‘ dd - . . D I . .O’l - s I ‘ 4 ‘ A ‘ ‘ Flu“! gflfi._u.f1v§5x_.uu1vnv§b3:§ wit... mans. 2.55%.}...133: .. .....é: ...‘LxfierBLEJQem-l. ; c ‘Isf. , v : . .Ihmn..nnu1.u.ln.r. ha}... “was. All. ..UIL... than? .. s.1.t..v|.t.fi§.fl...Influgaznavrflfivw.v. .311 t: ! lti Lit... t .1 1 . -. IIQV .IHYI ll.” 0.)!- » ..lcllt‘ll .4! ii I N 1....“ v... .r - . Vsu‘lpcr on 1112-! ‘.111.1.L | I! (.lrvKIt‘I III Ioln\. t‘ XWMIVvN-athvtiuxvflm‘l Jan—Ulnar}: ....vflPIhMI-chfiflo.gvu\ovb~ mum-v0.0.hfis‘x. but;‘\m ,1 .HH u ‘HI .MAW . ‘0'. v .tp‘ b I t. 0'.» . ‘ E I I 0 . ‘ Qua-Wt. a 0‘ ..Vv-OQ-AI‘ ‘ A . . In. I ‘ ‘ ‘1‘ n a .00.. r... - - Q . ‘- ‘. I 9. ..l . g \n :1 ‘ by luatIWN—l § cl iw D‘ --.. A $111.. . . I. I.‘ ‘13”— $1...- . [‘1‘] 1601: IIPrIHwk ..ll‘r - 3 _ v. 1A I I”\ “ NH»... 41‘. Ill Il’l l4. 1 14‘ ‘ I\ [I‘ll . Wm 3‘33"? 33,: 37”.: 24,92}qu l " 41?" 1 4‘”! ' l . 9 “H ‘g'! ’ ' ‘ ' Tr‘ MM} lllllllllllllllllllllllllllllllzllllllllll 3 1293 1048 M’s-1:515 ; r2222. fimte Lg?» no la n- m .2 «gs-s". a a F 5 Univemafiy J This is to certify that the dissertation entitled THE INFLUENCE OF COACHING BEHAVIORS ON YOUNG ATHLETES' PERCEPTIONS OF COMPETENCE AND CONTROL presented by THELMA STERNBERG HORN has been accepted towards fulfillment of the requirements for Ph.D. degree in Philosophy QOW'J A”. /jo~iefl Major professor Date 1 l 82 MSU is an Affirmative Action/Equal Opportunity Institution 0—12771 MSU LIBRARIES a; RETURNING MATERIALS: Place in book drop to remove this checkout from your record. FINES will be charged if book is returned after the date stamped below. THE INFLUENCE OF COACHING BEHAVIORS ON YOUNG ATHLETES' PERCEPTIONS OF COMPETENCE AND CONTROL By Thelma Sternberg Horn A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Health, Physical Education, and Recreation 1982 ABSTRACT THE INFLUENCE OF COACHING BEHAVIORS 0N YOUNG ATHLETES' PERCEPTIONS OF COMPETENCE AND CONTROL By Thelma Sternberg Horn Based on a review of the research from several fields, a theoretical model was developed which delineated the processes through which coaching behaviors may influence young athletes' perceptions of competence, control and motivation in athletic achievement areas. A field study was then conducted to test two components of the prOposed model. Specifically, the associative relationship between coaches' perceptions of players' ability and their subsequent behavior towards individual athletes was examined. Secondly, changes in players' psychological responses over the conpetitive season were examined as a function of exhibited coach-athlete interactions. Seventy-two female junior high softball players and their coaches participated as subjects in this study. The Coaching Behavior Assessment System (Smith, Smoll & Hunt, 1977) was used to record individual coach-athlete interactions. Pre-season and post season assessments of coaches' expectations concerning players' ability and players' perceptions of their competence and control were also conducted. Thelma Sternberg Horn Multivariate analyses indicated that coaches do exhibit differential patterns of behavior to individual athletes based on their perceptions concerning players' ability. However, the demonstrated effects were found only in relation to coaches' behavior in games. Examination of the direction of these effects suggested that these differential patterns of behavior may reflect coaches' attempts to individualize instruction rather than their biased behavior towards athletes with high ability. Multivariate regression analyses additionally revealed that a small but significant portion of the variance in players' psychosocial growth over the season could be predicted by measures of the players' attained skill level and the behaviors of their coach in response to that perfonnance. Although the level of attained skill was the most consistent predictive variable, certain coaching behaviors (e.g., reinforcenent, nonreinforcement, and punishment) were also influential in predicting changes in players' perceptions of conpetence and control. The salience of these particular coaching behaviors was discussed in light of their contingency to players' perfonnance and their role in providing players with clear, consistent evaluation of their performance. It was concluded that the proposed coaching effectiveness model may be a viable means of designing and formulating future research in coaching effectiveness. Cepyright by THELMA STERNBERG HORN 1982 ACKNOWLEDGMENTS Completion of the requirements for the doctoral degree represents, for me, the end of a long career as a student, but also the beginning of a new commitment to the field of education. I would like to acknowledge those individuals whose contributions assisted me in making these educational goals a reality. Dr. Dan Gould, my academic advisor, who has guided, directed, and encouraged my graduate work for four years. The perfonnance-contingent and appropriate feedback (liberal anounts of praise and_criticism) which he has always provided me was essential to my deveTOpment as a Sport psychologist. His own professional accomplishments as a researcher, writer, clinician, and teacher have been and always will be for me a standard of perfonnance towards which I will strive. Dr. Annelies Knoppers whom I have known as a coach, teacher, colleague, and friend, and who has in each of these roles directly influenced my thoughts and ideas I especially acknowledge her role in my graduate work as one of the finest and most knowledgeable professors I have had. Dr. Deb Feltz whose accomplishments as a researcher have encouraged my own interest and whose advice as a statistician and acadenic professional I have often sought. iii Dr. Vern Seefeldt whose commitment to the goals and phil050phy of the Youth Sports Institute has really strengthened my desire to continue working with youth in sport. Dr. Jere Brophy whose research and writing I greatly adnire, for providing me with essential direction in conducting this research and for always expressing interest in my work and my ideas. Dr. Maureen Weiss for sharing with me many personal and professional interests and for leaving me with a very high standard of graduate school perfonnance. My own parents, my very first teachers, for "giving" me my first college degree and for developing in me a respect and desire for a continuing education. My "second" parents who have encouraged and supported all our efforts in so very many ways but who seldan allow us an Opportunity to express our gratitude. My daughter, Joci, for helping me to learn that being a mother does not prevent me from fulfilling other professional and academic roles. My husband, Dave, for his work in data collection, keypunching and proofreading, for his role in making graduate school a tolerable experience, and finally, for reasons that have nothing to do with this project or this degree. iv TABLE OF CONTENTS LIST OF TABLES ......................... vii LIST OF FIGURES . ................. . . . . . . . ix CHAPTER I: REVIEW OF RELATED LITERATURE ............ 1 Research on Instructional Effectiveness . . . . . . . . . . 6 Instructional Behaviors and Children' s Achievement ............ . . . 9 Instructional Effectiveness in Motor Skill Settings . . 13 Instructional Behaviors and Children' s Psychosocial Devel0pment....................19 Psychosocial Growth in Motor Skill Settings . . . . . . 23 Instructional Effectivenss: Summary and Research Implications ..... . . . . . . ........ 27 Theoretical Models: Explaining Instructional Effectiveness . . . . . . . . . . . ....... 31 Expectation Theories . . . . . . . .......... 32 Perceptions of Control . . . ............ 47 An Integrated Model of Instructional Effectiveness . . 63 Overview of the Present Study ........ . . . . . . . 69 CHAPTER 11: METHOD ......... . ..... . . . . . . . . 71 Methological Overview . . ...... . . . ..... . . . 71 Subjects . . .......... . ............. 74 Instrumentation . . . . . . . . . . . . . . . . . . . . . . 78 Process Assessments . . . . . . . . . . . . . . . . . . 78 Product Assessments . . . ....... . . . . . . . . 86 Presage Variables . .................. 90 Research Procedures . . . . . . . . . . . . . . . . . . . . 93 Training of Coders . . . . . . . . . . . . . . . . . . 93 Data Collection Procedures ............. . 94 CHAPTER III: RESULTS . . . . . . . . . . . . . . . . . . . . . . 99 Statistical Analyses: Expectancy Effects ........ . 99 Data Preparation . . . . . . . . ......... . 103 Results: Expectancy Effects . . . . . ........ 110 V Statistical Analyses: Coaching Effectiveness ...... 134 Psychometric Analyses . . . . . . . . . . . ..... 134 Data Preparation . . . . . . ........... 137 Practice Behaviors and Players' Psychosocial Development .................. 140 Game Behaviors and Players' Psychosocial Development . . ............ 145 Analyses of Coaching Effectiveness: Smmnary . . . . 146 CHAPTER IV: DISCUSSION .................... 148 Expectancy Effects .................... 148 Coaching Effectiveness .................. 154 Future Research Directions in Coaching Effectiveness .................. 160 A Re-Examination of Coach-Athlete Interactions . . . 164 Conclusions and Implications ............... 168 Methodological Issues . ............... 168 Theoretical Implications .............. 171 APPENDIX A: LETTER TO SCHOOL PERSONNEL ............ 175 APPENDIX B: PARENT LETTER AND CONSENT FORM .......... 177 APPENDIX C: PSYCHOMETRIC TEST BATTERY ............ 180 APPENDIX D: COACHES“ DEMOGRAPHIC QUESTIONNAIRE ........ 188 APPENDIX E: STATISTICAL ANALYSES: PARALLELISM REGRESSION RESULTS ............. 191 APPENDIX F: DATA ....................... 195 REFERENCES NOTES . . ......... . ........... 215 REFERENCES .......................... 217 vi TABLE «h m N o o 0 IO. 11. 12. 13. 14. 15. 16. 17. 18. 19. LIST OF TABLES Model for Academic Expectancy Effects .......... 33 Expectation Communication Model . . . .......... 46 Breakdown by School of Project Participants . . ..... 76 Summary of Assessment Procedures . . . . . . ....... 98 Summary by School of Coaches' Practice Behaviors ..... 101 Summary by School of Coaches' Game Behaviors ....... 102 Computational Summary of Coaching Behaviors Indices . . . 105 Indices of Coaching Behaviors During Practice Sessions . . 107 Indices of Coaching Behaviors During Game Sessions . . . . 108 Positions Played by High and Low Expectancy Athletes . . . 113 Post-Season Expectancy Effects: Discriminant Function Results for Frequency of Coaching Behaviors ...... . . . . . . ........... 117 Pre-Season Expectancy Effects: Discriminant Function Results for Coaches' Game Behaviors ..... 120 Post-Season Expectancy Effects: Discriminant Function Results for Coaches' Game Behaviors ..... 122 Standardized Regression Coefficients for Prediction of Coaches' Game Behaviors . . . . . . . . . . . . . . 124 Canonical Loadings for First and Second Canonical Correlations: Coaches' Expectations and Coaching Behaviors . . . . . . . . . . . . ...... 127 Categorical Percentages of Coaches' Attribution for Players' Perfonnance . . . . . . . . . . . . . . . 130 Reliability Estimates for the Softball-Specific Perceptions of Control Subscles . . . . . . . ..... 136 Subscale Measures of Perceptions of Control and Competence . . . . . . . . . . . . . . . . ...... 139 Standardized Regression Coefficients: Coaches' Practice Behaviors and Players' Perceived Competence Gains . . . . . . . . . .......... 142 vii TABLE 20. 21. 22. 23. 24. Canonical Loadings: Coaches' Practice Behaviors and Players' Perceived Competence Gains ........ 144 Summary Results of the Test for Homogeneity of Regression Planes: Low and High Competency Athletes . . . .................... 191 Follow-Up Regression Analyses: Low and High Perceived Competence Athletes . . . . . ........ 192 Standardized Regression Coefficients: Coaches' Practice Behaviors and Perceived Control for Low Perceived Competency Players . . . . . . . . . . . 193 Canonical Loadings: Coaches' Practice Behaviors and Perceived Control for Low Perceived Competency Players .................. 194 viii LIST OF FIGURES FIGURE 1. Harter's (1981) Phase 3 devel0pmental model of intrinsic mastery motivation ............. 57 2. Proposed model for coaching effectiveness ........ 64 3. Dimensional categorization of sport-related attributions (Roberts & Pascuzzi, 1979) ........ 84 ix CHAPTER I REVIEW OF RELATED LITERATURE The growing pepularity of competitive youth sport programs has generated considerable controversy regarding the influence of organized sport participation on the psychological develOpment of young children (Martens, 1978). Proponents of youth sport programs assume that par- ticipation by children in competitive sport will "build character" by teaching young children responsibility, task persistence, cooperation, confonnity to team rules, and self-discipline. In contrast, critics contend that participation in competitive athletics by young children promotes the development of such negative psychological traits as competitive anxiety, fear of failure, low self-esteem, aggression, and anti-social behavior. Research, however, has failed to support either position. A number of investigators and reviewers have concluded that sport participation has not been consistently found to either facili- tate or deter positive psycho-social growth in youth sport competitors (Ash, 1978; Gelfand & Hartman, 1978; Scanlan & Passer, 1978; Simon & Martens, 1979). Several writers have suggested, however, that participation in competitive Sport has the potential to exert either a positive or negative effect on the psychological development of the young child, depending on the quality of the program itself (Alley, 1974; Martens, 1978). A key factor in determining whether sport participation will be a positive or negative experience is the quality of the adult leader- ship. Both Martens (Note 1) and Gould (Note 2), for example, have recently emphasized the role which coaches can play in influencing 1 young athletes' cognitive perceptions and attitudes towards athletic participation. Gould (Note 2) further recommended that coaches should actively plan and initiate coaching strategies designed to facilitate children's perceptions of competence and self-worth in athletic endeavors. While these individuals have suggested that relationships exist between coaching behaviors and the psychosocial growth of young athletes, Sport psychologists have just begun to empirically examine this issue. For example, a series of studies recently conducted by Smith and Small and their associates (Smith, Smoll 8 Curtis, 1979; Smith, Smoll & Hunt, 1977; Smoll, Smith, Curtis & Hunt, 1978) was designed to investigate the influence that coaching behaviors exert on young athletes' attitudes and self—esteem. The results of this longitudinal research project provided support for the contention that the success of a Sport program in facilitating positive player attitudes and increased measures of self-esteem is significantly influenced by the behavior of the coaches towards the players. This demonstrated saliency of coaches' behavior in regard to athletes' psychosocial growth is consistent with research results from developmental psychology concerning patterns of psychosocial deveTOpment in children. Most of this research indicates that the behaviors of parents, teachers, and other significant adults influence the develOpment of such attributes as achievement motivation, and self-concept in young children (Harter, 1978; Maccoby & Jacklin, 1974; Mussen, Conger, 3 Kagan, 1974; Stein & Bailey, 1973). It seems logical, then, to presume that coaches can and do influence the psychosocial development of their young athletes. However, much more specific information concerning this relationship needs to be acquired. From a practical standpoint, there exists a need to identify which coaching behaviors are most facilitative of the child's growth. Theoretically, it is necessary to ascertain how these particular coaching behaviors influence the child's development. These two issues provide the basis for the present investigation which was specifically designed to empirically assess the relationship between the instructional behaviors exhibited by coaches and the corresponding psychosocial reSponses exhibited by their young athletes. To provide a framework for this proposed relationship, a model, based on a review of the related literature, was develOped to delineate the process by which coaching behaviors may influence the growth of young athletes. This first chapter, then, contains a survey of the available literature which was used to fonnulate the coaching effectiveness model. The investigation itself was conducted to assess certain aspects of the developed model. Although very little research has been reported in the sport scientific literature relative to coaching effectiveness, researchers in the parallel field of teacher education have deveTOped, within the past decade, some very practical research techniques which have allowed them to acquire a sizable amount of infonmation concerning the instruc- tional behaviors which most effectively facilitate students' academic and psychosocial deveTOpment. Initial, but lhnited, attempts to apply these research methods to the study of instructional effectiveness in motor skill settings have also yielded some practical knowledge concerning effective instructional behaviors (Smoll et al., 1978; Tharp & Gallimore, 1976; Yerg, 1981). Therefore, the review of the relevant literature for this paper focused on the methodology and findings of the research which has been conducted in the field of teacher education, and to a lesser extent, in the area of physical education. Although it is recognized that coaches Operate in a considerably different setting than do academic and physical education classroom teachers, it is also apparent that the procedural methods develOped and utilized in these two parallel fields of study may be highly applicable to the study of coaching effectiveness even though the results may be unique to the particular situation. Additionally, coaches as well as teachers share a common goal -- to facilitate student learning and perfonmance. Therefore, due to the lack of previous research in coaching effectiveness, the literature relating to instructional effectiveness in both academic and motor skill settings was reviewed to provide the procedural and theoretical background for the present research investigation. In this context, instructional effectiveness was defined as those behaviors which have been shown to be most facilitative of student perfonnance and/or psychosocial deveTOpment. In addition, several theoretical models were reviewed, each of which may be utilized to explain the causal mechanisms underlying the empirical link between instructional behaviors and optimal student develOpment. This literature was examined because a complete understanding of teaching and coaching effectiveness must include infonmation concerning the processes through which instructional behaviors facilitate children's growth. Finally, the results of the empirical and theoretical literature examined in these first two sections were summarized and integrated for the purpose of deveTOping a theoretical model which can be used to guide present and future research in teaching/coaching effectiveness. This model was designed in order to both predict and explain the relationship between teaching/coaching behaviors and students'/athletes' psychosocial growth. Research on Instructional Effectiveness An historical examination of the research in teaching effectiveness reveals considerable variation among researchers in their conception of what constitutes effective instruction. Such variation in the definition of the area under study is reflected both in the methodology a particular researcher utilizes to determine the correlates of effective instruction and in the validity of the research results obtained. Medley (1979) has recently identified four broad but distinct conceptual definitions of teaching effectiveness which have been used to guide research in this area. Understanding these paradigms is necessary if one is to utilize past research in teaching effectiveness to guide future investigations. The first of these four conceptual definitions assumed that the effeCtiveness of a teacher in promoting students' achievement was dependent on the personality traits of the individual teacher. This conceptual orientation generated research designed to identify those personality traits which were characteristic of the "good" teachers and which distinguished than from the "poor" teachers. Because this orientation romanticized teaching as an art rather than a science, very little useful information concerning effective instruction was obtained. The second conceptual definition of teaching effectiveness was more oriented toward teaching method or style. Specifically, teaching effectiveness was investigated in terms of the comparable effects which certain teaching techniques exerted on students' achievement. Such research compared students' gains in learning as a function of teaching method (e.g., "Open" versus "structured", "teacher-dominant“ versus "guided discovery"). Because such research did not consider the effect of the individual teacher in the administration of such methods, this approach did not produce consistent results. The third conceptual definition of teaching effectiveness is the one which has been most utilized within the past several years. Measures of effectiveness focus on actual instructional behaviors and their effect on the learning rates of students. This research is usually referred to as "process-product" research and is oriented to the observation of teachers' behavior in actual classroom situations. This methodological and conceptual approach has produced some consistent and generalizable results concerning the relative effectiveness of teachers in promoting student learning. The fourth research orientation is an extension of the process-product paradigm and is the approach advocated by leading researchers in the field today. This approach views effective teaching as the acquisition of effective instructional competencies and differs from the third in recognizing that a specific set of teaching behaviors cannot be identified which will be effective in all situations and with all students. Rather, the effective teacher has any number of behavioral competencies but utilizes the apprOpriate one dependent upon the situation. Medley (1979) suggested that research in teaching effectiveness is currently entering this fourth phase. The methodology utilized with this type of conceptual orientation is the process-product paradigm with the added stipulation that effective instructional behaviors are identified according to the particular situation for which they have been demonstrated to be effective. Several major reviewers (e.g., Br0phy, 1979; Gage, 1979; Locke, 1977) have advocated the utilization of the process-product research paradigm as the most valid and reliable method for the investigation of teaching effectiveness. For this reason, the review of the research pertaining to instructional effectiveness was limited to those research projects which have employed process-product methodologies. Specifically, this methodology utilizes measures of classroom processes (i.e., actual, observable teaching behaviors), as well as measures of educational product (i.e., assessments of student gains in achievement and psychosocial growth) to detenmine the correlates of effective instruction. While a substantial amount of process-product research has examined students' academic growth as a function of teachers' behavior, few of the reported research projects have included assessments of students' growth in both areas -- academic achievement and psychological development. Those researchers who did include both types of assessment (Evertson, Anderson, Anderson & Brophy, 1980; Good & Grouws, 1977; Peterson, 1977; Solomon & Kendall, 1976), reported that the instructional behaviors which are most effective in promoting students' academic growth are ngt the same instructional behaviors as those which are facilitative