Development and validation of the sport character scale

A variety of qualitative and quantitative measures exist to measure character in sport. These measures, however, are characterized by numerous limitations. In order to address these shortcomings the purpose of this study was to: (a) develop measure of sport character based on the model of moral action and (b) validate the measure through examination of the relationships between character in sport and other psychological variables. Two preliminary phases and four studies were undertaken to address the study purposes. Item development and refinement occurred in the preliminary phases and was grounded in the model of moral action. Items targeting compassion, integrity, sportspersonship, and fairness were developed for the Sport Character Scale (SCS). In total 62 items were created. In study 1, the pilot study, the items were further clarified and refined by a representative sample of 50 Korean athletes. This resulted in two items being deleted. In study 2, Exploratory Factor Analysis (EFA) was conducted on responses from 332 Korean athletes (mean age = 19.86 years) participating in a variety (n = 31) of sports. Results of the EFA suggested that a five-factor, 27-item model best fit the data. The factors included Compassion, Fairness, Sportspersonship, Antisocial Attitude, and Integrity. In Study 3, a Confirmatory Factor Analysis was conducted on a unique sample of 322 Korean athletes. The validity of the five-factor solution was confirmed. Fit indices suggested a good fit. Lastly, in Study 4, the concurrent and iv construct validity of the Sport Character Scale were examined. The SCS was related to the Multidimensional Sportspersonship Orientation Scale and two subscales of the Attitudes to Moral Decision-making in Youth Questionnaire providing support for concurrent validity. Construct validity was supported by significant relationships that emerged between the SCS and the Caring Climate Scale and the Perceived Motivational Climate in Sport Questionnaire. The SCS appears to measure unique characteristics of sport character in Korean athletes. It is possible to educate coaches, parents, and athletes about the importance of compassion, sportspersonship, fairness, and integrity. Moreover, the findings of this study can be used to inform coaching and parent education programs to optimize the sport participation of youth.

DEVELOPMENT AND VALIDATION OF THE SPORT CHARACTER SCALE by Chang-Yong Jang A dissertation submitted to the faculty of The University of Utah in partial fulfillment of the requirements for the degree of Doctor of Philosophy Department of Exercise and Sport Science The University of Utah May 2013 Copyright © Chang-Yong Jang 2013 All Rights Reserved The University of Utah Graduate School STATEMENT OF DISSERTATION APPROVAL The dissertation of Chang-Yong Jang has been approved by the following supervisory committee members: Maria Newton , Chair 02/25/2013 Date Approved Barry B. Shultz , Member 02/25/2013 Date Approved Justine J. Reel , Member 02/25/2013 Date Approved Les Podlog , Member 02/25/2013 Date Approved Jonathan Butner , Member 02/25/2013 Date Approved and by Barry B. Shultz , Chair of the Department of Exercise and Sport Science and by Donna M. White, Interim Dean of The Graduate School. ABSTRACT A variety of qualitative and quantitative measures exist to measure character in sport. These measures, however, are characterized by numerous limitations. In order to address these shortcomings the purpose of this study was to: (a) develop measure of sport character based on the model of moral action and (b) validate the measure through examination of the relationships between character in sport and other psychological variables. Two preliminary phases and four studies were undertaken to address the study purposes. Item development and refinement occurred in the preliminary phases and was grounded in the model of moral action. Items targeting compassion, integrity, sportspersonship, and fairness were developed for the Sport Character Scale (SCS). In total 62 items were created. In study 1, the pilot study, the items were further clarified and refined by a representative sample of 50 Korean athletes. This resulted in two items being deleted. In study 2, Exploratory Factor Analysis (EFA) was conducted on responses from 332 Korean athletes (mean age = 19.86 years) participating in a variety (n = 31) of sports. Results of the EFA suggested that a five-factor, 27-item model best fit the data. The factors included Compassion, Fairness, Sportspersonship, Antisocial Attitude, and Integrity. In Study 3, a Confirmatory Factor Analysis was conducted on a unique sample of 322 Korean athletes. The validity of the five-factor solution was confirmed. Fit indices suggested a good fit. Lastly, in Study 4, the concurrent and iv construct validity of the Sport Character Scale were examined. The SCS was related to the Multidimensional Sportspersonship Orientation Scale and two subscales of the Attitudes to Moral Decision-making in Youth Questionnaire providing support for concurrent validity. Construct validity was supported by significant relationships that emerged between the SCS and the Caring Climate Scale and the Perceived Motivational Climate in Sport Questionnaire. The SCS appears to measure unique characteristics of sport character in Korean athletes. It is possible to educate coaches, parents, and athletes about the importance of compassion, sportspersonship, fairness, and integrity. Moreover, the findings of this study can be used to inform coaching and parent education programs to optimize the sport participation of youth. TABLE OF CONTENTS ABSTRACT ....................................................................................................................... iii LIST OF TABLES ........................................................................................................... viii LIST OF FIGURES ............................................................................................................ x ACKNOWLEDGMENTS ................................................................................................. xi Chapters I. INTRODUCTION .......................................................................................................... 1 Defining Character in Sport .......................................................................................... 2 Theoretical Perspectives of Character Development in Sport ...................................... 3 Measures of Character and Moral Development Research in Sport ............................. 6 Summary of Character Construct in Sport .................................................................. 11 Problem Statement ...................................................................................................... 12 Purpose of Study ......................................................................................................... 13 Research Questions ..................................................................................................... 13 Hypotheses .................................................................................................................. 13 Significance of Study .................................................................................................. 14 Limitations .................................................................................................................. 15 Delimitations ............................................................................................................... 15 Assumptions ................................................................................................................ 16 Definition of Terms..................................................................................................... 16 II. REVIEW OF LITERATURE...................................................................................... 19 The Definition of Character ........................................................................................ 19 Character as a Moral Perspective ................................................................................ 20vi The Definition of Sport Character .............................................................................. 21 Theoretical Framework ............................................................................................... 22 Theory Related Research ............................................................................................ 31 Character and Moral Development Research in Sport Settings .................................. 33 Measures in Character and Moral Development Research ......................................... 38 III. METHODS ................................................................................................................ 53 Phase I: Specifying the Domains of the Construct ..................................................... 53 Phase II: Item Generation and Development .............................................................. 54 Study 1: Pilot Study .................................................................................................... 63 Study 2, 3, and 4: Validation of the Sport Character Scale ........................................ 65 IV. RESULTS .................................................................................................................. 79 Study 1: Pilot Study .................................................................................................... 79 Study 2: Exploratory Factor Analysis ......................................................................... 81 Comparison of Two Models ....................................................................................... 89 Study 3: Confirmatory Factor Analysis ...................................................................... 91 Study 4: Validity of the Sport Character Scale (SCS) ................................................ 94 V. DISCUSSION ........................................................................................................... 116 Preliminary Phases and Study 1: Pilot Study ............................................................ 116 Study 2: Exploratory Factor Analysis ....................................................................... 119 Study 3: Confirmatory Factor Analysis Phase .......................................................... 121 Study 4: Validity Examination Phase ....................................................................... 123 Limitations ................................................................................................................ 127 Future Directions ...................................................................................................... 129 Applied Implications ................................................................................................. 131 Appendices A. SPORT CHARACTER SCALE (76 ITEMS) TO REVIEW (ENGLISH VERSION) ............................................................................................ 134 vii B. SPORT CHARACTER SCALE (76 ITEMS) TO REVIEW (KOREAN VERSION) ............................................................................................. 141 C. QUESTIONNAIRES (ENGLISH VERSION) ......................................................... 149 D. QUESTIONNAIRES (KOREAN VERSION).......................................................... 158 REFERENCES ............................................................................................................... 167 LIST OF TABLES 1. 12-Component Model of Moral Action ....................................................................... 52 2. A Sampling of Questions and Answers from the Focus Groups ................................. 76 3. Demographic Information of Participants ................................................................... 77 4. Frequency of Participants Based on Sport Type .......................................................... 78 5. Demographic Information of Participants ................................................................... 98 6. Frequency of Participants Based on Sport Type .......................................................... 99 7. Descriptive Statistics for the SCS .............................................................................. 101 8. Actual and Random Eigenvalues ............................................................................... 103 9. Total Variance Explained by the Four Extracted Factors of the SCS........................ 104 10. Rotated Factor Pattern Matrix for the 27-item SCS................................................. 105 11. Factor Correlations and Factor Alpha Coefficients for the SCS (N = 332) ............. 106 12. Total Variance Explained by the Five Extracted Factors of the SCS ...................... 107 13. Rotated Factor Pattern Matrix for the 27-item SCS................................................. 108 14. Factor Correlations and Factor Alpha Coefficients for the SCS (N = 332) ............. 110 15. Demographic Information of Participants ............................................................... 110 16. Frequency of Participants Based on Sport Type ...................................................... 111 17. Goodness-of-Fit Indices for the Five-factor CFA Model with the 27 Items ........... 112 18. Standardized Solution and Correlations of the Five-factor CFA Model ................. 114ix 19. Assessment of Concurrent Validity for SCS ........................................................... 115 20. Assessment of Construct Validity for SCS .............................................................. 115 21. Types of Sport .......................................................................................................... 133 LIST OF FIGURES 1. Hypothesized Four-factor CFA Model of the SCS ...................................................... 18 2. Boxplot for Mahalanobis Distance ............................................................................ 100 3. Scree Plot for the PCA ............................................................................................... 103 4. Scree Plot for the PAF ............................................................................................... 106 5. Five-factor CFA Model for the Sport Character Scale (SCS) ................................... 113 6. A Hypothesized Model of Types of Sport Character ................................................. 