .012 -.103 .000*** -.113 -.066 Honors list .317 .011 .332 .000*** .296 .339 Voluntary disenrollment .000 .017 .000 .987 -.035 .035 Involuntary disenrollment -.240 .027 -.083 .000*** -.294 -.187 Intercollegiate sports -.033 .010 -.045 .001*** -.052 -.013 Intramural sports .007 .010 .008 .505 -.013 .026 Competitive programs -.023 .009 -.024 .014** -.041 -.005 Mission support programs .021 .009 .024 .014** .004 .038 Professional programs .061 .020 .029 .002*** .023 .100 Club programs .040 .021 .018 .053* .000 .080 Recreational programs .042 .015 .025 .007*** .011 .072 Civil Air Patrol, Boy/Girl Scouts, -.006 .008 -.008 .434 -.022 .009 Campfire Girls Grit-S survey score .054 .006 .083 .000*** .042 .067 Probations—all types -.327 .008 -.445 .000*** -.342 -.312 Note. Model 2.4b R2 = .542, p < .001, Confidence Interval (CI) = 95%, *p < .05, **p < .025, ***p < .0125

For Model 2.4b, the regression results also showed that, on average, when controlling for the other covariates, for every one-point increase in Grit-S score, OPA was higher by .054 points (B = .054, p < .001). Comparing the practical significance or absolute value of the ß for all of the statistically significant variables in the model revealed that “probations—all types” (ß = -.445) had a medium to large effect on OPA

141

followed by honors list (ß = .332) with a medium effect. However, while statistically significant, Grit-S score (ß = .083) did not demonstrate a practically significant effect size.

These two models showed there were several variables that demonstrated statistically significant associations with OPA, including, in part, academic probation,

“probations—all types”, honors list, Grit-S score, passion subscore, and resilience subscore. However, neither Grit-S score, passion subscore, nor resilience demonstrated significant effect sizes on OPA.

RQ3 – Attrition vs Grit-S Score

The third research question explored the extent to which Grit-S score was associated with cadet attrition. To analyze attrition, all cadets were put into two groups: not disenrolled (or cadets who remained enrolled through graduation), and disenrolled which included those cadets who were either voluntarily disenrolled or involuntarily disenrolled. Logistic regression models were used to understand the odds that a specific cadet predictor would be associated with one of the three attrition categories. This research question is focused on the association between the dependent variables and Grit-

S score, passion subscore, and resilience subscore; thus, while the results show other statistically significant variables only interpretations for the three grit variables will be discussed.

Enrollment Status. These models used the dependent variable enrollment status to compare disenrollment and non-disenrollment with the Grit-S score and subscores and other cadet predictor variables as shown in Tables 23 and 24 below. For comparison,

Model 3.1a included the Grit-S survey passion and resilience subscores and Model 3.1b

142

used the combined Grit-S score. In a similar manner to the first two research questions, the first sub-model contains the separated performance and probation variables and the second sub-model combines them into OPA and “probations—all types”. The "enter" logistic regression method was used to determine which independent variables were predictors of non-disenrollment. Data screening identified a small number of outliers, which were all kept in order to preserve sample integrity. The race variables were dummy coded, and Caucasian was used as the control variable.

Regression results from Model 3.1a indicated the overall model fit of the predictors was highly questionable (-2 Log likelihood = 2694.317) but was statistically reliable in predicting cadets who were not disenrolled (p < .001) with one degree of freedom (df). A summary of regression coefficients is presented in Table 23 below and shows that eight of the 42 variables significantly contributed to the model, controlling for the other covariates.

143

Table 23

Model 3.1a Logistic Regression Coefficients Comparing Enrollment Status with Cadet Predictors

Odds Lower Upper Variables B SE Wald p Ratio CI CI Gender .216 .127 2.946 .090 1.241 .967 1.593 Black -.327 .197 2.800 .096 .721 .490 1.060 Hispanic -.329 .190 3.040 .083 .719 .496 1.044 American Indian -.055 .398 .023 .890 .946 .434 2.063 Asian -.042 .192 .051 .828 .959 .658 1.398 Native Hawaiian/Pacific Islander -.580 .454 1.650 .202 .560 .230 1.364 Race unknown .247 .241 1.060 .306 1.280 .798 2.052 Family education - high school .480 .692 3.455 .498 1.616 .372 7.017 Family education - some college .576 .657 4.140 .393 1.779 .442 7.149 Family education - Associate degree .547 .703 3.959 .449 1.727 .388 7.699 Family education - Bachelor's degree .644 .696 5.214 .370 1.905 .430 8.445 Family education - Graduate degree .723 .688 5.856 .310 2.060 .475 8.938 Family income - don't know -.437 .242 3.474 .071 .646 .402 1.038 Family income < 25K -.723 .302 6.150 .017 .485 .268 .878 Family income 25K-74,999K -.379 .206 3.613 .067 .685 .457 1.026 Family income 75K-124,999K -.289 .184 2.847 .116 .749 .522 1.075 Family income 125K - 174,999K -.335 .192 3.372 .082 .716 .491 1.043 Family members .026 .041 .477 .521 1.027 .947 1.113 From single parent home .068 .168 .352 .687 1.070 .769 1.490 First generation college student -.191 .176 2.194 .282 .826 .582 1.173 High school GPA percentage of max .010 .005 3.650 .066 1.010 .999 1.020 High school athletics -.072 .203 .133 .721 .930 .625 1.384 Recruited college athlete .372 .204 3.365 .068 1.451 .973 2.164 Honors list -1.864 .154 166.626 .000 .155 .115 .210 Grade point average (GPA) -.281 .154 4.451 .070 .755 .557 1.024 Military performance average (MPA) 2.048 .280 75.510 .000 7.750 4.454 13.487 Physical education average (PEA) -.304 .163 4.148 .063 .738 .535 1.017 Passion subscore -.114 .071 2.552 .111 .893 .776 1.027 Resilience subscore .055 .088 .399 .535 1.056 .888 1.256 Academic probation -.015 .145 .080 .915 .985 .740 1.310 Aptitude probation -.538 .294 3.450 .067 .584 .328 1.038 Athletic probation .478 .182 7.110 .009 1.613 1.129 2.304 Conduct probation .809 .296 7.509 .006 2.246 1.258 4.009 Honor probation .062 .278 .060 .824 1.064 .617 1.833 Intercollegiate sports -.723 .185 15.380 .000 .485 .338 .697 Intramural sports -.114 .144 .648 .429 .893 .674 1.183 Competitive programs -.435 .171 6.527 .011 .647 .463 .904 Mission support programs -.215 .153 2.002 .158 .806 .598 1.088 Professional programs -.280 .388 .522 .472 .756 .353 1.619

144

Model 3.1a Logistic Regression Coefficients Comparing Enrollment Status with Cadet Predictors

Odds Lower Upper Variables B SE Wald p Ratio CI CI Club programs -.235 .419 .319 .575 .791 .348 1.797 Recreational programs -.068 .293 .055 .816 .934 .526 1.659 Civil Air Patrol, Boy/Girl Scouts, .469 .114 17.269 .000 1.598 1.279 1.996 Campfire Girls Constant -1.985 1.601 2.181 .216 .137 .006 3.229

Note. N = 5,454, df = 1, Confidence Interval (CI) = 95%

These variables included family income < 25K (B = -.723, p = .017), honors list

(B = -1.864, p < .001), MPA (B = 2.048, p < .001), athletic probation (B = .478, p =

.009), conduct probation (B = .809, p = .006), intercollegiate sports (B = -.723, p < .001), competitive programs (B = -.435, p = .011), and Civil Air Patrol, Boy/Girl Scouts,

Campfire Girls (B = .469, p < .001). However, passion (B = -.114, p < .111) and resilience (B = .055, p < .535) subscores were not statistically significant predictors of cadets who were not disenrolled.

While not included in the analysis of this study, the odds ratio related to MPA variable in relation to non-disenrollment was particularly noteworthy. The odds ratio for

MPA (OR = 7.750, p < .001) indicated that for every one-point increase in MPA, a cadet was predicted to be 7.75 times more likely to remain enrolled through graduation. This is an interesting finding given that a one-point increase in MPA may simply be an increase from 2.50 to 3.50. This highlights the strong relationship between MPA and enrollment and may serve as an interesting opportunity for future research.

For Model 3.1b, the overall model fit of the predictors was also highly questionable (-2 Log likelihood = 2799.482) but was statistically reliable in predicting cadets who were not disenrolled (p < .001) with one degree of freedom (df). A summary

145

of regression coefficients is presented in Table 24 below and shows that 13 of the 35 variables significantly contributed to the model, controlling for the other covariates.

Table 24

Model 3.1b Logistic Regression Coefficients Comparing Enrollment Status with Cadet Predictors

Odds Lower Upper Variables B SE Wald p Ratio CI CI Gender .099 .122 .679 .415 1.104 .870 1.402 Black -.365 .191 3.697 .056 .694 .478 1.009 Hispanic -.330 .185 3.196 .075 .719 .500 1.033 American Indian -.099 .381 .075 .796 .906 .429 1.912 Asian -.037 .189 .042 .847 .964 .665 1.397 Native Hawaiian/Pacific Islander -.550 .447 1.531 .218 .577 .240 1.384 Race unknown .289 .239 1.485 .226 1.335 .836 2.131 Family education - high school .343 .662 2.709 .612 1.409 .347 5.725 Family education - some college .432 .592 2.802 .475 1.540 .444 5.339 Family education - Associate degree .433 .652 3.052 .515 1.542 .389 6.113 Family education - Bachelor's degree .645 .640 5.072 .329 1.906 .489 7.419 Family education - Graduate degree .719 .639 5.770 .277 2.053 .528 7.978 Family income - don't know -.465 .237 4.079 .049 .628 .395 .999 Family income < 25K -.852 .294 8.928 .004 .427 .240 .760 Family income 25K-74,999K -.473 .204 5.894 .021 .623 .418 .930 Family income 75K-124,999K -.396 .180 5.565 .028 .673 .472 .959 Family income 125K - 174,999K -.300 .189 2.835 .114 .741 .511 1.074 Family members .013 .040 .160 .744 1.013 .936 1.097 From single parent home .072 .171 .468 .672 1.075 .768 1.506 First generation college student -.587 .168 20.292 .001 .556 .397 .778 High school GPA percentage of max .010 .005 4.407 .044 1.010 1.000 1.021 High school athletics -.072 .200 .137 .719 .930 .629 1.377 Recruited college athlete .395 .199 3.963 .047 1.485 1.005 2.194 Honors list -1.713 .145 161.725 .000 .180 .136 .240 Intercollegiate sports -.571 .177 10.472 .001 .565 .399 .800 Intramural sports -.088 .137 .415 .523 .916 .701 1.198 Competitive programs -.363 .166 4.826 .029 .695 .503 .963 Mission support programs -.161 .150 1.169 .282 .851 .635 1.141 Professional programs -.332 .384 .753 .387 .718 .338 1.522 Club programs -.412 .420 .980 .326 .662 .291 1.507 Recreational programs -.102 .292 .128 .726 .903 .509 1.600 Civil Air Patrol, Boy/Girl Scouts, .443 .111 16.134 .000 1.558 1.253 1.936 Campfire Girls Overall performance average (OPA) .572 .255 8.555 .029 1.771 1.063 2.951

146

Model 3.1b Logistic Regression Coefficients Comparing Enrollment Status with Cadet Predictors

Odds Lower Upper Variables B SE Wald p Ratio CI CI Grit-S score -.074 .100 .585 .459 .929 .764 1.129 Probations—all types .020 .138 .129 .885 1.020 .778 1.338 Constant 2.453 1.425 3.808 .086 11.626 .704 192.099 Note. N = 5,454, df = 1, Confidence Interval (CI) = 95%

These variables included Black (B = -.365, p = .056), family income—“don’t know” (B = -.465, p = .049), family income < 25K (B = -.852, p = .004), family income

25K – 74,999K (B = -.473, p = .021), family income 75K – 124,999K (B = -.396, p =

.028), first generation college student (B = -.587, p = .001), high school GPA percentage of max (B = .010, p = .044), recruited college athlete (B = .395, p = .047), honors list (B

= -1.713, p < .001), intercollegiate sports (B = -.571, p = .001), competitive programs (B

= -.363, p = .029), Civil Air Patrol, Boy/Girl Scouts, Campfire Girls (B = .443, p < .001), and OPA (B = .572, p = .029). Grit-S score (B = -.074, p = .459) was not a statistically significant predictor of cadets who were not disenrolled.

The purpose of research question three was to understand the association between the attrition variables not-disenrolled and disenrolled and a list of cadet predictor variables, including Grit-S scores, passion subscores, and resilience subscores. The results of this research question show that there were several statistically significant variables associated with attrition, such as Black, family income—“don’t know,” family income < 25K, family income 25K – 74,999K, family income 75K – 124,999K, first generation college student, high school GPA percentage of max, recruited college athlete,

MPA, OPA, honors list, athletic probation, conduct probation, intercollegiate sports, competitive programs, and Civil Air Patrol, Boy/Girl Scouts, Campfire Girls. However,

147

the results also showed no clear association between the attrition variables and Grit-S score, passion subscore, and resilience subscore.

RQ4 – Changes in Grit-S Score Over Time vs Cadet Predictors

The fourth research question explored the changes in cadet Grit-S score over time, including understanding how the Grit-S, passion, and resilience gain scores were associated with cadet predictor variables. Separate models were completed using Grit-S gain score, passion gain score, and resilience gain score each separately as the dependent variable, and two sub-models were completed separating out each of the performance and probation statistics, and then combining those into the OPA and “probations—all types.”

Average Changes in Grit-S Score Variables. The first step in this analysis was to understand the change in Grit-S scores, passion subscores, and resilience subscores over time. It is important to note that for this trend comparison, the time interval between surveys and between the first and last survey were not factored in since this period varied from one month to nearly four years. The next three figures represent the grit variable gain scores from three different groups. Group one includes all the cadets who took multiple surveys and Figure 1 shows the changes that occurred between their first and second Grit-S survey.

148

Group 1 (1,137 cases)

3.95 4.00 3.96 3.90 3.80 3.70 3.60 3.60 3.60 3.50 3.40 3.30 Average Scores Average 3.27 3.20 3.10 Resilience 3.00 3.24 1 Grit-S

Passion 2 Number of Surveys Taken

Figure 3. Grit-S, passion, and resilience score trends for group one

The results of this comparison show that after this first group of 1,137 cadets took their second Grit-S survey, that on average, resilience subscore increased by .01 points (<

1%), passion subscore decreased by .03 points (< 1%), and Grit-S score remained the same. This comparison revealed very little change in any of the grit variable scores between the first and second survey.

149

Group 2 (171 cases)

3.89 3.90 3.85 3.86 3.70 3.57 3.50 3.51 3.30 3.52 3.24 3.10 2.90 3.17 3.17 Average Scores Average 2.70 2.50 Resilience 1 Grit-S 2 Passion 3

Number of Surveys Taken

Figure 4. Grit-S, passion, and resilience score trends for group two

The results of this comparison show that for the second group of 171 cadets who took three Grit-S surveys, that on average, the resilience subscore first decreased by .04 points (-1.0%) and then increased by .01 points (< 1%) for an overall decrease of .03 points (< 1%). Group two passion subscore initially decreased by .07 points (-2.2%) and then remained the same for an overall decrease of .07 points (-2.2%). Group two Grit-S survey score initially decreased by .06 points (-1.7%) and then increased by .01 points (<

1%) for an overall decrease of .05 points (-1.4%). This comparison revealed very little change in the resilience and passion subscores and only a slight change in the grit variable scores.

150

Group 3 (7 cases)

4.30 4.10 4.21 4.00 3.90 3.68 3.68 3.70 3.82 3.70 3.50 3.45 3.55

3.30 3.43 3.21 3.39 3.10 3.43 Average Scores Average 2.90 2.70 2.50 1 Resilience 2 Grit-S 3 Passion 4 Number of Surveys Taken

Figure 5. Grit-S, passion, and resilience score trends for group three

The results of this comparison show that for the third group of seven cadets who took four Grit-S surveys, that on average, the resilience subscore initially stayed the same, then increased by .32 points (8.7%), then increased again by .21 points (5.3%) for an overall increase of .53 points (14.4%). The passion subscore initially increased by .22 points (6.7%), then decreased by .04 points (-1.2%), and then increased by .04 points

(1.2%) for an overall increase of .22 points (6.7%). For group three Grit-S score initially increased by .10 points (2.9%), then increased again by .15 points (4.2%), and finally increased again by .12 points (3.2%) for an overall increase of .37 points (10.7%).

Comparing all three graphs show there was very little or only slight changes to the grit variable scores for the first two groups. None of the increases or decreases were

151

large enough to change categories, for example changing from “somewhat like me” worth three points, to “mostly like me” worth four points. However, in every group, the resilience subscore was consistently higher than the passion subscore by 7.3% to 22.7%.

This indicates that on average, cadet resilience subscores were consistently higher than their passion subscores.

As described in the variables section of chapter three, the differences in time intervals between first and last Grit-S score was used to standardize the gain scores.