of students' psychosocial growth. For these reasons, teaching and coaching effectiveness relative to each of these product measures was discussed separately. Instructional Behaviors and Children's Achievement Although many process-product research efforts were conducted during the 1970's (see reviews by Good, 1979; Good, Biddle & Br0phy, 1975; BrOphy, Note 3), integration of the accumulated findings was difficult due to methodological inconsistencies in study design and execution. However, several large-scale, correlational research projects (Brophy & Everston, 1976; Good & Grouws, 1976; Soar and Soar, 1976; McDonald and Elias, Note 4) conducted in the last half of the decade were specifically designed to provide more rigorous and controlled testing of effective teaching behaviors. Each of these reported field studies collected data through extensive observation of classroom processes. Brophy and Evertson (1976), for example, conducted a longitudinal project using 31 second and third grade teachers from both low and high socioeconomic school districts. Regular observation of teachers' and students' behaviors was conducted over a period of two years, and teachers' attitudes and cognitions were assessed via interviews and questionnaires. Similar methodolbgies were reported by additional investigators (Good & Grouws, 1977; Soar & Soar, 1976; McDonald & Elias, Note 4; Stallings & Kaskowitz, Note 5), each of whom collected infonmation concerning actual instructional behaviors through observation of classroom activities. Product measures, assessing students' achievement gains, were also collected, 10 and each set of researchers correlated these product measures with the observed teaching behaviors for the purpose of detennining those instructional behaviors which were most highly associated with students' achievement. Although each of these investigations have differed somewhat in methodology and instrumentation, there is enough similarity among them to provide some valid and replicative infonmation relative to teaching effectiveness. Integration of findings obtained from these process-product research projects has resulted in the identification of certain instructional behaviors which seem to be most effective in promoting students' achievement in elementary classrooms. Specifically, three categories of teaching behavior have been cited by a number of reviewers (BrOphy, 1979; Gage, 1979; Good, 1979; Rosenshine, 1979) as indicators of teaching effectiveness in the elementary classroom. These categories include: (1) efficient classroom managment skills which can generally be defined as those which keep students actively and constructively involved in academic work for the greatest share of classroom time; (2) direct instructional skills which include all teaching behaviors which maintain an academic and teacher-centered focus allowing little student choice of activity; and (3) positive teacher attitudes and expectations concerning both the instructor's ability to teach as well as the students' capability to learn the skills. While these three general categories of instructional behavior have been identified as most effective in facilitating students' perfonmance, two qualifications concerning these conclusions must be 11 made. First, the identification of these correlates of effective teaching has been made on the basis of correlational evidence (i.e., an associative relationship between these behaviors and high levels of students' achievement). A causal_relationship cannot be inferred from such findings, no matter how consistently they are found. However, experimental research has recently been initiated to detennine if such behaviors can cause significant increases in students' learning. Such studies, using the effective behaviors identified in early correlational field studies, have found that elementary classroom teachers trained in such behavioral techniques are more successful in producing high academic achievement than are control teachers who do not consistently employ such “effective" behaviors (Anderson, Evertson & Brophy, 1979; Good & Grouws, 1979). This initial evidence, then, does support the contention that these three categories of instructional behaviors are most facilitative of elementary students' academic achievement. The second qualification to be made regarding the identification of these effective teaching behaviors concerns the generalizability of these findings. Ahnost all of the process-product research on which these conclusions are based, have been conducted in elementary classrooms and have primarily investigated teaching effectiveness in tenns of students' acquisition of basic academic skills. In addition, this instructional effectiveness has been determined by using students' scores on standardized achievement tests as the product measure. Therefore, generalization of these findings cannot be made to 12 instructional effectiveness in upper grades, in teaching more adVanced academic skills, or in facilitating students' attitudinal or psychological growth. In fact, process-product research conducted in academic situations other than those previously researched has demonstrated that these identified effective behaviors cannot be completely generalized. That is, several sets of researchers have demonstrated that contextual factors (factors specific to certain classroom situations) systemmatically influence the effects of certain teacher behaviors (see, Good, 1979; Peterson & Walberg, 1979; Rosenshine, 1979). Of the contextual factors identified to date, most researchers cite students' socioeconomic status (SES) as the major variable that mediates the effectiveness of certain teaching behaviors. Based on the results of two correlational research projects (Brophy & Evertson, 1976; Soar & Soar, 1972), it seems that different instructional and motivational teaching behaviors are facilitative of academic success with low and high SES students. Similarly, students' grade level and the specific academic task requirements (i.e., the type of subject and the level of cognitive demand) have been identified as factors which influence teaching effectiveness (Flanders, 1970; Soar & Soar, 1976; Brophy, Note 3). In addition to the contextual factors already identified, Brophy (Note 6) has convincingly argued that additional factors may also be influential in detennining effective teaching behaviors. Several categories of contextual variables which BrOphy has suggested may 13 mediate instructional effectiveness include teacher-related factors (attitudes, perceptions, values). student-related variables (age, sex, race, cognitive style, personality attributes), and environmental variables (subject matter, task objectives, time of year). Some of these variables have already been investigated and have been found to influence variation in teaching effectiveness (BrOphy & Evertson, 1978). Based on this evidence, it is apparent that instructional effectiveness cannot be defined as the acquisition of a number of generic teaching behaviors which will be effective in all instructional situations. For this very reason, Medley (1979) suggested that teaching effectiveness has entered into its fourth conceptual stage. That is, effective teaching is now defined as the acquisition of a variety of teaching behaviors ang_the ability to employ the particular behavior appropriate for specific teaching situations. Therefore, research in teaching effectiveness is now oriented towards the identification of effective instructional behaviors within specific contextual situations. Instructional Effectiveness in Motor Skill Settings The reported literature concerning teaching and coaching effectiveness in motor skill instructional settings has resulted in very little valid infonmation. In fact, Locke (1979) referred to research in teaching physical education as a "dismal science." 14 Certainly some of this lack of valid information can be attributed to the fact that the process-product paradigm has been only sparingly used to investigate instructional behaviors. A review of the reported research concerning instructional behaviors in motor skill areas, for example, indicates that most of the studies can be classified into two groups on the basis of the methodology employed. These include teacher training/modification studies and descriptive studies concerning exhibited teaching behavior. The first of these two groups of research is oriented toward the training and modification of teachers' and coaches' behavior. These studies attempt to show that certain instructional behaviors can be induced, either through teacher training procedures or through behavior modification techniques (e.g., Darst, 1976; Rushall & MacEachern, 1977; Rushall & SiedentOp, 1976; Rushall & Smith, 1979; Mancini, Note 7). Although these studies have consistently demonstrated that teacher/coach behaviors can be modified, they have failed to show that such instructional behaviors are associated with optimal learning or perfonnance. The behaviors advocated in such studies are based either on theoretical discourses concerning effective teaching behaviors or on research concerning effective teaching in academic classrooms (e.g., behaviors recommended by Flanders, 1970). In either case, this class of studies fails to provide empirically-based information relative to those instructional behaviors which are most facilitative of students' perfonmance. 15 The second general class of studies in the literature concerning teaching/coaching effectiveness presents descriptive data identifying the instructional behaviors which are most commonly exhibited in motor skill instructional settings (e.g., Bain, 1976; Danielson, Zelhart & Drake, 1975; Nygaard, 1975; Keane, Note 8). The Danielson et al. (1975) study was one of the most comprehensive of this category. These researchers asked 160 male hockey players, ages 12-18 years, to list the coaching behaviors most commonly exhibited by their coaches. Data analyses resulted in the identification of a number of behavioral categories which players perceived their coaches to employ most frequently in game and practice situations. While studies such as these provide a clearer picture of the types of behaviors commonly emitted by coaches/teachers during instruction, they do not provide any information concerning the effectiveness of such behaviors, as the relationship between these instructional orientations and players' perfonnance gains was not examined. It is apparent that neither of these two research approaches utilizes the process-product paradigm which is so consistently adecated in the literature pertaining to effective instruction in academic classrooms. Locke (1977), who recommended the methodology associated with the process-product paradigm, suggested that the only research which deserves to be identified as research in teaching physical education is that which includes direct observation of teaching behaviors in naturalistic situations. 16 Using Locke's definition of research in teaching effectiveness, only two lines of investigation can be found in the reported literature. The first, a rather unusual approach to the study of instructional effectiveness, was that conducted by Tharp and Gallimore (1976). Identifying then-U.C.L.A. basketball coach, John Wooden, as one of the most successful coach-teachers in the history of college athletics, Tharp and Gallimore (1976) attempted to identify the teaching behaviors commonly employed by this "master teacher." Coding and analyzing behaviors exhibited by Wooden during practice sessions, Tharp and Gallimore recorded more than 2000 acts of teaching over the course of 30 hours of observation. Data analyses revealed that the teaching behaviors coded as "instructions" accounted for 50% of the total exhibited behaviors. In addition, it was found that at least 75% of Wooden's teaching acts carried information relating to skill development -- infonmation that tended to be highly repetitive. These researchers also found that praise as a teaching behavior was rarely given by Wooden. In fact, he was coded to give twice as many "scolds" as "praise." However, the majority of Wooden's scolds were accompanied by remarks containing additional instruction and were not equally distributed to the various athletes. Although the results of this study are based on the observation of only one teacher/coach, descriptive evidence is provided concerning Specific instructional behaviors which have been associated with consistent gains in players' perfonmance. However, Tharp and Gallimore (1976) emphasized that the generalizability of this infonmation is 17 certainly limited by the case study approach, as well as by the use of elite, college-level athletes who are both highly motivated and highly skilled. Perhaps the most valid of the reported research projects in teaching effectiveness in physical activity settings was conducted by Yerg at Florida State University. Her initial study (1977) was designed to measure the effectiveness of selected instructional behaviors on the motor perfonmance of children in grades three to six. On the basis of research from motor learning which differentiates the instructional needs of learners according to skill level (Gentile, 1972), Yerg selected three teaching behaviors which she assumed would have the most impact on the perfonmance of children at a certain stage of skill deveTOpment. The skill she chose to use in this study was the cartwheel. Assuming that children in grades three to six would have already mastered the basic skill components and were at the stage of refining these previously learned patterns, she selected the three instructional behaviors which would be most apprOpriate for this stage of skill development. These instructional behaviors included clarity of task presentation, guided and supported practice, and specific, task-related instructional feedback. The frequencies of these instructional behaviors were assessed through observation of a series of simulated instructional sessions during which teachers were assigned to teach a group of children the cartwheel. In addition to measuring children's post-instructional task perfonmance, Yerg also 18 assessed the pre-instructional perfonmance of the teacher and the child. Regression analyses indicated that the only Significant contributor to post-instructional perfonmance was the child's initial task ability. However, additional analyses revealed that the amount of practice time given students was positively correlated with the level of achievement attained while the amount of time spent in teacher talk was negatively correlated with attained perfonnance. Interestingly, Yerg (1977) also found that the lesser skilled teachers (those who themselves perfonned the skill poorly) tended to emit significantly more instructional talk than those who were more highly skilled. In a subsequent discussion concerning these findings, Yerg (1981) made some important observations concerning the investigation of teaching effectiveness in physical education settings. She indicated that the instructional behaviors she had selected for investigation had been chosen on the basis of their assumed impact on the perfonnance of students who had mastered the basic skill components of the cartwheel. Initial, pre-instructional assessment of the children's cartwheel abilities, however, showed that the majority of them were actually beginners. Therefore, Yerg (1981) hypothesized that the selected teaching behaviors may have been ineffective for this skill level. She further suggested that future researchers in teaching effectiveness must recognize the influence that skill level exerts on the effectiveness of teaching behaviors. This observation coincides with the conclusions made in the literature pertaining to instructional 19 effectiveness in academic classrooms (Br0phy, 1979; Gage, 1979), that contextual factors, as well as student variables, will mediate the correlates of teaching effectiveness. In summary, additional process-product research across a wide range of contextual situations is needed to identify the correlates of effective instruction. Recently, more interest has been generated in research areas related to instructional effectiveness (e.g., behavior modification, reinforcement and feedback, modeling, and teacher training programs) in physical activity settings, but these research findings are not based on actual observation of instructional behaviors in classroom situations. Generally, the research orientation which has been predominantly utilized in the teaching/coaching effectiveness literature corresponds to Medley's (1979) second stage of research, where teaching effectiveness is defined as the implementation of certain instructional styles or methods. This research orientation does not recognize that such factors as students' age, skill level, self-concept, and achievement motivation may influence the effectiveness of such methods. Instructional Behaviors and Children's Psthosoci T Devel0pment 0n the basis of the available literature concerning teaching effectiveness, it does appear that certain instructional behaviors have been identified which are consistently associated with maximum gains in academic achievement by students, although such effective behaviors are 20 limited by the contextual Situation. Unfortunately, considerably less process-product research has been conducted to investigate the types of teaching behaviors which are most facilitative of psychosocial growth in children. The lack of research in this area has most often been attributed to the inherent difficulties in attempting to measure change in self—concept, values or attitudes (Brophy, Note 6) and to the primary emphasis in American educational systems on students' attainment of basic academic skills rather than on their psychosocial devel0pment (Good, et al., 1975). A few investigators, however, have included measures of psychosocial development in their examination of effective teaching behaviors (Good & Grouws, 1977; Peterson, 1977; Solomon & Kendall, 1976; Stallings & Kaskowitz, Note 5). Results from these process-product studies have provided about the only information relevant to the association between instructional behaviors and students' psychosocial responses. The comprehensive, observational study of classroom processes conducted by Stallings and Kaskowitz (Note 5), for example, has provided some initial infonmation relating to teaching effectiveness and the affective responses of students. These two researchers collected observational data on children's achievement-related behaviors in the classroom and correlated these with observed instructional behaviors. Regression analyses indicated that the frequency with which students exhibited such positive behaviors as independence, task persistence, c00peration, and question 21 asking was significantly related to such instructional behaviors as individualized teacher attention (one-on-one instructional situations), adult response to student questions, and frequent and friendly teacher interaction with individual students. Thus, this research revealed that the instructional behaviors which were most closely associated with higher frequencies of achievement orientations in students can be characterized as indirect instruction. The effectiveness of indirect instructional behaviors in facilitating positive attitudes in students was also demonstrated in several other research studies (see reviews by Flanders, 1970; Peterson, 1979; Rosenshine, 1973), although the number of process-product research projects on which this conclusion is based is considerably smaller than that used to identify the correlates of effective instruction in tenns of students' academic achievements. A second category of instructional behaviors which has been examined in relation to its effect on students' attitudes and affective responses includes teachers' reinforcement patterns. C00per and Good (in press), for example, reported on the association between teachers' feedback and students' self-efficacy scores. They found that teacher criticality was negatively correlated with students' self-efficacy, while frequency of public, teacher-initiated interactions was positively correlated with such affective measures. Flanders (1969) and Rosenshine (1973) both concluded that teacher praise is significantly related to positive student attitudes and that critical teacher statements are negatively correlated with such attitudes. 22 Dunkin and Biddle (1974), however, caution that teachers' praise is not indiscriminately associated with positive affect, but that the use of discriminatory, perfonmance-contingent, positive instructional reinforcement may be influential in increasing students' attitudes. Furthennore, a few recent writers have indicated that the effectiveness of a teacher's praise may be dependent both on the Situational context (Brophy & Evertson, 1978) and on the teacher's intent in administering such reinforcement (Brophy, Note 9). Therefore, the actual value of a teacher's use of praise in facilitating positive attitudes among students has not been accurately measured. Because so little process-product research has been conducted to measure the association between instructional behaviors and students' psychosocial growth, very little consistent information is available. However, on the basis of initial evidence, it seems that the teaching behaviors which are most conducive to positive student attitudes and cognitions are those which are characteristic of the indirect style of teaching (e.g., teacher warmth, praise, individualized instruction). This is in contrast to those teaching behaviors identified as effective in promoting academic achievement (i.e., direct instructional techniques). Consequently, some reviewers (Peck, 1976; Brophy, Note 4) have suggested that this opposition may eventually result in the necessity for teachers to engage in "trade-offs" (e.g., being forced to choose between promoting students' academic achievement or their psychosocial growth). More research is needed, however, before the conclusion can be made that these goals are incompatible. 23 Additionally, future process-product research designs should provide for the Shmultaneous collection of numerous product assessments (i.e., measuring students' gains both in academic achievement as well as in psychosocial devel0pment) and the analyses of such results through multivariate procedures. Psychosocial Growth in Motor Skill Settings The importance of the athletic coach's role in facilitating positive psychosocial growth in young athletes has received considerable emphasis in the Sport psychological literature (Singer & Gerson, 1980; Martens, Note 1; Gould, Note 2). However, very little empirical research has been conducted to Specifically identify effective instructional behaviors in motor skill settings. Martinek (1981) has recently reported the results of a number of studies designed to investigate his theory that teachers' expectations influence both their actual instructional behaviors as well as the subsequent growth of students. In each of these studies, significant differences were found between the behaviors which physical education teachers exhibited toward their high ability as compared to their low ability students. Specifically, Martinek and Johnson (1979) demonstrated that students who were expected by teachers to attain the highest levels of physical achievement were given more praise and supportive encouragement than their low expectancy peers. 