133 ACKNOWLEDGMENTS There are many people without whom this long journey would not have been possible. First, it is with the deepest gratitude that I thank Dr. Maria Newton for her invaluable assistance in the completion of this project. As an advisor, mentor, and good colleague, she has taught me a great deal about the scientific endeavor and of a caring life. It is with her encouragement that I intend to put some "emotion" back into research and to keep "motivation and delight" in the center of my work. Special appreciation is also extended to Drs. Justine Reel and Jonathan Butner who provided great insight and encouragement for this frustrated researcher. In addition, I would like to thank the other members of my doctoral committee, Drs. Barry Shultz and Les Podlog, for their valuable assistance and constructive feedback. I also wish to thank my research "crew" at the University of Utah for offering advice and encouragement. Thanks to Seunghwan for helping me to do the fun job of entering data for my study, to Joseph and John for being my dissertation "co-writers" while dealing with my anxiety attacks and frustration. I would also like to thank Dr. Junsu Kim for offering his expertise and support during the formative stages of this project and Dr. Jun Kim for keeping me going with spiritual comfort. I could never have achieved this without the love and support of my family. I would like to thank Jihye, my wife, for her understanding and support. Thanks to all of my older brothers, Wonsook, Yoonsook and Chungsook, and their family for believing in, xii encouraging, and supporting me. Finally, I wish to dedicate this to my parents, Ki-Tae Jang, who has since passed away, and Goonja Jung, both of whom have supported me affectionately, emotionally, and financially throughout my life. Both of whom have encouraged me to be a good educator and researcher. Their compassion and sacrifice was the most important factor that sustained me. Father, I love and deeply admire you. I also believe that you are always with me.CHAPTER I INTRODUCTION The concept that sport builds character is a popular belief in our society (Kavussanu & Roberts, 2001; Sage, 1990; Sage & Kavussanu, 2007). Sport is regarded as a vehicle for learning to cooperate with teammates, negotiate and provide solutions to moral conflicts, and learn virtues such as fairness, sportspersonship, team loyalty, teamwork, responsibility, and subordination for the greater good (Kleiber & Roberts, 1981). According to Arnold (1984), sport embodies freedom and equality because individuals choose freely to participate and the rules are designed to be applied to all the players. Players are aware that sport is rule-governed activity, and they commit themselves to follow the rules. In this sense, participants can build acceptance of responsibilities and self-respect through sport. Participants benefit from sport by learning to overcome obstacles, develop self-control, cooperate with teammates, and behave with virtues such as respect, honesty, fair play, and compassion (Smith & Bar-Eli, 2007). Due to these aspects, sport can theoretically be a setting in which people practice moral virtue and form and display positive character. Although some researchers maintain that sport builds character, others counter the claim and suggest that sport builds ‘characters'. It is easy to find drug usage to enhance performance, antisocial behaviors, and cheating in the 2 sport situation (Bredemeier, 1994; Shields & Bredemeier, 1995). In addition, some make the critique that competition in sport reduces prosocial behavior (Kleiber & Roberts, 1981), increases antisocial behavior (Kohn, 1986), and an overemphasis on winning in competition generates moral problems (Orlick, 1990). In sum, divergent views exist relative to sport and character development. On one hand, sport is viewed as a powerful vehicle for character development. On the other hand, sport is perceived as an impediment to the display of virtuous behavior in sport. Such discrepancy justifies a more detailed examination of the concept of character. Defining Character in Sport It is widely recognized that character is a complex, multifaceted concept. The term ‘character' in sport is often used synonymously with morality in sport, sportspersonship, or fair play (Weiss, Smith, & Stuntz, 2008). According to Weiss et al. (2008), the word ‘moral development' is rarely used in daily language because moral development in sport is more an academic term and instead people tend to use more common terms, such as sportspersonship, fair play, or character. According to Rudd and Stoll (2004), there exists a discrepancy in how character is conceptualized. This notion is based on a view that the concept of character is classified by a social or moral perspective (Beller, 2002; Beller & Stoll, 1995). Athletes, coaches, sport administrators, and parents define the concept of character from a social perspective when they refer to values such as loyalty, dedication, sacrifice, teamwork, and good citizenship. Concurrently, sport scientists and sport philosophers continue to define 3 character from a moral perspective, which is regarded as the ability to consider a person's actions in relationship to moral values such as honesty, fairness, justice, and fair play (Camire & Trudel, 2010). A number of studies in sport have conceptualized character from a moral perspective. For example, Arnold (1999) stated that moral character involves a life that complies with such virtues as justice, honesty, and compassion. Shields and Bredemeier (1995) described character in sport as the possession of personal virtues such as compassion, fairness, sportspersonship, and integrity. Recently, Doty (2005) defined character as behaviors that show respect and integrity in a sport setting. In sum, sport character is comprised of a number of distinguishable components. Because every psychological quality, capacity, or process can influence a player's moral engagements and behavior, a complete list of sport character components is probably impossible. However, it is relatively easy to identify a number of clear and important components of character in sport contexts. In order to clarify the concept of character in sport, some main theoretical frameworks are considered. Theoretical Perspectives of Character Development in Sport Various theories have been proposed to explain moral development and behavior. The most prominent theories are social learning theory and structural developmental theory (Kavussanu, 2007; Weiss et al., 2008). According to social learning theory (Bandura, 1977), moral behavior is regarded as prosocial behaviors consistent with societal norms and conventions, such as responsibility, honesty, cooperation, and respect. 4 The social learning approach emphasizes behavioral outcomes. In contrast, according to structural developmental approaches, morality is defined as a concern for physical and psychological welfare of self and others (Kavussanu, 2007; Weiss et al., 2008). Structural developmental theory focuses on individuals' underlying reasoning or judgment processes associated with behaviors (Weiss et al., 2008). One useful framework for investigating character issues in sport and the basis of this study is Shields and Bredemeier's (1995) 12-component model of moral action. Shields and Bredemeier (1995) have contributed greatly to the understanding of moral and character development in sport and physical activity. Their 12-component model of moral action is a structural developmental approach that is theoretically based on the work of Piaget (1923), Kohlberg (1981, 1984), and Haan (Haan, Aerts, & Cooper ,1985), and draws heavily from Rest's (1984, 1986, & 1994) four-component model of moral action (Shields & Bredemeier, 1995). Shields and Bredemeier (1995) applied Rest's model to sport and suggested that the same four processes operate in sport contexts. According to the 12-component model of moral action (Shields & Bredemeier, 1995), each of the four processes (interpretation, judgment, choice, and behavior) is influenced by three sources of influences, namely, contextual factors, personal competencies, and ego processing variables. These contextual factors and individual differences (i.e., personal competencies and ego processing) influence specific morality processes or interact to influence several processes. In more detail, Shields and Bredemeier (1995) outlined several social contextual factors, which influence moral thoughts and behaviors in competitive sport contexts. 5 Contextual factors are the first source of influences and focus on those aspects of the environment that consistently and significantly affect moral action in sport and physical activity contexts. These aspects are related to situational ambiguity (i.e., time pressures, rules and norms, and the unobservable nature of motives), motivational climate or goal structure (i.e., cooperative or competitive structuring of the environment), moral atmosphere, domain reasoning cues (i.e., societal domain, moral domain), and power structures (e.g., leadership style). In particular, among the several contextual factors, moral atmosphere (i.e., prevailing team norms and coaching behaviors), and task-involving and ego-involving motivational climates are the two dominant factors that have received the most attention in previous research (e.g., Kavussanu, Roberts, & Ntoumanis, 2002; Kavussanu & Spray, 2006; Ommundsen, Roberts, Lemyre, & Treasure, 2003; Stephens, 2000). Second, personal competence influences include cognitive and affective competencies that make moral action possible and are essential to fully understanding morality. These include such things as role-taking and perspective-taking abilities, moral reasoning, motivational orientation, self-regulation, and social problem skills. To date, moral reasoning, goal orientations and legitimacy beliefs are the prominent personal competencies (Kavussanu, 2007; Weiss et al., 2008). Finally, ego processing factors, the third track of influences, are based on Haan's (1977) model of ego processing. Ego processes bifurcate into either coping processes or defending processes. Coping processes enable people to think clearly and coordinate 6 their feelings and ideals. In contrast, defending processes are used to distort reality in order to maintain a positive sense of self. In sum, Shields and Bredemeier's 12-component model undergirds structural developmental research in sport and physical activity, provides a comprehensive view of character issues, and is the foundation of this study. Measures of Character and Moral Development Research in Sport From a methodological perspective, most previous studies have used measures that reflect various aspects related to character constructs in the sport field. The assessments measure moral functioning (e.g., Bredemeier, Weiss, Shields, & Cooper 1986; Gibbs, Basinger, & Fuller, 1992; Hahm, Beller, & Stoll, 1989; Lee, Whitehead, & Ntoumanis 2007; Rest, 1979, 1986; Stephens, Bredemeier, & Shields, 1997), sportspersonship orientations (Vallerand, Briere, Blanchard, & Provencher, 1997), prosocial and antisocial behavior (Kavussanu, Seal, & Phillips, 2006), moral disengagement (Boardley & Kavussanu, 2007), and attitudes toward moral decision-making (Lee et al., 2007). These questionnaires use either qualitative or quantitative methods. There have been several well-known assessments used in the sport context to measure character issues: the Moral Judgment Interview (MJI; Colby et al., 1987), the Defining Issues Test (DIT; Rest, 1979, 1986), the Hahm-Beller Values Choice Inventory (HBVCI; Hahm et al., 1989), the Scale of Children's Action Tendencies in Sport (SCATS; Bredemeier, 1994), and the Judgments About Moral Behavior in Youth Sport 7 Questionnaire (JAMBYSQ; Stephens et al., 1997). These instruments consist of examples of dilemmas or common sport scenarios that arise in sport situations. For example, Kohlberg's Moral Judgment Interview (MJI; Colby et al., 1987) is an instrument used to assess the participant's stage of moral development, consisting of three parallel forms. Each form includes three hypothetical moral dilemmas indicating a conflict between two moral values. The MJI asks participants to respond to several hypothetical dilemmas. After presenting each moral dilemma, the interviewer asks 9 to 12 standardized probe questions to elicit the participant's rationale for saying why some line of action was more morally justified than another. The MJI is lengthy to administer, requires trained interviewers and scorers, and is laborious to score. Rest (1979, 1986) developed the Defining Issues Test (DIT) to assess moral reasoning maturity. The DIT is a multiple choice test with six hypothetical moral dilemmas. Each dilemma consists of a list of 12 items with short statements or questions that the participant is asked to rate in terms of importance in deciding the dilemma. Participants rank the four most important items at the end of each dilemma. When compared to the MJI, the DIT is easily administered and scored, but it still remains somewhat lengthy and cannot be used with those who have less than a 7th-grade reading ability. One of the more popular inventories, the Hahm-Beller Values Choice Inventory (HBVCI; Hahm et al., 1989) was designed to assess athletes' moral reasoning in sport situations. The HBVCI is composed of 21 common sport scenarios, each describing a behavior for which a moral rationale is provided. The moral issues within the scenarios 8 focus on justice, honesty, and responsibility. Participants are asked to respond on a 5-point Likert scale. The HBVCI assesses moral content related to sport ethics, but it does not reflect the processes of moral reasoning. The Judgments About Moral Behavior in Youth Sport Questionnaire (JAMBYSQ) was developed to measure youth soccer players' attitudes towards three types of nonmoral behavior: cheating, aggression, and lying to officials. Specifically, the JAMBYSQ assesses four constructs, players' (a) self-described fair play action tendencies; (b) legitimacy judgments concerning unfair play; (c) developmentally influenced moral motives, as they relate to temptations to engage in unfair play; and, (d) perceptions of team norms pertaining to unfair play behavior. The assessment consists of three soccer scenarios and each scenario has six items. Each item is designed to tap the participant's judgment. The very characteristic that make dilemma and scenario-based assessments valuable also limit their general usefulness. These assessments use dilemmas or scenarios to describe hypothetical situations taken from real sport contexts, but such specificity fails to account for the variety of sport situations in which issues related to character might arise. For example, scenarios in the JAMBYSQ only reflect soccer situations. In addition, in relation to Rest's theory and Shields and Bredemeier's model, these instruments do not reflect the processes of the moral action because they do not feature interactive relations between athletes' psychological processes and behaviors. Recently, two types of methodologies have been used to assess constructs relevant to sport character: (a) observations or evaluations of behavior by researchers or other 9 knowledgeable observers (e.g., Bredemeier & Shields, 1984; Kavussanu et al., 2006; Sage & Kavussanu, 2007) and (b) paper-and-pencil measures to assess moral attitudes or moral behaviors (e.g., Boardley & Kavussanu, 2007; Doty, 2005; Kavussanu, 2006; Kleiber & Roberts, 1981; Knight & Kagen, 1977; Lee et al., 2007; Sage, Kavussanu, & Duda, 2006; Vallerand et al., 1997). With respect to observed prosocial and antisocial behaviors, Sage and Kavussanu (2007) created 20 items to assess moral behavior, classified as prosocial and antisocial from a table soccer game, called foosball. The equipment used in this study was a soccer table and a video camera. The soccer table included 11 playing figures per team and two goals. Frequencies of prosocial and antisocial behavior were coded by observers through visual and auditory videotaped information from participants' two 10-minute games of table soccer. In another study, Kavussanu et al. (2006) generated 13 total items with three tapping prosocial behaviors and 10 that assessed antisocial behaviors. The items were created by video analysis through a total of 12 soccer games. However, these measures only indicate sport-specific contexts, making use in other sports difficult. Further, these observed measures focus on participants' behaviors without considering psychological processes. There have been a number of paper-and-pencil measures to assess moral attitudes or moral behaviors. The Attitudes to Moral Decision-making in Youth Sport Questionnaire (AMDYSQ; Lee et al., 2007) is an assessment of attitudes towards moral decision making in youth sport. The AMDYSQ consists of a three-factor nine-item instrument assessing one prosocial (i.e., keeping winning in proportion) and two anti-10 social attitudes (i.e., acceptance of cheating and acceptance of gamesmanship). However, the AMDYSQ focuses on social psychological approach without considering moral developmental approach. Furthermore, the AMDYSQ does not explain processes that behaviors in sport situations expose. In particular, the AMDYSQ has not been designed to address moral judgments. One of the more well-known instruments developed from the social psychological perspective, the Multidimensional Sportspersonship Orientations Scale (MSOS), measures athletes' orientations on the five sportspersonship dimensions with 