After analyzing the time interval between Grit-S scores, 73.86% of the cases showed a difference in time interval of only 12 months. Removing these cases (n = 198) reduced the total number of cases by 14.8% leaving a final group of 1,137 cases for gain score comparison. The gain score was calculated by subtracting the first Grit-S survey score from the last survey score. The same procedure was completed for the passion gain scores and resilience gain scores, and these scores were regressed on the list of independent variables to understand the relative associations. The next step in this analysis was to conduct a regression analysis comparing the gain scores to cadet predictors. The regression analysis descriptive statistics for research question four are presented below in Table 25 and include the mean and standard deviation for the variables used in the regression models.

152

Table 25

Descriptive Statistics Comparing Grit Variable Gain Scores with Cadet Predictors

Variables Mean / Proportion SD Grit-S gain score -.003 .517 Passion gain score -.035 .739 Resilience gain score .030 .602 Gender 30.00% Black 6.00% Hispanic 9.00% American Indian 2.00% Asian 8.00% Native Hawaiian / Pacific Islander 2.00% Race unknown 4.00% Family education 3.910 High school 7.10% Some college 12.20% Associate degree 7.80% Bachelor's degree 34.00% Graduate degree 33.20% Unknown 5.70% Family income 4.310 Unknown 5.20% <25K 2.20% 25-74,999K 17.05% 75-124,999K 29.30% 125-174,999K 24.00% ≥175K 22.25% Family members 4.320 1 2.20% 2 4.10% 3 15.70% 4 38.50% 5 25.20% 6 9.20% 7 2.90% 8 1.70% 9 0.10% ≥10 0.40% From single parent home 13.00% First generation college student 49.00% High school GPA percentage of max 94.11% .082 High school athletics 96.00% Recruited college athlete 17.00% Honors list 92.00% Grade point average (GPA) 3.092 .473 Military performance average (MPA) 3.334 .219 Physical education average (PEA) 2.780 .403

153

Descriptive Statistics Comparing Grit Variable Gain Scores with Cadet Predictors

Variables Mean / Proportion SD Voluntary disenrollment 4.00% Involuntary disenrollment 0.00% Academic probation 25.00% Aptitude probation 5.00% Athletic probation 7.00% Conduct probation 4.00% Honor probation 2.00% Intercollegiate sports 32.00% Intramural sports 81.00% Competitive programs 19.00% Mission support programs 22.00% Professional programs 4.00% Club programs 5.00% Recreational programs 6.00% Civil Air Patrol, Boy / Girl Scouts, Campfire Girls 30.00% Overall performance average (OPA) 3.081 .315 Probations—all types 33.00% Note. N = 1,137

Grit-S gain score. The first set of models compared Grit-S gain score with the list of cadet predictor variables as shown in Tables 26 and 27 below. For comparison,

Model 4.1a included the performance variables of cadet’s grade point average (GPA), military performance average (MPA), and physical education average (PEA), and Model

4.1b used the overall performance average (OPA) predictor variable since the GPA,

MPA, and PEA are combined to create the cadet OPA. The same procedure was applied to the probation variables, and Model 4.1a included academic, aptitude, athletic, conduct, and honor probations while Model 4.1b combined the five probation types into a

“probations—all types” variable. The race variables were dummy coded, and Caucasian was used as the control variable.

Regression results indicated that for Model 4.1a, the overall model did not significantly predict the Grit-S gain score (R2 = .039, p = .121). A summary of regression coefficients is presented in Table 26 below and shows that only two of the 34

154

variables significantly contributed to the model, controlling for the other covariates.

These variables included GPA (B = .142, p = .004) and mission support programs (B =

.091, p = .018).

155

Table 26

Model 4.1a Regression Results Comparing Grit-S Gain Scores with Cadet Predictors

CI Lower CI Upper Variables B SE ß p bound bound Gender -.009 .036 -.008 .814 -.080 .063 Black .099 .069 .045 .151 -.036 .235 Hispanic .079 .056 .043 .156 -.030 .189 American Indian .029 .113 .008 .797 -.192 .250 Asian .043 .057 .023 .453 -.069 .154 Native Hawaiian / Pacific Islander -.043 .100 -.013 .668 -.239 .154 Race unknown .093 .086 .033 .278 -.075 .261 Family education -.009 .019 -.022 .650 -.047 .029 Family income .015 .013 .037 .279 -.012 .041 Family members -.006 .013 -.015 .649 -.030 .019 From single parent home -.031 .052 -.021 .548 -.133 .071 First generation college student -.051 .038 -.049 .177 -.124 .023 High school GPA percentage of max -.001 .002 -.023 .460 -.005 .002 High school athletics .046 .085 .017 .586 -.121 .214 Recruited college athlete .025 .058 .019 .663 -.089 .140 Honors list .002 .066 .001 .970 -.127 .132 Grade point average (GPA) .142 .050 .130 .004 .045 .240 Military performance average (MPA) -.033 .095 -.014 .731 -.218 .153 Physical education average (PEA) .026 .047 .020 .578 -.066 .118 Voluntary disenrollment -.099 .084 -.036 .238 -.264 .066 Involuntary disenrollment -.144 .234 -.019 .538 -.603 .314 Academic probation .077 .045 .065 .089 -.012 .165 Aptitude probation .047 .131 .020 .720 -.210 .303 Athletic probation -.094 .066 -.046 .157 -.224 .036 Conduct probation -.002 .141 -.001 .991 -.279 .275 Honor probation .088 .110 .024 .427 -.128 .304 Intercollegiate sports -.023 .048 -.021 .631 -.118 .072 Intramural sports -.025 .050 -.019 .624 -.124 .074 Competitive programs .015 .041 .011 .719 -.066 .095 Mission support programs .091 .039 .073 .018 .016 .167 Professional programs .104 .081 .039 .199 -.055 .263 Club programs -.100 .076 -.040 .191 -.249 .050 Recreational programs .042 .067 .019 .527 -.089 .173 Civil Air Patrol, Boy/Girl Scouts, .028 .035 .025 .436 -.042 .097 Campfire Girls 2 Note. Model 4.1a R = .039, p = .121, Confidence Interval (CI) = 95%

156

The regression results for Model 4.1a show that, on average, when controlling for the other covariates, for every one-point increase in GPA, Grit-S gain score was higher by .142 points (B = .142, p = .004) and that the Grit-S gain score for cadets who participated in mission support programs was higher by .091 points (B = .091, p = .018) than for non-participants. Comparing the practical significance or absolute value of the ß for these two variables revealed that GPA (ß = .130) had a small effect size and that mission support programs (ß = .073) did not have a significant effect size, according to

Cohen’s broad guidelines. These guidelines were used throughout research question four to provide practical effect size comparisons. It should be noted that the significance of the model suggests a better model specification is possible.

For Model 4.1b, regression results also indicated that the overall model did not significantly predict the Grit-S gain score (R2 = .033, p = .111). A summary of regression coefficients is presented in Table 27 below and shows that only two of the 28 variables significantly contributed to the model, controlling for the other covariates.

These variables included OPA (B = .180, p = .010) and mission support programs (B =

.092, p = .017).

157

Table 27

Model 4.1b Regression Results Comparing Grit-S Gain Scores with Cadet Predictors

CI Lower CI Upper Variables B SE ß p bound bound Gender -.021 .035 -.018 .556 -.090 .049 Black .105 .069 .048 .125 -.029 .240 Hispanic .082 .056 .044 .142 -.027 .191 American Indian .042 .113 .011 .710 -.179 .262 Asian .048 .057 .026 .400 -.063 .159 Native Hawaiian / Pacific Islander -.056 .100 -.017 .577 -.252 .140 Race unknown .103 .085 .037 .230 -.065 .270 Family education -.007 .019 -.019 .697 -.045 .031 Family income .014 .013 .036 .301 -.012 .040 Family members -.005 .012 -.013 .694 -.029 .020 From single parent home -.031 .052 -.020 .555 -.133 .072 First generation college student -.057 .035 -.055 .102 -.126 .011 High school GPA percentage of max -.002 .002 -.024 .437 -.005 .002 High school athletics .052 .085 .019 .539 -.114 .219 Recruited college athlete .035 .058 .026 .545 -.079 .150 Honors list .001 .065 .000 .993 -.127 .128 Voluntary disenrollment -.095 .083 -.034 .255 -.258 .068 Involuntary disenrollment -.144 .234 -.019 .537 -.602 .313 Intercollegiate sports -.014 .048 -.012 .776 -.107 .080 Intramural sports -.023 .050 -.018 .641 -.121 .075 Competitive programs .017 .041 .013 .680 -.063 .097 Mission support programs .092 .038 .074 .017 .017 .167 Professional programs .099 .080 .037 .220 -.059 .256 Club programs -.105 .076 -.042 .167 -.253 .044 Recreational programs .046 .067 .021 .491 -.085 .177 Civil Air Patrol, Boy/Girl Scouts, .027 .035 .024 .449 -.042 .095 Campfire Girls Overall performance average (OPA) .180 .070 .110 .010 .043 .316 Probations—all types .018 .042 .016 .667 -.064 .100 2 Note. Model 4.1b R = .033, p = .111, Confidence Interval (CI) = 95%

In Model 4.1b, the regression results show that, on average, controlling for the

other covariates, for every one-point increase in OPA, Grit-S gain score was higher by

.180 points (B = .180, p = .010) and that the Grit-S gain scores for cadets who

participated in mission support programs were higher by .092 points (B = .092, p = .017).

158

Comparing the practical significance or absolute value of the ß for these two variables revealed that OPA (ß = .110) had a small effect size and that mission support programs (ß

= .074) did not have a practically significant effect size. It should be noted that the significance of the model suggests a better model specification is possible.

These two models showed that GPA, OPA, and mission support programs demonstrated statistically significant associations with Grit-S gain score. Both GPA (ß =

.130) and OPA (ß = .110) produced significant, albeit small effect sizes, while mission support programs did not demonstrate a significant effect size. Neither of these models demonstrated overall statistical significance so the results should be considered with caution.

Passion gain score. The second set of models compared passion gain score with the list of cadet predictor variables as shown in Tables 28 and 29 below. For comparison,

Model 4.2a included the performance variables of cadet’s grade point average (GPA), military performance average (MPA), and physical education average (PEA), and Model

4.2b used the overall performance average (OPA) predictor variable since the GPA,

MPA, and PEA are combined to create the cadet OPA. The same procedure was applied to the probation variables, and Model 4.2a included academic, aptitude, athletic, conduct, and honor probations while Model 4.2b combined the five probation types into a

“probations—all types” variable. The race variables were dummy coded, and Caucasian was used as the control variable.

Regression results indicated that for Model 4.2a the overall model did not significantly predict the passion gain score (R2 = .034, p = .258). A summary of regression coefficients is presented in Table 28 below and shows that four of the 34

159

variables significantly contributed to the model, controlling for the other covariates.

These variables included Black cadets (B = .225, p = .023), GPA (B = .180, p = .011), academic probation (B = .130, p = .045), and professional programs (B = .259, p = .025).

160

Table 28

Model 4.2a Regression Results Comparing Passion Gain Scores with Cadet Predictors

CI Lower CI Upper Variables B SE ß p bound bound Gender .071 .052 .044 .173 -.031 .173 Black .225 .099 .071 .023 .031 .419 Hispanic .139 .080 .053 .083 -.018 .295 American Indian .012 .162 .002 .939 -.304 .329 Asian .053 .081 .020 .511 -.106 .213 Native Hawaiian / Pacific Islander -.050 .144 -.010 .727 -.332 .231 Race unknown .077 .123 .019 .528 -.163 .318 Family education -.019 .022 -.035 .391 -.063 .025 Family income .028 .019 .050 .136 -.009 .064 Family members .001 .018 .002 .947 -.034 .036 From single parent home -.023 .080 -.010 .776 -.181 .135 First generation college student -.030 .058 -.020 .607 -.143 .083 High school GPA percentage of max .002 .003 .023 .471 -.003 .008 High school athletics .005 .122 .001 .965 -.234 .244 Recruited college athlete -.013 .083 -.007 .874 -.177 .150 Honors list -.079 .094 -.029 .403 -.264 .106 Grade point average (GPA) .180 .071 .116 .011 .041 .320 Military performance average (MPA) -.102 .138 -.030 .459 -.372 .168 Physical education average (PEA) .020 .067 .011 .768 -.112 .152 Voluntary disenrollment -.120 .121 -.030 .318 -.357 .116 Involuntary disenrollment -.196 .335 -.017 .560 -.853 .462 Academic probation .130 .065 .076 .045 .003 .256 Aptitude probation -.037 .187 -.011 .842 -.404 .330 Athletic probation -.122 .095 -.042 .196 -.308 .063 Conduct probation .024 .202 .007 .906 -.373 .420 Honor probation .115 .158 .023 .466 -.194 .424 Intercollegiate sports -.016 .069 -.011 .812 -.152 .119 Intramural sports -.069 .072 -.037 .337 -.211 .072 Competitive programs .003 .059 .001 .964 -.113 .118 Mission support programs .090 .055 .051 .104 -.019 .198 Professional programs .259 .116 .069 .025 .032 .487 Club programs -.110 .109 -.031 .313 -.324 .104 Recreational programs .017 .096 .005 .863 -.171 .204 Civil Air Patrol, Boy/Girl Scouts, .046 .051 .029 .361 -.053 .145 Campfire Girls Note. Model 4.2a R2 = .034, p = .258, Confidence Interval (CI) = 95% The regression results for Model 4.2a show that, on average, when controlling for

the other covariates, that passion gain scores for cadets who were Black were predicted to

161

be higher by .225 points (B = .225, p = .023) than white cadets. For every one-point increase in GPA, passion gain scores were predicted to increase by .180 points (B =

0.180, p = .011). Passion gain scores for cadets who were on academic probation were higher by .130 points (B = .130, p = .045) than those who were not on probation, and passion gain scores for cadets who participated in professional programs were higher by

.259 points (B = .259, p = .025) than non-participants. Comparing the practical significance or absolute value of the ß for these four variables revealed that only GPA (ß

= .116) produced a significant albeit small effect. Black (ß = .071), academic probation

(ß = .076), and mission support programs (ß = .051) did not have practically significant effect sizes. It should be noted that the statistical significance of the model suggests a better model specification is possible.

For Model 4.2b, regression results also indicated that the overall model did not significantly predict the passion gain score (R2 = .028, p = .288). A summary of regression coefficients is presented in Table 29 below and shows that three of the 28 variables significantly contributed to the model, controlling for the other covariates.

These variables included Black cadets (B = .233, p = .018), professional programs (B =

.246, p = .033), and OPA (B = .191, p = .055).

162

Table 29

Model 4.2b Regression Results Comparing Passion Gain Scores with Cadet Predictors

CI Lower CI Upper Variables B SE ß p bound bound Gender .053 .051 .033 .300 -.047 .152 Black .233 .098 .074 .018 .040 .425 Hispanic .145 .080 .056 .068 -.011 .302 American Indian .036 .161 .007 .823 -.280 .352 Asian .059 .081 .022 .466 -.100 .218 Native Hawaiian / Pacific Islander -.064 .144 -.013 .656 -.345 .217 Race unknown .092 .123 .023 .451 -.148 .332 Family education -.018 .022 -.032 .424 -.062 .026 Family income .026 .019 .047 .159 -.010 .063 Family members .001 .018 .002 .951 -.034 .036 From single parent home -.024 .080 -.011 .766 -.182 .135 First generation college student -.045 .054 -.030 .402 -.151 .061 High school GPA percentage of max .002 .003 .021 .497 -.004 .007 High school athletics .016 .122 .004 .896 -.222 .254 Recruited college athlete .000 .084 .000 .997 -.163 .164 Honors list -.083 .093 -.031 .373 -.266 .099 Voluntary disenrollment -.104 .119 -.026 .384 -.338 .130 Involuntary disenrollment -.179 .335 -.016 .593 -.835 .477 Intercollegiate sports -.002 .068 -.001 .974 -.136 .131 Intramural sports -.061 .072 -.033 .397 -.201 .080 Competitive programs .004 .059 .002 .947 -.111 .119 Mission support programs .094 .055 .053 .090 -.014 .202 Professional programs .246 .115 .065 .033 .020 .471 Club programs -.116 .109 -.033 .287 -.328 .097 Recreational programs .023 .096 .007 .809 -.164 .210 Civil Air Patrol, Boy/Girl Scouts, Campfire .047 .050 .029 .344 -.051 .146 Girls Overall performance average (OPA) .191 .100 .081 .055 -.004 .386 Probations—all types .028 .060 .018 .641 -.089 .145

Note. Model 4.2b R2 = .028, p = .288, Confidence Interval (CI) = 95%

For Model 4.2b, the regression results show that, on average, controlling for the

other covariates, passion gain scores for cadets who were Black were higher by .233

points (B = .233, p = .018) than for White cadets, that passion gain scores for cadets who

participated in professional programs were higher by .246 points (B = .246, p = .033)

than for non-participants, and that for every one-point increase in OPA, passion gain

163

score was higher by .191 points (B = .191, p = .055). Comparing the practical significance or absolute value of the ß for these three variables revealed that Black (ß =

.074), professional programs (ß = .065), and OPA (ß = .081) did not produce practically significant effect sizes. It should be noted that the statistical significance of the model suggests a better model specification is possible.