24 Based on such expectancy effects, Martinek and his associates then examined the relationship between these differential instructional behaviors and various psychosocial measures taken on both high and low expectancy students. It was found that high expectancy students, who received more praise and encouragement from their teachers, develOped greater gains in self-concept measures over the course of a 16-week semester than did their low expectancy classmates (Martinek & Johnson, 1979). It was also reported, however, that a significant teacher by expectancy group interaction occurred, indicating that expectancy group self-concept differences were specific to only three of the five classes. Therefore, the relationship between teachers' expectations and students' self-concept cannot be generalized to all classrooms and all teachers. In a related study, Martinek (1980) also found that third and fourth grade children's expectations concerning their own physical Skill ability could be predicted by assessing both the child's self-concept as well as the teacher's expectations concerning students' ability. Although Martinek's study was extremely limited in scape (one teacher and 63 students). his results suggested that teachers' expectations account for the greatest amount of the variability in young children's self-expectations concerning their motor perfonnance. Based on the results of these studies, Martinek (1981) has contended that teachers' expectations concerning students' ability may influence the instructional behaviors exhibited towards individual students. This differential behavior, in turn, affects both the 25 perfonnance and psychosocial responses of young children. Although there is sufficient evidence in the teacher education literature to support his proposal, the research findings revealed in his reported series of studies are based only on correlational data analyses, and a causal relationship between teaching behaviors and students' cognitions has not been established in the physical education classroom. A series of studies recently conducted by Smith and Smoll and their associates, however, has provided evidence to show that coaching behaviors do influence the cognitions and attitudes of young male athletes. This project began with the deveTOpment of the Coaching Behavior Assessment System (CBAS) (Smith et al., 1977). This instrument was designed to measure the behaviors of coaches towards their players in both contest and practice Situations. The CBAS consists of a number of behavioral categories which were derived on the basis of empirical observation of coaches as well as on theoretical principles identified in the Sport psychological and social psychological literature. Following the deveTOpment and validation of the CBAS, Smith and Small and various associates conducted a two-phase investigation examining the relationship between coaching behaviors and players' attitudes and self-concept (Smith, Smoll & Curtis, 1979; Smoll et al., 1978). In the project's first phase, the researchers observed and categorized the coaching behaviors of 51 Little League coaches over the course of an entire playing season. At the end of the season, the 542 male players, ages 8 to 15, were interviewed and their attitudes 26 towards their teammates, the sport, and their coaches were measured. In addition, both general and athletic self-concept measures were taken. Data analyses revealed that those coaches who exhibited the highest frequencies of behaviors categorized as "technical instruction," "reinforcement," and "mistake-contingent reinforcement" were evaluated more positively by their players than those coaches who scored low in these categories. Furthermore, players of these highly reinforcing and instructive coaches attained significantly higher post-season self-esteem scores and had more positive attitudes toward participation than did players of coaches who did not consistently exhibit these behaviors (Smoll et al., 1978). In Phase 2 of this project, these researchers (Smith et al., 1979) utilized an experimental approach by manipulating the behavior of some of the coaches and then measuring the subsequent effects of this manipulation on players' attitudes and perceptions. An educational program, administered to the experimental group of coaches, utilized three behavior-modification techniques, and was designed to teach coaches to exhibit the effective coaching behaviors identified in the previous study. Following this instruction period, behavioral data was collected in a manner very Similar to that done in Phase 1. Data analyses revealed that significant differences were evident between the coaching behaviors of the experimental and the control groups, with the experimental coaches exhibiting more of the desired behaviors (encouragement, reinforcement, technical instruction). The athletes who played for the experimental coaches also rated their coaches 27 significantly higher in knowledge and teaching technique and expressed a greater degree of enjoyment than did players of control group coaches. Finally, children who played for the trained coaches evidenced Significant increases in self-esteem scores over the course of the playing season while players of control coaches did not Show comparable changes. In summary, the results of this multi-year project, encompassing both observational and experimental research designs, demonstrated that Specific coaching behaviors are related to children's attitudes toward participation as well as their attraction towards coaches and teammates. In addition, the level of the athletes' self-concept was found to be influenced by Specific types of behaviors exhibited by the coaches. Thus, this series of studies provided support for the theoretical position that coaches' behavior can affect the psychosocial deveTOpment of young athletes. Instructional Effectiveness: Summary and Research Implications On the basis of infonmation obtained from a review of the research in instructional effectiveness, it is apparent that the behaviors exhibited by instructors in both academic and motor Skill instructional settings can influence the course of student growth in academic, physical, and psychosocial areas. Theoretically, then, it should be possible to identify those instructional behaviors which will most effectively facilitate the achievement perfonmance and psychosocial deveTOpment of young children. Numerous research projects conducted in 28 both academic and athletic settings, however, have shown that contextual factors such as student type, subject area, and grade level mediate the effectiveness of many instructional behaviors. Therefore, although a few generic instructional behaviors have been identified as effective across a variety of teaching situations, the majority of the effective instructional behaviors will have to be identified in situation-specific contexts. This specificity of instructional effectiveness implies that the course of future research lies in the empirical investigation of process-product relationships in specific contexts. Gage (1979) theorized that the identification of effective instructional behaviors will result in a hierarchial structure. At the base of the structure will be those few general behaviors which are effective across all contextual situations. However, the next hierarchical level will contain those instructional behaviors which are most effective in a more limited context (i.e., at a certain age level). Each successive hierarchical level will consist of those behaviors which have been identified as effective in increasingly more specific situations. Thus, the t0p of the structure will include those instructional behaviors which have been designated as effective for certain students in certain grade levels for a Specific academic skill. The identification of such a hierarchy of effective teaching behaviors will require considerable and continued empirical field work. A particular methodology apprOpriate for research in teaching effectiveness has been outlined by Yinger (Note 10) and has 29 been tenmed "grounded theory research." This methodology requires the researcher to conduct process-product, observational research in a specific contextual Situation. On the basis of these findings, the researcher begins to identify effective instructional behaviors and then tests whether such behaviors are generalizable to similar contexts (i.e., through replicative studies). Finally, theories relative to instructional effectiveness can be drawn. This approach to research in teaching, then, emphasizes the develOpment of theory through field-based research rather than "proving" theory through empirical investigation. As Gage (1979) has recognized, the task of identifying the correlates of effective instruction is fonmidable considering the variety of contextual factors which influence teaching effectiveness. Nowhere is this more true than in the motor skill instructional setting where a very small research base has so far been established. Locke (1977), in his editorial emphasizing the lack of research in teaching effectiveness, also recognized that the task facing researchers is monumental. To make the task considerably easier and less prone to “trial and error" influences, Locke suggested that physical educators and researchers should explore the literature in such related areas as sociology and psychology of learning to identify existing theories which may be used to guide process-product research in teaching physical education. Locke's recommendation concerning the use of theory-guided research was proffered with the cautionary note that such theories must guide research which is ultimately based on observation 30 of teachers in naturalistic Situations. This approach to the study of teaching motor skills seems a most logical and reasonable way to begin the search for effective instructional behaviors. Therefore, in the following section, some existing theories from the social and developmental literature were outlined. Each of these theories was chosen for its relevance to the study of instructional effectiveness in motor skill settings. 31 Theoretical Models: EXplaining Instructional Effectiveness In the previous section it was demonstrated that the instructional behaviors utilized by teachers and coaches influence students' academic and psychosocial develOpment. This link between instructor-related variables and student develOpment, however, has been demonstrated only through empirical methods. That is, correlational studies (Br0phy & Evertson, 1976; Good & Grouws, 1977; Soar & Soar, 1976; Yerg, 1977) Showed that certain teaching behaviors were associated with maximal gains in student learning. More recently several experimental studies (Anderson, Evertson & Brophy, 1979; Good & Grouws, 1979; Smith et al., 1979) have indicated that manipulation of instructional behaviors induces corresponding changes in students' attitude and/or achievement. In contrast to such demonstrated empirical relationships, theoretical relationships have not been generated to explain the link between classroom processes and educational product. AS several reviewers (Cooper, 1979; Good, 1981; Br0phy, Note 6) have indicated, little is known about why or how instructional behaviors influence student growth. The establishment of causal mechanisms to explain empirically-demonstrated links between instructional behaviors and student growth would not only contribute to the explanation of these relationships but would also serve as a guide for future research in instructional effectiveness. Moreover, unique theoretical mechanisms do not necessarily need to be generated to explain these relationships. 32 AS Locke (1977) has suggested, utilization of existing theories from related fields of study may result in more valid research in teaching effectiveness. Two such theoretical fonnulations are available which may apply to the study of instructional effectiveness in motor skill areas. One of these formulations can be found in the teacher behavior literature and suggests that teachers' expectations influence classroom processes to the extent that the course of student growth is affected. The other set of theories has been developed in the social and developmental psychological literature and postulates that differential rates of student achievement can be attributed to their individual perceptions of control. These perceptions, then, become the link between teachers' behavior and students' perfonnance. Expectation Theories Expectation theories postulate that the perceptions which teachers hold concerning the achievement potential of each of their students are communicated to individual students through differential teaching behaviors. Such communicated expectations influence students to behave in ways which confonm to the teacher's original expectations. Consequently, the expectations of teachers are predicted to influence students' perfonmance. Two specific models, each of which has Specified in sequential observable steps the relationship between teachers' expectations and students' growth, have been deveTOped. 33 Model for Academic EXpectancy Effects The first of these models was advanced by Brophy and Good (1974) and was utilized as a guide for the initiation of subsequent research concerning expectancy effects in the classroom. This model consists of the five sequential steps outlined in Table 1. TABLE 1. Model for Academic Expectancy Effects STEP 1: STEP 2. STEP 3. STEP 4. STEP 5. Teachers fonm differential expectations regard- ing the achievement potential of individual students. Teachers' behavior in the classroom towards individual students reflects these differential expectations. Differential instructional behaviors convey to the student the type of behavior and achieve- ment expected from him or her. Because this information is available to the student, it influences his or her motivation, self-concept, and level of aspiration. If such differential teaching behavior is consistent over time, and if the student does not resist, his or her subsequent perfonnance and behavior may confonm to the teacher's expectations. Brophy and Good, 1974 34 DeSpite the fact that this model outlines in clearly observable steps, the sequential connection between instructional expectations and students' subsequent classroom performance, most of the research conducted to test its behavioral predictions has focused on Steps 1 and 2. However, the sequential nature of the model was used to summarize the research pertaining to expectancy effects in instructional settings. Step 1: Instructional Expectations. Much of the early research in relation to expectancy theory was conducted to test whether expectations are viable in academic Situations. The results reveal that teachers do fonm expectations about the future academic achievement of their students and that these expectations are fonmed on the basis of infonmation from a variety of sources. These sources include such student-related variables as socio-economic status (Rist, 1970; Goodwin & Sanders, Note 11), sex (Kehle, 1974; Palardy, 1969), physical attractiveness (Clifford & Walster, 1973; Dion, 1972), and attained scores on standardized tests (C00per, 1979; Willis, 1972). Although teachers use such pre-observational information as SES, race, sex, and previous academic achievement as cues for predicting achievement potential, it is equally true that initial classroom contact with students causes the teacher to form first impressions concerning academic ability. Willis (1972) for example, asked first grade teachers who were unfamiliar with their in-coming students to rank order then in tenms of expected achievement after only one week of school. Teachers were found to use students' attentiveness, 35 self-confidence, and independent work habits as cues for the fonmation of ability expectations. It was also found that teachers used race, physical attractiveness, and physical size as infonmational cues, but to a lesser degree than the other factors. Finally, in another investigation, Long and Henderson (Note 12) found that students' attentiveness and attained scores on a school readiness tests constituted salient cues for teachers' expectations. The results of these studies indicate that teachers utilize infonmation from a number of sources, including such student-related variables as membership grouping (sex, SES, race), physical characteristics (size, attractiveness, dress), and attained perfonnance scores to make judgments concerning their students' abilities. But, the specific antecedents of teacher expectations may be very difficult to detennine. Finn (1977) has suggested that naturally-fanned teacher expectations cannot be simply explained by examining only one or two specific infonmation sources but are probably based on a complex interaction of factors and influenced by teacher as well as student characteristics. Assuming that teachers do fonm expectations concerning the academic potential of their students on the basis of infonmation from many sources, the next logical question concerns the stability and accuracy of such teacher perceptions. Willis (1972) examined this issue by measuring teacher expectations after only one week of classes, and then repeating this procedure several weeks later and again at the end of the semester. She found that the stability of teachers' 36 expectations was quite high over the course of the semester (correlations ranging from .56 to .86). In addition, teachers' expectations, measured at all three time periods, were also Significantly correlated with actual student test scores (.56 to .80), although these predictions increased in accuracy after teachers were provided with readiness test scores. Similar estimates of the stability of instructional expectations were reported by Evertson, Brophy, and Good (Note 13). Instructional expectations have also been examined in motor skill settings. Two laboratory studies (Hatfield & Landers, 1978; Rikli, 1976) revealed that expectations concerning children's perfonnance could be induced in "observer-evaluators" by providing them with false infonmation relating to the performer's motoric ability. These manipulated expectations influenced the performance rating which the observer-evaluator assigned to individual children, with high expectancy children receiving correspondingly higher perfonnance ratings. Expectancy effects have also been demonstrated in competitive athletic field settings. Scheer and Ansorge (1975) found that gymnastics judges used the order of performance in a meet (i.e., whether a performance occurred early or late in competition) as a source of infonmation by which to evaluate the perfonmance. It was suggested that the common practice of placing highly skilled perfonners later in the match order induces judges to expect better performance 37 from later competitors. Such expectations, then, influenced performance ratings. Student gender has also been found to be a viable factor in the fonmation of teachers' expectations concerning students' physical achievement potential. Crowe (1977) found that both male and female teachers expected significantly better performance from boys than from girls in their junior high physical education classes. However, Martinek and Johnson (1979) found that student gender had no effect on teacher expectations in third grade physical education classes. It may be that physical education teachers' expectations are influenced by the students' gender only at higher age levels and in certain sport Skill areas. In summary, it is apparent that academic and physical education teachers do fonm differential expectations concerning the academic and physical ability of individual students, but this does not automatically imply that such differential expectations adversely affect the student. If teachers' expectations are used to design and implement individually-based learning experiences, they can, in fact, assist the teacher to provide Optimal instruction for all students. The model advanced by Brophy and Good (Table 1), however, implies that such fonned expectations result in differential teacher- student interactions in the classroom which actually enhance, or at least maintain, disparities in students' achievement. Step 2: Differential Instructional Behaviors. The link between an instructor's expectations and his or her teaching behavior has been 38 more extensively studied than any Of the other components of the expectation model. Literature reviews by a number of writers (BrOphy & Good, 1974; COOper, 1979; Dusek, 1975; Good, 1981) have shown that teachers fonm expectations concerning the academic abilities Of students and that these expectations influence the interactions that teachers have with individual students. These expectancy effects are not universally found among all teachers, however. All teachers may fonm expectations concerning students' perfonnance, but not all teachers allow these expectations to influence their classroom behavior. Rosenthal's (1976) meta-analysis of the literature relating to expectancy effects gives some indication of the variability among teachers in relation to their susceptibility to expectancy effects. Of the expectation studies which reported individual teacher data, 70% of the participating 340 teachers showed Significant differential treatment Of high and low expectancy students. Some of the sampled teachers, then, did not allow their expectations to influence their behavior with individual students. Recognizing that not all teachers exhibit such eXpectancy effects, a survey of the literature indicates that many instructors do Show differential behaviors toward high and low expectancy students. It would be difficult to summarize all the process variables (teacher-student interactions) which researchers in the last decade have Shown to be associated with expectancy effects. However, Rosenthal (1974) has identified four categories of instructional behaviors (climate, input, output, and feedback) which can be used to 39 classify the teaching behaviors that have been associated with expectancy effects across a large number Of studies. First, several studies have reported that teachers tend to provide a warmer socioeconomic climate for their high, as compared to their low, eXpectancy students. This wanmer climate has been found to be facilitated by such non-verbal instructional behaviors as smiling, head nodding, and leaning towards the student (Chaikin, Sigler & Derlega, 1974; Kester & Letchworth, 1972; Page, 1971). Secondly, a number Of studies suggest that teachers provide more input, in tenms Of learning material for high expectancy as compared to low expectancy students. That is, teachers have been found to ask more difficult questions of the high expectancy students (Mendoza, Good 2 Brophy, Note 14), to allow bright students a longer time to respond with an answer (Rowe, 1974), and to persist longer in attempting to extract a correct answer from the bright student (BrOphy & Good, 1970). These studies Show, then, that the quality of the individual teacher-student interaction can be affected by the teachers' expectations. The third factor, verbal output, is defined as the frequency Of interactions between teachers and individual students. The majority of studies indicate that high expectancy students will seek more interactions with the teacher than do low expectancy students (see Brophy & Good, 1974). However, in studies which analyze teacher-initiated contact, conflicting expectancy effects have been found. Some teachers tend to initiate more contact with high 4O expectancy students (Kester & Letchworth, 1972; Good, 1970), while others interact more frequently with low expectancy students (Mendoza, Good & Brophy, Note 14). It appears, then, that teachers' expectations influence the frequency of teacher-student interactions, but there are differences between teachers in the direction Of the effect. As indicated by Brophy and Good (1974), these differences could be a function of grade level, as well as other situational factors such as teaching style and/or class structure. Finally, considerable research has been conducted to detenmine whether teachers' reinforcement patterns (evaluative feedback) towards individual students are influenced by their expectations. Although these studies have differed considerably in methodology and instrumentation, they consistently show that teachers tend to provide more praise per correct response for high expectancy students and more criticism per incorrect response for low expectancy students (BrOphy & Good, 1970; Cooper & Baron, 1977; Firestone & Brody, 1975; Meichenbaum, Bowers & Ross, 1969), thus demonstrating that the evaluative feedback that teachers provide to students may be dependent on their expectations concerning the students' ability. This