25 items. The five factors, each with five items, represent (a) commitment to participation, (b) respect for social conventions, (c) respect for rules and officials, (d) respect for opponents, and (e) a negative approach to participation. However, the work of Vallerand et al. on sportspersonship reflects dispositional tendencies toward behavior rather than actual behavior in sport situations (Kavussanu, 2006). That is, the MSOS assesses sport participants' attitudes or beliefs toward sport. In addition, compared to the concept of sport character defined by Shields and Bredemeier (1995), sportspersonship is just one part of sport character, namely, the third of the four components. Doty (2005) developed an instrument to assess character through sport in cadets at the United States Military Academy (USMA). The assessment consists of 33 items with two factors, including respect and integrity. Doty's sport character assessment has several strengths: it appears to be valid and reliable. It is simple and clear with 33 one sentence items, thus making it a very pragmatic instrument in comparison to the measures consisting of dilemmas or scenarios in sport. However, their study and instrument was 11 based on the USMA program of developing leaders with character. In other words, the instrument is an adequate measure only for USMA cadets and may not be proper to use in sport research. In summary, there have been numerous instruments to assess character issues. There are three clusters of measures and methods that researchers have developed and/or adapted to measure character issues in the sport domain. The first cluster consists of measures that utilize dilemmas or scenarios in sport. The second cluster centers on methods of assessing observed behavior. The final cluster comprises paper-and-pencil measures of assessing moral attitudes or behaviors. Each of these clusters of assessment methods and measures has strengths and weaknesses. Based on Shields and Bredemeier's 12-component model, the present study created a new instrument designed to assess multiple dimensions of character in the sport contexts. Summary of Character Construct in Sport There have been arguments for and against sport as a vehicle for developing character. Such discrepancy has stimulated researchers to initiate a detailed review of character in sport. According to Rest (1984), the inner processes that produce behaviors must be critically examined to understand moral behavior. Character involves an internal state, which is manifested in behavior (Shields & Bredemeier, 1995). On the basis of Shields and Bredemeier's model, character in this study is defined as the possession of virtues associated with compassion, fairness, sportspersonship, and integrity in the sport context. Based on previous literature and theoretical consideration, 12 sport character will be classified into four subscales: compassion, fairness, sportspersonship, and integrity. These four virtues reflect Rest's (1984, 1994) moral action model extended to the sport situation. First, compassion is the character virtue that corresponds to Process I of the moral action: interpreting the situation. Compassion is an ability to feel with others, reflecting the extensive use of empathy (Hoffman, 1990, 2000; Shields & Bredemeier, 1995). Second, fairness is the virtue that is connected to Process II of the moral action model: constructing a moral ideal. Fairness includes equal consideration and can develop through two component processes: encouraging an intrinsic value of the experience and promoting prosocial norms (Shields & Bredemeier, 1995). Third, sportspersonship is associated with Process III of the moral action model: selecting the moral value from among competing values. Sportspersonship involves good manners such as winning graciously, losing with dignity, and being polite to opponents and umpires (Shields & Bredemeier, 1995). Lastly, integrity is the character trait related to Process IV of the moral action model and is synonymous with behaviorally fulfilling intentions. Integrity, or implementing action, is a quality of character clearly relevant to sport experiences and is characterized by individuals acting on their own convictions, even if such action is negatively received by a coach, teammates, or fans (Shields & Bredemeier, 1995). Problem Statement Although studies on moral and character development and behavior in sport contexts have emerged using a variety of qualitative and quantitative measures, there are 13 certain limitations regarding the conceptualization of character as well as the structure of instruments on character in sport. First, there is no measure to reflect the combined concept of sport character, connecting internal decision-making processes and behavior. Second, little research has developed instruments to assess the multidimensional constructs of character in sport. This study has been designed to address these issues. Purpose of Study The purpose of this study was to: (a) create a measure to assess the multidimensional components of sport character; and, (b) validate the scale through examination of the relationships between character in sport and other psychological variables. Research Questions The following research questions were addressed by this study: 1. Can a reliable and valid measure of sport character be developed? 2. Is there evidence of criterion-related validity of the Sport Character Scale (SCS)? 3. Is there evidence of construct validity of the SCS? Hypotheses There were several hypotheses proposed for this study: 14 1. The Sport Character Scale (SCS) will consist of four factors based on Shields and Bredemeier's model. 2. As shown in Figure 1, there will be significant relationships among the four factors. 3. Criterion-related validity of the SCS will be supported by correlation analysis. a. Significant relationships with the Multidimensional Sportspersonship Orientation Scale will provide concurrent validity of the SCS. b. The SCS will be negatively related to the antisocial moral attitudes scale, providing evidence of concurrent validity. 4. Construct validity of the SCS will be supported by correlation analysis. a. The SCS will be significantly related to the Caring Climate Scale, providing evidence of construct-related validity. b. The SCS will be significantly related to the Perceived Motivational Climate Scale, providing evidence of construct-related validity. Significance of Study First, there is no consensus definition of character in sport and physical activity. By framing this study within the model of moral action there will be some measure of uniformity and consistency among the operational definition, theory, and measurement of character. Second, there is no combined assessment of character that taps both internal processes and behavior in the sport domain. This study will attempt to provide and test a reliable and valid measure of integrated character construct in sport. Third, little research 15 has explored the entire process of the model of moral action as outlined by Rest (1984) and Shields and Bredemeier (1995). This research examining a character scale in sport will provide a more complete picture of character to researchers and practitioners. Lastly, this research examines the relationships between motivational responses and character constructs. The results of this study will provide coaches and parents with valuable information about how to create an environment that will develop character in sport and the physical activity field. Limitations There were some limitations that could influence the results of this study: 1. Participants of this study were limited to high school athletes and university athletes located within South Korea. This may limit the generalizability of the results of the study. 2. Participants were selected using convenience sampling. 3. Variation may exist in types of sport. Variation may exist between individual sport and team sport. Delimitations 1. Participants were high school and university athletes in Korea, ranging in age from 15 years and older. 2. The selection of participants was limited to one geographic region of South Korea, which limits the result's use in generalization. 16 3. Participants answered the surveys on a voluntary basis. Assumptions 1. Participants understood and responded to the questionnaires with honesty and to the best of their ability. 2. Participants represented a normal population of high school and university athletes. 3. Participants were not influenced by the investigator or assistant. 4. Participants' responses were not influenced by their peers. Definition of Terms Antisocial behavior is defined as "behavior intended to harm or disadvantage another individual" (Kavussanu, 2006, p. 578). Character is defined as "the possession of those personal qualities or virtues that facilitate the consistent display of moral action, and is described as four virtues: compassion, fairness, sportspersonship, and integrity" (Shields & Bredemeier, 1995, pp. 192-193). Compassion involves empathy or a feeling with others (Shields & Bredemeier, 1995). Fairness is explained by adhering to the rules of game and the spirit of the rules while competing (Shields & Bredemeier, 1995). Integrity is the ability to maintain and act on one's morality and convictions (Shields & Bredemeier, 1995). 17 Moral character is defined as "the ability to act in compliance the universal moral values, of honesty, justice, and responsibility" (Rudd, 1998, p. 9). Moral judgment refers to "the person's decision about what ought to be done" (Kavussanu, 2007, p. 267). Moral reasoning refers to "the criteria that the person uses to form a moral judgment" (Kavussanu, 2007, p. 267). Moral value involves "the relative worth placed on one's actions, motives, and intentions as they impinge on others, including honesty, justice, respect, and responsibility" Rudd, 1998, p. 10). Prosocial behavior is defined as "behavior intended to help or benefit another" (Kavussanu, 2006, p. 578). Social character refers to the ability to express social values such as loyalty, dedication, sacrifice, cooperation, teamwork, good citizenship, and so on (Sage, 1998; Rudd, 1998). Social value refers to those values deemed by our society to be important, which include loyalty, dedication, sacrifice, cooperation, teamwork, good citizenship, and so on (Sage, 1998; Rudd, 1998). Sportspersonship is one of the character traits associated with process 3 of the moral action model (selecting the moral value from among competing values), and involves knowing the rules and standards of behavior in sport (Shields & Bredemeier, 1995). Structural developmental theory refers to "how individuals reason about or judge values and behavior" (Weiss & Smith, 2002, p. 247). 18 Figure 1. Hypothesized Four-factor CFA Model of the SCS CompassionFairnessSportspersonshipIntegrityitem1item2item3item4item5item6item7item8item11item9item10item121111111111111111CHAPTER II REVIEW OF LITERATURE The purpose of the present study was to develop and validate a character scale in the sport domain. In the present chapter, literature related to character issues in sport was reviewed. The Definition of Character Due to the vagueness and complexity of the term ‘character' (Hodge, 1988; Shields & Bredemeier, 1995), it has proved difficult to define. Thus, there is no consensus regarding the definition of character. For instance, "character is a complex, multidimensional construct composed of many personality traits which are mutually dependent and integrated and highly influences and conditioned by the culture in which one lives" (Shea, 1996, p. 181), and "character is having the wisdom to know what is right and having courage to do what is right" (Docheff, 1997, p. 34). In a conceptual analysis of character development in sport, Hodge (1988) identified theoretical and operational definitions of character: "Theoretically, character is defined as the coordination of those socioculturally conditioned traits prescribed within a given social grouping as being necessary for the effective functioning of both the individual and the 20 group. Operationally defined, character consists of moral reasoning, ego-identity, psychological maturity, and autonomy-assertiveness" (p. 72). As indicated above, researchers have identified that character contains multidimensional components. Character as a Moral Perspective Historically, theoretical backgrounds in empirical studies of character have been based on moral perspectives. According to Hogan (1973), character consists of five related dimensions: moral knowledge, socialization, empathy, autonomy, and moral judgment, labeled moral character. Hogan (1973) made an important distinction between moral knowledge and moral judgment, in that a person may possess knowledge of the proper response to a given situation, but needs empathy or autonomy to constitute proper moral judgment. However, as Hodge (1988) pointed out, Hogan did not provide specific theoretical relationships among aspects of character. Whiteley (1982) defined character as a combination of two elements: moral reasoning and ego development. That is, "character is (a) the understanding of what is the right, fair or good thing to do in a given circumstance and (b) the courage to act in accordance with one's understanding of what is right, fair and good" (Whiteley, 1982, p. 14). Although Whiteley (1982) outlined these two factors of character as being mutually dependent and interrelated with each other, his definition of character was not described in detail. A number of educators and researchers have recently been studying character in sport based on a moral perspective. Sport is a moral practice because it is played based 21 on written rules (Rudd, 1998). Thus, sport involves moral values such as respect, honesty, fairness, fair play, and responsibility. The Definition of Sport Character There have been multiple definitions cited for character in the sport context. Some researchers have classified character into two types: social and moral (Beller, 2002; Beller & Stoll, 1995; Rudd, 1998; Stoll & Beller, 1998). Social character refers to a set of social values such as loyalty, dedication, sacrifice, cooperation, teamwork, good citizenship, and so on (Lumpkin, Stoll, & Beller, 2002; Sage, 1998), whereas moral character is regarded as the ability to consider one's actions in relationship to moral values including honesty, fairness, fair play, justice, and responsibility (Beller, 2002; Rudd, 1998). Regarding the moral aspect of sport, Doty (2005) defined character in sport as behaviors that show respect and integrity. Doty's character study is based on moral development theory by Piaget, Kohlberg, and Rest. Doty (2005) believed that the result of moral thought and moral emotion is moral behavior, or character. Shields and Bredememeier defined character as "the possession of those personal qualities or virtues that facilitate the consistent display of moral action" (1995, pp. 192-193). Accordingly, Shields and Bredemeier (1995) suggested a concept of character that links the four processes of Rest's (1984, 1986, & 1994) model of moral action with four corresponding virtues (Shields & Bredemeier, 1995): (a) interpreting the situation, related to 22 Compassion or empathy or an ability to feel with others; (b) constructing a moral judgment, related to Fairness or adherence to the rules of game and the spirit of the rules while competing; (c) selecting the moral value from among competing values, related to Sportspersonship or knowledge of the rules and standards of behavior in sport; and, (d) executing and implementing what one intends to do, related to Integrity, or the ability to maintain and act on one's morality and convictions. In other words, moral character in sport involves the internal state, which is manifested in behavior (Shields & Bredemeier, 1995). The internal state guides decisions and behaviors while defining the qualities that consist of a person's nature (Marrella, 2009). In general, under the influence of situational and personal factors, individuals go through internal decision-making processes using cognitive and affective sources before displaying behaviors. This study focused on the moral aspect of character based on Rest's (1984) four-component model and Shields and Bredemeier's (1995) 12-component model of moral action. That is, the main construct of this study, sport character, was defined as the possession of virtues associated with compassion, fairness, sportspersonship, and integrity as displayed in sport. Theoretical Framework For the purposes of this study, character development was based on moral development theory by Piaget, Kohlberg, and Rest, as reviewed below. 