These two models showed that Black, GPA, OPA, academic probation, professional programs, and mission support programs demonstrated statistically significant associations with passion gain score. GPA (ß = .116) produced a significant, albeit small effect size, but Black, OPA, academic probation, professional programs, and mission support programs did not demonstrate significant effect sizes. Neither of these models demonstrated overall statistical significance so the results should be considered with caution.

Resilience gain score. The third set of models compared resilience gain score with the list of cadet predictor variables as shown in Tables 30 and 31 below. For comparison, Model 4.3a included the performance variables of cadets’ grade point average (GPA), military performance average (MPA), and physical education average

(PEA), and Model 4.3b used the overall performance average (OPA) predictor variable since the GPA, MPA, and PEA are combined to create the cadet OPA. The same procedure was applied to the probation variables and Model 4.3a included academic, aptitude, athletic, conduct, and honor probations while Model 4.3b combined the five probation types into “probations—all types”. The race variables were dummy coded, and

Caucasian was used as the control variable.

164

Regression results indicated that for Model 4.3a, the overall model did not significantly predict the resilience gain score (R2 = .042, p = .086). However, a summary of regression coefficients is presented in Table 30 below and shows that three of the 34 variables significantly contributed to the model, controlling for the other covariates.

These variables included gender (B = -.088, p = .037), high school GPA percentage of max (B = -.005, p = .031), and mission support programs (B = .093, p = .039).

165

Table 30

Model 4.3a Regression Results Comparing Resilience Gain Score with Cadet Predictors

CI Lower CI Upper Variables B SE ß p bound bound Gender -.088 .042 -.067 .037 -.171 -.005 Black -.026 .081 -.010 .744 -.184 .132 Hispanic .020 .065 .009 .760 -.108 .147 American Indian .046 .131 .010 .729 -.212 .303 Asian .032 .066 .015 .630 -.098 .161 Native Hawaiian/Pacific Islander -.036 .117 -.009 .760 -.265 .193 Race unknown .108 .100 .033 .276 -.087 .304 Family education .002 .024 .005 .933 -.047 .051 Family income .001 .016 .003 .937 -.030 .033 Family members -.013 .015 -.027 .388 -.041 .016 From single parent home -.040 .062 -.023 .521 -.161 .082 First generation college student -.072 .048 -.060 .133 -.166 .022 High school GPA percentage of max -.005 .002 -.068 .031 -.009 .000 High school athletics .088 .099 .027 .377 -.107 .282 Recruited college athlete .064 .068 .040 .345 -.069 .197 Honors list .084 .077 .038 .275 -.067 .234 Grade point average (GPA) .104 .058 .081 .073 -.010 .217 Military performance average (MPA) .037 .111 .014 .740 -.181 .255 Physical education average (PEA) .033 .055 .022 .553 -.075 .140 Voluntary disenrollment -.078 .098 -.024 .425 -.270 .114 Involuntary disenrollment -.093 .272 -.010 .733 -.626 .440 Academic probation .024 .053 .017 .646 -.079 .127 Aptitude probation .131 .152 .048 .389 -.167 .429 Athletic probation -.065 .077 -.027 .400 -.216 .086 Conduct probation -.027 .164 -.009 .870 -.349 .295 Honor probation .060 .128 .014 .638 -.191 .312 Intercollegiate sports -.030 .056 -.024 .593 -.140 .080 Intramural sports .020 .059 .013 .735 -.095 .135 Competitive programs .027 .048 .018 .573 -.067 .121 Mission support programs .093 .045 .064 .039 .004 .181 Professional programs -.051 .094 -.017 .590 -.236 .134 Club programs -.089 .089 -.031 .314 -.263 .084 Recreational programs .068 .078 .026 .382 -.084 .220 Civil Air Patrol, Boy/Girl Scouts, .009 .041 .007 .828 -.072 .090 Campfire Girls Note. Model 4.3a R2 = .042, p = .086, Confidence Interval (CI) = 95%

The regression results for Model 4.3a show that, on average, controlling for the

other covariates, resilience gain scores for females were lower by .088 points (B = .088, p

166

= .037) than for male cadets, for every one-point increase in high school GPA percentage of max, resilience gain scores were lower by .005 points (B = .005, p = .031), and that resilience gain scores for cadets who participated in mission support programs were higher by .093 points (B = .093, p = .039) than for non-participants. Comparing the practical significance or absolute value of the ß for these three variables revealed that gender (ß = -.067), high school percentage of max (ß = -.068), and mission support programs (ß = .064) did not have practically significant effect sizes. It should be noted that the statistical significance of the model suggests a better model specification is possible.

However, for Model 4.3b, regression results indicated that the overall model significantly predicted resilience gain score (R2 = .039, p = .037) and accounted for 3.9% of variance in resilience gain score. A summary of regression coefficients is presented in

Table 31 below and shows that four of the 28 variables significantly contributed to the model, controlling for the other covariates. These variables included gender (B = -.095, p

= .022), high school GPA percentage of max (B = -.005, p = .030), mission support programs (B = .091, p = .043), and OPA (B = .169, p = .037).

167

Table 31

Model 4.3b Regression Results Comparing Resilience Gain Score with Cadet Predictors

CI Lower CI Upper Variables B SE ß p bound bound Gender -.095 .041 -.072 .022 -.175 -.014 Black -.023 .080 -.009 .777 -.179 .133 Hispanic .018 .065 .008 .784 -.109 .144 American Indian .048 .131 .011 .716 -.209 .304 Asian .036 .066 .017 .582 -.093 .165 Native Hawaiian/Pacific Islander -.048 .116 -.012 .682 -.276 .181 Race unknown .113 .099 .035 .255 -.082 .308 Family education .003 .024 .008 .891 -.045 .052 Family income .001 .016 .003 .927 -.030 .033 Family members -.011 .015 -.024 .452 -.039 .018 From single parent home -.038 .061 -.021 .540 -.158 .083 First generation college student -.069 .044 -.058 .118 -.156 .018 High school GPA percentage of max -.005 .002 -.068 .030 -.009 .000 High school athletics .088 .099 .027 .371 -.105 .282 Recruited college athlete .070 .068 .044 .299 -.062 .203 Honors list .084 .076 .038 .265 -.064 .232 Voluntary disenrollment -.085 .097 -.026 .378 -.275 .104 Involuntary disenrollment -.110 .271 -.012 .685 -.641 .421 Intercollegiate sports -.025 .055 -.019 .653 -.133 .083 Intramural sports .014 .058 .009 .807 -.100 .128 Competitive programs .030 .048 .020 .530 -.063 .123 Mission support programs .091 .045 .063 .043 .003 .178 Professional programs -.049 .093 -.016 .603 -.231 .134 Club programs -.094 .088 -.033 .286 -.267 .079 Recreational programs .069 .077 .027 .374 -.083 .221 Civil Air Patrol, Boy/Girl Scouts, .006 .041 .004 .889 -.074 .085 Campfire Girls Overall performance average (OPA) .169 .081 .088 .037 .010 .327 Probations—all types .008 .048 .006 .868 -.087 .103 Note. Model 4.3b R2 = .039, p = .037, Confidence Interval (CI) = 95%

For Model 4.3b, the regression results show that, on average, controlling for the

other covariates, that resilience gain scores for females were lower by .095 points (B =

.095, p = .022) than for male cadets and that for every one-point increase in high school

GPA percentage of max, resilience gain scores were lower by .005 points (B = -.005, p =

.030). Resilience gain scores for cadets who participated in mission support programs

were higher by .091 points (B = .091, p = .043) than non-participants, and for every one-

168

point increase in OPA, resilience gain scores were higher by .169 points (B = .169, p =

.037). Comparing the practical significance or absolute value of the ß for these four variables revealed that gender (ß = -.072), high school percentage of max (ß = -.068), mission support programs (ß = .063), and OPA (ß = .088) did not have practically significant effect sizes, according to Cohen’s broad guidelines.

These two models showed that gender, high school GPA percentage of max,

OPA, and mission support programs demonstrated statistically significant associations with resilience gain score. However, none of these variables demonstrated significant effect sizes. Only model 4.3b demonstrated overall statistical significance so the results of model 4.3a should be considered with caution.

Summary of Research Question Results

This study analyzed the relationship between USAFA cadet resiliency and a variety of individual characteristics using quantitative regression analyses to determine the extent to which individual characteristics demonstrated statistically significant and practically significant associations with Grit-S, passion, and resilience scores. In all research questions, effect sizes were reported using Cohen’s broad guidelines (Cohen,

1988). This analysis was completed using four research questions that focused on different aspects of cadet character in order to better understand the connection between cadet resilience and a variety of predictor variables.

The purpose of RQ1 was to understand the association between Grit-S score, passion subscore, and resilience subscore, and a list of cadet predictor variables. The results demonstrated that for Grit-S score, 14 of the 33 cadet variables were statistically significant and included gender, Asian, number of family members, first-generation

169

college student, high school GPA percentage of max, honors list, GPA, PEA, OPA, intercollegiate sports, intramural sports, competitive programs, mission support programs, and recreational programs. Of these variables, only OPA demonstrated a practically significant effect size with Grit-S score. For passion subscore, 13 of the 33 cadet variables were statistically significant and included gender, Black, number of family members, high school GPA percentage of max, high school sports, recruited college athlete, honors list, GPA, PEA, OPA, intercollegiate sports, intramural sports, and mission support programs. Of these variables, only OPA demonstrated a practically significant effect size with passion subscore. For resilience subscore, 12 of the 33 cadet variables were statistically significant and included family education level, number of family members, first-generation college student, high school sports, honors list, GPA,

MPA, PEA, OPA, intercollegiate sports, mission support programs, and club programs.

Of these variables, only OPA demonstrated a practically significant effect size with resilience subscore. Comparing the statistically significant variables between the three models revealed seven variables common to all models including number of family members, honors list, GPA, PEA, OPA, intercollegiate sports, and mission support programs.

Notably, OPA was the only variable that was statistically significant and demonstrated a practically significant effect size, albeit small when regressed on Grit-S score (B = .263, ß = .172, p < .001), passion subscore (B = .229, ß = .105, p < .001), and resilience subscore (B = .295, ß = .173, p < .001), according to Cohen’s broad guidance on effect sizes. However, none of the other statistically significant variables demonstrated any practically significant effect sizes. Thus, the results of research

170

question one showed that several variables demonstrated statistically significant associations with Grit-S score, passion subscore, and resilience subscore, and only OPA demonstrated both a statistically significant and practically significant association with the same grit variables. A complete list of the statistically significant variables will be presented in chapter five.

The purpose of RQ2 was to understand the association between the USAFA performance variables GPA, MPA, PEA, and OPA and a list of cadet predictor variables that included, in part, Grit-S scores, passion subscores, and resilience subscores. The result was that Grit-S score, passion subscore, and resilience subscore demonstrated statistically significant associations with both GPA and OPA. Grit-S score and resilience subscore also demonstrated statistically significant associations with MPA and PEA, but passion subscore did not. However, neither Grit-S score, passion subscore, nor resilience subscore produced practically significant effect sizes when regressed on the performance variables, according to Cohen’s broad guidelines. Thus, the results of research question two demonstrated that Grit-S score and resilience subscore demonstrated statistically significant associations to GPA, MPA, PEA, and OPA; passion subscore demonstrated a statistically significant association to GPA and OPA; and that the grit variables produced no practically significant effect sizes.

The purpose of RQ3 was to understand the association between USAFA attrition variables and the cadet predictor variables, including Grit-S scores, passion subscores, and resilience subscores. While the results of this research question showed several statistically significant variables from the list of cadet predictors, there was no association between resilience subscore, passion subscore, overall Grit-S score, and the attrition

171

variables in either model. Thus, the analysis of research question three demonstrated no meaningful associations between Grit-S score, passion subscore, resilience score, and the attrition variables.

The purpose of RQ4 was to focus on the concept of growth by understanding the relationship between cadet predictors and the gain score produced by comparing the first and last Grit-S survey taken within a specified time window. The intent with this question was to understand not just the level of the grit variables, but to analyze how the variables changed over time. The three graphs showing the grit variable score trends revealed that for groups one and two scores changed very little with -.05 points (-1.4%) being the largest change in both groups. However, group three revealed larger changes in the grit variables with resilience subscore showing the largest overall change by increasing a total of .53 points (14.4%) from the first to the last survey. The graphs also revealed that, on average, resilience subscores were consistently higher than passion subscores by .25 to .78 points (7.3% to 22.7%) showing that cadets’ resilience subscores were 18.77% higher than their passion subscores.

The regression results demonstrated that for Grit-S gain score, three of the 34 cadet variables were statistically significant and included GPA, OPA, and mission support programs, and that GPA (B = .142, ß = .130, p = .004) and OPA (B = .180, ß =

.110, p = .010) produced practically significant, albeit small effect sizes. For passion gain score, five of the 34 variables were statistically significant and included Black, GPA,

OPA, academic probation, and professional programs, and GPA (B = .180, ß = .116, p =

.011) produced a practically significant, albeit small effect size. For resilience gain score, four of the 34 variables were statistically significant and included gender, high school

172

GPA percentage of max, OPA, and mission support programs, but none of the variables demonstrated practically significant effect sizes. Comparing the statistically significant variables between the gain score models revealed that OPA and mission support programs were common to all the models.

Thus, the results of RQ4 showed that several variables demonstrated statistically significant associations with the Grit-S, passion, and resilience gain scores. GPA and

OPA demonstrated practically significant effect sizes when regressed on Grit-S gain score, and GPA also demonstrated a practically significant effect size when regressed on passion gain score. A complete list of the statistically significant variables will be presented in chapter five. From the six models used in this research question, only model

4.3b demonstrated overall statistical significance, so the results of the other models are somewhat questionable and should be considered with caution.

The last chapter will present the analysis and findings related to RQ5 along with a discussion of the implications of these results, recommendations for policy changes and development, and opportunities for future research that will benefit from the groundwork this study has provided. This baseline of information provides the critical first step in exploring the valuable data USAFA continues to collect in the ongoing effort to build leaders of character.

173

CHAPTER V

DISCUSSION AND CONCLUSION

The purpose of this study was to understand the relationship between USAFA cadet resiliency and a group of individual characteristics ranging from the background and pre-admission competency of cadets to their performance at USAFA and their participation in a variety of programs designed to develop them into leaders of character.

Since a successful measure of individual resiliency is included as a subscore of the overall Grit-S survey score, the analysis of resiliency was done by analyzing the Grit-S survey scores, comprised of passion and resilience subscores, from 5,454 USAFA cadets who completed 6,974 surveys from 2009 until the Fall of 2018. This large sample size had the statistical power sufficient to analyze a large number of cadet variables to understand their association with Grit-S score, passion subscore, and resilience subscore.

The purpose of this chapter is to discuss the results of each research question in terms of each of the three grit variables; discuss the implications of those results; propose recommendations for policy and practice modifications using the USAFA leadership growth model (LGM) as a framework; and to identify future research opportunities to develop further understanding into the nature of resilience, passion, and overall grit.

While this chapter will discuss the research results in regard to resilience, passion, and grit, the implications and recommendations set forth will focus on resilience.

While discussing the results and implications of the study, it is important to note that this study did not seek to identify the causal relationships between the various independent variables and Grit-S score, passion subscore, and resilience subscore; instead, this study looked for the direction and magnitude of the associations between

174

them. Association simply implies that when one particular behavior is found, an associated resilience subscore, passion subscore, or Grit-S score will also be found without making a claim as to which factor caused the other (Spatz, 2008). Thus, while the results demonstrate clear associations between the grit variables and other predictor variables, additional research conducting pre- and post-tests using control and comparison groups would be necessary to make causal inferences.

It is also important to recognize that even though a particular variable may have a statistically significant association with the dependent variable in question, the standardized coefficient (ß) may demonstrate such a small effect size that it fails to be practically significant. Additionally, even if an independent variable is statistically significant and demonstrates a practically significant effect size, the unstandardized coefficient (B) of the dependent variable generated when the independent variable increases by one point may be such a small change that the difference may not be noticeable or meaningful. What results is a difference without a distinction. Future research to understand the magnitude of meaningful changes of B values is needed to better interpret the overall impact of these changes. Examples of this will be pointed out during the discussion of the results.

Discussion

Discussion of Results

RQ1 – Grit-S Score vs Cadet Predictors. The first research question studied the extent to which cadet characteristics—including demographics, background, performance at USAFA, and participation in various USAFA clubs and programs—were associated with Grit-S score, passion subscore, and resilience subscore. The results identified

175

several statistically significant variables related to each of the grit variables. The results showed that 14 statistically significant variables were associated with Grit-S score.

These included gender, Asian (negative), number of family members (negative), first- generation college student (negative), high school GPA percentage of max (negative),

USAFA honors list, grade point average (GPA), physical education average (PEA), overall performance average (OPA), and participation in USAFA programs including intercollegiate sports, intramural sports, competitive programs, mission support programs, and recreational programs (negative). The results also showed that 13 statistically significant variables were associated with passion subscore. These included gender, Black, number of family members (negative), high school GPA percentage of max (negative), participation in high school sports (negative), recruited college athlete,

USAFA honors list, GPA, PEA, OPA, and participation in intercollegiate sports, intramural sports, and mission support programs. Finally, the results showed that 12 statistically significant variables were associated with resilience subscore. These included family education level (negative), number of family members (negative), first- generation college student (negative), participation in high school sports, USAFA honors list, GPA, PEA, OPA, and participation in intercollegiate sports, mission support programs, and club programs (negative).