brief summary of the research pertaining to expectancy effects in the classroom indicates that the quantity as well as the quality of individual teacher-student interactions occurring in the classroom may reflect the teachers' expectations concerning individual students. It is important to note, however, that these four categories represent a summary of the preferential behaviors which teachers across 41 a variety of studies have been found to exhibit. A comprehensive meta-analysis by Smith (1980) provides more Specific infonmation relative to the strength of the demonstrated expectancy effects as investigated in a variety of instructional situations. This analysis revealed that a modest relationship exists between teacher expectations and teacher behavior. Specifically, high expectancy students were consistently found across all studies to receive more learning Opportunities while low expectancy students, in contrast, were ignored more often than their classmates. In regard to the effect of teachers' expectations on product measures, Smith reported that student achievement and affect were related to teachers' expectations but few expectancy effects were found relative to students' 10 scores. Within the past few years, several research attempts have been initiated to apply these expectation theories and procedures to the study of expectancy effects in physical education classrooms and athletic settings. The results of these studies suggest that the same processes identified in the academic setting are also viable in motor skill instructional settings. For example, Crowe (1977) asked four junior high physical education teachers to rank their students according to expected physical achievement potential. Observation Of teacher-student interactions showed that high achievers received significantly more affinnation and praise than did low achievers and were generally treated more warmly and given greater Opportunity to respond to teacher-initiated contacts. Such tendencies on the part of the instructor to give more praise and encouragement to high expectancy 42 students were also found by Martinek and Johnson (1979) in elementary school physical educators. Only one study has been reported in the literature which examines expectancy effects in the athletic setting. Rajeski, Darracott, and Hutslar (1979) Observed individual coach-athlete interactions in practices and game settings, using 71 children and 14 coaches in a youth sport athletic league. Comparison between high and low expectancy athletes, in tenms of coach-athlete interactions, revealed two significant findings. First, high expectancy athletes received more reinforcement from their coaches than did low expectancy athletes. Secondly, coaches also showed a significant tendency to provide more general technical instruction to low expectancy athletes as compared to their high expectancy peers. Although researchers have been slow to investigate teacher behavior in physical education/athletic settings, initial evidence indicates that expectancy effects exist and do influence the quantity and quality of the teacher—student interactions in the gymnasium. It seems likely that the expectation-perfonmance process in motor skill settings may parallel that found in the literature pertaining to expectancy effects in the academic classroom. Step 3, 4, and 5: Student Growth. The last three steps in the model proposed by BrOpny and Good (1974) postulate that the differential behaviors exhibited by instructors influence the academic and psychosocial growth of the students. This link between instructional behaviors and students' achievement-related growth has 43 been discussed at length in the previous sections. Enough research does exist to Show that the behaviors of teachers and coaches influence the course Of student growth. What has net been demonstrated, however, is that the differential instructional behaviors which are attributable to expectancy effects actually influence the academic and psychosocial growth of children. The independent links between expectations and instructional behaviors, as well as those between instructional behaviors and students' responses have been established, but the entire causal chain has not been adequately examined. While the relationships implied by steps 3, 4, and 5 have not received much investigatory attention, considerable research interest has been paid to examination of the direct link between Step 1 (teacher expectations) and Step 5 (student achievement). Interest in this relationship was generated by a much publicized study conducted by Rosenthal and Jacobson (1968) apprOpriately titled, "Pygnalion in the Classroom." These investigators induced positive but false expectations in classroom teachers concerning the academic potential of certain of their students. When such students exhibited significantly higher gains in academic achievement over the course of the school year than did their classmates, Rosenthal and Jacobson concluded that these gains were due to the expectations induced by their teachers. Following the publication of this study, many investigators began to examine whether teachers' expectations can actually bias student perfonmance. However, extensive reviews of this research by BrOphy and Good (1974) and Dusek (1975) revealed considerable inconsistencies in 44 the reported literature concerning the validity of expectancy effects in the classroom. These reviewers indicated that teachers' expectations can influence student achievement, but that expectancy effects cannot be adequately investigated without also assessing classroom processes. That is, the relationship between teachers' expectations and student perfonmance cannot be reliably assessed unless it can be demonstrated that high and low expectancy students do receive differential educational experiences in the classroom. The Brophy and Good model for the investigation of teacher expectations in the classroom was based on this valid Observation. However, as Good (1981) indicated in his recent review of expectancy effects, very little research has been generated to assess the relationship between all facets of this model. Summary While considerable research has been initiated to investigate expectancy effects in instructional situations, the influence which such expectations exert on students' growth has not been clearly detenmined. It seems apparent that teachers' expectations do affect their instructional behavior in the classroom. However, the causal link between these differential instructional behaviors and the disparity in the performance of low and high expectancy students has not been adequately assessed. COOper (1979) suggested that this relationship has been difficult to determine because a causal mechanism has not been identified to explain this link. The model Specified by BrOphy and Good (1974) postulates that differential instructional 45 behaviors affect student achievement both directly and indirectly. As a direct effect, high expectancy students may actually be provided with a better Opportunity to learn (i.e., higher performance standards set, more difficult materials given, more Opportunity for input provided). Indirectly, teachers may communicate differential expectations Of students. Low expectancy students who perceive the teacher's lower perceptions of their academic potential may exert less effort after failure, accept failure more quickly, and subsequently aSpire to lower achievement standards than their high expectancy classmates. However, the sequential explanation of this relationship does not specify exactly how students' perceptions of teachers' expectations actually influence their performance. Re-formulation of the Expectation Model Cooper (1979), in an effort to identify more clearly the causal mechanisms underlying the expectation process, has reformulated the Brophy and Good model. Drawing on principles derived from the social psychological literature, COOper (1979) hypothesized that the concept of personal control can be used to explain much of the variation in human behavior, including that exhibited by both students and teachers in instructional settings. The Expectation Communication Model consists of five sequential items and is outlined in Table 2. Cooper and his associates (COOper & Good, in press) have begun research efforts designed to validate the relationships between the model components. However, in general, these efforts have met with only limited success, particularly in regard to the influence which 46 TABLE 2: Expectation Communication Model STEP 1. STEP 2. STEP 3. STEP 4. STEP 5. Teachers fonm differential expectations concerning students' perfonmance potential. These fonmed expectations, along with classroom contexts, influence teachers' perceptions of control over students' learning behavior. Teachers' perceptions of control influence instructional behavior towards individual students in the classroom. Because teachers perceive less control over the performance of low expectancy students, the teacher tends to discourage interactions with such students in low control situations (i.e., in public, student-initiated Situations). Conversely, the teacher encourages such Situational interactions with high expectancy students because he or she perceives greater control over their perfonmance. Therefore, teachers give significantly different feedback to low than to high expectancy students. These differences in teachers' feedback influences the students' own perception of control concerning their perfonmance outcome. Low expectancy children receive significantly more non-contingent feedback which reduces their perception Of personal control. In contrast, high expectancy students receive perfonmance-contingent feedback because teachers do not need to use feedback to control public interactions. Low perceptions of personal control result in decreased achievement task effort, persistence, and performance. COOper, 1979 47 teachers' perceptions of control are predicted to exert on their classroom behavior. In commenting on this lack Of support, Good (1981) observed that teachers' interactional contacts with individual children in the classroom are influenced by any number of teacher beliefs and attitudes which may be specific to the individual child or may reflect a more generalized attitude concerning teaching and children. Therefore, Good suggests that more precise information might be Obtained if a greater number of instructional attitudes, in addition to teacher perceptions of control are measured. Although Cooper's theory may not be entirely functional, one facet of the model -- that pertaining to students' perceptions of control -- does deserve further investigation. COOper's model hypothesizes that differential teacher feedback to individual students may influence students' perceptions of perfonmance control. Moreover, Cooper and Good (in press) have demonstrated that aspects of teachers' feedback to students are associated with students' perceptions concerning their control and academic success and failure. Therefore, to better understand how instructional feedback influences student achievement, a second set of theoretical fonnulations focusing on the deveTOpment and maintenance of an individual's perceptions of control in achievement situations was also examined. Perceptions of Control Perceptions of control can best be defined as the extent to which an individual perceives him or herself to be a causal agent in the 48 effort-outcome sequence. In an achievement setting, them, the individual who perceives high levels of personal perfonmance control assumes that effort and outcome will co-vary (e.g., "If I try hard, I will succeed."). In contrast, lower levels of personal control result in perceptions of independence between effort and outcome (e.g., "If I try hard, it won't make any difference."). Two recent developments in the literature concerning children's perceptions of perfonmance control may be used to provide the necessary causal link between instructional behaviors and subsequent growth by children in both achievement and psychosocial areas. The first research development links children's perceptions Of control to various achievement behaviors. This literature implies that the degree to which the child assumes responsibility for his or her academic success or failure may predict many other achievement-oriented behaviors, such as success expectancy, increased effort and persistence at a task. Secondly, evidence also exists to show that children's perceptions of control can be influenced or affected by the feedback which adults provide relative to children's achievement task perfonmance. While initial research efforts in these areas seem promising, both of these issues need to be examined for their applicability to instructional effectiveness. Perceptions of Control and Achieving Behaviors Research from the social psychological literature has consistently revealed an associative relationship between an individual's perceptions of personal control and various achieving behaviors. 49 Specifically, an internal perception of control (i.e., the belief that effort and outcome covary) has been correlated with high levels of achievement orientation, whereas low levels of achievement motivation were found to be associated with lower internal perceptions of performance control. Further evidence concerning the relationship between perceptions of control and achievement-oriented behaviors comes from the psychological literature, which implies that the deveTOpment of an individual's perception of control is associated with his or her cognitions concerning the causes of achievement task success and failure. That is, attribution of successful perfonmance to ability (an internal cause) induces the individual to assume that he or she is personally responsible for that success and to expect success in future achievement attempts, whereas attribution of success to luck or task ease reflects an external perception of outcome control and low expectancy of future success (Weiner, 1974). Similarly, in the case of failure, ascription to lack of effort is associated with the belief that perfonmance failure is still under personal control, and success can be achieved in future performance attempts (McMahan, 1973; Weiner, Nierenberg & Goldstein, 1976). Attribution to lack of ability, however, induces the individual to perceive that he or She cannot avoid future failure no matter howlnuch personal effort is exerted (Covington & Omelich, 1981). This association between attributional patterns and perceptions of perfonmance control has been demonstrated in the research pertaining to 50 learned helplessness, a state of depression arising from an individual's perception that his or her response will be ineffectual in obtaining reinforcement (Abramson, Seligman & Teasdale, 1978). The most recent theory pertaining to the develOpment of learned helplessness suggests that attribution of continued failure at an achievement task to lack of ability (e.g., "trying does not work because I lack ability“) is a causal factor in the develOpment of feelings of helplessness (i.e., low perceptions of personal control) (Abramson et al., 1978; Covington & Omelich, 1981). Such perceptions of helplessness have been shown to result in decreased effort and persistence at the particular achievement task, in addition to significant decreases in the individual's self—concept (see review by Abramson et al., 1978). This relationship between perceptions of control and various achievement behaviors has also been examined in academic settings. Dweck and Reppucci (1973), for example, found that children who tended to "give up" following achievement task failure could be identified by their low scores on the Intellectual Achievement Responsibility (IAR) test (Crandall, Katovsky & Crandall, 1965). That is, children who scored low on items assessing personal responsibility for academic success and who attributed failure to lack of ability were significantly more apt to respond to failure with decreased effort than those children who reported higher levels of personal responsibility for academic success and who ascribed failure to lack of effort. Similarly, Johnson (1981) has recently demonstrated that children's 51 perceptions concerning the degree to which they were responsible for achievement failure were significant predictors Of their academic self-concept. In general, these studies do provide support for the contention that children's cognitions concerning the degree to which they can control academic outcome influences such achievement-oriented behaviors as task persistence, increased effort, high success expectancies, and increased task perfonmance. Perceptions of Control and Instructional Behaviors Based on this demonstrated association between children's perceptions of control and their subsequent academic behavior, a number of writers have begun to emphasize the need for teachers to implement instructional strategies designed to increase the individual student's perception of control in academic achievement areas (Covington & Beery, 1976; Thomas, 1980; Weiner, 1976). However, the specific instructional behaviors which may be most effective in facilitating children's perceptions of internal control have not been identified. Nevertheless, there is some infonmation available which suggests that students' cognitions concerning perfonmance control in instructional situations may be highly influenced by the infonmational and evaluative feedback which they receive from significant adult figures. First, results from laboratory studies pertaining to the phenomenon of learned helplessness indicate that non-contingent reinforcement (reinforcement given in a random manner) can be used to induce the perception Of independence between effort and outcome in 52 subjects (see reviews by Abramson et al., 1978; Maier, Seligman & Solomon, 1969; Seligman, 1975). Similarly, Dweck and Reppuci (1973) showed that behaviors associated with learned helplessness (e.g., decreased perfonmance, lessened personal responsibility for perfonmance outcome, decreased expectancy for future success) could be induced in some children by subjecting them to continued and non-contingent failure at an achievement-oriented task. Dweck (1975), however, also found that this process could be reversed. That is, children identified as learned helpless (those who perceived academic failure to be insunmountable), could be trained by an adult to assume responsibility for failure through attribution of such failure to lack of effort. Such a training program resulted in significant increases in actual academic perfonmance as well as greater persistence following failure. Although the contingency of reinforcement administered by adult evaluators has been shown to influence students' perceptions of control, other instructional behaviors have also been associated with such perceptions. COOper (1977) reported that students who received the greatest amount of criticism in classroom situations exhibited lower perceptions of effort-outcome covariation than their peers who received less teacher criticism. Although there is some controversy regarding the effects Of teachers' praise and criticism on student growth (Brophy, Note 9), Cooper's results indicated that the reinforcement patterns used by teachers need to be further investigated 53 in regard to their influence on students' perceptions of perfonmance control. Finally, some evidence exists to show that the content Of teachers' evaluative feedback may also influence students' perceptions of control. Dweck, Davidson, Nelson, and Emma (1978), who were interested in sex differences in achievement-oriented behavior, observed actual classroom events for a period Of time and recorded the content and contingency of the evaluative feedback given by teachers. These researchers found that male and female students received equal amounts of failure feedback; however, the boys' failure was more often accompanied by attributions to lack of motivation whereas girls' failure was almost exclusively attributed to lack Of competence or ability. Dweck et al. (1978) used this evidence to theorize that teachers' evaluations in the classroom contribute to the lower academic achievement orientation of females and also explains the greater tendency of girls to exhibit learned helpless behaviors in academic situations. Additional research to support this contention has not yet been reported, although Cooper (1979) indicated that teachers' attributional feedback may be important in determining instructional behaviors which influence children's perceptions of control. Summary Recent evidence from the literature in social and develOpmental psychology reveals that the behaviors exhibited by significant adult evaluators do influence children's perceptions of perfonmance control. Specific adult behaviors identified which have been found to be related 54 to children's perceptions of control include: (a) contingency of reinforcement; (b) degree of teacher praise and criticism; and (c) attributional ascriptions for student perfonmance. Enough evidence also exists to Show that individuals' perceptions Of control are associated with various achievement-oriented behaviors. This evidence provides support for the contention that the link between instructional behaviors and the academic, physical, and psychosocial growth Of young children may, in large part, be explained by perceptions of control. That is, the instructional behaviors exhibited by teachers and coaches may influence the individual child's perception of performance control in a particular academic or athletic area. This perception, then, determines the child's subsequent behavior in future task perfonmance, as well as his or her cognitions concerning personal ability and efficacy. These causal links between adult behaviors. children's perceptions of control, and positive psychosocial growth have been specified by Harter (1981) in her recently fonnulated model explaining the develOpment of achievement motivation in children. Harter has designed a three-stage develOpmental model based on White's (1959) theories of competence motivation. The third stage Of Harter's Model identifies the child's perception of control as the middle link in the chain between adult evaluation and various correlates of children's motivation in achievement task situations. Thus, this model draws together the constructs just identified in the previous section, and examination of the model may be particularly useful in providing a 55 theoretical framework for investigations designed to identify those coaching behaviors which will facilitate athletes' psychosocial growth. Harter's DevelOpmental Model Harter's (1981) major concern in fonnulating this develOpmental model of competence motivation was to provide an explanation for the differential levels of motivation exhibited by children in achievement situations. That is, there are differences among children both in the degree to which they choose tO participate in highly achievement-oriented activities and the affective and cognitive reactions they exhibit following success or failure at the activity. Harter theorized that these differences among children in motivational orientation can be attributed to certain antecedent conditions in the child's social environment which have either facilitated or attenuated the development of such intrinsic motivation. The three-stage model which Harter designed to explain the processes by which children develop an intrinsic achievement orientation is based, in part, on social learning theory which contends that such behaviors are learned by the child through the observation of role models and through the reinforcement Of social agents, such as parents and teachers. Although this model actually consists Of three stages, only the third is detailed in this review as the first two stages pertain to the correlates of motivation for infants and very young children. 