23 Kohlberg's Theory of Moral Development Piaget (1932), pioneer of cognitive development, introduced a theory of moral development and Kohlberg (1981, 1984) has achieved the most theoretical and empirical work in this area. Kohlberg's study of morality paralleled Piaget's general approach in the following ways: first, like Piaget, Kohlberg focused on cognition, which is the thinking process by which people construct reality and meaning; second, they both assumed that there would be stage processes in the organization of moral judgment (Rest, 1994). In addition, according to Sapp (1986), Kohlberg's theory of moral development embraced Piaget's assumptions that cognitive development emphasized moral development and that justice was the only acceptable basis for moral reasoning. Piaget (1932) identified a two-stage process of moral development: (a) the heteronomous stage where rules are viewed as sacred obligations that cannot be changed and actions are either right or wrong depending on the literal interpretation of the rule; and (b) the autonomous stage, which is characterized by the individual's ability to develop a more subjective sense of autonomy and reciprocity. Kohlberg extended Piaget's theory, offering that moral development is a continual process that occurs throughout the lifespan (Rich & DeVitis, 1994). According to Kohlberg (1981, 1984), the six moral stages are classified into three developmental levels: preconventional (Stages 1 and 2), conventional (Stages 3 and 4), and postconventional (Stages 5 and 6): "The preconventional level is the level of most children under age 9, some adolescents, and many adolescent and adult criminal 24 offenders. The conventional level is the level of most adolescents and adults in American society and in most other societies. The postconventional level is reached by a minority of adults and usually only after the age of 20-25" (Colby & Kohlberg, 1987, p. 16). Kohlberg (1984) suggested that the three levels are each associated with three types of relationships between the self and society's rules and expectations. Kohlberg's six stages comprise a developmental sequence such that all people begin using Stage 1 reasoning, and then move to the next stage (Kohlberg, 1981, 1984; Rest, 1994). At Stage 1 (heteronomous morality), a person is most impressed with the power of others, so the individual avoids breaking rules to avoid punishment. A person at Stage 2 (individualism, instrumental purpose, and exchange) realizes that all people have their own needs or interests, and this necessitates rules. Stage 3 (mutual interpersonal expectations, relationships, and interpersonal conformity) serves the need for the individual to be a good person in his/her own eyes and those of others. A person at Stage 4 (social system and conscience) fulfills the actual duties that a society has agreed on to keep it going and prevent a breakdown in the system. A person at Stage 5 (social contract or utility and individual rights) upholds a sense of obligation to law for the protection of all people's rights and values. At Stage 6, the individual follows self-chosen ethical principles that are universal principles of justice related to equality of human rights and respect for the dignity of human beings as individuals. However, there have been criticisms of Kohlberg's theory of moral development. According to Gilligan (1982), Kohlberg's study was biased against women because his 25 research was done with white men and boys. Gilligan (1982) insisted that women's moral judgment differs from men, mentioning women are more likely to consider caring and responsibility in response to moral dilemmas, whereas men are more likely consider rules and justice. Second, Kohlberg's theory overemphasizes justice as the only aspect of moral reasoning (Rest, Narvaez, Bebeau, & Thomas, 1999; Sapp, 1986). This critique brings about a question, "Is justice the only aspect of moral reasoning people consider when making moral choices?" When people make moral judgments and choices, they likely consider many other factors such as caring (Gilligan, 1982), empathy (Hoffman, 2000), compassion (Shields & Bredemeier, 1995), and other interpersonal feelings. Third, Kohlberg does not view moral development in terms of behavior. That is, there is no clear statement as to whether moral reasoning leads to moral action or not. Rest's Four Component Model Rest's (1984, 1986, & 1994) four-component model has been formulated to indicate the major psychological processes determining moral behavior. Rest's view is that there are more components to morality than just moral judgment as indicated in Kohlberg's six stages of moral development (Rest, 1994). Thus, Rest's four-component model of morality describes the factors that influence the relationship between moral thoughts and behaviors. The four-component model begins with a question, "What must we suppose happens psychologically in order for moral behavior to take place?" (Rest, 26 1994, p. 23). The four major psychological components of moral behavior are (a) moral sensitivity, (b) moral judgment, (c) moral motivation, and (d) moral character (Rest, 1984, 1986, & 1994). The first component, moral sensitivity, indicates interpreting the situation in terms of how a person's actions affect other people. The second component, moral judgment, is deciding which action is morally right or wrong among competing moral values. Moral motivation, the third component, is to select moral values among other competing values. The fourth component involves executing and implementing what a person intends to do, called moral character. Component 4 involves figuring out the sequence of behaviors, working out unexpected difficulties, overcoming fatigue and distractions, and persisting in tasks (Rest, 1986). Lack in any one component can cause moral failure (Rest, 1994). For example, if a person is easily distracted or discouraged, breaking down under pressure, even if the person has good moral sensitivity (component 1), makes great moral judgments (component 2), and chooses high moral values (component 3), the person is still regarded as having weak character, with an indication that moral failure occurred because of deficiency in component 4 (Rest, 1994). Although the four processes are presented in a logical sequence, the four components are interactive and influenced by other factors, as well (Kavussanu & Ntoumanis, 2003; Shields & Bredemeier, 1995). For instance, the components were used both as dependent variables before and after a moral education program (Gibbons, Ebbeck, & Weiss, 1995), and as correlates of variables such as perceived social approval 27 of antisocial behaviors (Stuart & Ebbeck, 1995). Kavussanu and Roberts (2001) examined the role of achievement goals in three of the four components. The term "moral functioning" referred to three components, namely moral judgment, intention, and behavior. Kavussanu and Ntoumanis (2003) also extended previous work that identified relationships between sport involvement and various aspects of morality. Based on Rest's model (1984), they examined moral judgment, intention, and behavior as indices of moral functioning. For example, the process of making a moral decision is affected by social norms and motivational factors, whereas behavior is influenced by personal condition (e.g., fatigue or distraction) as well as other factors that physically prevent someone from executing a plan of action (Kavussanu & Roberts, 2001). Rest's model suggests that multiple processes of morality are significant to the understanding of moral behavior. Moreover, his model allows researchers to consider cognitive-affective interactions within and among component processes, rather than divide cognition and affect. Sport studies based on Rest's model have identified moral judgment, intention, and behavior as indices of moral functioning (e.g., Kavussanu & Ntoumanis, 2003; Kavussanu & Roberts, 2001; Kavussanu et al., 2006; Romand, Pantaleon, & Cabagno, 2009). For example, Kavussanu and Roberts (2001) examined the role of achievement goals in three of the four components of Rest's model, namely moral judgment, intention, and behavior. This study revealed that high ego orientation corresponded to lower levels of moral judgment and intention in athletes, and greater acceptance of intentionally injurious acts. 28 In summary, Rest's four-component model focuses on the decision making process prior to executing moral behavior. Before taking action, for instance, individuals tend to go through psychological processes involving cognitive and affective factors. In addition, this model considers cognitive-affective interactions among four component processes, instead of separating cognition from affect. A 12-Component Model of Moral Action Shields and Bredemeier (1995) have suggested a 12-component model of moral action to provide more practical and comprehensive factors that influence moral behavior in sport and physical activity contexts (see Table 1). The model is based on the four processes addressed in Rest's four-component model. Shields and Bredemeier (1995) explain a moral action as the consequence of four interrelated processes. Simply, before a moral action can take place, a person must (a) interpret the situation and the action possibilities, (b) make a moral judgment about what should be done, (c) choose to act on a value (moral or nonmoral), and (d) carry out intended action to completed action (Shields & Bredemeier, 1995). The 12-component model of moral action (Shields & Bredemeier, 1995) addresses influences on the four components that Rest's model does not explicate; each of the four processes is described as being influenced by three sources of factors: (a) contextual factors, (b) personal competencies, and (c) ego processing variables. 29 As shown in Table 1, contextual factors, personal competencies, and ego processing influences may impact specific morality processes or may interact to impinge on several processes. To begin with, the contextual factors focus on those aspects of the environment that consistently and significantly influence moral action in sport and physical activity contexts (Shields & Bredemeier, 1995). The contextual influences most associated with Process 1 (interpretation) are the degree of situational ambiguity and goal structure or motivational climate. Motivational climate is the prevailing situational goal structure created by significant others, such as teachers, parents, and coaches (Ames, 1992). In Process 2 (judgment), the salient contextual influence is moral atmosphere, which indicates prevailing moral norms recognized in a group, such as coaches' and team-mates' normative beliefs. In Process 3 (choice), the contextual influences are divided into three domain cues: moral, social-conventional, and prudential reasoning (Shields & Bredemeier, 1995). Sport has many moral domain cues, such as the presence of physical risk and the saliency of issues of fairness. As sports are multidimensional social contexts, they include many social-conventional domain cues (e.g., hierarchical organization and institutional roles and rules). In Process 4 (implementation), the main contextual influences are power structures (e.g., leadership styles). According to Shields and Bredemeier (1995), power structures play a significant role in determining how gender, ethnicity, social status, and class influence an individual's ability to execute an intended action. In the next set of influences, personal competencies include the cognitive and 30 affective competencies that make moral action possible and are essential to more fully understanding morality. Role taking and perspective taking abilities are the main personal competency influences in Process 1. Perceptual role taking requires the ability to understand that others have visual fields different from one's own and then to comprehend the nature of these differences (Shields & Bredemeier, 1995). Social perspective taking is defined as how the individual differentiates one's perspective from other perspectives and relates these to one another (Selman, 1980). For example, when an individual wants to accurately interpret moral situations, she or he must be able to infer from relevant cues how the situation appears to others (role taking) and how these multiple views are related and coordinated (perspective taking). In Process 2, the main competence is moral reasoning, which includes moral beliefs, attitudes, and values. This process reflects a person's moral stage and yields judgments that impinge on the production of moral action. The competence influence in Process 3 is the self-structure, indicating the person's organized perceptions and evaluations of self. The self-structure is a dynamic psychological organization that makes moral judgments and consists of two dimensions: motivational orientation and the moral self. According to Shields and Bredemeier (1995), motivational orientation refers to whether one is oriented primarily to demonstrate competence relative to others or relative to one's own performance; the moral self means the moral qualities that are used to define oneself. In Process 4, social-cognitive capacities play an important role in carrying out one's selected values. Psychological autonomy and social problem-solving 31 skills are critical competence influences that reflect social-cognitive capacities. Finally, ego processing factors, based on Haan's (1977) model of ego processing, include 10 generic functions, each reflecting a process, function, or regulation that is required for constructive-integrative psychological activity (Shields & Bredemeier, 1995). Ego processes mediate and coordinate among psychological structures and between the psychological world and the environment (Bartek, Krebs, & Taylor, 1993; Haan et al., 1985; Shields & Bredemeier, 1995). Examples of ego-processes are empathy and the ability to focus attention. Thus, a number of variables have been proposed to influence each process. Theory Related Research A number of morality studies conducted in sport have been guided by the theoretical framework of Kohlberg and Rest. Based on Kohlberg's theory, several studies have examined moral reasoning as an indicator of moral development (Power, Higgins, & Kohlberg, 1989). In contrast, research based on Rest's theory has investigated multiple components of morality, including judgment, intention, and behavior (Kavussanu et al., 2002; Kavussanu & Spray, 2006). Shields and Bredemeier's 12-component model has played a significant role to connect numerous sport studies with morality. For example, several studies have investigated perceptions of motivational climate in moral or character variables in sport (e.g., Kavussanu, 2006; Miller, Roberts, & Ommundsen, 2004; Ommundsen et al., 2003) and the relationship between moral 32 atmosphere and tendencies toward aggressive behavior in competitive sport (e.g., Kavussanu et al., 2002; Kavussanu & Spray, 2006; Stephens, 2000, 2001, 2004; Stephens & Bredemeier, 1996). For example, a perceived mastery climate (a task-involving climate) emerged as a positive predictor of prosocial behavior, whereas a perceived performance climate (an ego-involving climate) positively predicted antisocial behavior (Kavussanu, 2006). A mastery climate was positively related to sportspersonship orientations (Miller et al., 2004; Ommundsen et al., 2003). In addition, the moral atmosphere of the team indicated strong influence on athletes' moral functioning (Kavussanu et al., 2002; Kavussanu & Spray, 2006). Besides, the moral atmosphere of the team (e.g., athletes' perceptions of their team's proaggressive norms, their coach's goal orientation) was a significant influence on youth participants' beliefs about aggression and unsportsmanlike play (Stephens, 2000, 2001; Stephens & Bredemeier, 1996). Other research has supported these findings on personal competence influences on moral or character issues in sport. For example, basketball players higher in ego orientation reported greater approval of unsportsmanlike play (Duda, Olson, & Templin, 1991) and aggressive acts (Duda et al., 1991; Kavussanu & Roberts, 2001). Ice hockey players higher in ego orientation were more likely to approve aggressive behaviors and less likely to show respect for rules and officials (Dunn & Causgrove-Dunn, 1999). In contrast, higher task orientation was associated with greater sportspersonship (Dunn & Causgrove-Dunn, 1999; Lemyre, Roberts, & Ommundsen, 2002). 