Combined, 20 of the 33 variables demonstrated statistical significance with the grit variables, and Table 32 shows each of the variables listed beneath the grit variable they were associated with. In the table, each predictor variable is labeled to show non- academic variables, variables with significant effect sizes, and variables that were negatively associated with any of the grit variable scores.

176

Resilience subscore Resilience † family education level (negative)* level education family (negative)* members of family number (negative) student college generation first sports* school high USAFA honors list* GPA MPA* PEA* OPA* sports* intercollegiate support programs* mission (negative)* programs club ize Passion subscore † gender* Black* (negative)* members of family number GPA school (negative) high % of max sports school (negative)* high athlete* college recruited USAFA honors list* GPA PEA* OPA* sports* intercollegiate sports* intramural support programs* mission Grit-S surveyGrit-S score †

Table 32 Table variables grit with associated predictors significant Statistically gender* Asian (negative)* (negative)* members of family number (negative) student college generation first GPA school (negative) high % of max USAFA honors list* GPA PEA* OPA* sports* intercollegiate sports* intramural programs* competitive programs* support mission (negative)* programs recreational s effect significant with variables † denotes variables, non-academic *denotes Note.

177

This first finding is noteworthy because it demonstrates conclusively that the grit variables had statistically significant associations with a variety of predictor variables, including USAFA performance variables. These results help to clarify the significant but contradictory findings about the relationship between grit and performance in other university students and West Point cadets (Bazelais et al., 2016; Chang, 2014; Duckworth et al., 2007) by confirming the statistically significant associations between the grit variables and performance variables.

The second noteworthy finding is that the majority of the statistically significant variables identified in this study were related to non-academic types of activities and performance. As shown on Table 32, 11 of the 14 variables significantly associated with

Grit-S score were non-academic, 11 of the 13 variables significantly associated with passion subscore were non-academic, and 10 of the 12 variables significantly associated with resilience subscore were non-academic. Non-academic variables common to all three grit variables included number of family members (negative), honors list, PEA,

OPA, intercollegiate sports, and mission support programs. The only academically related variables that showed statistically significant associations with all three grit variables was GPA, and this confirms other research findings that identified close correlations between grit and academic performance (Duckworth et al., 2007). The non- academic variables specific to USAFA include PEA, OPA, intercollegiate sports and mission support programs, and a review of the activities included in the mission support program category reveals a wide variety of activities that cadets can choose to engage in according to their interests, skills, or in preparation for future jobs in the Air Force. The mission support group lists 37 different activities that cross a wide variety of interests

178

including cadet first responders, show choir, combat pistol, cyberwarfare, drum and bugle corps, falcon handler, forensics, honor guard, mock trial, soaring and several others.

With such a large number and wide variety of activities being associated with the grit variables, the results indicate that resilience, passion, and overall grit aren’t limited to academics or even to a specific type of activity. They are associated with individuals who participate in a variety of service groups, performance groups, competitive groups, individuals with interests in investigative and judicial professions, animal care, and asymmetric warfare. These activities are very different from each other.

The association between these programs and the grit variables demonstrate that academic performance, non-academic qualities such as athletic ability, teamwork, leadership, and participation in a variety of different activities are associated with each of the grit variables. Statistical significance between these non-academic variables and the grit variables suggests there are a variety of non-academic factors that influence individual grit, passion, and resilience scores. When interpreting these associations, it is important to realize that this study could not ascertain whether the cadet variables influenced the associations with the grit variables or whether people who already had resilient or gritty character traits were associated with the specific cadet variables.

Exploring the possible causal relationships between resilience, passion, grit, and the activities individuals are engaged in would be an excellent opportunity for future research.

Third, the results revealed variables both positively and negatively associated with the grit variables meaning that for each of these variables, membership in the group, higher performance scores, and participation in these programs were predictably

179

associated with higher grit variable scores. The nine variables positively associated with

Grit-S score included gender, honors list, GPA, PEA, OPA, intercollegiate sports, intramural sports, competitive programs, and mission support programs, and the five variables negatively associated with Grit-S score included Asian, number of family members, first-generation college student, high school GPA percentage of max, and recreational programs. The 10 variables positively associated with passion subscore included gender, Black, recruited college athlete, honors list, GPA, PEA, OPA, intercollegiate sports, intramural sports, and mission support programs, and the three variables negatively associated with passion subscore included number of family members, high school GPA percentage of max, and high school sports. The eight variables positively associated with resilience subscore included high school sports, honors list, GPA, MPA, PEA, OPA, intercollegiate sports, and mission support programs, and the four variables negatively associated with resilience subscore included family education level, number of family members, first-generation college student, and club programs.

Condensing this into an overall list of positively and negatively associated variables revealed that gender, Black, high school sports, recruited college athlete, honors list, GPA, MPA, PEA, OPA, intercollegiate sports, intramural sports, competitive programs, and mission support programs were positively associated with the grit variables, and Asian, family education level, number of family members, first-generation college student, high school GPA percentage of max, high school sports, club programs, and recreational programs were negatively associated with the grit variables. Resiliency theory mentioned that family support is a protective factor that can help to increase

180

individual resilience (Fleming & Ledogar, 2008), but no mention is made about the effects of the number of family members or the levels of parental education. Notably, participation in high school sports was positively associated with resilience subscore but negatively associated with passion subscore. While the results of this study connecting resilience with athletics is supported with resilience research of Olympic athletes

(Fletcher & Sarkar, 2012), there appears to be no research explaining why passion may be negatively associated with participation in high school sports.

Prior grit research studies have primarily focused on academic and athletic performance but with some mixed results. While no literature has been found addressing the types of activities included in mission support programs, research studies have produced evidence that grit is a strong predictor of academic performance in military academies and high schools (Duckworth et al., 2007; Maddi, Matthews, Kelly, Villarreal,

& White, 2012), has a positive influence on athletic performance with Olympians in training (Fletcher & Sarkar, 2012), and is associated with gender in college students

(Chang, 2014). However, research also exists with evidence that grit is not predictive of academic performance, gender, or race/ethnicity and that only perseverance of effort or resilience was a positive predictor of the academic performance of high school and college students (Bazelais et al., 2016; Bowman et al., 2015; Credé et al., 2017). One difference between the opposing research findings may be in the context of each of the studies. The studies involving students at military academies or military training programs (Bartone et al., 2008; Duckworth et al., 2007; Maddi et al., 2012) found significant relationships between the grit variables and performance while studies previously mentioned involving students at non-military university and pre-university

181

institutions (Bazelais et al., 2016; Chang, 2014; Credé et al., 2017) did not find significant relationships between grit and performance. This may suggest that compared to participation in non-military programs, participation in programs involving the military may significantly influence the resilience, passion, and overall grit of the participants.

The results of this study of USAFA cadets would support this finding; however, additional research would be necessary to determine the extent of this relationship.

While no known literature seeks to make comparisons between the performance of military and non-military students, I can say from personal experience that the constant challenges and learning-under-pressure conditions common in military contexts significantly influenced my own resilience, passion, and overall grit as it did for those I trained with. Thus, since little is known about the variables that may have positive or negative impacts on the grit variables, further research investigating the reasons behind the positive and negative associations would add clarity to these findings. This and other research opportunities and will be included in a larger discussion of possible future research opportunities later in this chapter.

Finally, results revealed that seven of the variables had statistically significant associations with all three grit variable scores. These variables included the number of family members, honors list, GPA, PEA, OPA, intercollegiate sports, and mission support programs. The number of family members was the only variable negatively associated with the grit variables, meaning that an increased numbers of family members was predictably associated with lower scores in resilience, passion, and overall Grit-S.

While studies on grit have included the relationship with parent income (Bowman et al.,

2015), family socioeconomic status (Borman & Overman, 2004), and the value of

182

nurturing parents (Luthar, 2006), there is nothing in the literature to explain why the number of family members would be negatively associated with the grit variables.

Resiliency theory actually includes family support as a key aspect of the protective factors as family support assists individuals experiencing disruptive adversities (Fleming

& Ledogar, 2008). This family support is one of the essential elements of resiliency theory. However, the results of this study may suggest that because children in larger families have more people available to help them overcome their challenges, children from smaller families are forced to become more self-reliant. While family support represents a protective factor in resiliency theory, children from large families may actually learn less about how to problem solve or become self-reliant and learn more about relying upon others to solve their problems or simply relying upon other to do things for them. While attending the National Character and Leadership Symposium

(NCLS) held at USAFA in 2016, I listened to a young cadet share the challenges he and his brother faced at a very young age after their caregivers abandoned them for months at a time, forcing them to essentially raise themselves and rely only upon each other. These children learned to be self-reliant problem solvers at a young age and may have much different experiences than children from large families. However, future research would need to examine this finding to discover why the number of family members was negatively associated with resilience, passion, and overall grit.

Conversely, the remaining six cadet variables were positively associated with all three grit variables, indicating that inclusion on the honors lists; increased GPA, PEA, and OPA scores; and participation in intercollegiate sports and mission support programs was predictably associated with higher Grit-S, passion, and resilience scores. Apart from

183

GPA, which has already been discussed, there is no known literature discussing similar results. It is also notable that, out of the overall 33 variables included in this analysis and the 20 statistically significant variables, only OPA, which is a measure of the combination of academic, leadership, and athletic performance, was statistically significant, positive, and demonstrated practically significant effect sizes for resilience subscore, passion subscore, and overall Grit-S score. This suggests both the uniqueness of this variable among the other variables and the strength of its association with the grit variables compared to the other significant predictors. I recognize that since Cohen’s broad guidelines were used to evaluate the effect sizes, the interpretation of the OPA effect size may not be accurate, but since a more appropriate method is not available, the results suggest OPA has a unique and meaningful association with the grit variables.

This suggests other opportunities for future research to confirm the associations of these variables, verify the meaningfulness of the effect sizes, and understand why OPA demonstrates such unique results.

RQ2 – Performance vs Grit-S Score. The second research question examined the extent to which Grit-S scores, passion subscores, and resilience subscores were associated with cadet GPA, MPA, PEA, and OPA. However, different from RQ1, RQ2 used the grit variables as independent variables and regressed them on the USAFA performance variables that were used as dependent variables. Different from the first research question, which was designed to understand the effects of the performance variables on the three grit variables, this question was designed to understand the effects of the grit variables on the performance variables. In other words, RQ2 analyzed what statistically significant associations and/or practically significant effects resilience,

184

passion, and Grit-S scores had on the performance variables GPA, MPA, PEA, and OPA.

Two main results were found.

First, results demonstrated that all three grit variables demonstrated significant and positive associations with GPA and OPA. This is similar to other research findings that found that resilience predicted academic performance and completion of military training academies (Duckworth et al., 2007; Eskreis-Winkler et al., 2014; Maddi et al.,

2012). Grit-S score and resilience subscore were also positively and significantly associated with MPA and PEA, but passion subscore was not associated with MPA or

PEA. MPA and PEA have never been used in resilience research, so there is no mention of them in the literature, but researchers have identified the significant connection between hardiness, resilience, and leadership (Bartone, Barry, et al., 2009; Bartone et al.,

2013) and suggested that resilient leaders were good at developing resilient units

(Bartone, 2006). While this does not directly support the connections between the grit variables and MPA in this study, it does support the connection between the grit variables and leadership development. This shows that all three grit variables are significantly associated with GPA and OPA and that resilience subscore plays a larger part of leadership performance and athletic performance than passion subscore.

Additionally, the statistical figures demonstrate that while passion subscore (B =

.032, p < .001) had a slightly larger unstandardized coefficient than resilience score (B =

.023, p = .014) when regressed on GPA, that resilience subscore (B = .043, p < .001) had a slightly larger unstandardized coefficient than passion subscore (B = .015, p < .001) when regressed on the OPA. Since OPA is an overall performance score made up of

GPA, MPA, and PEA, this may suggest that resilience subscore was a more significant

185

contributor to OPA than passion subscore. However, since no known literature has established benchmarks to identify meaningful changes in grit scores or performance scores, interpretations of the changes to the unstandardized coefficients is only speculative and should be considered with caution.

The results of RQ2 demonstrate a statistically significant association between

Grit-S scores and subscores and the USAFA performance scores and further recognizes the significant and positive relationship between the grit variables and performance. The research studies that were mentioned support the connection between the grit variables and USAFA performance variables although the extent of this relationship is difficult to identify due to the lack of meaningful benchmarks for the grit variables. This provides several research opportunities to understand the diverse effects of Grit-S scores and subscores on performance and to identify the benchmarks by which changes in grit variable scores and performance scores may be evaluated.

RQ3 – Attrition vs Grit-S Score. The third research question studied the extent to which Grit-S score, passion subscore, and resilience subscore were associated with cadet attrition from USAFA. This question used the variables “disenrolled from

USAFA” or “not disenrolled from USAFA” to discover if the grit variables were predictably associated. However, after conducting multiple regression analyses, results demonstrated no statistically significant associations between the grit variables and the cadet attrition variables. While the results produced no significant findings in relation to the grit variables, that in and of itself is a significant finding. Research studies have found individual resilience to be one of the elements that causes students to remain in nursing school (Williamson, Health, & Proctor-Childs, 2013) and that mindfulness and

186

mindfulness-based healing programs, identified in chapter two as a promotive factor in the resilience theory construct, demonstrate positive effects on Navy soldiers trying to recover from injuries (Udell, Ruddy, & Procento, 2018). Thus, the results of this study, finding no association between attrition and the grit variables, diverges from the literature. While the results would suggest that cadets who disenroll from USAFA either voluntarily or involuntarily do so for reasons not associated with their Grit-S scores, passion subscores, and/or resilience subscores, additional research should investigate why these results might diverge from the literature and provide greater insight into understanding the reasons cadets disenroll from USAFA.

RQ4 – Changes in Grit-S Score Over Time vs Cadet Predictors. The fourth research question studied the extent to which cadet characteristics were associated with

Grit-S, passion, and resilience gain scores. The intent of this question was to track the associations between cadet characteristics and grit variable gain scores over time to identify any patterns or trends. Selecting a suitable way to track the grit variable scores over time posed a significant challenge since cadets took Grit-S surveys randomly and at different times in their careers at USAFA. This resulted in a sample of cadets from all the different classes who had taken the surveys at a variety of different ages, different class years, during various times of the year, and at a variety of time intervals spanning 3 months to almost 4 years between surveys. One of the only possible and plausible ways to track changes was to select cadets from the overall sample with the most similar time interval between Grit-S surveys. While this did not account for age, class level, class year, or time of year, it did focus on the change of the overall Grit-S score within a

187

common time period, allowing for analysis of the cadets who took the survey, experienced a similar growth time period, and then took the survey again.

When combined, 8 of the 33 variables demonstrated statistically significant associations with the grit variable gain scores, and Table 33 shows the Grit-S gain score, passion gain score, and resilience gain score categories with the corresponding variables listed below each category. In the table, each predictor variable is labeled to show non- academic variables, variables with significant effect sizes, and variables that were negatively associated with any of the grit variable scores.

Table 33

Statistically significant predictors associated with grit gain score variables

Grit-S gain score Passion gain score Resilience gain score gender (negative)* Black*

high school GPA % of max (negative) GPA† GPA† OPA*† OPA* OPA* academic probation mission support mission support programs* programs* professional programs* Note. *denotes non-academic variables, † denotes significant effect size

The results from this research question produced several interesting and significant findings. First, the results revealed several variables with statistically significant associations with the grit variable gain scores. Three variables were associated with Grit-S gain score, and these included GPA, OPA, and mission support programs. Five variables were associated with passion gain scores, and these included being Black, GPA, OPA, academic probation, and professional programs. Four variables

188

were associated with resilience gain scores, and these included gender (negative), high school GPA percentage of max (negative), OPA, and mission support programs. The results also showed that OPA and mission support programs were common between all the grit variable gain scores again highlighting the significance of non-academic variables. As has already been discussed, OPA is a combination of performance scores in academics, leadership, and athletics. Mission support programs include a variety of activities most of which are neither academic nor athletic.

While no known studies have looked at grit variable gain scores or tracked how resilience, passion, and/or Grit-S scores have changed over time, there are studies that examined learning in stressful conditions outside of the normal academic context and found positive correlations between grit and hardiness and the candidates attending special forces operations training (Bartone et al., 2008) and cadets attending West Point

(Duckworth et al., 2007; Maddi et al., 2012). While cadets at USAFA may have different experiences than West Point cadets or special forces candidates, the findings of this study nonetheless suggest that much of what is associated with resilient and gritty behavior is learned outside of the classroom through a variety of experiences and real-world scenarios under stressful conditions. These findings are also similar to studies that found that college students participating in leadership development programs focused on service activities improved their personal growth and development (Woelk & Weeks, 2010).

However, none of these studies tried to track the grit variables or other non-cognitive characteristics over time. It is important to note that this study did not seek to identify the causal relationships between the grit variable gain scores and cadet characteristics, so it is

189

unknown whether participation in these programs influenced the grit variable scores or if cadets who were already resilient and gritty participated in these programs.