56 Stage 3 (detailed in Figure 1), is applicable, as Harter indicates, for children who have attained the Piagetian stage of concrete Operational thought (beginning around ages 5 or 6). Stage 3 Of Harter's (1981) develOpmental model (see Figure 1) begins with the child's attempted perfonmance at an achievement task which is labelled in Figure 1 as mastery behaviors (Component A). The product of the child's mastery behavior is generally identifiable as either a success or a failure (Component 8 of the model), and the particular outcome has two implications. First, the performance outcome induces certain affective reactions (Component C). such as pride, shame, and anxiety. Additionally, the Outcome of the child's perfonmance often generates an evaluation (Component 0) of the perfonmance by an Observer. If the Observer is a "Significant other" figure in the child's social environment, this evaluation is presumed to directly influence the child's sense Of competence or ability at this achievement task (Component E). However, evaluation of the child's performance also indirectly affects his or her perceived competence through a developmental process which Harter identifies as the internalization of cognitive-informational structures (Component F). That is, Older children (above the age of 6), who have develOped the capacity to think logically and in terms of cause-effect relationships begin to internalize aspects of the evaluative feedback. Thus, through socialization processes (modeling, instruction, and direct reinforcement), the child begins to adOpt the perfonmance standards of significant others (particularly parents and teachers) in 58 his or her environment. This internalization process actually consists of the develOpment of cognitive systems which allow the child to judge how much he or She values achievement in a particular area (system Of mastery goals) and what level of performance connotes success (system Of self-reward criteria). Contingent on the consistency and apprOpriateness of the acquired, internalized perfonmance standards, the child also develOps a perception concerning the degree to which he or She can control perfonmance outcome (Component G). Harter (1981) theorized that a strong internal perception of control (i.e., the belief that he or she is responsible for and in control of personal perfonmance) is dependent upon the type of internalization structures gained through socialization processes. If a child has been given clear, consistent, and realistic evaluation about his or her perfonmance, then the child will develop consistent and realistic internalization structures and will perceive an internal source Of perfonmance control. Inconsistent or unclear evaluation of performance will lead the child to perceive that control for his or her perfonmance lies with powerful others or with an unknown source. The next step in this model Specifies that the child's perception of perfonmance control directly influences the degree of competence (belief that he or She has capability for performance success) the child perceives in relation to achievement task perfonmance (Component E). A high degree of perceived control results in high perceptions of competence, whereas the belief that powerful others are responsible for 59 his or her performance leads to low perceptions of ability. For a younger child (below the age of 5), the link from adult evaluation to the child's perception Of competence is a direct one. However, with increasing cognitive maturity, the influence of adult evaluation is mediated through internalization structures (Component F) which includes perceptions of control. For Harter (1981), the resulting strength of the child's perception of competence is central to the develOpment and maintenance Of various other achievement-oriented traits and behaviors, including intrinsic motivation, self-efficacy, self-worth, and anxiety controls. If a child perceives that he or She is capable in a particular activity, then he or she will develop strong feelings Of self-worth (Component H), exhibit lower levels of performance-related anxiety (Component C), and will be intrinsically motivated to continue to pursue success in this achievement activity. Harter has redefined this intrinsic motivation as competence motivation (Component 1) to reflect the connection between the child's perception of competence and his or her subsequent desire or motivation to continue to demonstrate competence in this activity. High levels of competence motivation are reflected in continued participation in the activity at higher levels of skill (i.e., continued exhibition Of mastery or achievement behaviors). and the achievement process continues. One aspect of Harter's (1981) model that distinguishes it from many of the other theories relating to achievement motivation is that this model has been designed to be applicable in four independent 60 domains. That is, the various components which constitute the model (e.g., perceptions of control, perceived competence, and competence motivation) do not represent global or unitary psychosocial traits. Rather, each of these constructs has been defined in four domains: cognitive, which pertains to academic competence; physical, which refers to sports and outdoor game participation; social, describing competence in peer relationships; and general, which measures a general feeling of personal worth rather than an actual competency. The model, as specified in Figure 1, then can be applied to competence in each of the three independent achievement domains. Because none of the existing devices for assessing psychological development were adequate for testing these concepts of competence and motiviation, Harter and Connell (in press) began the process of developing and validating instruments to assess the various psychosocial attributes implied in their model. Each of the developed instruments (e.g., Perceived Competence Scale) actually consists of four subscales -- each one measuring perceived competence in a different achievement domain (cognitive, social, physical, and general). Harter has recognized, then, that a child may feel highly competent in one achievement area but not in another, and that measures of competence and control perceptions must reflect such differentiated motives. Following the development and validation of these instruments, a series of research studies designed to test the implied links between the model's components was initiated. Although many of these studies 61 are still in progress, initial results indicate that the implied relationship between children's perceptions Of competence and their intrinsic motivation in achievement situations has been empirically validated. High levels of perceived competence have been shown to be statistically associated with measures Of children's motivation in academic activities (Harter, 1981). Additionally, a causal relationship between perceptions of control and perceptions of competence has been identified. That is, the degree to which children perceive control over their academic perfonmance detennines their sense of competence in that achievement domain (Harter, 1981; Harter & Connell, in press). The link between childrens' perceptions of competence and their motivational orientation has also been tested in the competitive athletic setting by Roberts, Kleiber, and Duda (1981). These researchers found that young sport participants obtained significantly higher scores on the perceived competence scale as measured in the physical domain than did their non-participant peers. In addition, children who exhibited higher levels of perceived competence also exhibited a greater tendency to respond to competitive athletic situations with such achievement oriented behaviors as continued persistence following failure and higher expectancies of success than did their peers who exhibited lower perceptions of competence. This study, then, has provided initial support for the viability of Harter's model in competitive athletics and indicated that perceptions of 62 competence may be a central concept in terms Of an athlete's psychosocial status. That is, the degree to which the child feels competent in a given area may be a detenminant of his or her achievement behavior in that situation. Therefore, it would seem highly desirable to detenmine how perceptions of competence can be facilitated in the young athlete. According to the model's specifications, a child's level of perceived competence is directly and causally influenced by his or her perceptions of perfonmance control in a particular achievement situation. Furthennore, Harter's (1981)1nodel Specifies that the evaluation of significant others concerning the child's perfonmance attempts contributes to the child's internalization of control perceptions. To date, however, no reported research has been conducted to specifically investigate the extent of the influence which social agents are presumed to exert on the child's perceptions of control and competence. Nevertheless, the application of this model to the investigation Of coaching effectiveness appears quite tenable. Harter (1981) has recently emphasized the contributions that adult evaluation and feedback can make to the develOpment of the child's perceived competence in the physical domain. She noted that although children Often link their perceptions of physical competence to basic mastery of sport skills, adult evaluation of their perfonmance can either facilitate or denigrate the child's sense of competence and control in athletic areas. 63 Because perceived competence has been demonstrated to be an important mediator of many other achievement behaviors in both academic and athletic situations, the present investigation was designed to identify those facets of coaching behavior which facilitate the development Of the young athlete's perceptions Of control and competence. Harter's model, (1981) was used as a framework for the investigation of coaching effectiveness. However, since the review of the previous literature revealed that the link between instructional effectiveness and children's psychosocial growth is complex and influenced by a number of factors not discussed by Harter, a revised model was developed. This model integrates facets Of the literature previously reviewed (e.g., expectation theory, attribution theory) which appear to be applicable to the investigation Of coaching effectiveness. An Integrated Model Of Instructional Effectiveness The integrated model designed to guide this investigation of coaching effectiveness is contained in Figure 2. In a manner Similar to Harter's fonmulation, the processes specified in this model begin with the child's perfonmance behavior in an achievement domain (Component A). This perfonmance attempt can be categorized as either a success or a failure (Component 8) and influences the child's state of affect (pride, Shame, joy) and also his or her perceptions of anxiety (Components C and D). Again, as in Harter's model, the child's 64 .nmocozauoto mew—88 .5”. 35:. 1308.:— .~ 3:9..— >._.—.=m< Wmm>ememm wzuem_xz< E a. 9.0.313 / .55on e2mzm>m_:u<;mmem<2 >0. ‘ do mzoEmumm: m2f individual players. Thus, in addition to categorizing coaches' 81 behaviors toward individual athletes according to type (i.e., appropriate CBAS category). each behavior was also classified according to the initiator (coach or athlete) of the particular interaction. The rationale for using this categorical assessment was based on Cooper's (1979) review of the literature concerning expectancy effects. In this review, he concluded that consistent expectancy effects have been demonstrated in relation to frequency Of teacher-student interactions, with high expectancy students initiating significantly more such interactions than their low expectancy classmates. However, as indicated in the previous review of the literature, there is considerable controversy concerning the direction of expectancy effects in relation to frequencies of teacher-initiated interactions. Therefore, this issue was examined in the present study. An initiation by the player was coded when the trained Observer judged that the individual player verbally initiated (through direct questioning or comment) a subsequent coach-athlete interaction. This most often occurred when the individual athlete specifically (verbally) requested the coaches' responding behavior (e.g., "Coach, what did I do wrong?"; “Should I play closer to the bag?") or in such non-performance related contexts as a general communication or an organizational concern. In addition to adding an initiator component to the basic CBAS, a second component which assessed the specific attributional content of the coach's feedback following players' performance was also included. The rationale for including this component was based upon research from 82 the social psychology literature which indicates that the attributions which teachers use in ascribing causes for students' performance may influence the students' perceptions of perfonmance control (e.g., Dweck, 1975; Dweck et al., 1978). Although such attributional or evaluative statements made by coaches in athletic situations have not been empirically investigated, recent writers in the Sport psychological literature (Singer & Gerson 1980; Martens, Note 1; Gould Note 2) have all emphasized the contribution that coaches' feedback and behavior can play in encouraging players to perceive personal control over their own sport performance. It seems logical, therefore, that coaches' attributional feedback should be included in a coding system designed to assess the effect of coaches' behavior on players' psychosocial develOpment. Because little research has been conducted to examine attributional feedback in an instructional setting, no existing coding schemes are available. Dweck and her associates (1978). who conducted one of the few reported studies to investigate this issue in the academic classroom, utilized a coding scale which was Simply based on Weiner's (1974) four-factor attributional model (i.e., effort, ability, luck, and task difficulty). However, the use of this four-category system may not be apprOpriate in athletic Situations as recent research (Bukowski & Moore, 1980; Roberts & Pascuzzi, 1979) has shown that individuals use different attributions to eXplain success and failure in athletic situations than they do in other achievement task situations. 83 The particular attributions which seem to be most valid in sport skill settings have been determined through the use of Open-ended responses. Roberts and Pascuzzi (1979), for example, demonstrated that Open-ended attributional responses by 346 college students could be categorized into 11 categories. These categories included the four groupings identified by Weiner which are comprised Of ability, effort, luck and task difficulty (defined for athletic situations as the competence or performance of the Opponent). In addition, however, Roberts and Pascuzzi (1979) found such categories as teamwork, psychological factors (motivation, arousal, anxiety), practice, unstable ability (played well or poorly today), coaching, and officials to be salient factors in the ascription of athletic performance. These researchers also indicated that the four traditional causal attributions advocated by Weiner only accounted for 45% of all attributions made. However, 100% of the responses obtained from subjects in this study could be classified along the two dimensions (stability and control) prOposed by Weiner (1974) if the attributions were carefully diagnosed in relation to the athletic situation (see Figure 3). Therefore, for the present study, coaches' attributional statements to individual players following successful or unsucessful performance were classified according to the 11 categories indicated by Roberts and Pascuzzi. For purposes of data analyses, however, the attributions were categorized as indicated in Figure 3. It should be noted that the categorization Of attributional coaching behaviors was considerably more high-inference (i.e., required STABLE. STABILITY UNSTABLE 84 LOCUS OF CONTROL INTERNAL , EXTERNAL ABILITY COACHING EFFORT LUCK PSYCHOLOGICAL TASK DIFFICULTY FACTORS (e.g., (Competence of Motivation. opponent) Anxiety) TEAMWORK UNSTABLE ABILITY OFFICIALS PRACTICE Figure 3. Dimensional categorization of sport-related attributions (Roberts & Pascuzzi, 1979). 85 more judgment to code reliably) in nature than the other CBAS components which were more objectively scored. Raters, however, were trained and instructed to code all coaching statements made toward individual players which were oriented toward ascription or explanation for player perfonmance. During the training Of raters, as well as during pilot testing sessions, example athletic statements for each category were identified. Each coder was then trained to recognize and categorize such statements. Finally, two behavioral categories were added to the basic CBAS instrument. A new reinforcement category was used to record coaching behaviors (in response to a player's desirable perfonmance) which contained a positive reinforcement component along with a technically instructive statement (e.g., "Good swing, Sally, you kept your elbow up that time"). This response category was included in the present study because observation of the coaches used in the pilot testing suggested that there was a real distinction between reinforcement which was only an evaluative response to player perfonmance and reinforcement which was positively evaluative but also included some technical instruction concerning the player's performance. The second category added to the CBAS was one labeled as “Uncodable” and included those coaching behaviors which were not codable due to situational factors (e.g., coach-pitcher conversations on the mound, coach-batter conferences). 86 Product Assessments Three written questionnaires were utilized to assess the psychosocial response which were of interest in this study. These three instruments constituted a battery of psychometric tests which were administered prior to and at the end of the season (see Appendix C for entire test battery). Perceived Competence Scale for Children. This psychosocial assessment instrument was develOped by Harter (Note 15) and consists of four sub-scales, three of which measure the degree to which the child feels competent (i.e., believes that he/she is capable of performing successfully) in a particular achievement domain. The three domains include: (a) cognitive competence which is oriented to academic achievement; (b) social competence which assesses social skill; and (c) physical competence which measures perceived ability in Sport and games. The fourth sub-scale assesses the child's general feelings of worth or self-esteem, independent of any particular domain. Thus, a child may not feel competent in a certain domain but may still have positive feelings about him or herself. The fonnat of each of these sub-scales was designed for the purpose of minimizing social desirability effects. Therefore, a "structured alternative fonmat" was used in which the alternatives are both perceived as socially legitimate. The child is instructed to select the statement which is most like him or her. Then, the child indicates whether that chosen statement is really_true for him or her or only sort of true. 87 Each of the four sub-scales contains seven items, and each item is scored from 1 (low competence) to 4 (high competence). The perceived competence score for each child in each domain is computed by adding up the seven individual items and dividing by seven, thus obtaining an average competence score across all seven items comprising that subscale. All sub-scales are computed separately, so that each athlete will have four perceived competence scores - one for each of the four domains, cognitive, social, physical, and general. The Perceived Competence Scale for Children was validated by Harter using several large samples of children from four states in grades three to nine. Item scores from all samples were subjected to factor analytic techniques, and the internal consistency of each sub-scale was assessed using the Kuder-Richardson fonmula. These reliability estimates ranged from .73 to .83. Multidimensional Measure Of Children's Perceptions of Control. This self-report instrument was designed to assess the degree of responsibility children feel for both successes and failures in each of Harter's (Note 15) four competence domains (Connell, Note 16). This psychometic measure actually consists of four subscales, one for each of the achievement domains, and each subscale assesses the degree to which the child believes that each particular source (Self, Powerful Others. Unknown) is responsible for his or her perfonmance outcome. These three sources of control were identified, through extensive validation procedures, as those which explain the greatest amount of variation in children's perceptions of who or what is responsible for 88 their achievement task success and failure. Obtained scores on each of these subscales represent the strength of the child's belief in each of these three sources Of control: Self or internal (I am responsible). Powerful Others (Someone else is responsible), and Unknown (I don't know what is responsible). Additionally, children's perceptions of the degree to which each source is responsible for their perfonmance are measured independently (i.e., the sources of perceived control are not measured on a single, uni-dimensional scale). The fonnat Of this assessment device is very similar to that used in the Perceived Competence Scale in that social desirability effects are minimized. The total instrument consists Of 96 items which includes measures of control in all four domain-specific achievement areas. The complete set Of scales was administered to 521 children in grades three through nine for the purpose of assessing the instruments' psychometric properties. Median subscale reliability estimates for each of these grade levels were Obtained, and acceptable estimates for all possible subscales were demonstrated (Connell, Note 16). For this particular study, only the subscales which measure children's perceptions of control in the physical domain were used. To supplement this subset of 12 items, an additional 12 items were written which reflected the same scale factors represented by the original scale but pertained specifically to success/failure in the perfonmance Of softball Skills. These additional items were appended to the original scale and constitute the last 12 items in the Perceptions of Control instrument (see Appendix C for the complete instrument). 89 Connell (Note 16) indicates that such additional items can increase the ecological validity of the measure but that the newly-created items must accurately represent the same dimension as the original items and that reliability estimates for these new subscales should be Obtained. These reliability assessments were conducted and the results are presented in Chapter 3 of this paper. Generalized Expectancy of Sport Sucess Scale. The coaching effectiveness model, as well as Harter's model for the develOpment of achievement motivation, suggest that high levels of perceived competence are also reflected in the child's expectancy of future success in a particular achievement activity. Therefore, an instrument Specifically designed to measure the strength of an individual's generalized expectation for athletic success was also used in this study (Coulson & Cobb, Note 17). This instrument utilizes the semantic differential scaling technique and consists of 20 bipolar adjective items, each of which is rated on a five-point scale (see Appendix C for a sample test scale). The reliability and validity of this scale were assessed using three independent samples of college students (!?593) (Coulson & Cobb, Note 17). Reliability procedures indicated high levels of reliability across time (test-retest, rf.90) as well as good internal consistency (I?