33 In sum, according to Shields and Bredemeier's (1995) 12-component model of moral action, before the occurrence of a moral action a person must (a) interpret the situation, (b) decide on a course of moral action, (c) choose to act on a value, and (d) move from intended action into completed action. Each of the four processes is influenced by contextual factors, personal competencies, and ego-processing variables. The 12-component model of moral action is a useful framework for research on the relationship between the psychological world and behavior in sport and physical activity domains because of its internal-external approach. Character and Moral Development Research in Sport Settings Shields, Bredemeier, and their colleagues pioneered the field of character and moral development in sport domains. They have examined the relationship between sport participation and various perspectives of character or morality, such as moral reasoning, aggressive tendencies, and judgments. Eventually, they developed two methods to determine the relationship between sport participation and moral issues. In a study exploring the relationship between athletic aggression and moral reasoning levels, Bredemeier and Shields (1984) used Rest's Defining Issues Test (DIT; 1979) to find that college basketball players' moral reasoning maturity was lower than that of college athletes' norms. In another work to investigate the relationship between sport participation and maturity of moral reasoning regarding general social problems and sport-specific dilemmas, Bredemeier and Shields (1986) extended the comparison of 34 athletes to nonathletes. They tested the moral reasoning maturity of male and female basketball players and non-athletes at both high school and college levels. Results revealed that athletes' moral reasoning was less mature than nonathletes' within the college sample. Also, females' moral reasoning at both high school and college levels was more mature than that of males in sport. However, there was no statistically significant difference between athletes and nonathletes at the high school level. The second method to examine the relationship between sport participation and morality is to consider the interplay between sport involvement and sport character issues. According to Bredemeier, Weiss, Shields, and Cooper (1986), the extent of boys' involvement in high contact sports (e.g., football, wrestling, judo), and girls' involvement in medium contact sports (e.g., soccer, basketball) indicated less mature moral reasoning and greater aggressiveness in both sport and daily life. In another study, Bredemeier, Weiss, Shields, and Cooper (1987) found that boy's legitimacy judgments were significantly related to their moral reasoning, aggression tendencies, and involvement in high contact sports; extensive participation in high contact sports among boys corresponded to the judgment that aggressive behaviors in sport are legitimate. Hahm (1989) investigated moral reasoning and development among general university students, students majoring in physical education, and student athletes in Korea and the United States. Two questionnaires in this study were used: the Defining Issues Test (DIT) and the Hahm-Beller Values Choice Inventory (HBVCI). Eighty-seven American students and 249 Korean students completed the DIT and 197 American 35 students and 283 Korean students completed the HBVCI. The results indicated that in both Korea and America, general students had significantly higher scores in moral reasoning and more principled moral thinking than physical education majors and student athletes. Rudd (1998) conducted a study to more clearly define character and to measure character among nonathletes, team sport athletes, individual sport athletes, and military cadets. This research identified two types of character: moral and social character. Moral character was defined with the qualities of honesty, justice, and responsibility whereas social character was defined with the qualities of teamwork, loyalty, and self-sacrifice. The Rudd-Stoll-Beller-Hahm Value Judgment Inventory (RSBH-VJI) was developed to measure the two types of character. The outcome was that sport may build social character, but not moral character. Stevenson (1998) developed the Stevenson-Stoll Social Resposibility Questionnaire (SSSRQ) to measure the cognitive moral judgments of collegiate student-athletes, and compared the moral judgments of 415 participants (202 general students and 213 student-athletes) at selected NCAA Division I institutions. The results indicated that males' cognitive moral judgment scores were overall lower than females', and team sport athletes' cognitive moral judgment scores were lower than nonathletes' or individual sport athletes' scores. Proios (2010) developed and validated an instrument for the assessment of moral content judgments in sport called the Moral Content Judgment in Sport Questionnaire 36 (MCJSQ), a model with five factors: normative order, fairness, social consequences, harmony-serving consequences, and egoistic consequences. This study was designed to examine the predictive validity of the MCJSQ, the subscales of the MCJSQ showed significant correlations with task and ego orientation. Task orientation was positively related to the subscales of normative order, fairness, social consequences, and harmony-serving consequences. Ego orientation was positively associated with the subscale of egoistic consequences and negatively correlated with the dimensions of normative order, fairness, social consequences, and harmony-serving consequences. These findings are consistent with previous studies that identified the relationship between goal orientations and moral or character variables (e.g., Duda et al., 1991; Dunn & Causgrove-Dunn, 1999; Kavussanu, 2006; Kavussanu & Ntoumanis, 2003; Kavussanu & Roberts, 2001; Lemyre et al., 2002; Sage et al., 2006). Duda et al. (1991) examined the association between goal orientations and sportsmanlike attitudes among high school basketball players using six scenarios depicting aggressive acts in basketball. Athletes high in ego orientation legitimated injuring opponents so that they missed a game or were out for the season and nonphysically intimidating the opponents. Dunn and Causgrove-Dunn (1999) reported that elite male ice-hockey players higher in ego orientation were more likely to approve of intentionally injurious acts and less likely to show respect for rules and officials. Players higher in task orientation indicated greater respect for social conventions, rules, and officials. However, there was 37 no significant relationship between task orientation and legitimacy of injurious acts. Kavussanu and Roberts (2001) examined the role of achievement goals in indices of moral functioning (i.e., moral judgment, intention, and behavior), unsportsmanlike attitudes, and judgments about the legitimacy of intentionally injurious acts among college basketball players. The outcomes indicated that college basketball players higher in ego orientation reported that intimidating opponents, risking injury to an opponent, and faking injury were appropriate behaviors and showed greater acceptance of intentionally injurious acts. In addition, male athletes revealed higher ego orientation, lower task orientation, lower levels of moral functioning, and greater approval of unsportsmanlike behaviors. Lemyre et al. (2002) investigated the effects of achievement goal orientations and perceived sport ability on sportspersonship among competitive male youth soccer players. Findings revealed that highly task oriented players showed greater sportspersonship and ego-oriented players reported lower sportspersonship, including lower respect for social conventions, rules, officials, and opponents. This study also suggested that ego orientation may have a strong influence on antisocial beliefs. Among 221 college athletes, Kavussanu and Ntoumanis (2003) found that direct effects of sport participation on moral functioning were not statistically significant when the variable of ego orientation was included as a mediator variable. This study also indicated that task orientation corresponded to high levels of moral functioning. Sage et al. (2006) examined the relationships of goal orientations and moral identity with 38 prosocial and antisocial judgments and behaviors among 210 football players. The results revealed that task orientation was a positive predictor of prosocial judgments. Ego orientation related positively to antisocial judgments and behaviors. Measures in Character and Moral Development Research The first job of researchers in the empirical validation of a theory is to develop methods to measure constructs (Rest, 1994). There have been numerous measures and methods that researchers have studied to assess moral functioning and character in sport and physical activity contexts. The Moral Judgment Interview Kohlberg developed his Moral Judgment Interview in 1958 to assess moral stage development (Colby & Kohlberg, 1987). Kohlberg's measure was modified over the years and its final version is called The Standard Issue Moral Judgment Interview (Colby & Kohlberg, 1987). The Moral Judgment Interview (MJI; Colby & Kohlberg, 1987; Colby et al., 1987) consists of three parallel forms, each comprising three hypothetical moral dilemmas, followed by 9 to 12 standardized probe questions designed to draw justifications, elaborations, and clarifications of the subject's moral judgments. The MJI is administered as an oral, tape-recorded interview. To begin with, responses are transcribed and then the subject's responses are compared to examples and criteria in a scoring guide. The guide lists arguments at the various stages, and the 39 scorer's job is to match a subject's utterances with the criteria in the scoring guide. Over the course of a complete interview, typically about 50 matches are found between the subject's responses and the manual's examples. Each subject gains an overall stage score using summary rules. A typical interview lasts about 45 minutes, and scoring takes at least 30 minutes for a well-trained scorer. The scoring system of the MJI is detailed in an 800-plus page manual. The MJI has proven to be a highly reliable instrument (Colby & Kohlberg, 1987; Colby et al., 1987), indicating great test-retest, interrater, and alternate form reliability (.96 to .99, .98, and .82 to .95, respectively) and internal consistency (Form A, .92; Form B, .96; and Form C, .94). With regard to validity, construct validity for the MJI is supported by a 20-year longitudinal study in 84 boys, including cross-cultural studies. However, the MJI has the practical limitations that considerable training is required for moral interviewing and scoring, and the procedure of interviewing and scoring is laborious and expensive. The Defining Issues Test Based on Kohlberg's moral developmental theory, Rest developed the Defining Issues Test (DIT; 1979, 1986), which has been used in numerous studies (Rest, 1979, 1986, & 1994), including a number of studies in the sport domain (Beller & Stoll, 1992; Bredemeier & Shields, 1984; Hahm, 1989). Whereas in the Moral Judgment Interview (MJI) procedure a trained rater analyzes a subject's interview responses, the DIT is a 40 multiple choice, paper-and-pencil test that can be group-administered and computer-scored. The DIT presents the subject with six hypothetical moral dilemmas (some of the same moral dilemmas in the MJI are used in the DIT), each consisting of 12 items that the subject is asked to rate on a 5-point scale for greatest concern in response to the dilemmas. Then, their task is to rank the four most important items for each dilemma. Most of the 72 items for the six stories included in the DIT represent reasoning designed to appeal to a particular stage of moral thinking. Other items are designed to catch attempts at a high score, making the DIT resistant to social desirability responding. The DIT has been examined to have significant validity and reliability (Rest, 1986, 1994). Its test-retest correlations average in the .80's, and internal consistency scores of the instrument also average in the .80's. Longitudinal studies indicated that the DIT scores have positive correlations with age and educational experience and cross-cultural studies have further supported the instrument's reliability (Rest, 1994). Although the DIT is easily administered and scored, it is lengthy and difficult for children to read, requiring a significant attention span (Gibbs et al., 1992). The Sociomoral Reflection Measure-Short Form Gibbs designed the Sociomoral Reflection Measure-Short Form (SRM-SF; Gibbs et al., 1992) to assess the developmental status of moral judgment (Gibbs et al., 1992). The SRM-SF includes 11 items addressing sociomoral values. The respondents are asked to make a moral judgment and then to justify the importance of each value. The 41 justificatory responses are scored for stages of moral reasoning. The SRM-SF has examined acceptable levels of reliability and validity in participants ranging in age from fourth graders through adults (Gibbs et al., 1992). The test-retest correlation for the entire sample (n = 384) was .88, and the internal consistency value was "great" (Cronbach's alpha = .92). Concurrent validity was examined between the SRM-SF and the MJI: the correlation was significant (r = .69). Convergent validity was shown through positive correlations among the theoretically relevant variables of age (r = .66), verbal intelligence (r = .49), and socioeconomic status (r = .20). The SRM-SF also demonstrated discriminant validity by showing a statistically nonsignificant correlation with a measure of social desirability; construct validity was supported by its discrimination among diverse age samples and between delinquent and nondelinquent adolescent males. The Prosocial Play Behavior Inventory The Prosocial Play Behavior Inventory (HPPBI; Horrocks, 1979) was developed for use at the upper elementary school level to assess prosocial rather than moral behavior thought. Horrocks believed that the two were correlated. The HPPBI contains 10 behavioral items, including arguing, showing off, complaining, teasing, sharing, disobeying rules, and so on. The observer rates each participant on each behavior using a 4-point Likert-type scale. The study's ratings had moderate correlation with moral reasoning (r = .55), 42 evidencing moderate construct validity (Horrocks, 1979) and the checklist indicated a great level of reliability, with high internal consistency scores for two separate samples (.96 and .98). The HPPBI directly checks behavior, however, the moral meaning behind the rated behaviors is unclear because the observed behaviors are not asked for their own interpretations of their behaviors. Hahm-Beller Values Choice Inventory The Hahm-Beller Values Choice Inventory (HBVCI; Hahm et al., 1989) is the only inventory specific to the sport milieu, developed to evaluate moral reasoning based on three