There is a lack of literature on the subject of tracking and evaluating changes in

Grit-S scores, and the results of this study provide some of the first insights into the influences that, over time, may effect changes to Grit-S score, passion subscore, and resilience subscore. However, additional research needs to identify which activities within the USAFA activity categories are statistically and practically significant, to determine to what extent they influence Grit-S, passion, and resilience gain scores, and to understand why OPA and mission support programs have such a significant association to grit variable gain scores.

Second, the results showed there were variables both positively and negatively associated with the grit variable gain scores. This means that for each of these variables, membership in the group, higher performance scores, and participation in these programs was predictably associated with higher grit variable gain scores. The three variables associated with Grit-S gain score were all positive and included GPA, OPA, and mission support programs. The five variables associated with passion gain score were all positive and included being Black, GPA, OPA, academic probation, and professional programs.

However, two of the variables significantly associated with resilience gain score were positive and two of the variables were negative. The two variables positively associated with resilience gain score were OPA and mission support programs and the two variables negatively associated with resilience gain score were gender and high school GPA percentage of max. Since the control variable for gender was male, this means that being female and having a higher high school GPA percentage of max had a statistically

190

significant association with lower resilience gain scores. This finding differs somewhat from other studies that found no significant differences between gender and Grit-S scores

(Bowman et al., 2015) and that found significant and positive associations between high school GPA and Grit-S scores (Duckworth et al., 2007), but no literature has demonstrated findings in relation to Grit-S, passion, or resilience gain scores over time.

Thus, additional research needs to investigate the effects of gender on changes to Grit-S, passion, and resilience gain scores over time.

Third, the results showed that both GPA and OPA produced practically significant and positive effect sizes when regressed on Grit-S and passion gain scores, according to

Cohen’s guidance on effect sizes. GPA showed practically significant and positive effect sizes when regressed on both passion gain score and overall Grit-S gain score, and OPA showed a significant and positive effect size when regressed on overall Grit-S gain score.

This finding supports the results of RQ1 and RQ2 that both academic performance and participation in non-academic activities are significantly associated with resilience, passion, and overall grit. These results are also significant because they help to begin understanding how grit variable scores change over time and which programs, activities or other factors had the most influence on those changes. With a lack of research on the factors that influence changes to resilience, passion, and grit over time and a lack of research on grit variable gain scores, these results provide some of the first steps upon which to launch additional research to understand these associations. Future research in this area may begin to identify specific activities and/or programs that predictably increase resilience, passion, and overall grit. In relation to this finding, it is important to mention again that the effect sizes found in this study were evaluated based on Cohen’s

191

guidance on effect sizes, and while these effects sizes may not represent the most appropriate effect sizes to measure grit variable gain scores, they provide some indication of relative effect size when no better method has been identified. However, this new information provides a foundation upon which to base a myriad of research studies on understanding changes to resilience, passion, and Grit-S scores as they transform over time and discovering new effect size benchmarks against which grit variable gain scores may be evaluated.

Finally, the graphs in chapter four near the beginning of the results section for

RQ4 show the changes to the grit variable scores over time for the three different groups of cadets. For group one (n = 1,137), who took the Grit-S survey twice, the changes to the grit variable scores were less than one percent, revealing no significant change in the grit variables after taking the second survey. For group two (n = 171), who took the Grit-

S survey three times, the changes to the grit variables were both negative to positive but were also very small (< 1% to -2.2%) after taking the third survey. For group three (n =

7), who took the survey four times, there were larger and mostly positive changes in the grit variables (-1.1% to 14.4%) after taking the fourth survey, but since the sample size of this group represents less than one percent of the whole group, the value of this finding is very low. These results demonstrate that while the gain scores may be both statistically and practically significant, the changes to the unstandardized raw scores, on average, may be so low as to have little value in supporting modifications to policies or practice.

While no known literature exists comparing the changes in resilience subscores to passion subscores, this may be related to grit studies of university or pre-university students that found resilience or perseverance was correlated to academic performance

192

but also found that passion was either only minimally or not at all related (Bowman et al.,

2015; Chang, 2014; Credé et al., 2017; Wolters & Hussain, 2015). I recognize that for cadets, completing the Grit-S surveys is voluntary, making it difficult to isolate the reasons for the changes in grit variable scores. Changes in these numbers could also be a result of cadets disenrolling from USAFA and different cadets with higher or lower levels of resilience, passion, and grit completing the surveys. However, the large sample size used to analyze the grit variable gain scores (n = 1,137) gives the analysis a high level of statistical power, which reduces the overall effect of these differences. Nonetheless, these results demonstrate the strong influence of resilience as a subcomponent of overall grit and is similar to other findings that found resilience, synonymous with perseverance of effort, to be a stronger subcomponent of grit than passion, synonymous with consistency of effort (Bowman et al., 2015; Chang, 2014; Wolters & Hussain, 2015).

This finding emphasizes the value of resilience as an important aspect of grit and supports the value of focusing efforts on the development of resilience throughout all aspects of academic, leadership, and athletic development.

The final research question wraps together all the findings of this research study and asks about recommendations for future research, policy development, and practices to build cadet resiliency at USAFA. After discussing the implications of this study, presenting policy and practice recommendations, and identifying opportunities for research, I will summarize those concepts at the end before providing some concluding comments and insights.

193

Addressing Earlier Criticisms

Earlier in chapter two when discussing the strengths and weaknesses of resiliency theory, several criticisms of resiliency theory were presented arguing that 1) individuals can become over-resilient, 2) there is little to no evidence linking Grit-S scores to performance, and 3) strategies to build resiliency are only available to the privileged.

First, while this study did not attempt to address the first criticism, the idea that a person can be overly resilient to the detriment of themselves and others (Chamorro-Premuzic &

Lusk, 2017; Kashdan, 2017) was addressed previously in the literature review by showing how resiliency theory is based on promotive factors that include self-efficacy, mindset, and mindfulness. These concepts lead individuals to examine themselves and their circumstances and take actions that would lead to the improved resilience of themselves and others. Thus, these resilient individuals will not become “overly” resilient since their self-efficacy and mindfulness will lead them to seek a state of resilient reintegration after overcoming adversity (Richardson, 2002). This new state is not achieved in spite of others or themselves but in consideration of the sometimes overwhelming challenges they, and others, may face on a regular basis.

Second, researchers have claimed there is a lack of support correlating resilience and grit with performance indicators, particularly academic achievement among students in university and pre-university settings (Bazelais et al., 2016; Credé et al., 2017). In contrast, this study conducted at a military academy with over 5,400 USAFA cadets over nine years to examine over 30 variables demonstrated a positive, statistically significant, and practically significant relationship between several performance variables and the resilience, passion, and grit variables. These results confirm the findings of studies

194

conducted with West Point cadets that also found significant correlations between the grit variables and overall performance (Duckworth et al., 2007; Maddi et al., 2012). One noticeable difference between the two groups of studies is the element of military versus non-military educational training environments. While untested, this military element may be a key factor in understanding the differences in individual performance levels since students in military environments may experience more stress and challenging conditions than students in non-military environments. Nonetheless, the results of this study show that among USAFA cadets, there is in fact, a significant and positive association between resilience and grit, and overall performance. This finding suggests the existence of other conditions and variables in addition to academic performance that may significantly influence changes to individual resilience, passion, and overall grit.

Thus, while academic performance is a key indicator of grit, other variables should be included in the analysis to give a more detailed scope of the factors that influence the grit variables. Doing this may produce results that include a wider array of grit-influencing factors, which, in turn, may provide key leaders with the information they need to modify policies and programs to better support and develop the resilience, passion, and grit so beneficial to their students.

Others have claimed building resiliency is limited only to those with the resources necessary to engage in resiliency-building activities, creating an environment where only those privileged enough to take part in the activities can have a chance at improving their resilience (Schreiner, 2017). However, the findings of this research study with cadets whose annual family income ranged from less than $30K to over $175K, and whose family education level ranged from high school to graduate degrees demonstrated that

195

neither of these variables was significantly associated with resilience. From the time researchers started looking into stress-resilience in the 1970s until the present day, most of the research related to resilience and later to grit has focused on the ability of at-risk individuals and poor families to display resilient behaviors (Garmezy, 1974; Orthner,

Jones‐Sanpei, & Williamson, 2004). However, no known studies make comparisons between the resilience or grit of individuals from poor versus wealthy families, although reference to wealthy parents who bail out their children is mentioned but was not significantly associated with teenager maladjustment (Luthar & Barkin, 2012).

Additionally, while studies have found that, among educated adults grit increases with age (Duckworth et al., 2007), no known studies have linked parental education levels with the resilience or grit of their children. Thus, this study did not find any evidence to support the idea that building resiliency is available only to the privileged or to individuals from higher socio-economic backgrounds, so there appears to be no evidence to support this position.

Implications

This study revealed a variety of statistically and practically significant variables both positively and negatively associated with Grit-S score, passion subscore, and resilience score. However, while these variables may be statistically significant, practically significant, and demonstrate changes in the unstandardized coefficients, it is important to know whether the results have identified any real distinctions or noticeable differences in outcomes. In other words, does any of this really matter? As an active- duty infantry officer and then military intelligence officer in the Army, I was always mindful of the critical necessity for the “actionable intelligence” that could inform

196

decision-makers before they issued orders. If the information I had to offer them could not inform their ability to act, then the information was largely useless. After retiring from the military, I still respect this need to provide decision-makers with the information they need to make informed decisions, and thus, the purpose of this section is to review the critical takeaways from this research that may enable further action.

The purpose of this research study was to understand the relationship between

USAFA cadet resiliency and a variety of individual characteristics using quantitative regression analyses to determine the extent to which individual characteristics were associated with individual resiliency scores. A careful and comprehensive analysis of a long list of variables were compared with Grit-S survey scores completed by 5,454 cadets over a nine-year period. The results found several significant variables and a few key concepts that may help bridge the gap between resiliency theory and actual practice and provide the actionable intelligence key leaders need to begin applying the concepts immediately. The hope is that the application of these concepts will bring about noticeable results, and the recommendations section that follows after this section is an attempt to give key leaders in any organization a place to start. This research study arguably identified several takeaways, but the most important—and the most easily actionable takeaways—are outlined below.

Implication 1 – Academic and non-academic variables are associated with grit

This study demonstrated that resilience, passion, and grit are associated with more than just academic performance. While this study primarily focused on cadet resiliency, the path to that end included using the Grit-S survey since it includes a reliable test for resiliency (Duckworth & Quinn, 2009). The results from this study demonstrated the

197

close relationship between grit, passion, and resilience and a variety of academic and non-academic factors. However, it became immediately apparent that, while much of resiliency and grit research looks for correlations between those variables and academic performance, the majority of variables significantly and practically associated with resilience, passion, and grit were non-academic factors. Since resiliency theory posits that the promotive factors intrinsic to an individual have a large impact on individual resilience (Braverman, 2001; Zimmerman, 2013; Zimmerman et al., 2013), it makes sense that these characteristics would apply to more than just academic performance and spread into all aspects of an individual’s life.

This connection between the grit variables and non-academic activities was demonstrated best by the findings that the Grit-S, passion, and resilience scores were significantly and positively associated with USAFA honors list, grade point average

(GPA), physical education average (PEA), intercollegiate sports, mission support programs, and also statistically and practically associated with the overall performance average (OPA) of USAFA cadets. GPA is obviously an academic performance indicator, but the remaining variables were much more related to physically demanding activities, leadership development, and participation in programs more reflective of adult life challenges than leadership development, academics, or sports. Cadets can be included on the different USAFA honors lists based on their superior performance in academics, leadership, physical performance, or a combination of all three. PEA and intercollegiate programs distinguish cadets based on their ability to physically perform, and OPA is the overall performance indicator that combines their academic, leadership, and athletic performance into a single performance score. For a cadet, their OPA can determine their

198

leadership positions while at USAFA, their future jobs in the Air Force after graduating from USAFA, and whether or not they are selected for graduate school or pilot training.

Thus, the results would suggest USAFA has developed a more complete way to measure individual success as OPA is a reflection of the complete individual and not just their academic performance. OPA was one of six variables with statistically significant associations with resilience, passion, and overall grit, but it was also the only variable that showed a practically significant association with the grit variables. While I recognize that this practical significance was determined using Cohen’s guidance for effect sizes and that these effect size benchmarks may not be the best benchmarks to use, they provide at least some indication as to the significance of OPA since it was the only variable to do so. The bottom line is that OPA, the measure of overall cadet performance, was statistically, practically, and positively associated with resilience, passion, and overall grit. This means that cadets with a high OPA are predicted to also have high resilience, passion, and overall Grit-S scores. Thus, cadets who demonstrate a high level of performance throughout all aspects of their lives, including academics, leadership opportunities, and athletic and physically demanding activities, also have high levels of resilience and the other grit variables. The research also found mission support programs to be significantly associated with resilience, passion, and grit. These programs are less reflective of performance in academics, leadership, or athletics and more reflective of participation in programs, organizations, and activities that engage individual interests, skills, and talents. This emphasizes the value of having experiences outside of the classroom and more reflective of the challenges faced in everyday living.

199

It also lends evidentiary support to the value of experiential learning or learning experiences outside of the formal education process.

Although finding associations between the grit variables and specific activities within mission support programs was beyond the scope of this study, the findings suggest that participation in organizations, programs, and activities with challenging goals outside of the academic environment presents a way to improve resilient skills and behaviors.

This is supported by researchers who suggest using challenging tasks and recognition as a way to build resilience (Bartone, Barry, et al., 2009). In recent years, several programs have demonstrated success in building resilience in veterans suffering from post- traumatic stress (PTS) by focusing on challenging outdoor experiences that range from surfing, rock climbing, and mountaineering (Donaldson, 2016) to fly-fishing (Project

Healing Waters Fly Fishing, 2017), horseback riding at USAFA (Wilson, 2018), wilderness therapy (Alpern, 2016), and engaging in activities like archery, equine therapy, adaptive water sports, and winter sports such as adaptive skiing, snowboarding, and ice hockey (National Ability Center, 2019). An adventure-based experiential learning program at USAFA is also underway and provides cadets with a variety of challenges to overcome while operating in the outdoors (Rivezzo, 2019).

This concept encourages leaders, teachers, parents, and mentors to focus on the whole-person performance instead of mere academic performance alone. Striving to experience and demonstrate success in a variety of environments in addition to academics can build this concept of demonstrating successful overall performance. At home, parents can encourage their children to participate in local sports, music programs, or community events, any of which can provide leadership and service opportunities. In the

200

classroom, teachers can find projects that get the kids out of the classroom and working together to solve real-life problems. From personal experience, I witnessed a math teacher challenge their students’ math skills with a competition to see who could most accurately determine the area, volume, and mass of the water in a nearby pond by the end of the period. The students were wading through the water figuring out the best way to take measurements without falling completely in, engaging in challenging problem- solving with other students, critically thinking on their own, and developing new theories that they thought could help them win the competition. This is one place where these young people were developing their resilience, perseverance, and grit to succeed. I recognize that the focus of this study was not to identify the effectiveness of specific practices in building or supporting resilience. However, this study did identify certain principles that can be put to use using a variety of different practices. Thus, the results of this study suggest there is great value in using non-academic, real-world scenarios and challenges to support and develop resilient qualities and turning those into practices that will demonstrate and teach these ideas.

Implication 2 – Attrition is not associated with grit

It is a logical assumption that if successful performance is positively associated with resilience, passion, and grit, that unsuccessful performance is probably negatively associated with the same variables. In other words, if higher performance is related to higher levels of grit, then lower performance or quitting is probably related to lower levels of grit. However, this research study found no evidence to support such a claim.

The research results demonstrated no significant association between probation and attrition and individual resilience or grit. This may seem a surprising finding since

201

research has shown grit to be a factor in retention of West Point cadets in military training programs (Duckworth et al., 2007). However, this finding is also loosely related to research that found no correlation between retention in nursing students and the non- cognitive characteristic of conscientiousness that is sometimes related to grit (Deary,

Watson, & Hogston, 2003). These differences aside, it is important to recognize that this study analyzed the relationship between the grit variables with the characteristics of cadets who disenrolled from USAFA and not those who remained at USAFA. This distinction is important because this study examined the characteristics of cadets who disenrolled from USAFA and not of those who remained enrolled. Thus, this study makes no claims as to the relationship between the grit variables and retention; instead, it suggests no relationship exists between the grit variables and attrition. This suggests both that the grit variables do not influence the probation and attrition of USAFA cadets and that beliefs suggesting a lack of resilience or grit as a basis for quitting or otherwise disenrolling from USAFA lack any empirical evidence.