°95)' Construct validity procedures indicated that those individuals who had been involved in an organized competitive athletic program (high school varsity, junior varsity, or club team) Obtained Significantly higher scores in terms of general success expectancy than 90 did those individuals who had little or no past competitive experience. In addition, collegiate physical education majors exhibited significantly higher scores on this scale than did non-physical education majors. Although this scale has been validated for a population considerably Older than the athletes to be used in this study, it is the only reported instrument designed Specifically to assess athletes' perceptions concerning their competence in competitive athletic Situations. Moreover, the wording of the items seemed apprOpriate for a wide age range, and pilot testing revealed that young athletes of junior high age could easily understand and respond to all items. This scale, therefore, was used in this investigation as a sports-specific expectancy assessment device. Presage Variables Presage variables are defined as those background characteristics which teachers and students bring into the learning situation and which may influence the process-product relationship. By identifying and measuring those relevant presage variables, it may be possible to remove some of the superfluous variation attributable to unknown factors and thus to obtain a better statistical estimate of the relationship between coaching behaviors and student growth. In the present study, two presage variables, coaches' expectations concerning players' ability and players' actual ability, were identified and assessed. 91 Coaches expectation scale. Research and theory have already been reviewed to Show that much of the variation in instructional behaviors can be attributed to the expectations which teachers and coaches fonm concerning the ability of their students/athletes. Therefore, coaches' perceptions of their players' athletic potential was assessed by administering a questionnaire to each coach at the beginning of the season. Immediately after team tryouts, each coach was given this questionnaire which required her or him to rank all players on the team from highest to lowest according to their potential softball ability. The following instructions were given to each of the coaches: "Please rank-order all players on your team according to your expectations concerning their potential softball ability.“ Because infonmation concerning the stability and accuracy of coaches' perceptions of players' ability was desired, this expectation ranking scale was also given to coaches at the end Of their competitive season, and estimates of the reliability of such rankings were then statistically obtained. Ability assessments. Although Harter's scale was utilized to measure each player's perceived level of competence, additional infonmation was also collected to provide an estimate of player's actual competence and was employed in data analyses to control for the assumed correlation between players' ability and observed coaching behaviors, as well as to measure the influence which pptppl_competence contributes to the player's level of perceived competence. Because valid and reliable performance statistics (e.g., batting and fielding 92 averages) were not available for all players who were included in the sample, individual athletes' playing ability was assessed through evaluation by their teammates. At the end Of the season, each player was asked to rank all Of her teammates (herself excluded) from the highest to lowest according to softball Skill. On the basis of all of these rankings, an average "ability ranking" was calculated for each player on each of the four teams and was then used as a rough estimate of the players' relative skill level. Demographic information. Each Of the coaches and players involved in this study was also asked to complete a questionnaire designed to provide infonmation concerning relevant background characteristics. Coaches were asked for information concerning the extent Of their coaching experience as well as their competitive playing experience. In addition, such demographic information as age, educational and occupational status was collected. Similarly, players were asked to detail their previous softball experience as well as the extent of the related competitive experience they had (i.e., other interscholastic Sport participation, community leagues, instructional leagues). 93 Research Procedures Training of Coders Although all of the observational data was collected by only two coders, a total Of five individuals were trained so that more extensive reliability checks could be made and to serve as an available "pool" of substitute Observers. Each of these individuals was trained through the use of the CBAS Audio Visual Training Module developed by Smith, Smoll, and Hunt (1976). This training module employs programmed videotaped instruction, written tests, and a videotaped proficiency test to develOp and assess each coder's competence in identifying and recording coaching behaviors. Training sessions were conducted over a oneqnonth period, and each coder attained at least a 95% accuracy score on both the written and proficiency tests. In addition to this training, coders were also required to demonstrate scoring competence in actual field situations (i.e., pilot testing sessions). Assessment of inter-rater reliablity was conducted during these field training sessions and each rater was required to attain an average reliability coefficient (as based on the percentage Of behaviors for each CBAS category) of .80 or higher which indicated high agreement with an expert coder. All of the trained raters reached this criterion. In addition, periodic assessments of inter-rater reliability were conducted during the course of the study. These assessments required two raters to code a game/practice session Simultaneously but 94 independently. Category percentages were then compared through correlational analyses to determine the degree of agreement between trained raters. Obtained correlation coefficients ranged from .78 to .92 over all assessment sessions and all CBAS categories. Data Collection Procedures Extensive procedures were followed to collect the necessary data for this study. These procedures included the collection of data prior to, during, and after the completion of the competitive season. Pre-season. Following the coaches' final selection of team members, an explanation of the experimental procedures was given to all players. and consent forms were Obtained from each of the coaches and athletes. In addition, several assessment measures were also administered. Specifically, each coach was asked to rank order all his or her players from highest to lowest in tenns Of expected softball ability, and a battery of pencil-and-paper tests were administered to all players for the purpose of assessing their baseline or entry level scores on a variety of psychosocial measures, including perceptions of control and cOnpetence and expectancies for athletic success. Finally, both coaches and players were given a questionnaire to complete which provided demographic and other background infonmation relevant to this study. During the season. Process variables (i.e., coaching behaviors) were recorded at periodic intervals during the course of the entire season. Each team was observed a total of seven times - four practice 95 and three game Situations. Such Observations occurred for each coach and team at the rate Of approximately one day per week (although the particular day for each team was randomized over the season). Because the purpose of this study was to identify coaching behaviors towards individual players, only those coaching behaviors which were actually directed to individual players were recorded. In order to equalize the contextual situation across all recording sessions, an observational limitation, adapted from COOper and Good (in press). was imposed. This observational principle limited behavioral coding to only those Situations in which pll_athletes had an EBEEL chance for interactions with the coach. That is, the only coach-athlete interactions recorded were those which occurred when the entire team was working together or in small group workouts where the coach could move freely between groups. Therefore, coach-athlete interactions which occurred during practice Situations in which the coach worked exclusively with a small group of athletes (i.e., working with just the pitchers) were not utilized in data collection. However, it should be noted that this Situation seldom occurred during the course of data collection. Each Of the five coaches ran a very "coach-directed" practice and most often worked with all athletes Simultaneously. The observer began data collection when the coach "officially" began practice or, similarly, when the game was officially begun. Although the observer recorded coach-athlete interactions throughout the entire practice or game session (except when the coach worked 96 exclusively with a small group of athletes). a minimum of 60 minutes was set for each Observational session. Therefore, if a practice session lasted less than one hour, it was pp; counted as an observation session. Individual coach-athlete interactions were recorded on data Sheets by using players' numbers for game observations. Practice interactions were recorded on data sheets which contained pictures Of individual athletes. Therefore, each of the coders had no difficulty identifying or differentiating between individual players. While recording coaches' behaviors, the observer stationed him or herself in a position which allowed the most accurate and complete observation Of coach-athlete interactions while also trying to maintain an unobstrusive role. This limitation was deemed necessary in order to ensure that "reactivity" effects did not bias the collected data. Coaches and players were infonmed at the beginning Of the season that the purpose of the study was to collect some descriptive data concerning fenales' achievement motivation in competitive athletics. Therefore, coaches and players were not aware of the specific nature Of the behavioral observation. However, all coaches were interviewed at the end of the season, and at that point were still unaware of the exact type of information collected. After being infonmed of the specific purpose of the study, they indicated that they did not think that the observers' presence, in any way, affected the players' behavior. 97 Post-season. After the playing season had been completed, the measures used to assess coaches' expectations and players psychosocial status were again adminstered. That is, coaches were asked at the end Of the season for their perceptions of the athletic and softball ability of each of their players, and all players were administered the same battery of tests which assessed the psychosocial variables of research interest (e.g., perceptions of competence, control and expectancies of athletic success). A complete summary Of the data collection procedures is listed in Table 4. 98 TABLE 4. Summary of Assessment Procedures Pre-Season Procedures 1. Consent Fonm (parents, coaches, athletes) 2. Coaches' Expectation Scale 3. Psychometric Test Battery a. Perceived Competence Scale b. Multidimensional Measure of Children's Perceptions of Control c. Generalized Expectancy of Sport Success Scale 4. Demographic Information Questionnaire During Season Procedures Observation and Categorization of Exhibited Coaching Behaviors using CBAS and attributional coding scale Post-Season Procedures 1. Coaches' Expectation Scale 2. Psychometric Test Battery a. Perceived Competence Scale b. Multidimensional Measure of Children's Perceptions of Control c. Generalized Expectancy of Sport Success Scale 3. Players' Ability Rankings CHAPTER III RESULTS This study was designed to statistically address two issues in relation to coaching effectiveness. First, patterns Of coach-athlete interactions were examined as a function of coaches' expectations concerning athletes' Skill level. That is, are the behaviors exhibited by coaches to individual players influenced by coaches' perceptions Of players' softball ability? Secondly, the influence of coaching behaviors on players' perceptions of competence and control was also examined. Are there certain coaching behaviors which are most conducive to players' develOpment of competence and performance control perceptions? The results Obtained from of each of these statistical analyses will be discussed separately. Statistical Analyses: Expectancy Effects Over the course of the 1982 competitive season, coach-athlete interactions were recorded and categorized according to the Coaching Behavior Assessment System (CBAS). These observation procedures resulted in the collection of a total of 6478 individual coach-athlete interactions, as summed across all coaches and observation sessions. Of this total, 3662 behaviors were recorded during team practice sessions (four practices for each team) while 2816 of them were coded 99 100 during games (three games per team). Tables 5 and 6 present a summary of the data obtained for each CBAS category and for each of the five coaches in practice and game situations. The total amount Of observed practice time for gppp_tean ranged from 325 total minutes (summed across four sessions) to 350 minutes, with a mean of 338 minutes. All games were at least five innings long, and although the amount of observed time varied considerably from game to game, the total number Of minutes of game time under Observation for each team ranged from 294 minutes (as summed across three games) to 387 minutes with an average of 358.20 minutes. Because the individual player was to be used as the unit of analysis, the observed coaching behaviors in both game and practice Situations were recorded and coded as the frequency of each type Of behavior which was received by a player. Summary statistics, used to describe the distribution of coaching behaviors towards individual players, revealed that the averge number of coaching communications received by an individual player across all four Observed practice sessions was 50.86 (§Q§20.25). However, the range of this value indicates that there was considerable variation among individuals in relation to the number of interactions with their coach. That is, some players received as few as 12 coaching communications over four practices while others received as many as 114 communications. In game situations, the average number of coaching behaviors directed towards individual players was 39.11 (Spf36.01). Again the range (6 to 199) 101 .auo«>ozon veganscxo Co songs: amuse on» xn xCOJOHeu cues ca mucu>azon Co genes: may vascu>mu 5n wouefisofiuo new: museum camucoouom “can: won wo5 com awo 5&5 a. c o o m.— w— ~. — n. c 0.5 on 0.5 5« n.5 c5 w.c QN o.. @— o.m mo c.o mo ~.a— Om. 5.0— no 5.p— cm 0.5 an 0.5 ma 5.~ 5N ~.~ a. ... o «.05 Na ..n cw e.n aw o.~ a. o.a co m. 0 —.~ 0— o. c o o a." p— o. N c.~ o~ n.— n— n. ~ o.m as o.m no o.~ n— o.— o— a. m m.« on m.a mo 5.n an m.5 55 c.n pN w.o— no ~.e —~ n.0 we o.c no m.n - 0.9 @— c.o Nn o.c— 50— w.- —- 5.oN $~p ..5N n—N c.m 5a m.« 5~ 0.5 A5 ..N n— 5.0— mm— a.c~ oN— «.mn n5~ 5.- ANN 5.Nc oQN m._— mm c. N c. n o.- m- ..p 5 o. n brooms; xocvsoowu broaden xocoscmwu wmwuuoni Auzvacsyu wmmommml xocosdwuu wmmuemm 50cmsvo~u m. a (M N — Iuozom coupon»: ~aa0— 2 02:65 u:o§u3au:00:g undocuu caduuzuumc_ unuwczoo_ aauocuu cOwumoMcseeou meuocou seque~aceauo deducou oeuaoox 2253.5; cum: 53339: ueuficcoo_ ucoaCMuc001oxwumwz ucocfinqcam assume: a:«~oca~ acosoomuaoocu accusaucouuoxaumwz ceauosuunc_ ~auqczuo— acuvaucootoxouuwx acoEOOHOCcMOucoz u:o§ouuoCcfioe ccuuusuumc_ Hquczoo— cum: assauouochoz n80w>m£om oowuoeum .mosomou Co ~oocom xn xumeesm .a usm<_ 102 333:3 ..O 03.5: on. NNn 0n0 000 005 Nnn 0.. m N.. 0. 0.N 0 0. 0 0.N .. N.0N 00 N.0n nmN 0.NN 00 «.0. 5n. 0.5. 00 ..0N 00 0... 5o 0.0N N.. 0.0m ..N N.0n ..N m. n 5.. 0. 0.N 0 0. 0 0.. 0 5.. 0 n.. .. 0.. 0 0. 0 0.. m 0 0 ... a N. . .. . 0.. 0 0.0 0. m. 0 N. . 0. 0 5.N m. 5.0 m. 0.. 0 n. N 0. 0 N. . 0.0 NN 0.0 00 N.0 5. n.0 0n 0.0 5N 0.. a 0.0 50 n.0 0. ..n NN 0.0 NN ..0 0N 0.0 5n 0.0 0. m.m 5n N.n 0N 5.0. an 5.. 0. 0.N .. 0.n .N 0.n Nn 5.N. .0 5.Nn 05N n.0N 0.. n.0n n.N 0.0. m0 a. . 0 0 0 0 0. n 0. N wmmoOmnl Nucosaouu greased Nucmsdouu wrouuml. Nocoacwwu wmouuoh Nocosvmmu “cocoon Nocsscomh m 0 n . 1.0000 63.523 03. 2.3 ..0 doses: .253 on» 50 300.300 come 5. .330 50 0333.00 oufl. meadow Kinsman—om "nan! 9316.30 850 .30. 032505 “cues—.8335 .3800 €032.ch— neofiloo. .388 cos-W328 .3930 causgoms .Ouucou veaaoox 0:923:30 5!. c0383?— ~eowccou. ace—"Ewucoocoxmunqx €055.50 nexeunwx 053:0. 008000 .5005 «coat. 0:00.. 9.0 an... :3. .0339: 13.5.00. acuaczcooéxmuawx 0:23.30 c. 3:2 ascended; c... 0.. 33.350:— .3123. 5... nauseous can: 232500 9.60 32.0000 .o .0038 .3 5.55.50 .0 0400. 103 indicates considerable variation among players in the absolute amount of coaching communications received. Data Preparation Computational Indices Although data collected with the CBAS instrument has been and can be utilized in a number of ways, Smoll et al. (1978) suggest that the most useful and reliable behavioral index has been shown to be the percentage of behaviors across all Observations which fall within each of the coding categories. However, percentage scores may not be valid measures to use for the type of analyses to be employed in the present study because such percentage scores for individual athletes will be highly influenced by the absolute amount of desirable/undesirable perfonmances which that athlete exhibits. That is, because high, as compared to low, ability players exhibit relatively higher proportions of successful as compared to unsuccessful performances, their percentage scores will reflect a disprOportionate amount Of reinforcement Opportunities. Because the purpose of this study was to examine whether individual athletes receive different coaching behaviors as a function of coaches' expectations, percentages were judged to be unreliable as a measure of coaches' biases because they may actually be reflecting only differences in absolute amounts of success/failure attained by players. Therefore, in addition to the calculation Of 13 percentage scores (one for each of the CBAS categories), nine index scores were also computed for each athlete. These indices provided a measure of the relative frequency of the 104 coach's reSponse to individual players following their successful/unsuccessful perfonmance. The divisor for each index score was not the total number of behaviors but the total number of desirable perfonmances or the total number of undesirable perfonmances. These indices made it possible, then, to assess differential treatment by coaches of high and low expectancy athletes when these athletes were in similar Situations (i.e., after success/failure). Of the nine calculated indices, three measured relative frequencies of the coaches' reSponses to a desirable perfonmance and five indices measured frequencies of the coach's response to players' mistakes. The ninth index represented the relative frequency with which some type of technical instruction (e.g., reinforcement with technical instruction, mistake-contingent technical instruction, mistake-contingent technical instruction with punishment, or general technical instruction) was provided the individual athletes. The divisor for this index score was the total number of instances in which technical instruction could have been provided. This index score was specifically chosen for inclusion in the analyses because the developmental psychology literature suggests that perceptions of performance control may be influenced by the presence of clear, consistent and perfonmance-contingent feedback. A computational summary of each of these indices is presented in Table 7. Although percentage scores as well as indices were calculated for each player , the majority of the data analyses was conducted using the indices because they provide a more accurate and reliable measure of 105 200 + 0.. + 22.. + 2.. + ..m + a + 2. + 20 + 22 + mv\aw.. + 22.. + 2.. + 0.20 n 2.. xoucH :o..u=c.m=. .wo.:;om. 2.. 22.. + a + 2H + 20 + 2..\22.. u 02.. .cwssm.c=a 20.: co..oac.mc. .eu.::um. .comc..cou-oxeum.2 a2.. 22.. + a + 2. + 20 + 2..\a u a pcmsgm.::a a 220. + a + z. + gm + zH.\zH . :0 mosesm.z m:..o=m0 20 220. + a + :0 + 2m + zs.\zd . =0 seasomatsoucm seam=_p=ou-osepm.z 2m 22.. + a + 2. + z“ + zH.\zH. . EH. co.oo=2om=H ..u_=;om. pcoaceocou-oxesmsz so. 22 + m + ..m\mz u mz .cosmoco.c.mmucoz mz mz + a + H.m\m u m ucm=ouco.=.mm m 22 + m + ..m\..z u 0.2 co..o=L.mcH .oo.:;om. g..3 acoewuco.=.mm ..m co..m.=aeou xouc. .mgo.>m;mm .0000 o>.pa.cummo mou.ucH .eco.>mgmm ac.;ueoo .o accessm .m:o..eu=quu .5 000<. 106 individual coach-athlete interactions for the purpose of this study. Tables 8 and 9 contain a descriptive summary of the behavioral indices of each coach in both practice and game situations. Contextual Variables Thirteen percentage and nine index scores for each athlete were calculated separately for practice and game situations. To test whether the Observational data for games and practices could legitimately be combined, Hotellings 1? analysis, using pairwise comparisons, was conducted to assess the similarity of the patterns of individual coach-athlete interactions across both types of situations. Results indicate that there was a significant difference between the coaching behaviors received by players in practice sessions and those recieved by players in game situations, I? = 57.80, 5(7,65) = 7.56, p < .0001. Additionally, canonical correlation analyses, a multivariate statistical technique designed to assess relationships between complex behavioral data sets, revealed that no significant canonical variates could be computed for these two sets of data. Thus, while practice and game behaviors significantly differed, these differences did not follow any clear pattern. On the basis of these two statistical tests, it was concluded that coaches' behaviors towards individual athletes differs significantly from game to practice Situations. Therefore, all of the analyses were conducted separately for games and practices. Unit Of Analysis In their study of coaching effectiveness Smith et al. (1979) have examined the relationship between coaches' behavors and player 107 00.0w 00.0w .N.m0 N0.mm .m.m0 xmoz. zo..0=m.mz. 000.2200. 0m.. 00.0 mm.m N... 00... mN.mN .0.0 M5.N m5.N .0.5 00.00 00.0. 00..N M5... .0..N 0N.0. .m.0N om.N. 0N.N. 00.0 55..N M5.m0 00.00 50.N5 50.00 mmuHozH .zuwzH5zouumx<5mH2 .0.5N 50.0 mN.mm 00.0 00.00 00..5 .N.00 N0..5 .0.mm .N.mm 0... N0. M5.m m0.N 0N.N mmuch. .zmzmomoqumm m 0 m N H 20<00 xmu:. co..u=g.mc. .eu.:;om. ucwexm.:=a :p.3 =o..u=g.m:. .mu.cnuw. acoe:m.:=a axeum.2 ac.cocm. .cwewoecaoocm :o..u:cum:. .mu.:cum. .cmemucowc.mmucoz .cogouco.=.mm co..o=g.m:0 .eu.c;um. g..3 newswoco.=.mm m00.><200 02.:0<00 mco.mmmm do..umca m:.c=o mco.>ezmm m:.sumou .o moo.uc. .0 msm<. 108 00.00 00.50 00.00 00.50 00.00 x0020 20000000020 0<0sz00~ 50.50 00.0 00.0 00.0 00.00 00.00 00.0 05.0 00.0 00.0 00.00 00.00 00.00 05.00 05.00 00.0 00.00 00.00 05.00 00.00 H5.H0 05.00 00.00 00.00 00.00 0000020 0200200200n0¥<0002 00.00 00.0 00.0 00.0 00.50 00.00 0H.00 00.00 50.00 00.05 00.0 0 0 00.0 05.0 0000020 0202000002000 0 0 0 0 H :0<00 xmucH :o_pu=gum:0 Pou_:=om5 acmezmwcaa new; copuusgumca qupcgum5 ucmpgmwcaa mxmamwz mewgocmH acmewmmgaou=0 copuuagpmcH qupcgump pcmgmugoucrmaucoz acmswugoycpmm coppusgpmcH pwuwcsum5 ;p_z acmewugowcvmm 0000>wzmm mapgumou mo mmu_uc0 .0 00005 109 attitudes by treating the team as the unit of analyses. These researchers then coded all coaches' behaviors without reference to specific coach-athlete interactions. Coaching effectiveness was detennined by calculating the relationship between observed coaching behaviors and the team's average score on psychosocial measures. However, this method of analyses ignores the importance of differences between individuals on the same team. To use the individual player as the analytic unit also entails statistical problems in that it wrongly assunes that players' responses are independent of the general setting in which they occur (Cronbach, 1976). Several researchers (COOper & Good, in press; Martin & Veldnan, Note 18) have suggested that a viable solution to this problen is to analyze this relationship in two ways - first by looking at between-group differences (class or team) and secondly by analyzing within-group (individual) differences. Although this approach appears to be an excellent statistical solution, it is not feasible in this study due to the lack of a sufficient number of groups (n75). Therefore, for the present study, the individual player was used as the unit of analysis, but, a statistical procedure was employed to control for the variance attibutable to differences between teams. That is, all of the behavioral data (i.e., percentage and index scores) were converted to standard scores which reflect the nunber of standard deviations that each player is fran the team mean on each behavioral category. These standardized scores, then, allowed conparison of 110 individual players but also controlled for the influence of the team setting. Results: Expectancy Effects To assess the stability of coaches expectations over the course of the competitive season, Kendall's coefficient of concordance was used to compare coaches' ranking of their team members at pre-season with similar rankings given at post-season. The results of this correlational analysis indicated that coaches' expectancies did renain relatively consistent over the season, H)71)=°85'.E.