moral values: honesty, responsibility, and justice. The HBVCI is a self-administered, pencil-and-paper inventory consisting of 21 short common sport scenarios that participants are required to comment on based on a 5-point Likert scale (strongly agree - strongly disagree). To date, the HBVCI has assessed over 40,000 individuals from the ninth grade through adult populations. Internal reliability coefficients ranged from .74 to .88. Concurrent validity was established by comparison with Rest's Defining Issues Test, with correlation at .82. The HBVCI has been translated into six different languages. The Scale of Children's Action Tendencies in Sport Bredemeier (1994) developed the Scale of Children's Action Tendencies in Sport (SCATS), a self-report instrument to assess children's behavioral responses to situations 43 of conflict in sport. The SCATS stories are set in game or sport contexts. The measure itself consists of 10 items divided into six physical and four nonphysical aggression subscales. Scoring the SCATS requires summing the number of times assertive, aggressive, and submissive alternatives are selected. Using the KR-20 formula for dichotomous responses, the SCATS's internal consistency reliability within subscales was found adequate (Assertion = .68, Aggression = .85, and Submission = .66). Four psychologists and four physical educators identified each of the three action tendencies for face validity. The Judgments About Moral Behavior in Youth Sport Questionnaire The Judgments About Moral Behavior in Youth Sport Questionnaire (JAMBYSQ) (Stephens et al., 1997) was developed for use with upper elementary school girls participating in youth soccer. The JAMBYSQ is a multidimensional measure of moral functioning constructed to assess players' (a) self-described fair play action tendencies, (b) judgments concerning the legitimacy of unfair play, (c) developmentally influenced moral motives (as these relate to temptations toward unfair play), and (d) perceptions of team norms pertaining to unfair play behavior. The study consisted of three scenarios of unfair behavior in soccer: (a) lying to an official, (b) hurting an opponent to stop her from scoring, and (c) violating a critical game rule. Each scenario has six items, parallel in format and content. Item 1 assesses the respondent's deontic judgment by asking the respondent 44 whether Sally should or should not use her hands to deflect the soccer ball. Item 2 taps one dimension of moral atmosphere by prompting an estimation of the number of teammates who realistically would engage in the unfair behavior. Items 3 and 6 ask for the respondent's self-described action tendencies. Finally, Items 4 and 5 assess dimensions of moral motivation. The JAMBYSQ showed good internal consistency and construct validity. However, the scenarios of the JAMBYSQ are only soccer specific, limiting its generalizability. Sportspersonship Measures Sportspersonship instruments are designed to provide a comprehensive view of respondents' prescriptive beliefs or attitudes toward sport (Bredemeier & Shields, 1998; Shields & Bredemeier, 1995). Past research on sportspersonship and moral behavior has been influenced by two theoretical perspectives: social learning theory and moral developmental theory (Weiss & Bredemeier, 1991). Social learning theory (Bandura, 1977) focuses on models and reinforcement as determinants of appropriate and inappropriate behaviors. The moral developmental approach emphasizes that moral reasoning, which grows through several levels of development, is the major determinant of behavior. Vallerand and his colleagues have conducted a study to operationally define and measure sportspersonship (Vallerand, Deshaies, Cuerrier, Briere, & Pelletier, 1996). In the first phase of their study, 60 male and female athletes were asked to present their 45 definition of sportspersonship and to provide examples from various sport situations and behaviors. Twenty-one situations were identified to potentially indicate the meaning of sportspersonship, which were then presented to 1,056 French-Canadian athletes ranging in age from 10 to 18. A 21-item exploratory factor analysis suggested a five-factor model of sportspersonship. In the first dimension, athletes display full commitment toward sport participation by showing up and working hard during all practices and games and acknowledging one's mistakes, trying to improve. In the second, athletes respect social conventions found in sports such as shaking hands after a game, recognizing good performance by the opponent(s), and being a good loser. In the third, athletes emphasize respect and concern for the rules and officials, even when the official appears incompetent. In the fourth, athletes show true respect and concern for the opponent by lending one's equipment to the opponent, agreeing to play even if the opponent is late, and refusing to take advantage of an injured opponent. Finally, in the fifth, athletes have a negative approach toward sport, take a win-at-all-costs approach toward play, show a temper after a mistake, and compete for individual prizes and trophies. This study is important because the findings provide a definition of sportspersonship that can be captured in a scale assessing individual differences or orientations (Vallerand et al., 1997). The multidimensional sportspersonship orientations scale (MSOS) was developed to operationalize a multidimensional definition of sportspersonship (Vallerand et al., 1997). The MSOS is based on sportspersonship as defined in the previous study (Vallerand et al., 1996) and serves to measure athletes' 46 orientations in the five sportspersonship dimensions. The psychometric properties of the MSOS were examined in two studies by phases. To begin with, 20 items were prepared for each of the five dimensions. Two sport psychology researchers assessed the MSOS's preliminary version for content validity and subsequently modified these 20 items to 13 items for each of the five dimensions. Fifteen amateur athletes, ranging in age from 12 to 16 years, completed this refined version of the MSOS with 65 items. The 65-item MSOS was then administered to 132 athletes from various sport areas. The results from an exploratory factor analysis (where the best five items of each subscale were retained) led to a 25-item version of the MSOS. The MSOS was then administered to a new sample of 362 athletes with a mean age of 14.40 years, selected from six sports: badminton, basketball, hockey, swimming, track and field, and volleyball. Confirmatory factor analysis found the five-factor model had an acceptable fit for the data; internal consistency was adequate with values ranging from .71 (the commitment subscale) to .86 (the social conventions subscale) except for the negative approach subscale, which had an alpha value of .54. Correlations among the five MSOS subscales supported the construct validity of the MSOS; in Study 2, test-retest reliability over a 5-week interval was assessed with 53 athletes from the same types of sports surveyed in the other studies (mean age = 14.8 years). The resulting correlations varied from .56 to .76, with a mean test-retest correlation of .67, evincing the reliability of the MSOS. However, there are some limitations to the MSOS. First, the MSOS was validated 47 with specific athletes: young French-Canadian athletes from a limited number of sports. In order to generalize the concept of sportspersonship, other groups such as coaches, fans, parents, administrators, and so on (including other countries' populations) should be considered as samples. There was no study to examine a general consensus about the meaning of sportspersonship according to cultural values and behavioral norms. Second, the negative approach subscale revealed a low alpha value (.54), indicating that subscale's unacceptable reliability. These limitations can be interpreted as the difficulties of defining sportspersonship. The Moral Disengagement in Sport Scale The Moral Disengagement in Sport Scale (MDSS; Boardley & Kavussanu, 2007) is a sport-specific measure of moral disengagement developed in two studies. In Study 1, 59 items were developed and tested by 308 athletes from 5 team sports. In Study 2, a 6 factor, 32-item questionnaire was confirmed with 305 athletes from the same 5 sports. The construct validity of the MDSS was examined through evidence for concurrent and convergent validity for the overall scale and support for the discriminant validity of some subscales. Concurrent validity was established by the strong correlation between sport moral disengagement and antisocial behavior (r = .60) and the moderate negative correlation between sport moral disengagement and prosocial behavior (r = -.34). Convergent validity of the MDSS was provided by a strong positive correlation between sport and societal moral disengagement. Also, discriminant validity of the 48 MDSS subscales was evidenced by the correlations among the factors representing the six subscales and the relationships between the six subscales and prosocial and antisocial behavior. However, the MDSS only focuses on negative aspects of sport and does not consider internal procedures of behavior as displayed in sport. The Attitudes to Moral Decision-making in Youth Sport Questionnaire The Attitudes to Moral Decision-making in Youth Sport Questionnaire (AMDYSQ; Lee et al., 2007) was developed to assess attitudes toward moral decision-making in youth sport (as the title suggests) and described levels of ethical attitudes in young competitors. The AMDYSQ is a 3 factor, 18-item instrument assessing one prosocial (i.e., winning in proportion) and two antisocial attitudes (i.e., acceptance of cheating and acceptance of gamesmanship). The prosocial subscales of the MSOS indicated significant positive correlations with Keeping Winning in Proportion and significant negative correlations with Acceptance of Cheating and Acceptance of Gamesmanship, establishing concurrent validity (Vallerand et al., 1997). The AMDYSQ assesses different facets of sportspersonship than the MSOS (Vallerand et al., 1997) in that it addresses both a prosocial and two antisocial attitudes (Lee et al., 2007). It also addresses different dimensions of moral attitudes than the JAMBYSQ (Stephens et al., 1997), even though both have the subscale of cheating in common. 49 The AMDYSQ is potentially valuable for the investigation of unexplored dimensions of moral attitudes in youth sport and appears to be a sound instrument. However, the AMDYSQ has not been designed to address moral judgments or moral atmosphere. In other words, the AMDYSQ takes a social psychological approach without considering the moral developmental approach. The AMDYSQ does not reflect decision-making processes for moral attitudes as generated in various sport contexts. Therefore, the AMDYSQ is deficient to represent sport character issues. An Instrument to Measure Character in Sport at the U.S. Military Academy Doty (2005) developed an instrument to measure character through the sport participation experience of cadets at the United States Military Academy (USMA). This assessment to measure sport character is a clearer, more direct, simplistic, and pragmatic alternative to the HBVCI (Hahm et al., 1989) and consists of 33 items with two factors (respect and integrity). As mentioned above, the HBVCI is composed of 21 scenarios, which means it takes a long time to complete. Also, its long items tend to confuse some respondents, requiring rereading. However, Doty's instrument is composed of simple and clear, one-sentence items written at a 5th grade reading level, taking only about 8 minutes to complete. Results indicated evidence of reliability across the different samples and items, and evidence of convergent validity through correlations with the Rudd-Stoll-Beller-Hahm-Value Judgement Inventory (RSBH-VJI). 50 Although this instrument is psychometrically sound, respondent friendly, and measures some important behaviors that can be defined as character in sport, it has some limitations. Because this instrument was designed to measure character through sport in USMA cadets, it is doubted that it is adequate for use on general participants such as youth or college athletes. In addition, the concept of character in this study focused on behaviors that display respect and integrity in sport contexts. The question remains whether behaviors alone fully reflect sport character, without consideration of internal functions. The Assessment of Moral Content Judgment in Sport The Moral Content Judgment in Sport Questionnaire (MCJSQ; Proios, 2010) was developed to assess moral content judgments in sport and consists of a 24-item scale with five factors (normative order, fairness, egoistic consequences, social consequences, and harmony-serving consequences). In this study, moral judgment was defined as an individual's differentiation between ethical and unethical, right and wrong. The development of the MCJSQ items was based on the framework of cognitive-developmental theory. The 293 athletes in this study participated in three questionnaires including the MCJSQ: the Task and Ego Orientation in Sports Questionnaire (TEOSQ; Duda & Nicholls, 1992) and the Multidimensional Sportspersonship Orientation Scale (MSOS; Vallerand et al., 1997). Factor analyses suggested a five-factor model to be an adequate 51 fit for the MCJSQ, and indicated high internal consistency in four factors and moderate consistency in one (normative order; α = .63). This study also examined two subcomponents of criterion-related validity: predictive validity and concurrent validity. Analysis of its predictive validity found that task orientation was a positive predictor of four components of moral content (normative order, fairness, social consequences, and harmony-serving consequences) and a negative predictor of the dimension of egoist consequences. Further, ego orientation was shown to be a positive predictor of the factor of egoist consequences and a negative predictor for the components of the four moral contents. The five factors of the MCJSQ revealed significant relationships with four subscales of the MSOS, evidence of the concurrent validity of the MCJSQ. Though the MCJSQ is a reliable and valid measure of athletes' motives in decision making, the MCJSQ does not reflect combined processes, from interpreting situations to executing action. In terms of Rest's model, the items developed by the MCJSQ correspond to only the second process of Rest's model of moral action. In sum, there have been numerous instruments to assess character issues. Each of these assessment methods and measures has strengths and weaknesses. Due to the deficiencies of previous measures of sport character, this study developed and validated a new instrument designed to assess multiple dimensions of character in the sport context, based on Shields and Bredemeier's 12-component model. 