Disenrollment from USAFA can occur for both voluntary and involuntary reasons, and little is truly understood about the characteristics that may lead to disenrollment. In the nine years of data collected for this study, 404 cadets voluntarily disenrolled from USAFA, and 82 cadets were involuntarily disenrolled. However, it is nevertheless revealing to find no statistically significant results connecting the probation or attrition variables with the grit variables at all. I recognize that this small population of disenrollees may render quantitative research ineffective due to a lack of statistical power in the sample size. And since cadets leaving USAFA either voluntarily or involuntarily are typically not interested in completing surveys or conducting interviews,

202

the possibility of a qualitative study is daunting. However, while this presents a significant research challenge, interviewing cadets to examine the individual characteristics and self-reported motivations of those who voluntarily or involuntarily disenrolled from USAFA may uncover any significantly associated characteristics. The takeaway here is that key leaders should not look to a lack of resilience, passion, or overall grit as the reason for a cadet leaving USAFA; instead, they should look for other indicators that may show why cadets leave prior to graduation. With this increased understanding, USAFA leadership may be able to identify other cadet challenges that affect cadet disenrollment from USAFA and provide necessary support that may influence cadets to stay, thereby reducing attrition.

Implication 3 – Resilience is critical to leadership development

Grit-S score is comprised of resilience subscore and passion subscore. In the literature on grit, resilience is referred to as perseverance of effort, and passion is referred to as consistency of effort (Duckworth, 2016; Duckworth et al., 2007; Perkins-Gough,

2013). From this literature, resilience generally refers to the ability to learn, grow, and bounce back from setbacks and failures, and passion generally refers to the ability to maintain the drive to complete long-term goals. Together, these concepts result in a gritty individual who bounces back from setbacks and remains determined to achieve long-term goals. Both concepts are important aspects of grit, but I found myself wondering if one or the other on its own was more closely related to the successful performance grit was focused on predicting. For this reason, I included an analysis of the resilience and passion subscores as well as the overall grit score, and the results

203

demonstrated that some of the most significant variables were more closely associated with resilience than with passion, particularly in reference to leadership.

In the findings for RQ1, participation in high school sports was positively associated with resilience but negatively associated with passion. All four of the USAFA performance variables, GPA, MPA, PEA, and OPA, were significantly associated with resilience while only GPA, PEA, and OPA, and not MPA were associated with passion.

MPA is USAFA’s performance indicator that measures leadership performance, and it was not associated with passion subscore. In the findings for RQ2, resilience was again significantly associated with GPA, MPA, PEA, and OPA, and passion was associated with only GPA and OPA. In this case, passion was not associated with MPA or PEA, indicating that passion did not demonstrate a significant association with leadership performance or athletic performance. In figures 3-5 at the beginning of RQ4, an examination of the changes in the grit variables over time revealed no significant differences. However, in these same tables, resilience score was consistently higher than passion subscore by an average of 18.77%. Thus, when considering the contribution of resilience and passion to the overall concept of leadership development, resilience demonstrated a more significant relationship with leadership than passion, and cadets consistently had higher resilience scores than passion scores.

This is supported by other research studies with university students that found resilience to be significantly correlated with performance indicators while the correlation to passion was either non-existent or weak at best (Bowman et al., 2015; Chang, 2014;

Wolters & Hussain, 2015). It is also supported by a study that found no real correlation between passion or grit and performance indicators but found significant correlations

204

between perseverance, synonymous with resilience, and performance variables (Credé et al., 2017). This suggests that instead of developing policies and programs to build resilience, passion, and grit, resources may be better utilized by developing and supporting resiliency-building activities that may demonstrate a more positive overall impact leadership development. What this means for key leaders is a shift of focus to specifically emphasize resilience and resilience-building programs. Program evaluations can include an evaluation of the effectiveness of specific programs designed to foster and develop resilience in cadets. These evaluations can also determine the extent to which cadets experience the kinds of activities that cause them to engage in the resiliency- building process, shown earlier on page 26 of this dissertation, and the reintegrative state that the cadets achieve when emerging from this process. Individuals desiring to improve their ability to overcome or adapt to challenging circumstances may shift focus to developing resilience as it may prove more effective at dealing with individual trials than passion.

Recommendations

RQ5 – Recommendations for Future Research, Policy, and Practice

The final research question asks about recommendations for future research, policy development, and practices to build cadet resiliency at USAFA. This study has presented several discoveries related to resilience, passion, and overall grit and has used research-based evidence to discuss the implications of the results. However, it is important to move past the discussion of principles and ideas and to put them to work in the form of policy development and practice modifications that represent concrete courses of action to improve the development of resilient behaviors and tools. One of the

205

main goals of building resilience is to enable the individual to become better at employing their individual promotive and protective factors when facing the disruptive influences of adversity and to reintegrate or stabilize as a more resilient individual. The previous section mentioned three key concepts discovered through this study, and each concept is followed by a short discussion of how to put the concept into practice.

Additionally, the following discussion of policy recommendations and practice modifications can inform the immediate implementation of first steps to start the process of putting principle to practice and enabling leaders to begin to immediately effect their own organizations. The following two sections provide the answer to RQ5 and outline the recommended changes to USAFA policies and practices and the future research opportunities that may further inform cadet leadership development goals and processes.

Changes to Policy and Practice

First, with regard to USAFA, it is recommended that the current list of courses used to assess cadet grit at USAFA be augmented with additional programs identified to be significantly associated with resilience. Currently, the evaluation of Proficiency 5:

Exhibit Grit is assessed using four aquatics courses and three combatives courses to evaluate cadets’ ability to perform various skills and academic tasks. This list of courses could be expanded to cover a broader array of physical, leadership, and academic challenges by involving more demanding activities from the various program categories, such as boxing, rugby, the leader reactionary course (LRC), and the newly developed outdoor experiential learning programs. This study has shown that these activities are associated with grit, and their inclusion would provide greater insight into the development and evaluation of cadets’ abilities to exhibit resilience and grit.

206

Second, since the leadership growth model (LGM) is the framework currently practiced by USAFA to develop both leaders and followers, it is appropriate to match several recommendations with the specific LGM phase they relate to. The LGM identifies five phases in the growth process, and recommendations based on the results of this study will be made for each phase of the model.

1. Initial evaluation. Prior to and during the initial situation phase, leaders

should add a critical evaluation of their own knowledge of the role of

resiliency and their ability to effectively apply the principles of resiliency

theory throughout the entire process. This is a critical step in ensuring the

leader understands the concepts of resilience and how to apply them prior to

using them to mentor followers.

2. Expectations and inspiration. When setting proper expectations for their

followers, leaders should include a need to develop a knowledge of and an

ability to apply the principles of resiliency theory to various aspects of the

developmental process. Experiential learning is critical, and a plan to include

resiliency-building growth experiences into the process is likewise critical.

3. Instruction. When providing essential instructions to enable the subordinate

to meet expectations and objectives, leaders should review and teach about the

various aspects of resiliency theory, including the use of vignettes or examples

to demonstrate appropriate application. Armed with the expectations,

inspiration, and proper instruction, the follower can begin to take action and

apply what they have learned to the situations they experience at USAFA.

207

4. Feedback. When the time for feedback arrives, it is important to include both

a leader evaluation and a self-evaluation by the follower of how they have

demonstrated their ability to apply principles of resiliency theory to their

personal experiences. The evaluations should include experiences in

leadership, academics, athletics, and other non-academic activities at USAFA.

Discussion of resiliency in all these environments is important since resiliency

is only partly related to academics and very much includes leadership and

athletic experiences as well as real-world scenarios.

5. Reflection. Finally, during the reflection period designed to review

expectations, inspiration, instruction, and feedback, both leaders and

subordinates should include reflecting on their own understanding of

resiliency theory and the various ways it may be applied when setting new

goals and expectations. As the LGM implies, this process is continuously

repeated, providing for the continuous development of resilient behaviors and

resiliency-building tools in both leaders and followers.

Third, it is recommended that training and instruction be provided to USAFA cadre, faculty, and staff on the principles, framework, and processes of resiliency theory along with the depth and breadth of its effects and implications. Traditionally, USAFA, along with many other military service organizations, includes discussions of resiliency as it relates to equal opportunity, the prevention of sexual harassment, and having respect for others’ beliefs and ideals. However, while resilience is indeed an important part of these discussions, resiliency theory also includes a much larger and farther-reaching concept than these alone. A larger discussion of resiliency needs to take place discussing

208

the several elements of resiliency theory, the steps involved in the process of developing resilient behavior, and the diverse nature of activities that affect individual resiliency, and as a result, organizational resiliency.

Finally, and on a scale beyond just USAFA, it is recommended that all organizations, both military and non-military find ways to develop experiential programs where individuals can take the principles they have learned in courses, seminars, and conferences and apply them to real-life scenarios. These real-world situations provide the complexity and problem-solving opportunities that enable individuals to experience their own personal changes in resilient behavior instead of just talking about them. This study identified that successful performance in a variety of categories including academics, leadership development, and physically demanding activities is more closely associated with resilient behaviors than mere academics alone.

Future Research

In the discussion of both the findings from each research question and the implications of the research results, the need for future research in several areas was identified but only briefly discussed. A few of these research opportunities are presented below in greater detail and in no particular order. Of course, this list is by no means exhaustive, and researchers will undoubtedly develop and conduct future research on similar topics not mentioned here.

First, the depth of this study can be expanded by adding a qualitative research study using focus groups and interviews to further understand the relationship between the grit variables and a variety of cadet characteristics. Using similar research questions from this study to conduct a mixed-methods study would provide a more complete

209

research picture by taking the results of this study and adding the qualitative research data from current USAFA cadets to see if they converge or diverge from each other. Doing either a quantitative or qualitative study alone may only tell half the story, whereas doing both to identify converging or diverging themes may paint a more accurate picture.

Second, additional research focused on understanding the nature and influence of the newly identified variables found in this study may identify additional methods to build and strengthen resilient behaviors. Developing an understanding of these variables and their association with resilience may uncover additional factors leading to the development and strengthening of other non-cognitive behaviors that improve individual versatility and adaptation. In today’s military, the ability to quickly adapt and overcome challenging circumstances in rapidly changing environments is critical to mission success, and research supporting this effort may represent an effective use of resources.

Third, additional research could be conducted to understand how specific activities and/or sports within each of the USAFA program categories are associated with resilience, passion, and overall Grit-S scores. Since the types of activities included under mission support programs are so different from competitive programs or intramural sports, additional research may identify which activities are significantly associated with resilience and grit, as well as the reasons behind the associations. Understanding this would enable program developers to focus resources on the development or modification of these programs and to find ways to make them more accessible to cadets.

Fourth, additional research could study various groups of USAFA graduates to understand how resilience scores are related to career success, promotion rates, and obtaining command positions. This could also include examining the cadets who

210

disenrolled from USAFA to determine the longer-term effects of attrition from USAFA.

Providing further research on this subject could provide USAFA cadre and cadets with examples of outcomes for both military officers and civilians who disenrolled from

USAFA and identify the skills and characteristics that influenced their varying levels of success as leaders.

Finally, research could begin focusing on the effects of outdoor activity-based learning opportunities currently underway at USAFA. Routine and methodical program evaluation is critical to understanding the effectiveness and value of leadership development programs. USAFA directs a great deal of its resources to ensuring it produces the most versatile, adaptive, and capable Air Force officers prepared to lead forces and engage challenges on a variety of global fronts. Experiential leadership development programs can be effective contributors to individual and team success, and research to understand their overall impact may provide great benefit to USAFA.

Conclusion

In summary, the purpose of this study was to understand the relationship between the resilience of USAFA cadets and a variety of individual characteristics, some confirmatory and others exploratory, to understand the extent to which resilience was involved with performance. The results reviewed in this chapter speak to the significant associations between the grit variables, the academic and non-academic activities and performance, leadership performance, and athletic performance in support of the overall performance average (OPA) USAFA uses to evaluate the academic, leadership, and athletic performance of its cadets. This study followed the pattern of valid and sound arguments by posing a set of true premises in both the literature review and the data and

211

then showing how the premises led to logical conclusions related in part to prior research studies. In this manner, this study has added to the body of knowledge related to the effects non-cognitive factors like resilience, passion, and overall grit have on the actual performance and success of individuals. Studying this is essential to understanding how to build resiliency in individuals, particularly future military officers, as they will undoubtedly be faced with overwhelming challenges that they must overcome in order to ensure the safety and security of our great nation, as well as our allies and those around the world who simply need our help. As leaders and followers, we can and we must become more resilient, more adaptable, and better at overcoming challenges and rising again each time we fall as we help others do the same. In this, we achieve success.

Final Thoughts

The message of this study is that continuing to do our best, despite past failures, current setbacks, or anticipated obstacles, is closely related to achieving success. If this research has demonstrated anything, it has highlighted the significant connection between resilience and performance, though without making a claim as to which causes the other.

But it does not need to because, in reality, they affect each other in symbiotic ways, and thus to strengthen one is to strengthen the other. We can strengthen our resilience and improve our performance and achieve success in whatever way we choose because each of us has the ability to learn, grow, adapt, change, and improve in facing life’s adversities. We will always have our peaceful states of homeostasis unexpectedly disrupted. That is to be expected. However, how we utilize our past experiences and current resources to react to those changes will lead us to achieve new levels of resilience that may range from resilient reintegration in the best cases to getting back to where we

212

were before to, in the worst cases, dysfunctional reintegration. We have the freedom to choose which endstate we will work towards and dedicate our efforts to that end.

We sometimes hear about resilient people in the news who have overcome tremendous challenges to achieve some significant task. Online searches for resilient people list athletes who get back up after falling and finish the race, famous people who overcame their humble beginnings to achieve greatness, or others who succeed in business, television, the arts, or politics after overcoming multiple obstacles and failures.

Indisputably, these people exhibit resilience, and their publicized achievements leave no doubt. But many truly resilient people will never be recognized or remembered outside their own homes or social circles. I remember the faces of the soldiers I served with as they geared up repeatedly to head out on patrol despite past losses. I remember the face of a friend who shared with me her childhood memories of her mother who worked several jobs so her kids could become educated, go to dance school, and later attend college. I remember images of the dusty faces of firefighters in New York City who sifted through the rubble looking for survivors. I remember the high school players on the football team I helped coach who played with bruised bodies, broken fingers, and sprained ankles. I will always remember the story of my own grandfather, Loren A.

Stoddard, who was shot down during WWII and imprisoned in a Japanese POW camp for

15 months, spending time with Louie Zamperini and many others who never gave up, never quit, and never gave in despite the overwhelming opposition they faced. I will also remember the story of how my grandmother, while seeing renderings of my grandfather’s crash and receiving official notification from the U.S. War Department that her husband

213

was officially missing in action, continued to faithfully wait for him for 15 months despite never knowing if he was alive or not and how she rejoiced the day he came home.

I remember these and the stories of many others who just kept showing up, moment to moment and day after day, facing the challenges in front of them head on with a resilient determination to achieve. Success to them was not in winning a prize, becoming famous, or achieving some important honor or recognition. They never gave up and never quit because success meant being there for the children who needed them, the friends who were relying on them, the people in trouble who could not help themselves, and others who just needed a friend or a smile. They committed to a cause, refused to quit despite the adversities, and measured success in their ability to move forward. With these examples to guide us, anyone and everyone can learn to be more resilient.

214

References

Adversity. (2018). In Merriam-Webster’s online dictionary (11th ed.). Retrieved from

https://www.merriam-webster.com/dictionary/adversity

Ahern, N. R. (2006). Adolescent resilience: An evolutionary concept analysis. Journal of

Pediatric Nursing, 21(3), 175-185. doi:10.1016/j.pedn.2005.07.009

Allan, J. F., McKenna, J., & Dominey, S. (2014). Degrees of resilience: Profiling

psychological resilience and prospective academic achievement in university

inductees. British Journal of Guidance & Counselling, 42(1), 9-25.

Allender, D. B. (2006). Leading with a limp. Colorado Springs, CO: WaterBrook Press.

Alpern, G. (2016). Vets for vets: Harnessing the power of vets to heal. United States:

Psychological Development Publications.

Ancona, D., Bresman, H., & Caldwell, D. (2009). The X-Factor: Six steps to leading

high-performing X-Teams. Organizational Dynamics, 38(3), 217-224.

Avery, G. C. (2004). Understanding leadership: Paradigms and cases. Thousand Oaks,

CA: SAGE.

Avolio, B. J. (2007). Promoting more integrative strategies for leadership theory-

building. American Psychologist, 62(1), 25-33; discussion 43-27.

doi:10.1037/0003-066X.62.1.25

Bandura, A. (1997). Self-efficacy: The exercise of control. New York, NY: W. H.

Freeman.

Bandura, A. (2012). On the functional properties of perceived self-efficacy revisited.

Journal of Management, 38(1), 9-44. doi:10.1177/0149206311410606

215

Bartone, P. T. (1995). A short hardiness scale. Retrieved from Walter Reed Army

Institute of Research: http://www.dtic.mil/get-tr-doc/pdf?AD=ADA298548

Bartone, P. T. (2006). Resilience under military operational stress: Can leaders influence

hardiness? Military Psychology, 18(S), S131.

Bartone, P. T. (2007). Test-retest reliability of the dispositional resilience scale-15, a

brief hardiness scale. Psychological Reports, 101(3 Pt 1), 943-944.

doi:10.2466/pr0.101.3.943-944

Bartone, P. T., Barry, C. L., & Armstrong, R. E. (2009). To build resilience: Leader

influence on mental hardiness. Defense Horizons(69), 1.

Bartone, P. T., Eid, J., Johnsen, B. H., Laberg, J. C., & Snook, S. A. (2009). Big five

personality factors, hardiness, and social judgment as predictors of leader

performance. Leadership & Organization Development Journal, 30(6), 498-521.