< .0001. In addition to assessing the stability of coaches' perceptions of players' ability over the season, the accuracy of these perceptions was also exanined by correlating than with a measure of players' actual ability. Because no performance data (i.e., batting or fielding averages) were available, the skill level of all players was assessed through a teammate ranking system which was adninistered at the end of the season.1 Correlation of players' rankings with coaches' rankings indicated moderately high agreement, although the association between coaches' post-season rankings and the players' assessments (37'87'.E < .001) was stronger than the correlation between coaches' pre-season rankings and the players' assessments (5?.71, p.< .001). This finding 1Although it would have been most appropriate to assess the consistency or stability of the rankings supplied by the players themselves, Kendall's coefficient of concordance could not be conputed with this data because of the small number of cases relative to the nunber of judges (rankers) used. 111 suggests that coaches' perceptions of players' ability may have increased in accuracy as more infonnation was made available to then. That is, coaches' pre-season rankings were probably not based on knowledge of players' past softball performance as most of the athletes (gf49) had not played the previous season. Therefore, coaches’ initial perceptions were modified as actual infonnation concerning players' perfonnance was acquired. Selection of Low and High Expectancy Players To exanine the influence of coaches' expectations on their behaviors towards individual athletes, the study design required that coaches' behavioral indices be conpared for low and high expectancy athletes. Therefore, the results of the pre-season coaches' rankings were used to classify 40 of the 72 athletes as either high expectancy (2&20) or low expectancy (2&20) players. Because the total number of athletes on each tean who participated in this study varied fran tean to team, the actual number of athletes designated as high and low on each team also varied. Approximately 1/4 of the athletes on each team were designated as high expectancy and 1/4 of them were classified as low expectancy players. Schools 1 and 2 had five high and five low expectancy athletes each. School 4 had four athletes in each expectancy group, and the remaining 2 schools had only three designated high and three designated low expectancy players. This partitioning of players was repeated using the coaches' rankings which were obtained at the end of the season. Comparison of the pre-season groupings with the post-season groupings revealed that 13 of the 20 pre-season high expectancy athletes were similarly 112 designated as high at post-season. 0f the 20 pre-season low expectancy athletes, ten of then were assigned the same low expectancy grouping at post-season. One pre-season high expectancy individual ended up as a low-expectancy athlete at post-season, and similarly one pre-season low expectancy individual finished up the season as a high expectancy designate. Descriptive comparisons between these high and low expectancy athletes (both pre-and post-season groupings) revealed no differences between then in relation to the number of years of past experience either in competitive softball (Pre-season, tfi38) = 1.04, p_< .31; Post-season, tj38) = 1.37, p_< .18), or in sport participation in general (Pre-season, tfi38) = .66, E < .51; Post-season, t}38)=.74, pf.46). Similarly, no differences were found between the groups in practice attendance, but high expectancy athletes (both the pre-season and post-season groups) did play significantly more innings in the games which were under observation than did their low expectancy teammates (Pre-season, tf38)=4.82,.ps.0001; Post-season, ti38)=7.60, 5.001). Finally, descriptive statistics indicated that high expectancy athletes were more often placed in infield positions while low expectancy players were predominantly outfielders. Table 10 shows the breakdown of the 40 expectancy group athletes (both pre-and post-season) by position played. Although the results of these analyses do show that coaches' expectations renained relatively stable over the season, there was still some discrepancy between those players designated as high or low 113 00 H 0 HH 0 0 aucmuuonx0 300 zucmuumawm :00: 002000000 200<0mihmom 00 0 upmwwpzo 0 00 upmrwcm 0 0 gozuum05meuppa xucmwwqu0 200‘ aucmuumaxm :00: 002000000 200<0mu000 mopopcu< zucmpumax0 2o; can :00: 00 umxmpa mcowpvmom "00 000(5 114 expectancy at pre-season and those so designated at post-season. Therefore, the statistical analyses used to examine expectancy effects were conducted twice - once using the pre-season groups and once using the post-season ranked groups.2 Specific comparisons between these high and low expectancy players included examination of expectancy effects on three types of criterion variables: (1) the number of individual coach-athlete interactions; (2) the type of coach-athlete interactions, and (3) the number and type of attributions made by coaches concerning players' performance. Expectancy Effects on Number of Interactions To analyze the differences between high and low expectancy athletes in relation to the number of interactions with their coach, four criterion scores were calculated for each athlete. (1) the average number of practice interactions which were initiated by the coach (total number of coach-initiated behaviors in four practice session divided by the number of practices attended by the athlete) 2The researcher recognizes that the use of data collected after the season for the purpose of predicting behavior which occurred during the season is somewhat unorthodox. However, the correlational ana yses conducted to assess the stability and accuracy of coaches' expectations indicated that such perceptions or expectations were modified during the season as infonmation concerning players' ability was acquired. Because no infonnation was available to determine exactly when coaches' expectations were changed, the best alternative was to use post-season expectancy rankings as a measure of coaches' modified perceptions. However, it is recommended that additional research be conducted to determine when coaches' perceptions do change (i.e., after first few practices, after the first game, or at mid-season). 115 (2) the average number of practice interactions which were initiated by the player (total number of player-initiated behaviors divided by the number of attended practices) (3) the average number of game interactions which were initiated by the coach (total number of coach-initiated interactions divided by the number of innings played). (4) the average number of game interactions which were initiated by the player (total number of player-initiated behaviors divided by the number of innings played). Because these criterion measures were inter-related, multivariate t tests were conducted to assess differences between high (2720) and low (2720) expectancy athletes in relation to these four measures. Although no expectancy effects were found using pre-season expectancy groupings, 1205.31, 5(4,35)=1.45, 55.24, a significant multivariate effect was found for post-season groups, 1?=13.47, {(4,35) = 3.10, p < .03, thus indicating that the vector of means did differ for high as compared to low expectancy athletes. As a follow-up to the multivariate t test, a stepwise discriminant function analysis was employed to specifically identify which criterion variables were most responsible for this group difference (Huberty, 1975). This analytic technique selects in stepwise fashion those criterion variables which maximize the statistical difference between the two groups. In addition, one or more discriminant equations can be calculated, each of which is actually a weighted linear combination of the variables selected in the step-wise procedure. The contribution which each of 116 the criterion variables contributes to the equation is a measure of the importance of that variable in discriminating between the groups. The results of the discriminant function analysis using the four separate measures of frequency of coach-athlete interactions, indicated that three of the four criterion variables combined to fonn a single, significant discriminant function equation (Nilks Lambda = .7441, .EF-021' Comparison of the standardized discriminant coefficients (see Table 11) indicates that the number of behaviors initiated by the athletes in both practices and games are the two most powerful discriminators of expectancy group differences, although the number of behaviors initiated by the coach in game situations is also a significant contributor. Examination of the group means (also shown in Table 11) shows that low expectancy athletes initiated fewer interactions in practice situations but significantly more interactions in game situations than did their high expectancy teaiimates. Conversely, coaches tended to interact more often with their high-expectancy athletes in game situations than with their low ability players. This behavioral expectancy effect was not evident in practice situations, as it appeared that coaches did not interact with high expectancy players more frequently than with their lower skilled athletes. Interestingly, when the source or initiator of the interaction is ignored (i.e., all interactions were included in the frequency count regardless of initiator). significant expectancy effects for post-season groupings were found 3(38)=2.28, p_< .03) which indicate 117 Ho.wMW# 00.v0« copumacm.mowuu==w pcmcpsweumwu mg» ops? umgmucm mm: mpnwwgm> :umm sows: pm amum m:» o» ccoammggou wruvmmgoammw 00.i 55. 05.i pemwepm0aeu ucocpepgumpo em~vecaeeegm 000.0 «00.0 «*00.5 an 000.- 000. mgor>mgmm mama umpwpuwcfiizumou m0ogmz< 500. 000. meow>msom memo uopavuPcHigmxopa m0mgm>< 000.- 000. mgov>msmm mupuumea uwumwawcHigozmpa o0agmz< cow: com: meowvcfi Pagov>mgw0 zucmuuqu0 3o; xucmuumnx0 :00: mgo_>mgm0 0cpzoco0 mo aucmscmgu Lo; mupzmmm compucam aca:_ewgomwo “muumwm0 accouumnx0 commmmiumoa "Ha 000<~ 118 that coaches interacted with their high expectancy athletes significantly more often (flf.42) than with their low expectancy players (Me-.35), in practice situations. However, as indicated above, this significant expectancy effect in practice situations disappeared when player-initiated interactions were partialed out. Expectancy Effects and the Quality of Coach-Athlete Interactions For this section of the data analyses, multivariate t tests were conducted to assess whether coaches' expectations influenced their reactions and responses to players' desirable and undesirable perfonmances. For this analyses, then, the nine indices were used and expectancy effects (using both pre-season groupings and post-season groupings) were examined for game and practice situations separately. Expectancy effects and practice behaviors. Multivariate analyses revealed that coaches' expectations concerning their players' ability, as measured both prior to and after the season, did not significantly influence their behaviors towards individuals in practice situations. Hotellings 1?, comparing players designated as low or high expectancy athletes based on pre-season rankings, indicated no difference in the type of behaviors these athletes received from their coaches, I? = 5.82, [(9,30)=.51,.ps.86. Similarly, post-season expectancy rankings also revealed no effects on coaching behaviors, 1?=8.86, f(9,30)=.78,25.64. Expectancy effects and game behaviors. In contrast, both pre-season and post-season expectancy rankings were associated with differential coaching behaviors in game situations. Multivariate 119 analyses revealed significant differences, 1?=34.78,E(8,31) = 3.55,pf.0006, between high and low expectancy (pre-season groupings) players in the type of behaviors received from their coaches. Discrim- inant function analyses indicated that six categories of coaching behaviors contributed significantly to the difference between expectancy groups (Nilks-Lambda = '57,.EF°004)' Examination of the standardized discriminant function coefficients (see Table 12) clearly indicates that reinforcement and non reinforcement are the most powerful discriminators. Comparison of group means shows that high expectancy athletes received considerably less reinforcement for positive performance than did low expectancy athletes. Corresponding- ly, these high ability athletes also were ignored more often (i.e., received no reinforcement) after a successful perfonmance than were their low ability teammates. Coaches' responses to players' game mistakes were also significantly influenced by their perceptions of players' ability. That is, low expectancy athletes were ignored more often following an undesirable perfonmance while high expectancy athletes received more punishment (criticism) and more technical instruction delivered in a critical manner. Finally, however, these discriminant analysis results also indicated that low expectancy athletes were given more technical instruction in game situations than were high expectancy players. The previously presented expectancy effects were based on analyses using pre-season expectancy rankings. However, very similar effects were found using expectancy groupings based on rankings provided by coaches at the end of the season. Again, this analyses indicated 120 830 .. 00.v « coppmaam.mopuu=:w acacvswgumwu as» oucw umgmucm we: mpnmwgw> zoom curs: um amum mg» on ccoammggou mpg vmugoammn 50. **00.0 000. 000.- xmucu corpusgumcm pachsuwp 00.- **00.0 000.- 000. acmegmrcza saw: cowuuzgpmcm pmovcguop pcmmcvpcou-mxmummz 00.- **00.0 000.- 000. acmszmwcam 00. «¥0o.0 H00. 050. mmxaumwz 0:_Lo:00 05. 000.0 00~.- 000. acm=5ugou=_mm-coz 00.0 «00.5 050. 000.- acmgmugomcwmz ucwwupmwwou i. com: com: mmu_c=0 Pagow>wgmm acmcwg;cum_o mu aucwuumax0 :00 xucmuumax0 :00: vaVugeccmum mgov>msm0 wemu .mmsumou Lo» mppammm cappucsu acm:*swgumwo ”muumwm0 aucmpumax0 comamm-mga “0H 000<~ 121 significant differences in coaches' behaviors during games as a function of their expectations, 12=25.31,§_(8,31)=2.58,p<.03. Follow-up discriminant analysis revealed that a weighted linear combination of four behavioral indices provided effective group discrimination (Nilks Lambda = .6091, 25.002). The most significant contributor to this equation was the relative frequency with which the coaches provided technical instruction (see Table 13), with low expectancy athletes again receiving more such perfonmance-related information. Similarly, when performance mistakes were committed, the low expectancy athletes received more technical instruction, and, correspondingly, less technical instruction accompanied by punishment than did the high ability athletes. In regard to desirable perfonmance, the only significant differences between the feedback received by these two expectancy groups was the frequency with which such perfonmance was ignored or not acknowledged by the coach (non-reinforcement), with the high expectancy players experiencing more of this type of response than did low expectancy players. Expectancy Effects: Predictive Strength The statistical tests utilized to assess expectancy effects which have thus far been described have only used 40 (20 high and 20 low) players out of the 72 who were observed. Therefore, in order to include all of the subjects in the analyses for expectancy effects and also to obtain a numerical estimate of the influence of coaches' expectations on their behaviors, a multivariate multiple regression analysis was also conducted. Predictor variables were the pre-season 122 N8 . out. moo.va* cowpmzcm.mowpu==» acmcwswgomwv mg» can? cmgmpcm we; mpnmwcm> seem news: we amum mg» op ucoammggou m.u vmugoammw 00.- ¥*00.0 000. 00H.- copausgumcm Pmuwcguw» ucmmcwucou-mxmpmpz 00. «*00.0 000. 000.- xmccH cowuuzgumcfi Pauwcgumh 05.- «00.0 000.- 000. acm=:m_:=a new: cowuuagumcH Pmu_=;ump pcw0=_u:o0-mxoum_z H5.- *0H.0 0H0.- 000. acmemugomcwmg-coz pcmpuw$0mo0 i. com: com: mauvucH pagop>mzmm pcmc_ewgumwo m0 mocmpuoax0 300 00ccuumax0 :00: umuwuecucepm mgow>mzmm mama .mmcumoo L00 mppzmmm compuczm ucmcmsweumwo "muummw0 zucmuumax0 comwmm-pmom "00 000<~ 123 and post-season rankings (standardized by team) which coaches had assigned to each player. Criterion (dependent) variables were the behavioral indices which were again divided into practice and game, and behavioral sets. The results of these two separate analyses indicated no association between coaches' expectations and their practice behaviors, [(18,122)=.74, p <.76. However, a significant relationship between coaches' expectations and their behaviors in game situations, [(16, 124) = 2.87, pf.0006, was revealed. Standardized regression coefficients, presented in Table 14 provide a measure of the relative importance which pre- and post-season expectancy rankings contribute to the prediction of coaches' behaviors in game situations. Comparison of these beta weights suggests that post-season expectations have the most predictive influence on the frequency with which non-reinforcment is given as well as on the frequency of mistake-contingent technical instruction accompanied by punishment or criticism. In contrast, pre-season expectancy rankings were most highly associated with the frequency of reinforcement given. To more specifically examine the influence of expectancies on coaching behaviors, canonical correlation analysis was also conducted. This type of analysis, which was designed for the multivariate analysis of relationships between complex sets of behaviors (Hotelling, 1935), statistically defines pairs of linear combinations of weighed variables. One set of each pair represents the predictor variables and the other set within the pair represents the dependent variables. Each 124 050.- 000.- 000. 000.- 000. 000.- 000.- 000. 000. 000. 000.- 050.- 500.- 550. 500. 000. xcmm xcmm xocmpumax0 aucmuumax0 commmm-pmoa comwmm-oga ucmszmwcsa new: copuusgumcH Paupczumh pcmmcvaco0-mxoumwz ucmezmwcza mmxoumpz 0:000:00 acmewmogzoucm pcmmwucoo-mxmumwz cowpuacamcH Paupcgum5 pcmmcvpcou-wxmumwz pcmsmueomcvmm-coz ucmsmueomcwom copuusgamcm paupcsump mmuwucm Pacow>mgmm mgow>msm0 mama .mmgueou mo cowuuwumca Lo» mucmwup$$mou cowmmwgmmm uwNpugmucoum "0H 000 Louuwuoga 000.- 000.- cowuuzepmcH PmuPccum» pcwmw”&WMWHquwwnr 000. Hmo.- newsgmwcaa So. So. 303:: 9.7.83 000.- 000.- unwewmmgsouc0 acmmcwucou-mxmampz 000.- 000. cowpuaeumcH Pouwczump acm0:_p:o0-mxwpmpz 050. 000.- ucmsmugomcwmm-coz 500.- 000. acoswugowcwmm 000.- 000. cowpusgumcfi quwczump mpamwgm> covgmpwgo mm0:_uoo0 "0 mmcwuoo0 ”H mpnmwgm> copumpwggou Paumcocwu cowuwpmggo0 PMUwcocwu meow>msmm 0:_;umo0 ucm mcowpmuuoax0 .mmgumou ”mcopampmgcou Pmuwcocmu ucouwm uco umcmu so» mmcwvwo0 Pmuwcocmu .00 000

00 m0nouamoom zomou go: 000 mo0womnam no you m000 ho» unanswumo 500000m00ou uocwounom can: cooommiumom coo: commom-oum muozuo 0:0uoxom i 0ocuouc0 xouc0 0wcuouc0 oouaom cxocxcn i woounom crocx o00o0xocx go acoux0 mouzom cxocxca mounom muoguo 0:0uoxom ouuaom 0ocuouc0 QC0aeoo 00onuuom muocuo 0:0uoxoa i 0ocuouc0 «xocc0 0mcuouc0 oouaom :xocxca i woouaom cxocx "00000xocx ho acoux0 oouzom crocxca oouzom muozao 0:0uoxom oouaom 0ocuouc0 cwmeoo 0oown>zm 0ou0cou oocmeuouuom to acowunoouoa oocouoaeou o>0uwc0o0 oocouoaeou 0ouocm0 oucouoaeou 0m0oom oocouoaeou 0a00a>cm xocoaooax0 uncooam uuonm oucouoaeou 0o ocoquoouom mocmumaeou new 0ouuco0 0o ch0uamouo0 go mmuammo: 00000030 "05 0000— 140 study is based hypothesizes that players' actual success/failure rates in canbination with coaches' evaluation of that perfonnance will influence their perceptions of competence, a measure of players' ability was entered into each regression equation as well. This measure of players' actual ability was obtained by averaging the ranks assigned to each player by all of her teammates relative to her overall softball ability. This assessment of players' skill level, although subject to measurement error, represented the only infonnation concerning players' attained skill over the season, as perfonnance scores (e.g., batting and fielding averages) were not available for all 72 players. Practice Behaviors and Players' Psychosocial Development Three separate regression analyses were conducted to assess the relationship between coaching behaviors (as exhibited in practices). players' attained skill level and thier psychosocial growth over the season. The first set of analyses used players' perceptions of competence as the dependent set of variables, and the last two analyses used gains in players' perceptions of control, in both the general physical domain and in the more softball specific domain. Changes in Perceived Competence A significant multivariate relationship was found to exist between the behaviors exhibited by coaches across four practice sessions and changes in the players' perceptions of competence, EiSO, 263) = 1.73, ‘p§.004. Examination of the univariate E_tests for each dependent 141 measure indicated that the predictor variables significantly influence three of the measures of perceived competence (Success Expectancy, [(10, 61) = 2.38, 25.02; Physical Competence, EJ10'51)=2°40’.EF-023 and Cognitive Competence, E!10,51)=3°13’.Ef°003) while no significance was found for perceived social competence, [(10,61)=1.23,.p§.29, or perceived general competence, EXIO, 61)=1.88, £5.07. Examination of the standardized regression coefficients (Table 19) consistently shows, for each measure of perceived competence, that the coaching behaviors exhibited in response to a desirable perfonnance (i.e. reinforcement and nonreinforcement) were the most influential detenninants of players' psychosocial develOpment. The negative sign associated with each of these coefficients suggests that both reinforcement and non-reinforcement were inversely associated with positive increases in players' perceptions of competence. Because no a priori ordering of either criterion or predictor variables was advanced, the use of both the step-down and the step-wise regression analyses to test the significance of each of the variables in contributing to the predictive equation was not deemed apprOpriate (Finn, 1974). However, based on the coaching effectiveness model, it had been hypothesized that coaching behaviors would contribute to the prediction of players' perceptions of canpetence and control above that accounted for by players' actual ability. Therefore, for this and all succeeding regression analyses, a covariate grouping key was used to test, in step-wise fashion, the additional contribution which all coaching behaviors in combination would provide to the regression 142 00m.- woo.- mmo.- Ham.- H00. 000. men. moo. 000. 00c.- ome. H00. omo.- 000. 000. 000. 000. m00.- 000. H00.- 000. 00m. 000. 000. 0mH.- mom. ~00. omH.- 00¢. 00m.- oo~.- 000.- 000.- Hfim.- com.- ¢m0.- mmo.- oHc.- 000.0- 000.- mmH.- oefi.- meo.- 00m.- 000.- Nem.- 000.- moH.- H0~.- 0mm.