52 Table 1 12-Component Model of Moral Action Processes Influences 1. Interpretation 2. Judgment 3. Choice 4. Implementation A. Contextual factors Goal structure (motivational climate); situational ambiguity Moral atmosphere Domain cues Power structure B. Personal competencies Role taking; perspective taking Moral reasoning Self-structure Autonomy and social problem-solving skills C. Ego processing Intraceptive processes (e.g., tolerance of ambiguity, empathy) Cognitive ego processes Affective impulse-regulating processes Attention-focusing processes Adapted from Shields and Bredemeier, 1995, Character development and physical activity (Champaign, IL: Human Kinetics), 92. CHAPTER III METHODS The purpose of this study was to develop and validate a character scale in the context of sport. This study consisted of two preliminary phases during which the sport character construct was specified and questionnaire items were developed. This was followed by four studies to examine the validity and reliability of the measure. Study 1, a pilot study, was conducted to determine if the initial items were appropriate for the participants. Exploratory and confirmatory factor analyses were explored in Studies 2 and 3, respectively to analyze the factor structure of the measure. Lastly, in Study 4 the construct and concurrent validity of the questionnaire was examined. Phase I: Specifying the Domains of the Construct The purpose of phase one of the study was to specify the definition and constructs of character in the sport domain. It is imperative that major concepts be defined clearly before generating a pool of items for a questionnaire (DeVellis, 2003; Streiner & Norman, 2003). This process allowed the researcher to delineate the sport character constructs based on a clear and succinct definition (DeVellis, 2003). Character does not have 54 physical properties, thus this study used four latent traits to make inferences and predictions about character in sport. For the purpose of this research, sport character was defined as the possession of the virtues associated with compassion, fairness, sportspersonship, and integrity in the sport context. These virtues represent the latent traits of sport character as articulated in Shields and Bredemeier's (1995) model of moral action. Therefore, based on previous literature and theoretical consideration, sport character in this study was conceptualized as having four constructs or subscales: compassion, fairness, sportspersonship, and integrity. These were operationally defined as follows. Compassion is the ability to feel with others and value sport participants. Fairness is adhering to the rules of game and the judgment that sport participants should be treated with equality according to the rules of the game. Sportspersonship is choosing to uphold standards of behavior in sport with an intense desire to win. Lastly, integrity is acting on one's morality and convictions in any given situation. Phase II: Item Generation and Development The second step in developing the sport character scale was to generate a pool of questionnaire items. Various theoretically and empirically based interpretations of the concept of sport character were reviewed and summarized in order to intellectually frame the construct. Before generating items, a number of focus groups, including Korean high school and university athletes (total n = 127), were formed and asked to brainstorm synonyms 55 and sport-related terms and examples to accompany the definitions of compassion, fairness, sportspersonship, and integrity. A sample of example questions and answers of the focus groups are included in Table 2. The responses from the focus groups were categorized based on the frequency of similar answers. Detailed domains branching out from each component were labeled. The category of compassion showed three detailed domains: (a) valuing participants (e.g., respecting others, caring for each other), (b) game manners (e.g., being courteous, shaking hands), and (c) verbal expression (e.g., encouraging opponents, genuinely congratulating the winner). Fairness included five detailed domains: (a) following the rules, (b) judging equally, (c) doing the best they can, (d) not cheating, and (e) accepting the outcome. Sportspersonship was categorized into six detailed domains: (a) doing the best they can, (b) following the rules, (c) game manners, (d) verbal expression, (e) expressing emotions, and (f) accepting the outcome. Last, integrity was classified into six detailed domains: (a) doing the best they can, (b) following the rules, (c) accepting the outcome, (d) complying with the referee's decision, (e) game manners, and (f) verbal expression. Generating Initial Draft Items Based on information from the focus groups, the literature, and discussions with experts in the sport field, 76 draft items for measuring sport character were generated. The content of each item was primarily intended to reflect the construct of sport character, including compassion (17 items), fairness (19 items), sportspersonship (19 items), and 56 integrity (21 items), respectively. The initial grouping of the items into separate categories was based on the item content and the definition of each construct. According to DeVellis (2003), multiple items constitute a more reliable test than individual items and each item should be sensitive to the essence of the latent variable. However, it is impossible to specify the number of items that should be included in an initial pool (DeVellis, 2003). Ideally a set of items is chosen randomly from the large number of items to measure the construct of interest. Regarding larger pools of initial items, DeVellis (2003), for example, suggested beginning with a pool of items three to four times larger than the final scale. Allen and Yen (1979) argued generating 1.5 to 3 times as many items as the final scale will include. Furthermore, if Likert scale items are used (e.g., the scale range is 1-5), 10 to 15 initial items per suspected subscale might be sufficient because such a number of items is more reliable (Nunnally & Bernstein, 1994). Ultimately, item analysis determines the number of items needed to obtain acceptable reliability (Pett, Lackey, & Sullivan, 2003). Internal consistency is a function not only of how strongly the items correlate with one another, but also how many items there are in the scale (DeVellis, 2003). Although reasonable internal consistency reliabilities can be obtained with as few as three items, scales with few items may lack content, construct, and internal validity (Nunnally & Bernstein, 1994). Therefore, it is better to have a large pool of initial items because the size of the correlations among the items cannot be predicted in early phases of scale development. 57 The 76 initial draft items were written in both Korean and English. Initially, the items were generated in Korean and subsequently translated into English with discussions between the researcher and three Korean-Americans who were fluent in both Korean and English. One of the translators was a football athlete at a high school in Salt Lake County. The other two were college students who played on a high school tennis team in Salt Lake County. Wording and Structuring of the Items In addition to the quantity of items, it also is essential to write good items. According to Pett et al. (2003), there are many problems that can occur when writing items that can contribute to measurement error. In order to decrease measurement error, items should be stated as clearly and unambiguously as possible. In addition, the language should be simple, straightforward, and appropriate for the reading level of the respondents (Clark & Watson, 1995; Pett et al., 2003). Therefore, the goal of this step was to keep the items short, simple, easy to read, and constructed such that adolescents are able to understand them. Moreover, throughout the expert and student review processes, the researcher excluded items that were redundant, meaningless, or confusing. Expert Review and Student Review: Content Validity The 76 initial items of Sport Character Scale (SCS) were tested for content validity. Content validity involves item sampling adequacy regarding "the extent to 58 which a specific set of items reflects a content domain" (DeVellis, 2003, p. 49). To assess content validity, the preferred method is to employ a panel of experts in the field (Clark-Carter, 1997). In this study, the assessment of content validity for the generated items of SCS was conducted through a review by a panel of experts in the area of sport as well as a small representative group selected from the target population. The items in English were reviewed by a panel of experts including two professors and four graduate students who had conducted research in sport and exercise psychology at the University of Utah. The items written in Korean were reviewed by Korean experts, consisting of three professors and two doctoral students who were experienced in scale development as well as two instructors who had played judo and golf. Additionally, seven Korean students who were competing in high school, college, and/or recreational levels in Utah were included for student review. These included five students recruited at the University of Utah and two students recruited from a high school in Salt Lake County. Three of seven were native English speakers and reviewed the initial items in English whereas the other four, who were Korean, reviewed the Korean items. The review process included e-mails and face-to-face meetings with the experts and representative group. They were provided the questionnaire that included the construct name (e.g., sport character), the definition of the construct (e.g., compassion is empathy or a feeling with others) and subscales (e.g., compassion, fairness, sportspersonship, and integrity), and the items included in each subscale (see Appendices 59 A and B). The experts and representative group were asked to provide feedback, make suggestions, and rate each of the items in domains. In particular, each item was rated relative to its content and clarity on a five-point scale from 1 (not clear) to 5 (very clear). Testing Equivalence of Items between Two Different Languages The main purpose of developing the SCS was for it to be used in Korean sport settings with Korean athletes. Thus, the items had to be in Korean. The questionnaire was translated into English for the purpose of reporting. Furthermore, both English and Korean versions had to be examined for equivalence between the two languages. Questionnaires in cross-cultural research should be equivalent in different languages (Hansen & Fouad, 1984). Applying a back-translation procedure among different languages is one method that can be used to assess the quality of the translation (Sinaiko & Brislin, 1973). In this study, a back-translation method was applied to assess the quality of translation between English and Korean. The procedures undertaken for a back-translation were as follows: first, the initial items were translated from Korean into English by the investigator and three Korean-Americans who were fluent in both Korean and English; second, the translated items were back-translated into Korean by different individuals. The researcher, a Korean faculty member studying sport psychology, and a doctoral student who taught Korean classes at the University of Utah were included as the back-translation committee members. 60 The back-translation committee individually translated the items from English into Korean. After translating the items, the items were evaluated at a meeting with the back-translation committee. The translated items were compared and several items were corrected and modified to maximize clarity and simplicity. Finally, in order to test the conceptual equivalence of the items written in two different languages, five individuals evaluated the items in Korean and the items translated into English for appropriate translation. To prevent any bias in this process, the five individuals were those who had not participated in the previous translation or back-translation process. They were all graduate school students; one masters student majoring in applied linguistics, two doctoral students majoring in sport science, and two doctoral students majoring in social science. The five students were asked to rate the translation of each item on a four-point scale (1 = absolutely different, 2 = different, 3 = the same, and 4 = absolutely the same). The items with mean values greater than or equal to three were regarded as showing the same meaning between Korean and English version. Four items had a mean below 2, indicating differing opinions among five individuals. These items were reviewed for possible inclusion and modified in order to clarify the intended meaning in Korean. In this process, the evaluation of translated items focused on conceptual equivalence, cultural meaning, and specific content of the sport situation rather than a literal translation. 61 Item Revision During the entire process of item development, the items were written and re-written as often as necessary to reflect critical feedback by the panel of experts and the small representative group. The suggestions from the experts who know the Korean language and are familiar with the culture of Korean sport and characteristics of Korean athletes were crucial in this study. Taking the expert feedback into account, a couple of items were reworded or modified for clarity and several items were eliminated due to the redundancy and ambiguity of the items. Consequently, 62 out of 76 items were retained for the subsequent analysis. More specifically, the 62 items retained through final revision process included 16 items for assessing compassion, 13 items for measuring fairness, 17 items for reflecting sportspersonship, and 16 items for measuring integrity. Response Format There are some issues to consider when determining a response option: the desired format of the items, need for variability of responses, number of response options desired, and reverse coding of some items (Pett et al., 2003). The two dominant response formats in assessment are dichotomous responding (e.g., 1 = true, 2 = false and 0 = yes, 1 = no) and continuous scales with three or more options (e.g., a 5-point Likert-type scale that ranges from 1 = very much to 5 = not at all) (Clark & Watson, 1995). Likert-type scales are used with a number of different response formats, such as frequency (never to 62 always), degree or extent (not at all to very much), similarity (like me to not like me), and agreement (strongly agree to strongly disagree) formats. Regarding item variability, if the items in an instrument do not have variability of response, inter-item correlations would be restricted and weak. Thus, a sufficient range of responses is important in item development. A significant source for obtaining variability of responses is the number of response options available to the participants (Pett et al., 2003). In the case of Likert scaling, the number of scale steps generally recommended is five to seven (Sudman & Bradburn, 1982). An odd number (typically, five or seven) allows for the midrange scale to be the neutral or indifferent point. Meanwhile, an even number of response options (typically, six) eliminates the problem of an odd number (uncertainty) but forces respondents to either agree or disagree to some extent (Clark & Watson, 1995; Pett et al., 2003). A six-point Likert-type scale in this study was selected for obtaining optimal conditions of reliability and validity. All responses ranged from 1 (strongly disagree) to 6 (strongly agree). By eliminating a neutral option of items, participants could be encouraged to think more carefully and thoroughly about each item. This might lead to maximum precision in respondents' answers. Foil Items Three foil items were written and randomly placed within the generated items. Foil items are statements commonly known to be incorrect that can be identified 63 immediately by a respondent who completes the scale carefully (Chen, 1998). The purpose of foil items is to identify those participants who do not complete the scale carefully (Chen, 1998). The respondents with a mean greater than two on three foil items were eliminated from data entry because a mean greater than two indicates that the participants did not carefully respond to the items. The three foil items used in this study were as follows: 1. Basketball is a non-contact sport like swimming. 2. The shape of a soccer ball is rectangular. 3. The shape of a basketball is a triangle. Study 1: Pilot Study Pilot tests of a newly developed instrument should be conducted with respondents selected from the same population from which the participants in the original study will be selected (Lackey & Wingate, 1998). Participants involved in the pilot study should not be contained in the original study. Additionally, the sample for a pilot study of an instrument is recommended to be one tenth the size of the sample for the primary study (Pett et al., 2003). Pilot testing an instrument allows researchers to identify items that are misunderstood or are not being answered in the way that researchers want (Pett et al., 2003). Based on these recommendations, a pilot study was performed to examine the reliability of the initial questionnaire and to filter out the poor items. 