Bartone, P. T., Hystad, S. W., Eid, J., & Brevik, J. I. (2012). Psychological hardiness and

coping style as risk/resilience factors for alcohol abuse. Military Medicine,

177(5), 517-524.

Bartone, P. T., Kelly, D. R., & Matthews, M. D. (2013). Psychological hardiness predicts

adaptability in military leaders: A prospective study. International Journal of

Selection and Assessment, 21(2), 200-210.

Bartone, P. T., Roland, R. R., Picano, J. J., & Williams, T. J. (2008). Psychological

hardiness predicts success in U.S. Army Special Forces candidates. International

Journal of Selection and Assessment, 16(1), 78-81.

216

Bazelais, P., Lemay, D. J., & Doleck, T. (2016). How does grit impact college students’

academic achievement in science? European Journal of Science and Mathematics

Education, 4(1), 33-43.

Beck, C. D. (2014). Antecedents of servant leadership: A mixed methods study. Journal

of Leadership & Organizational Studies, 21(3), 299-314.

doi:10.1177/1548051814529993

Bender, A., & Ingram, R. (2018). Connecting attachment style to resilience:

Contributions of self-care and self-efficacy. Personality and Individual

Differences, 130, 18-20.

Bishop, S. R., Lau, M., Shapiro, S., Carlson, L., Anderson, N. D., Carmody, J., . . .

Velting, D. (2004). Mindfulness: A proposed operational definition. Clinical

Psychology: Science and Practice, 11(3), 230-241.

Bonanno, G. A. (2004). Loss, trauma, and human resilience: Have we underestimated the

human capacity to thrive after extremely aversive events? American psychologist,

59(1), 20.

Borman, G. D., & Overman, L. T. (2004). Academic resilience in mathematics among

poor and minority students. The Elementary School Journal, 104(3), 177-195.

Bowman, N. A., Hill, P. L., Denson, N., & Bronkema, R. (2015). Keep on truckin’or stay

the course? Exploring grit dimensions as differential predictors of educational

achievement, satisfaction, and intentions. Social Psychological and Personality

Science, 6(6), 639-645.

Braverman, M. T. (2001). Applying resilience theory to the prevention of adolescent

substance abuse. Focus, 7(2), 1-12.

217

Brown, N. J. L. (2015). A critical examination of the U.S. Army’s comprehensive soldier

fitness program. The Winnower, 2. doi:10.15200/winn.143751.17496

Carlos, R., & Enfield, R. P. (1998). Resiliency and assets: Understanding the ecology of

youth development. FOCUS Monograph, Spring. University of California at

Davis.

Center for Character & Leadership Development. (2016). Retrieved from

http://www.usafa.edu/Commandant/cwc/?catname=cwc

Chamorro-Premuzic, T., & Lusk, D. (2017). The dark side of resilience. Retrieved from

https://hbr.org/2017/08/the-dark-side-of-resilience

Chang, W. (2014). Grit and academic performance: Is being grittier better? (Doctor of

Education Dissertation). University of Miami, Coral Gables, Florida. ProQuest

LLC database. (3681407)

Chemers, M., Hu, L., & Garcia, B. (2001). Academic self-efficacy and first year college

student performance and adjustment. Journal of Educational Psychology, 93(1),

55.

Chen, X. (2013). STEM attrition: College students’ paths into and out of STEM fields.

Statistical analysis report. NCES 2014-001. National Center for Education

Statistics.

Clawson, J. G. (2009). Level three leadership: Getting below the surface (4th ed.). Upper

Saddle River, NJ: Pearson Prentice Hall.

Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.).

Hillsdale, NJ: Erlbaum.

218

CollegeData. (2016). United States Air Force Academy admissions information.

Retrieved from

http://www.collegedata.com/cs/data/college/college_pg02_tmpl.jhtml?schoolId=7

CollegeFactual. (2018a, 2013-02-20). United States Air Force Academy graduation &

retention. Retrieved from https://www.collegefactual.com/colleges/united-states-

air-force-academy/academic-life/graduation-and-retention/

CollegeFactual. (2018b). United States Military Academy graduation & retention.

Collishaw, S., Hammerton, G., Mahedy, L., Sellers, R., Owen, M. J., Craddock, N., . . .

Thapar, A. (2016). Mental health resilience in the adolescent offspring of parents

with depression: A prospective longitudinal study. The Lancet Psychiatry, 3(1),

49-57.

Covey, S. R. (1992). Principle-centered leadership. New York, NY: Simon and Schuster.

Credé, M., Tynan, M. C., & Harms, P. D. (2017). Much ado about grit: A meta-analytic

synthesis of the grit literature. Journal of Personality and Social Psychology,

113(3), 492-511.

Creswell, J. W. (2003). Design: Qualitative, quantitative, and mixed methods approaches

(2nd ed.). Thousand Oaks, CA: SAGE.

Creswell, J. W., & Clark, V. L. P. (2011). Designing and conducting mixed methods

research (2nd ed.). Thousand Oaks, CA: SAGE.

219

Cross, T. M. (2014). Staying the course: Grit, academic success, and non-traditional

doctoral student. Paper presented at the SoTL Commons Conference, Georgia

Southern University, Statesboro, GA. Report retrieved from

http://digitalcommons.georgiasouthern.edu/sotlcommons/SoTL/2014/95

Deary, I. J., Watson, R., & Hogston, R. (2003). A longitudinal cohort study of burnout

and attrition in nursing students. Journal of advanced nursing, 43(1), 71-81.

Deichmann, D., & Stam, D. (2015). Leveraging transformational and transactional

leadership to cultivate the generation of organization-focused ideas. The

Leadership Quarterly, 26(2), 204-219. doi:10.1016/j.leaqua.2014.10.004

Donaldson, D. S. (2016). Wounded veterans: Reintegration through adventure-based

experience; A narrative inquiry. (Doctor of Education Doctoral Dissertation).

California State University, Long Beach, Long Beach, California. ProQuest

dissertations and theses full text database. (10155607)

Duckworth, A. L. (2016). Grit: The power of passion and perseverance. New York, NY:

Simon and Schuster.

Duckworth, A. L., Peterson, C., Matthews, M. D., & Kelly, D. R. (2007). Grit:

perseverance and passion for long-term goals. Journal of Personality and Social

Psychology, 92(6), 1087-1101. doi:10.1037/0022-3514.92.6.1087

Duckworth, A. L., & Quinn, P. D. (2009). Development and validation of the short grit

scale (grit-s). Journal of personality assessment, 91(2), 166-174.

doi:10.1080/00223890802634290

220

Duckworth, A. L., & Yeager, D. S. (2015). Measurement matters: Assessing personal

qualities other than cognitive ability for educational purposes. Educational

Researcher, 44(4), 237-251.

Duncan, R. D., & Pinegar, E. J. (2002). Leadership for saints. American Fork, UT:

Covenant Communications.

Durant, W. (1961). The Story of Philosophy: The Lives and Opinions of the Great

Philosophers of the Western World. New York, NY: Simon and Schuster.

Dweck, C. (2006). Mindset: The new psychology of success. New York, NY: Random

House.

Dyer, J. G., & McGuinness, T. M. (1996). Resilience: Analysis of the concept. Archives

of Psychiatric Nursing, 10(5), 276-282.

Eid, J., Johnsen, B. H., Bartone, P. T., & Nissestad, O. A. (2008). Growing

transformational leaders: Exploring the role of personality hardiness. Leadership

& Organization Development Journal, 29(1), 4-23.

Eskreis-Winkler, L., Duckworth, A. L., Shulman, E. P., & Beal, S. A. (2014). The grit

effect: Predicting retention in the military, the workplace, school and marriage.

Frontiers in Psychology, 5, 36. doi:10.3389/fpsyg.2014.00036

Fergus, S., & Zimmerman, M. A. (2005). Adolescent resilience: A framework for

understanding healthy development in the face of risk. Annual Review Public

Health, 26, 399-419. doi:10.1146/annurev.publhealth.26.021304.144357

Fleming, J., & Ledogar, R. J. (2008). Resilience, an evolving concept: A review of

literature relevant to aboriginal research. Pimatisiwin, 6(2), 7-23.

221

Fletcher, D., & Sarkar, M. (2012). A grounded theory of psychological resilience in

Olympic champions. Psychology of Sport and Exercise, 13(5), 669-678.

Fletcher, D., & Sarkar, M. (2013). Psychological resilience: A review and critique of

definitions, concepts and theory. European Psychologist, 18(1), 12-23.

doi:10.1027/1016-9040/a000124

Forbes. (2018). No. 5: United States Air Force Academy—Top STEM Colleges 2016.

Retrieved from

https://www.forbes.com/pictures/5774230031358e0aa22da837/no-5-united-states-

air-fo/

Gaddy, J. W., Gonzalez, S. P., Lathan, C. A., & Graham, P. K. (2017). The perception of

authentic leadership on subordinate resilience. Military Behavioral Health, 5(1),

64-72.

Garmezy, N. (1974). The study of competence in children at risk for severe

psychopathology. In E. J. Anthony & C. Koupernik (Eds.), The child in his

family: Children at psychiatric risk (Vol. 3, pp. 77-97). New York, NY: John

Wiley & Sons.

Garmezy, N. (1993). Children in poverty: Resilience despite risk. Psychiatry, 56(1), 127-

136.

Garmezy, N., & Masten, A. S. (1986). Stress, competence, and resilience: Common

frontiers for therapist and psychopathologist. Behavior Therapy, 17(5), 500-521.

Garmezy, N., Masten, A. S., & Tellegen, A. (1984). The study of stress and competence

in children: A building block for developmental psychopathology. Child

Development, 55(1), 97-111.

222

Garza, K. K., Bain, S. F., & Kupczynski, L. (2014). Resiliency, self-efficacy, and

persistence of college seniors in higher education. Research in Higher Education

Journal, 26, 1.

George, B., Sims, P., McLean, A. N., & Mayer, D. (2007). Discovering your authentic

leadership. Harvard Business Review, 85(2), 129-138.

doi:10.1371/journal.pone.0064247

Goleman, D. (2000). Leadership that gets results. Harvard Business Review, 78(2), 78-

90.

Greitens, E. (2015). Resilience: Hard-won wisdom for living a better life. New York, NY:

Houghton Mifflin Harcourt.

Grossman, P., Niemann, L., Schmidt, S., & Walach, H. (2004). Mindfulness-based stress

reduction and health benefits: A meta-analysis. Journal of Psychosomatic

Research, 57(1), 35-43.

Gutman, L. M., Sameroff, A. J., & Eccles, J. S. (2002). The academic achievement of

African American students during early adolescence: An examination of multiple

risk, promotive, and protective factors. American Journal of Community

Psychology, 30(3), 367-399. doi:10.1023/A:1015389103911

Hewitt, K. K., Davis, A. W., & Lashley, C. (2014). Transformational and transformative

leadership in a research-informed leadership preparation program. Journal of

Research on Leadership Education, 9(3), 225-253.

doi:10.1177/1942775114552329

223

Hochanadel, A., & Finamore, D. (2015). Fixed and growth mindset in education and how

grit helps students persist in the face of adversity. Journal of International

Education Research, 11(1), 47-50.

IBM Corp. (2017). IBM Statistical Package for the Social Sciences (SPSS) for Macintosh

(Version 25.0.0.1). Armonk, NY: IBM Corp.

Johnson, Thom, N. J., Stanley, E. A., Haase, L., Simmons, A. N., Shih, P.-a. B., . . .

Paulus, M. P. (2014). Modifying resilience mechanisms in at-risk individuals: A

controlled study of mindfulness training in Marines preparing for deployment.

American Journal of Psychiatry, 171(8), 844-853.

Johnson, C. E. (2017). Meeting the ethical challenges of leadership: Casting light or

shadow (6th ed.). Thousand Oaks, CA: Sage Publications.

Jowett, B. (1989). Plato: The republic and other works. New York, NY: Anchor Press.

Kabat‐Zinn, J. (2003). Mindfulness‐based interventions in context: Past, present, and

future. Clinical Psychology: Science and Practice, 10(2), 144-156.

Kashdan, T. B. (2017). How I learned about the perils of grit. Retrieved from

https://www.psychologytoday.com/blog/curious/201704/how-i-learned-about-the-

perils-grit

Kiswarday, V. (2012). Empowering resilience within the school context. Methodological

Horizons, 7(14), 93-103.

Ledesma, J. (2014). Conceptual frameworks and research models on resilience in

leadership. SAGE Open, 4(3), 1-8. doi:10.1177/2158244014545464

224

Lester, P. B., Harms, P. D., Herian, M. N., Krasikova, D. V., & Beal, S. J. (2011). The

comprehensive soldier fitness program evaluation report—3: Longitudinal

analysis of the impact of Master Resilience Training on self-reported resilience

and psychological health data. Retrieved from

http://digitalcommons.unl.edu/mwg-

internal/de5fs23hu73ds/progress?id=EOddq80yGfz0Kh5pamZfaaeOYarLtHLHu

1yed-w7Yf0,&dl

Losey, S. (2018a). Ending the pilot exodus: Air Force rolls out new bonuses,

incentives—will it work this time? Air Force Times.

Losey, S. (2018b). The military’s stunning fighter pilot shortage: One in four billets is

empty. Military Times.

Luthar, S. S. (2006). Resilience in development: A synthesis of research across five

decades. In Developmental psychopathology: Risk, disorder, and adaptation (2nd

ed., pp. 739-795). Hoboken, NJ: John Wiley & Sons Inc.

Luthar, S. S., & Barkin, S. H. (2012). Are affluent youth truly “at risk”? Vulnerability

and resilience across three diverse samples. Development and Psychopathology,

24(2), 429-449.

Luthar, S. S., & Cicchetti, D. (2000). The construct of resilience: Implications for

interventions and social policies. Development and Psychopathology, 12(04),

857-885.

Luthar, S. S., Cicchetti, D., & Becker, B. (2000). The construct of resilience: A critical

evaluation and guidelines for future work. Child Development, 71(3), 543-562.

225

Maddi, S. R., Matthews, M. D., Kelly, D. R., Villarreal, B., & White, M. (2012). The role

of hardiness and grit in predicting performance and retention of USMA cadets.

Military Psychology, 24(1), 19.

Masten, A. S., Herbers, J., Cutuli, J., & Lafavor, T. (2008). Promoting competence and

resilience in the school context. Professional School Counseling, 12(2), 76-84.

Mbindyo, M. N. (2011). Relationship between academic resilience and college success:

Cross-national experiences of low-income/first-generation students. (Unpublished

doctoral dissertation). The Pennsylvania State University, University Park, PA.

Retrieved from https://etda.libraries.psu.edu/files/final_submissions/4274

Meadows, S. O., Miller, L. L., & Robson, S. (2016). Airman and family resilience:

Lessons from the scientific literature. RAND Health Quarterly, 5(3), 10.

Meredith, L. S., Sherbourne, C. D., Gaillot, S. J., Hansell, L., Ritschard, H. V., Parker, A.

M., & Wrenn, G. (2011). Promoting psychological resilience in the U.S. military.

RAND Health Quarterly, 1(2).

Morales, E. E. (2014). Learning from success: How original research on academic

resilience informs what college faculty can do to increase the retention of low

socioeconomic status students. International Journal of Higher Education, 3(3),

92.

National Ability Center. (2019). Retrieved from https://discovernac.org/

Northouse, P. G. (2016). Leadership: Theory and practice (7th ed.). Thousand Oaks, CA:

SAGE.

Orthner, D. K., Jones‐Sanpei, H., & Williamson, S. (2004). The resilience and strengths

of low‐income families. Family Relations, 53(2), 159-167.

226

O’Brien, K., & Lomas, T. (2017). Developing a growth mindset through outdoor personal

development: Can an intervention underpinned by psychology increase the impact

of an outdoor learning course for young people? Journal of Adventure Education

and Outdoor Learning, 17(2), 133-147. doi:10.1080/14729679.2016.1232199

Pashak, T., Hagen, J., Allen, J., & Selley, R. (2014). Developmental assets: Validating a

model of successful adaptation for emerging adults. College Student Journal,

48(2), 243-248.

Pearce, C. L., & Conger, J. A. (2002). Shared leadership: Reframing the hows and whys

of leadership. Thousand Oaks, CA: SAGE.

Perez, W., Espinoza, R., Ramos, K., Coronado, H. M., & Cortes, R. (2009). Academic

resilience among undocumented Latino students. Hispanic Journal of Behavioral

Sciences, 31(2), 149-181.

Perkins-Gough, D. (2013). The significance of grit: A conversation with Angela Lee

Duckworth. Educational Leadership, 71(1), 14-20.

Project Healing Waters Fly Fishing. (2017). Retrieved from

https://projecthealingwaters.org/

Prudente, T. (2014, 08/24/2018). Superintendent: Fewer dropouts seen at Naval

Academy. Capital Gazette. Retrieved from

http://www.capitalgazette.com/news/naval_academy/ph-ac-cn-usna-attrition-

1202-20141201-story.html

Raisman, N. (2013). The cost of college attrition at four-year colleges & universities: An

analysis of 1669 U.S. institutions. Policy Perspectives.

227

Ramanaiah, N. V., Sharpe, J. P., & Byravan, A. (1999). Hardiness and major personality

factors. Psychological Reports, 84(2), 497-500. doi:10.2466/pr0.1999.84.2.497

Reivich, K. J., Seligman, M. E., & McBride, S. (2011). Master resilience training in the

U.S. Army. American Psychologist, 66(1), 25-34. doi:10.1037/a0021897

Reivich, K. J., & Shatté, A. (2003). The resilience factor: 7 keys to finding your inner

strength and overcoming life's hurdles. New York: Broadway Books.

Resilience. (2018). In Merriam-Webster’s online dictionary (11th ed.). Retrieved from

https://www.merriam-webster.com/dictionary/resilience

Richardson, G. E. (2002). The metatheory of resilience and resiliency. Journal of Clinical

Psychology, 58(3), 307-321.

Rivezzo, C. (2019). USAFA educator develops programs based on mountaineering.

Retrieved from https://www.usafa.af.mil/News/Features/Article/1424802/usafa-

educator-develops-programs-based-on-mountaineering-specops-background/

Rutter, M. (1985). Resilience in the face of adversity: Protective factors and resistance to

psychiatric disorder. The British Journal of Psychiatry, 147(6), 598-611.

Rutter, M. (1987). Psychosocial resilience and protective mechanisms. American Journal

of Orthopsychiatry, 57(3), 316-331. doi:10.1111/j.1939-0025.1987.tb03541.x

Schreiner, L. A. (2017). The privilege of grit. About Campus, 22(5), 11-20.

Shi, Q., Pavey, E. S., & Carter, R. E. (2012). Bonferroni‐based correction factor for

multiple, correlated endpoints. Pharmaceutical statistics, 11(4), 300-309.

Spatz, C. (2008). Tales of distributions: Basic statistics. Belmont, CA: Thomson

Wadsworth.

228

Staffaroni, L. (2018). Historical SAT percentiles: New SAT 2016 and 2017. Retrieved

from https://blog.prepscholar.com/historical-percentiles-new-sat

Strauss, A. L., & Corbin, J. (1998). Basics of qualitative research: Techniques and

procedures for developing grounded theory (4th ed.). Thousand Oaks, CA:

SAGE.

Tate, G. L. (2016). Effects of character development on fourth-class cadets at the United

States Air Force Academy. (Doctor of Philosophy). University of Colorado

Colorado Springs, Colorado Springs, CO.

The United States Air Force Academy. (2014a). Cadet military education and training

plan. Colorado Springs, CO: Headquarters United States Air Force Academy

The United States Air Force Academy. (2014b). The USAFA institutional outcomes.

Colorado Springs, CO: Headquarters United States Air Force Academy. Retrieved

from http://www.usafa.edu/df/usafaoutcomes.cfm?catname=Dean_of_Faculty

The United States Air Force Academy. (2016a). Center for character and leadership

development. Retrieved from http://usafa.smartcatalogiq.com/en/2015-

2016/Catalog/Center-for-Character-and-Leadership-Development

The United States Air Force Academy. (2016b). Center for character and leadership

development—honor directorate. Retrieved from

http://www.usafa.edu/Commandant/cwc/cwch.cfm?catName=cwc

The United States Air Force Academy. (2016c). Warrior ethos as airmen and citizens.

Colorado Springs, CO: Headquarters United States Air Force Academy.

The United States Air Force Academy. (2017a). Essence of USAFA. Retrieved from

http://www.usafa.af.mil/Leadership/The-Superintendent/Essence-of-USAFA/

229

The United States Air Force Academy. (2017b). USAFA Curriculum Handbook 2017-

2018. Retrieved from https://www.usafa.edu/app/uploads/CHB.pdf

The United States Air Force Academy. (2017c). USAFA pamphlet 36-3527—The officer

development system: Developing officers of character. Colorado Springs, CO:

Headquarters United States Air Force Academy

The United States Air Force Academy. (2018a). Majors & minors—United States Air

Force Academy. Retrieved from https://www.usafa.edu/academics/majors-

minors/

The United States Air Force Academy. (2018b). National character & leadership

symposium. Retrieved from https://www.usafa.edu/character/national-character-

leadership-symposium-ncls/

Thomas, K. H., & Taylor, S. P. (2015). Bulletproofing the psyche: Mindfulness

interventions in the training environment to improve resilience in the military and

veteran communities. Advances in Social Work, 16(2), 312-322.

Timmons, R. F. (2013). The United States Army comprehensive soldier fitness: A critical

look. Retrieved from Army War College:

www.dtic.mil/docs/citations/ADA590241

U.S. Department of the Air Force. (2014). AFI 90-506: Comprehensive airman fitness.

Washington D.C.: U.S. Department of the Air Force

U.S. Department of the Army. (2012). ARDP 6-22: Army Leadership. Washington D.C.:

U.S. Department of the Army. Retrieved from

http://cape.army.mil/repository/doctrine/adrp6-22.pdf

230

U.S. Department of the Army. (2014). AR 350-53: Comprehensive soldier and family

fitness. Washington D.C.: U.S. Department of the Army. Retrieved from

https://www.army.mil/e2/downloads/rv7/r2/policydocs/r350_53.pdf

Udell, C. J., Ruddy, J. L., & Procento, P. M. (2018). Effectiveness of acceptance and

commitment therapy in increasing resilience and reducing attrition of injured US

Navy recruits. Military medicine, 183(9-10), 603-611.

USAFA. (2016). Academy Military Training Non-Commissioned Officers (AFSC

8B200). Retrieved from http://www.usafa.edu/Commandant/cws/

VanBreda, A. D. (2001). Resilience theory: A literature review: With special chapters on

deployment resilience in military families & resilience theory in social work

(MPI/R/104/12/1/4, dd October 2001). Retrieved from Pretoria, South Africa:

http://vanbreda.org/adrian/resilience/resilience_theory_review.pdf

Vie, L. L., Scheier, L. M., Lester, P. B., & Seligman, M. E. (2016). Initial validation of

the U.S. Army Global Assessment Tool. Military Psychology, 28(6), 468-487.

Walumbwa, F. O., Avolio, B. J., Gardner, W. L., Wernsing, T. S., & Peterson, S. J.

(2007). Authentic leadership: Development and validation of a theory-based

measure. Journal of Management, 34(1), 89-126.

doi:10.1177/0149206307308913

Wenger, J. W., O’Connell, C., & Cottrell, L. (2018). Examination of recent deployment

experience across the services and components. Santa Monica, CA: RAND

Corporation.

231

Williamson, G. R., Health, V., & Proctor-Childs, T. (2013). Vocation, friendship and

resilience: a study exploring nursing student and staff views on retention and

attrition. The Open Nursing Journal, 7, 149.

Willink, J., & Babin, L. (2015). Extreme ownership: How U.S. Navy Seals lead and win.

New York, NY: St. Martin's Press.

Wilson, L. (2018, 2018-02-06). USAFA's Equestrian Center helps wounded warrior

‘thrive in life’. Retrieved from https://www.usafa.edu/news/usafas-equestrian-

center-helps-wounded-warrior-thrive-life/

Windle, G., Bennett, K. M., & Noyes, J. (2011). A methodological review of resilience

measurement scales. Health and Quality of Life Outcomes, 9(1), 8.

doi:10.1186/1477-7525-9-8

Woelk, C., & Weeks, P. P. (2010). The student success leader program: College-level

service enhances learning outside the classroom. NACTA Journal, 54(2), 18-20.

Wolters, C. A., & Hussain, M. (2015). Investigating grit and its relations with college

students’ self-regulated learning and academic achievement. Metacognition and

Learning, 10(3), 293-311.

Yeager, D. S., & Dweck, C. (2012). Mindsets that promote resilience: When students

believe that personal characteristics can be developed. Educational Psychologist,

47(4), 302-314. doi:10.1080/00461520.2012.722805

Zaccaro, S. J. (2007). Trait-based perspectives of leadership. American Psychologist,

62(1), 6-16. doi:10.1037/0003-066X.62.1.6

Zajacova, A., Lynch, S., & Espenshade, T. (2005). Self-efficacy, stress, and academic

success in college. Research in Higher Education, 46(6), 677-706.

232

Zhang, F. (2016). ACT score percentiles and ranks (High-precision 2016). Retrieved

from https://blog.prepscholar.com/act-percentiles-high-precision-2016

Zimmerman, M. A. (2013). Resiliency theory: A strengths-based approach to research

and practice for adolescent health. Health Education & Behavior, 40(4), 381-383.

doi:10.1177/1090198113493782

Zimmerman, M. A., Stoddard, S. A., Eisman, A. B., Caldwell, C. H., Aiyer, S. M., &

Miller, A. (2013). Adolescent resilience: Promotive factors that inform

prevention. Child Development Perspectives, 7(4), 215-220.

Zubeck, P. (2018). Air Force Academy loses men, women cadets at different rates.

Retrieved from https://www.csindy.com/coloradosprings/air-force-academy-

loses-men-women-cadets-at-different-rates/Content?oid=4247542

233

APPENDIX A

USAFA IRB Approval

DEPARTMENT OF THE AIR FORCE

HEADQUARTERS UNITED STATES AIR FORCE ACADEMY

31 October 2018

MEMORANDUM FOR MR. JUSTIN STODDARD

FROM: HQ USAFA/A9O (USAFA IRB)

SUBJECT: IRB approval for the use of human subjects in research

1. Protocol title: USAFA Cadet Resiliency Study

2. Protocol number: FAC20190002H

3. Risk: Minimal

4. IRB Approval date: 26 October 2018

5. Expiration date: 25 October 2019

6. Continuing Review Report due: 01 October 2018

7. Type of review: Expedited per HHS Category (7) Research on individual or group characteristics or behavior (including, but not limited to, research or perception, cognition, motivation, identity, language, communication, cultural beliefs or practices, and social behavior) or research employing survey, interview, oral history, focus group, program evaluation, human factors evaluation, or quality assurance methodologies.

8. Approved number of subjects: ~7,091

9. Assurance Number and Expiration Date: DoD Assurance 50046, expiration 31 December 2019

10. Training Expiration Dates: Stoddard, 3 May 2019

11. The above protocol has been reviewed and approved by the USAFA IRB using expedited review category 7. Waiver of informed consent was also reviewed and approved. All requirements, as set by the IRB and its legal counsel, have been fully complied with. Please note that the USAFA Authorized Institutional Official, HQ USAFA/CV and the Surgeon General's Research Oversight & Compliance Division, AFMSA/SGE-C review all USAFA IRB actions and may amend this decision or identify additional requirements. The study is minimal risk.

12. Any adverse reactions or issues resulting from this study should be reported immediately to the IRB Chair or Administrator. Instructions and forms are at: http://www.usafa.af.mil/Leadership/InstitutionalReviewBoard/ProtocolViolationsandAdverseEvents.aspx

13. Amendments to the protocol and/or revisions to informed consent documents must have IRB approval before they are implemented. Please retain both a hard copy and electronic copy of the final approved protocol and informed consent document. Instructions and forms are at:

Developing Leaders of Character

234

http://www.usafa.af.mil/Leadership/InstitutionalReviewBoard/Amendments.aspx

14. All inquiries and correspondence concerning this protocol should include the protocol number and name of the primary investigator. Please ensure the timely submission of all required progress and final reports. Please note that any reminders reference upcoming expiration dates are a courtesy and it is the investigators' responsibility to keep track of their expiration dates and submit their documents to the IRB on time.

15. Per DoDI3216.02_AFI40-402, Enclosure 2, 11.f., you must retain all research records (e.g., protocol, signed informed consent documents, IRB correspondence, and data) for at least three years after the research ends or for the length of time specified in applicable regulations, or institutional or sponsor requirements, whichever is longer. You must transfer research records to another PI or keep them with you and provide new contact information if you leave USAFA before the 3 years is over. In either case, you must inform the HRPP office that you are leaving USAFA.

16. If you have any questions or if I can be of further assistance, please don't hesitate to contact me at 333-6593.

LAURA J. NEAL HQ USAFA IRB Administrator

235

APPENDIX B

USAFA VOLLUNTEER SERVICE AGREEMENT (VSA)

236

APPENDIX C

GRIT-S Survey (Also called the “Short Grit Scale”)

Directions for taking the Grit Scale: Please respond to the following 8 items. Be honest – there are no right or wrong answers!

1. New ideas and projects sometimes distract me from previous ones. * q Very much like me q Mostly like me q Somewhat like me q Not much like me q Not like me at all

2. Setbacks don’t discourage me. q Very much like me q Mostly like me q Somewhat like me q Not much like me q Not like me at all

3. I have been obsessed with a certain idea or project for a short time but later lost interest. * q Very much like me q Mostly like me q Somewhat like me q Not much like me q Not like me at all

4. I am a hard worker. q Very much like me q Mostly like me q Somewhat like me q Not much like me q Not like me at all

5. I often set a goal but later choose to pursue a different one. * q Very much like me q Mostly like me q Somewhat like me q Not much like me q Not like me at all

237

6. I have difficulty maintaining my focus on projects that take more than a few months to complete. * q Very much like me q Mostly like me q Somewhat like me q Not much like me q Not like me at all

7. I finish whatever I begin. q Very much like me q Mostly like me q Somewhat like me q Not much like me q Not like me at all

8. I am diligent. q Very much like me q Mostly like me q Somewhat like me q Not much like me q Not like me at all ______

Scoring: 1. For questions 2, 4, 7 and 8 assign the following points: 5 = Very much like me 4 = Mostly like me 3 = Somewhat like me 2 = Not much like me 1 = Not like me at all

2. For questions 1, 3, 5 and 6 assign the following points: 1 = Very much like me 2 = Mostly like me 3 = Somewhat like me 4 = Not much like me 5 = Not like me at all

Add up all the points and divide by 8. The maximum score on this scale is 5 (extremely gritty), and the lowest score on this scale is 1 (not all that gritty). the lowest score on this scale is 1 (not at all gritty).

238

Grit Scale citation

Duckworth, A.L, & Quinn, P.D. (2009). Development and validation of the Short Grit Scale (Grit- S). Journal of Personality Assessment, 91, 166-174. http://www.sas.upenn.edu/~duckwort/images/Duckworth%20and%20Quinn.pdf

Duckworth, A.L., Peterson, C., Matthews, M.D., & Kelly, D.R. (2007). Grit: Perseverance and passion for long-term goals. Journal of Personality and Social Psychology, 9, 1087-1101. http://www.sas.upenn.edu/~duckwort/images/Grit%20JPSP.pdf

239

APPENDIX D

USAFA Intercollegiate Sports

Varsity Junior Varsity Baseball* Baseball* Basketball, Men's* Basketball, Men's* Basketball, Women's* Basketball, Women's* Boxing* Cheer Leader, Men's Cheer Leader, Men's* Cheer Leader, Women's Cheer Leader, Women's* Cross Country, Men's Cross Country, Men's* Cross Country, Women's Cross Country, Women's* Fencing, Men's* Diving, Men's* Fencing, Women's* Diving, Women's* Football* Fencing, Men's* Golf Fencing, Women's* Ice Hockey Football* Lacrosse Freestyle Swimming, Men's Rifle, Men's Freestyle Swimming, Women's Rifle, Women's Golf* Soccer, Men's* Gymnastics, Men's* Soccer, Women's Gymnastics, Women's* Swimming, Men's Ice Hockey* Swimming, Women's Lacrosse* Tennis, Men's Rifle, Men's* Tennis, Women's* Rifle, Women's* Track, Men's Soccer, Men's* Track, Women's Soccer, Women's* Volleyball Stroke Swimming, Men's Wrestling Stroke Swimming, Women's Swimming, Men's* Swimming, Women's* Tennis, Men's* Tennis, Women's* Track, Men's* Track, Women's* Volleyball* Water Polo* Wrestling* Note. * indicates that cadets included in this study participated in these sports.

240

APPENDIX E

USAFA Intramural Sports

Aerobics Basketball* Bowling Boxing, Coed Boxing, Men's* Boxing, Women’s* Cross Country* Flag Football, Coed* Flag Football, Women's* Flicker Ball* Foreign Exchange Futsal* Mountain Cycling* Racquetball* Reconditioning Rugby, Women's* Rugby, Men's* Soccer* Softball* Team Handball* Tennis* Ultimate Frisbee* Volleyball* Wallyball* Water Polo Weight Training CIC Wing Duty

Note. * indicates that cadets included in this study participated in these sports.

241

APPENDIX F

USAFA Competitive Programs

Aerobics Instructors Nordic Skiing* Aikido Club for Year Parachute / Skydiving Alpine Skiing* Performing Arts Arnold Air Society (AAS) Pistol Bluebards Reconditioning CICs Boxing, Women's* Rodeo* Cycling* Rugby, Coed Drum & Bugle Rugby, Men's* Fastpitch Softball, Coed Rugby, Women's* Fastpitch Softball, Women's* Science Fiction Forensics Ski & Snowboard Gospel Choir Team Handball, Coed Handball, Coed Team Handball, Men's* Handball, Men's Triathlon Club* Judo* Ultimate Frisbee* Lacrosse, Coed* Volleyball, Coed Lacrosse, Women's* Volleyball, Men's* Marathon* Water Polo, Women's* Mock Trial