64 Using Respondents Similar to the Target Group A convenience sample of 50 athletes (31 males and 19 females) in Korea was surveyed for the pilot study. The sample consisted of high school athletes (n = 23) and university athletes (n = 27), ranging from 16 to 25 years (M age = 18.96, SD = 2.60). The athletes represented a variety of sports, including baseball (n = 16), soccer (n = 3), judo (n = 3), golf (n = 8), tennis (n = 4), taekwando (n = 2), archery (n = 2), handball (n = 2), and others (n = 10). The athletes were quite experienced in their sport (M = 6.02 total years played, SD = 2.46). The questionnaire for the pilot test contained demographic information and a total of 65 items. The 65 items to assess sport character contained 16 items on Compassion, 13 items on Fairness, 17 items on Sportspersonship, 16 items on Integrity, and 3 foil items. Descriptive statistics (mean, variance, and standard deviation) and reliability estimates (Cronbach's alpha and item-to-total correlation) were utilized to evaluate the quality of items from the initial survey questionnaire. In order to evaluate the items in scale development, item-total correlations, item variances, and item means are generally used to detect certain characteristics of the items and construct (DeVellis, 2003). It is not desirable to include items if the variance is close to zero or the mean of an item is near one of the extreme scores of the range (DeVellis, 2003). In addition, items indicating low item-total correlations (r < .4) with other items in a scale should be removed from the item pool (Blunch, 2008). The item-total 65 correlation refers to the correlation between the given item and the total scale score without the given item from the total score (Pett et al., 2003). Study 2, 3, and 4: Validation of the Sport Character Scale The purpose of these studies was to validate the items of the Sport Character Scale. Participants According to Tabachnick and Fidell (2007), at least 300 cases are needed for factor analysis. Hair and colleagues suggest a sample size of at least five times the number of tested items (Hair, Anderson, Tatham, & Black, 1998). Thus, the sample size should be quite large and is influenced by the number of items that are included in the measure. A convenience sample of 734 athletes in Korea was recruited for the main study. Participants were from 13 institutions (seven high schools and six universities) in Korea. However, 49 respondents reported a mean greater than two on the three foil items and were removed from the total sample of 734, leaving 685 participants. The final sample included 685 athletes, ranging from 15 to 41 years (M age = 19.86, SD = 2.50). More men (n = 477) participated in the study than women (n = 208). Further, the sample consisted of high school athletes (n = 291) and university athletes (n 66 = 394) (see Table 3). The athletes represented 31 sports (see Table 4) and the average years played in their sport was 6.42 years (SD = 2.93), ranging from one to 16 years. Procedure for Data Collection Permission to recruit participants in the study was obtained from the athletic program directors and coaches of six high schools, one college, and five universities in Korea. Each Korean Institution submitted a letter indicating its approval of the research. After obtaining approval from the University of Utah's Institutional Review Board (IRB) for Human Subjects Research, the investigator contacted the athletic director and coaches at each institution to schedule a visit to collect data. Upon receiving permission, the researcher met the athletes to explain the study and ask for volunteers to participate. At an initial meeting with athletes under the age of 18 (during a regularly scheduled class or practice time), the investigator explained the study, and distributed parental permission forms. The researcher returned a few days later, collected the parental permission forms, and asked the athletes to sign the informed consent if they so desired. Only those athletes who had both forms completed participated in this study. Meanwhile, at a meeting with athletes 18 years and older, athletes were asked to sign and return the consent form to the researcher with the survey. Prior to asking for the athletes' consent, the researcher provided the purpose and procedures of the study. Questionnaire packets were handed out to participating athletes. No coaches were allowed in the room during data collection to provide a more comfortable atmosphere for 67 more honest answers. Participants were informed that their responses would be both anonymous and confidential. They were encouraged to ask any questions that arose during the collection process and instructed that they might choose to skip any questions they did not want to answer. The total time taken to complete the questionnaires was approximately 15 to 20 minutes. Measures Demographic information including age, gender, type of sport enrolled in, and year in school were collected to describe the personal characteristics of the athletes. The newly developed Sport Character Scale (SCS), with a 6-point Likert-type Scale (1 = strongly disagree; 6 = strongly agree), was used to measure athletes' character in a sport context. In order to address the validity of the SCS (Study 4), additional measures including the Caring Climate Scale (CCS, Newton et al., 2007), the Perceived Motivational Climate in Sport Questionnaire (PMCSQ; Walling, Duda, & Chi, 1993), the Multidimensional Sportspersonship Orientation Scale (MSOS; Vallerand et al., 1997), and the Attitudes to Moral Decision-making in Youth Questionnaire (AMDYSQ; Lee et al., 2007) were administered. Like the SCS, all of the questionnaires used in this study were translated into Korean for Korean athletes. The questionnaires translated into Korean were reviewed for validation by a panel of experts: three Koreans, including one faculty member who has expertise in sport psychology, a master student majoring in applied linguistics, and a 68 doctoral student majoring in sport psychology. To begin with, the researcher and two Korean-Americans who were fluent in both Korean and English translated the English version of the instruments into Korean. After completing translation, both the questionnaires including the original English version and the translated version in Korean were reviewed by three Korean experts. During this process, the translated Korean version of the questionnaires was modified to reflect conceptual equivalence with the original English version while remaining sensitive to the Korean culture. Some ambiguous items were revised after discussing with a panel of experts. Perceived motivational climate. The Perceived Motivational Climate in Sport Questionnaire (PMCSQ; Walling et al., 1993) was used to examine construct validity with the SCS. The PMCSQ assesses players' perceptions of the motivational climate on their team and consists of 21 items that measure mastery (9 items) and performance climate (12 items). The stem for each item is "On my team…" Examples of items tapping a perceived mastery climate included "trying hard is rewarded" and "the focus is to improve each game." Examples of items indicating a perceived performance climate included "players feel good when they do better than their teammates" and "players are punished when they make mistakes." Participants were asked to respond using a 5-point Likert response scale (1 = strongly disagree; 5 = strongly agree). Subscale scores were created by calculating the mean of item scores within the respective climate dimensions. The PMCSQ has been shown to be valid and reliable with youth athletes (Walling et al., 69 1993). In the present study, the alpha coefficients were .74 and .75 for performance and mastery climates, respectively. Caring climate. The Caring Climate Scale (CCS; Newton et al., 2007) is a 13-item questionnaire that assesses the extent to which the respondents feel the social and interpersonal context is caring. The participants responded to items on a 5-point Likert scale. The CCS has exhibited high internal consistency with a Cronbach alpha coefficient of .92 and strong convergent and discriminate validity (Newton et al., 2007). In the current study, this measure was used to examine construct validity with the SCS and its Cronbach alpha coefficient was .93. Sportspersonship. The Multidimensional Sportspersonship Orientation Scale (MSOS; Vallerand et al., 1997) was used in this study to evaluate concurrent validity with the SCS. The MSOS is a 25-item inventory consisting of five subscales to assess sportspersonship: (a) respect for social conventions of sport (e.g., I shake the opponent's hand after a game regardless if we have lost or win); (b) respect for rules and officials (e.g., I respect the rules); (c) respect and concern for the opponent (e.g., When an opponent gets hurt, I ask the official to stop the game so that he/she can be helped); (d) respect for one's commitment toward participation in sport (e.g., It is very important for me to be at every practice); and, (e) negative approach toward the practice of sport (e.g., After competition, I make excuses for a poor performance). All responses were made on a 5-point scale ranging from 1 (doesn't correspond to me at all) to 5 (corresponds to me exactly). The reliability of the MSOS has been supported (Vallerand et al., 1997). In the 70 present study, the alpha coefficients for the five dimensions of the MSOS were α = .65 for social conventions, α = .72 for rules and officials, α = .83 for commitment to sport, α = .64 for respect for opponent, and α = .45 for negative approach to sport. The low level of internal consistency on the negative approach subscale has also been reported in previous research (Proios, 2010; Vallerand et al., 1997). Thus, the negative approach subscale was excluded from subsequent analysis. Antisocial moral attitudes. Two subscales of the Attitudes to Moral Decision-making in Youth Questionnaire (AMDYSQ; Lee et al., 2007) were used to assess the participants' antisocial moral attitudes (i.e., acceptance of cheating and acceptance of gamesmanship) with 6 items. This measure was utilized to evaluate concurrent validity with the SCS. An example item for the subscale of acceptance of cheating is "If other people are cheating, I think I can too", whereas an example item for the subscale of acceptance of gamesmanship is "I sometimes try to wind up the opposition." Participants were asked to respond to the items on a 5-point scale ranging from 1 (strongly disagree) to 5 (strongly agree). The AMDYSQ has adequate concurrent validity, factorial validity, and internal consistency (Lee et al., 2007). In the present study, the alpha coefficients were .88 and .82 for acceptance of cheating and acceptance of gamesmanship, respectively. 71 Data Analysis The data collected from the survey were analyzed by using the SPSS 18.0 and AMOS 18.0 version. Descriptive statistics, correlation analysis, reliability analysis, exploratory factor analysis, and confirmatory factor analysis were conducted to analyze the data. Descriptive statistics. Descriptive statistics including the mean, standard deviation, range, skewness, and kurtosis were used to describe each item of the Sport Character Scale and to assess missing data, outliers, and normality. Exploratory factor analysis. The primary purpose of Study 2 was to examine the factor structure underlying the items. Exploratory Factor Analysis (EFA) was used for exploring the underlying dimensions of the construct of interest (Pett et al., 2003). Thus, EFA was conducted to determine how many latent variables underlie the set of items contained in the SCS and create appropriate factors. A number of extraction and rotation methods were performed to determine the most appropriate factor solution. In the SPSS package, numerous methods for factor extraction and rotation are available. Among the extraction techniques, PCA and PAF are the most commonly used (Tabachnick & Fidell, 2007). After extraction, rotation is used to improve the interpretability and scientific utility of a particular solution (Tabachnick & Fidell, 2007). The researcher must decide between orthogonal and oblique rotations. Orthogonal rotations such as varimax, quartimax, and equamax assume that the factors are uncorrelated, whereas oblique rotations such as direct oblimin, promax, and procrustes 72 offer a continuous range of correlations between factors (Pett et al., 2003; Tabachnick & Fidell, 2007). Criteria used to identify the correct number of factors in this study were as follows: (a) the Kaiser-Guttman rule: retaining only those factors for which the eigenvalues are greater than 1.00; (b) scree test, a graphical method for determining the number of components; and (c) a Parallel Analysis (PA), comparing the obtained eigenvalues from the actual data with those one would expect to obtain from a random data set. Of these, the PA approach is considered one of the most accurate methods in research for deciding the appropriate number of factors to retain (Fabrigar, Wegener, MacCullum, & Strahan, 1999; Hayton, Allen, & Scarpello, 2004). A number of studies have shown that the PA approach is better than Kaiser-Guttman rule or scree test as factor retention approach. Specifically, the Kaiser-Guttman approach tends to over-factor, that is, encourages the selection of too many factors (Fabrigar et al., 1999; Zwick & Velicer, 1986). The scree test is open to subjectivity and ambiguity, especially when there are either no clear breaks or two or more apparent breaks (Hayton et al., 2004) in the plot. In addition, factor loadings and factor interpretability were used as criteria for item reduction. Factor loadings greater than |.30| are considered to meet the minimal level, factor loadings of |.40| are regarded as more important, and factor loadings of |.50| or greater are considered practically significant (Hair et al., 1998). The generally acceptable reliability value is above .70 (Nunnally, 1978). In this study, items having a 73 factor loading less than .40 and high loadings on more than one factor were deleted from the item pool. Confirmatory factor analysis. Confirmatory Factor Analysis was used in Study 3 to confirm that factor structure identified through Exploratory Factor Analysis in Study 2. In contrast to EFA, "Confirmatory Factor Analysis (CFA) is appropriately used when the researcher has some knowledge of the underlying latent variable structure" (Byrne, 2001, p. 6). According to Nunnally and Bernstein (1994), CFA is conducted to assess the extent to which the hypothesized factors fit the data. In other words, CFA is used to test the utility of the underlying dimensions of a construct identified through EFA, to compare factor structures across studies, and to examine hypotheses concerning the linear structural relationships among factors related to a specific theory or model (Pett et al., 2003). For the purpose of this study, CFA using AMOS 18.0 program was conducted to confirm the construct validation after the underlying structure had been tentatively established by EFA. The overall model fit, the specific indicator loadings, and construct validity and reliability for the factor model were examined in the CFA. In the present study, CFA was conducted using maximum likelihood (ML) estimation method. ML is the most commonly used estimation method and requires a sample of at least 200 participants, continuous data, and a normal distribution of the test variables (Kline, 2005). ML performs, on average, better than most other estimation methods, even when its assumptions are violated (Chou & Bentler, 1995). 74 Subsequently, model fit indices were used to identify the factors extracted by EFA. The issue of goodness-of-fit focuses on the question of whether the model is consistent with or fits the sample data. The most basic goodness-of-fit index is the chi-square stat