- mucopmasou mucmpmasou mucmpmasou mucmumanu xucmpumaxu m>wumcmoo Fmgmcmw —mwoom pmuwmaza mmmuusm vm>mmugmnT imw>wmugom wwwwmugma wm>wmugmu .memxmpa new mgo_>wsmm mowuuwga .mmgumou 0000 zommmpHmu xmucH copuuzgpmcm FoUwcsumh ucoezmpcaa new: copuuagumcfi quwczumh pcwchpcou-mxmumvz acmsxmwcaa mxmumwz acwgocmH pcmEmmmgsoucm pcomcwucou-mxmumpz cowpuzgumcfi peevccuwh pcmmcwpcou-mxmmeZ acmewugomcpoe-coz ucmpmugo$cwmm coppozgpmcH Pmu_:sump zuwz acmemugow:_mm xuwpwn< 0000 meromommm mcwmo mucoumasou um>wmugm¢ ”mucmwuww0mou cowmmmgmmm chwcgmucmpm .00 04m gouupvmga mem.- mucmumasou m>0pvcmou um>vmugma cam.- mucmumqsou pogmcmo um>wmugma mom.- mucmawgsau meuom cm>pmugmq mm0.- mucmpwasou qu_mxga uo>vmugoa 00m.- xucmuuwaxu mmwuusm mwcwumoJ mmpampgm> cowgmppgo upmuwcocmo mucmumqsoo um>wmugma .mngmpa cam mgov>ozmm mcwec muwuumea .mmsumou ”mm:_umo4 pmuP=ocmu .00 udm<~ 145 perceptions of competence decreased over the course of the season. In addition, however, two particular coaching behaviors also contribute to the correlation between the two sets of data. The frequency of nonreinforcement suggests an inverse relationship with conpetence perceptions while the extent of punishment or criticisn received as a response to a player's mistake was positively associated with the set of competence measures. Changes in Perceived Control In addition to assessing the strength of the relationship between coaching behaviors and players' gains in perceived competence, their perceptions of performance control were also analyzed as a function of practice behaviors and ranked ability. These results indicated no association between coaching behaviors, measured ability, and perceptions of control as assessed in the physical domain, §_(50, 263)=1.31,p$.09, or between coaching behaviors and perceptions of control as measured through a softball-specific control scale, [(50, 263)=.85, £5.85, suggesting, then, that the set of predictor variables (players' ability measure and observed coaching behaviors) cannot accurately predict changes in players' perceptions concerning who or what is responsible for their performance success or failure. Game Behaviors and Players' Psychosocial Development None of the tested relationships between the behaviors exhibited by coaches in game situations and measures of players' psychosocial development was found to be significant. Specifically, game behaviors 146 could not be accurately used to predict changes in players' perceptions of competence, f]45,263)=1.02,pf.44, or in players' perceptions of control either in the physical domain, {(45, 263)=1.14, pf.27, or in the more softball-specific domain, [(45, 263)=1.25,pf.15. Generally, then, it seens that coaching responses to players' performance during competitive play were either not as salient to these young athletes as were the coaches' behaviors in practice sessions or that the contextual situation (e.g., the relatively lesser amount of playing time and coaching communications received by some players in game situations) may have reduced the chances of finding a significant relationship between the two sets of data. Analyses of Coaching Effectiveness: Summary In summary, the results of the multivariate correlational analyses revealed that the set of predictor variables, which included measures of coaching behaviors as well as a single assessment of players‘ ability, accounted for a significant portion of the variance in players' canpetence perceptions. The redundancy index indicated that about 15% of the changes in players' perceived competence which occurred over the season can be accounted for by measures of their coaches' behavior and their own attained softball ability. In contrast, changes in the players' perceptions of perfonnance control however, were not found to be a function of this set of predictors. Secondly, this associative relationship between the two behavioral sets of data was significant only when the coaches' behaviors 147 collected during practice sessions were used. The comparable analyses which examined changes in players' psychosocial growth as a function of coaches' game behaviors was not found to be significant. Finally, the most salient coaching behaviors (those which contributed most highly to the relationship) appeared to be those which are evaluative in nature (i.e., reinforcement, nonreinforcement, and punishment). More specifically, both of the reinforcing indices were inversely associated with the canpetence measures while the frequency of mistake-contingent punishnent or criticisn was postively associated with such psychosocial measures. CHAPTER IV DISCUSSION Expectancy Effects One of the primary purposes of this study was to assess both the strength and direction of expectancy effects as they may occur in junior high interscholastic athletics. The results indicated that coaches exhibited differential patterns of behavior towards their high as compared to their low expectancy athletes. Specific differences were evident both in coaches' responses to players' successful performance as well as in their response to athletes' skill errors. Low expectancy athletes received a higher frequency of reinforcenent following a desirable skill perfonnance while high expectancy players experienced a higher proportion of nonreinforcenent. In response to players' skill errors, coaches were more apt to provide low expectancy athletes with mistake-contingent technical instruction or to ignore the error. High expectancy athletes, in contrast, received more punitive or critical coaching responses to mistakes, this criticisn occurring either alone or in combination with technical instruction. Finally, low expectancy athletes generally received more technical instruction as measured across all coaching communications than did high expectancy athletes. Although these results indicate that coaches in this investigation exhibited differential patterns of feedback to individual athletes, 148 149 these expectancy influences were primarily demonstrated in game situations. Comparable expectancy influences were not found in relation to coaching behaviors exhibited solely in practice sessions. This contextual limitation in regard to demonstrated expectancy effects was not only true for the quality/type of interactions provided athletes but also for the relative frequency with which such coach-athlete interactions took place. Coaches were found to direct significantly more communications to high expectancy players during games, even when the frequency of such interactions was controlled for the number of innings athletes actually played. In contrast, no differences were found in practice sessions. That is, coaches interacted with low expectancy players in practices as frequently as they did their high expectancy group. The denonstrated expectancy effects in relation to frequency of coach-athlete interactions in game situations may be attributed to the field positions to which high and low expectancy athletes were assigned. Infield positions were predominantly given to high expectancy athletes while most of the low expectancy athletes (Pre-season, g_= 15; Post-season,.n_= 14) were outfielders. Since much of the action during game situations takes place in the infield, it is likely that the greater nunber of coaching communications to high expectancy athletes may be partially explained by their position as infielders. Similarly, it is also possible that coaches direct more performance-related communications to high expectancy athletes during competitive situations because such athletes were perceived by coaches 150 to be more crucial to game success, and thus their perfonnance was more salient to the coaches. In general, then, the results of these analyses indicated that coaches' behavioral patterns towards their low and high expectancy athletes differed depending on the contextual situation. Interestingly, pair-wise comparisons, using all 72 subjects, indicated that coach-athlete interactions markedly differed fron game to practice sessions for all athletes. In fact, multivariate correlational analyses revealed that no associative relationship could be found between the behavioral patterns exhibited to individual players in each of these two types of situations. These results suggest, then, that the contextual situation will certainly influence the type of information collected through observation of coaching behaviors, and further implies that a complete assessment of coaching effectiveness cannot be obtained unless observational data from both settings (i.e., practices and games) are examined. Although the results of this investigation indicated that coaches' behaviors towards individual athletes may, at least in some part, be predicted by their expectations concerning the athletes' ability, caution must be exercised in regard to the interpretation of such demonstrated expectancy findings. The traditional conception of expectancy effects as they occur in instructional contexts assumes that the behaviors exhibited by the instructor (coach) towards individual students (athletes) will be biased by the instructor's expectations concerning the children's ability. Such bias is reflected in 151 instructional behaviors which provide greater learning Opportunities for high expectancy children and which ultimately enhance or increase the disparity between the achievement of high and low expectancy groups. Although coaches in this study were observed to employ different patterns of behavior, the direction of these differences was ngt_consistent with the traditional conception of expectancy effects. That is, low expectancy players actually received more technical instruction and feedback, both in general as well as in mistake-contingent situations. This finding seems to indicate that coaches were trying to provide the most information to those players who, in their Opinion, had the lowest amount of skill. Similarly, such low expectancy athletes also received more reinforcement after a successful perfonnance than did high expectancy athletes whose successes were more often ignored (Nonreinforcement). Coaches, then, displayed a tendency to "make the most" of the low ability players' successful perfonmances, presunably as compensation for their lower success rate and to motivate them to continue practice efforts. Therefore, in this study, it appears that the differential coaching behaviors actually represented instructional techniques consciously employed by the coach to meet the needs of the individual player. Coaches' expectations, then, may be more accurately interpreted as perceptions of players' ability which induced these coaches to provide differential treatment to athletes based on their assessment of the athletes' needs. This interpretation seems especially credible considering the results of the analyses concerning the frequency of 152 coach-athlete interactions. Although it was demonstrated that high expectancy athletes initiated more interactions with their coach in practice sessions, there were actually no differences in the total nunber of coach-athlete interactions, indicating that coaches may have compensated for the greater tendency of the high ability players to interact with them by deliberately initiating more such interactional situations with the low expectancy athletes. Therefore, it does seem likely that the expectancy effects encountered in this investigation may actually reflect differential instructional techniques employed by coaches to facilitate the perfonnance and motivation of their low ability athletes, rather than discriminatory or biased behavior towards the high ability player. To hypothesize that the expectancy effects demonstrated in this study are actually forms of individualized instruction does not automatically imply, however, that these differential patterns of behavior are in actuality facilitative of players' performance and motivation. Because very little Sport science research has been reported relative to effective coaching behaviors, either in terms of players' skill performance or in relation to their psychosocial growth, we do not know what instructional behaviors are most beneficial for young athletes. Therefore, the relatively greater anounts of reinforcement and technical instruction given by coaches in this study to their low expectancy athletes may or may not be conducive to gains in their perfonnance, attitude or cognitions. Until further 153 information is available concerning effective coaching behaviors, the implications of the denonstrated differential patterns of behavior to selected groups of athletes cannot be adequately assessed. However, the results of the second part of this study may provide some additional infonnation concerning this issue. 154 Coaching Effectiveness The second major purpose of this study was to determine if changes in players' psychosocial develOpment over the course of a competitive season could be predicted by a measure of their skill ability in combination with measures of the coaches' evaluation of their skill perfonnance. This predictive relationship was found to be significant, but in a somewhat limited context. That is, this relationship was statistically significant at (or less than) an adequate level of confidence (i.e., pf.05). However, calculation of the redundancy index indicates that only 15% of the variation in the criterion variables (changes in players' perceptions of competence) is accounted for by a combination of actual player ability and observed coaching behaviors, leaving about 85% of the variance unexplained. It is equally true, however, that 15% of the variation in players' psychosocial growth over the course of the season represents a meaningful and substantial portion of that growth, especially when considering the myriad of other influences which certainly may be contributing to the child's self-perceptions. Therefore, the theoretical and practical importance of this relationship for further study can be justified. Secondly, although the statistical step-wise hypotheses testing indicated that coaching behaviors contributed to the prediction equation above and beyond that provided by the ability measure, the 15% of the variance accounted for cannot all be attributed to the observed measures of 155 coaching behavior. With these limitations in mind, then, a few general discussion points concerning the denonstrated results appear to be valid. First, the results suggest that players' ability and coaching behaviors influenced changes in the perceived competence of young athletes, but correspondingly seemed to have little effect on changes in their perceptions of performance control. In contrast to these findings, Harter's (1981) model theorizes that adult evaluation should indirectly influence children's perception of competence through direct effects on their perceptions of performance control. However, it may be that the athletes in this study were more dependent on coaches' evaluation of their perfonmance in order to make judgments concerning their competence due to their lack of previous experience in softball and in sport in general. Certainly, these hypothesized path relationships should be explored with an additional group of athletes varying in age and past sport experience. Secondly, although the relationship between the two behavioral sets was tested in two independent analyses using practice and gane behaviors as predictor variables, the results suggest that coaching behaviors exhibited in practice situations provided the strongest and most significant prediction for changes in players’ perceptions of competence. This may be due to the relatively greater nunber of practice sessions as compared to game situations (approximately a 2.5 to 1 ratio), or may indicate that players perceive coaches' behaviors 156 in practice as more salient indicators of their ability than coaches' game behaviors. In addition to providing some support for the general relationship between player ability, coaching behaviors, and players' psychosocial growth, the results of the canonical correlation analyses provided some infonnation concerning the individual predictor variables which most contribute to the regression equation. These results indicated that players' ability (as measured by teammate evaluation) was a consistent predictor of changes in players' perceptions of competence. This associative relationship between individuals' actual ability with their perceptions of ability has also been advocated by Bandura (1977) in his theory of self-efficacy. He suggested that past related perfonmance accomplishments (personal mastery experiences) play a central role in determining the strength of the individual's belief in his/her ability to successfully execute an achievement task. Research designed specifically to examine this theory in sport-related activity has supported the predictive influence of perfonnance attainment on measures of participants' self-efficacy (Feltz, 1982; Weinberg, Gould & Jackson, 1980). Therefore, the results of this investigation, which indicated that players‘ attained skill level was a significant predictor of changes in their perceptions of personal competence is consistent with the previous sport psychological research. The demonstrated importance of actual ability as a detenninant of competence perceptions emphasizes the necessity for including in future research attempts, the assessment of coaching behaviors both in tenns 157 of their facilitation of players' skill acquisition as well as in tenns of their influence on players' psychosocial growth. The results of this study suggest that coaching behaviors may exert their greatest influence on players' perceptions of competence through their facilitation of players' actual skill acquisition. Although perfonmance accomplishment (post-season ability ranking) was identified as the primary predictor of players' psychosocial growth over the season, step-wise hypothesis testing in this study additionally indicated that coaching behaviors contributed to the prediction of changes in players' perceptions of competence over and above that provided by the ability measure. Those coaching behaviors which were specifically identified as influential contributors were those which contained an evaluative component (i.e., frequencies of reinforcement, nonreinforcement, and criticism or punishment). Interestingly, coaches' reSponses to players' successful performance (e.g., reinforcenent and nonreinforcement) quite consistently contributed a negative influence to the equation, suggesting that high frequencies of either behavior were not facilitative of players‘ development of perceived competence, while the punishment or criticism conponent seemed to be positively associated with gains in perceived competence. Although exact and definitive interpretations of these findings are certainly difficult without additional information and further replication, these results can best be explained with respect to the contingency dimension of performance feedback. Both Harter's model, as 158 well as the prOposed coaching effectiveness model, hypothesize that clear, consistent and perfonnance-contingent feedback from significant adult evaluators will lead to significant increases in children's perceptions of competence and control. However, results from the teaching effectiveness literature indicate with some consistency that teacher praise as an instructional behavior is either not an influential contributor to students' performance or is even negatively correlated with such perfonnance gains (see reviews by BrOphy, 1979; Good, 1979). Brophy (1979) has commented that this negative relationship may exist because reinforcement is not used by teachers as a perfonmance-contingent and apprOpriate evaluation of the child's perfonnance but is often utilized for motivational and disciplinary purposes. In this particular study, the results of the expectancy analyses indicated that low expectancy (ability) athletes received a higher proportion of reinforcement for positive perfonnance than did high ability players. Given the lower skill level of the low expectancy players, many of these reinforcements may have been inappropriate and thus non-contingent to perfonnance. As a result, reinforcement as an instructional/coaching behavior may not be facilitative of players' perceptions of competence because it does not represent a contingent and apprOpriate mode of performance feedback. In contrast, punishment, given by coaches as a response to errors in skill perfonnance, may actually have been given in a more contingent manner. High expectancy athletes, in this study, did receive more punishnent-oriented response to their skill errors, and it may be that 159 these young players, then, perceived punishment as an apprOpriate and infonnative evaluation of their perfonnance and an indication that their coaches expected them to perform at a higher level, thus facilitating higher perceptions of competence. Therefore, it is suggested that the contingency/apprOpriateness of the coaching behavior may be a more salient source of information for the player than the actual behavior itself. Although the accuracy of this contingency interpretation cannot be directly assessed in this investigation, this explanation is consistent with two particular research findings fron the teaching behavior literature. First, several researchers have demonstrated that inappropriate reinforcement of low ability students is one of the means through which expectancy effects are exhibited in the academic classroon (Kleinfeld, 1975; Rowe, 1974; Heinstein, 1976). This inapprOpriate reinforcement often takes the fonm of providing praise or reward for incorrect performance or accepting lower quality of perfonnance from low ability students. Secondly, Meyer and his colleagues (Meyer, Bachmann, Biennann, Hempelmann, Ploger, & Spiller, 1979) conducted a series of six laboratory studies to examine whether differential patterns of evaluative feedback (praise and criticism) given to students actually provided observers and participants with infonnation concerning the students' ability. These researchers consistently found that students who received praise after success and neutral reaction after failure at very easy tasks were perceived to be low in task ability. In contrast, students who were given neutral 160 reactions after successful perfonnance and criticism after failure were perceived as high ability students. These patterns of behavior identified by Meyer et al. confonn very closely to the differential patterns of behavior exhibited by the coaches in the present study. That is, low ability athletes were given relatively more reinforcement or praise for success and were ignored more often after a mistake. In contrast, high ability athletes experienced higher frequencies of nonreinforcenent after success but more criticism following failure. Based on the results of Meyer et al., it seems likely, then, that the contingency or appropriateness of coaches' responses to players' performance may have influenced players' perceptions of competence. Future Research Directions in Coaching Effectiveness Because the present study represented an exploratory attenpt to identify some correlates of effective coaching behaviors, these results should primarily be used to delineate some profitable avenues for future research. Based on the initial results from this study, one of the major research needs would be to further investigate the contingency and apprOpriateness of coaches' instructional feedback. Specifically, do coaches enploy perfonnance-contingent patterns of feedback and consistent standards of perfonnance for all players, or do they selectively reward/respond to players on the basis of their expectations concerning players' ability or their desire to motivate lower-skilled players? 161 Correspondingly, of course, even if the feedback provided by coaches to young athletes (especially low ability players) is not appropriately and contingently given, it would also be necessary to detenmine if such evaluative feedback does differentially influence the performance and psychosocial reponses of these athletes. Therefore, players' perceptions of the information provided by coaches in relation to their skill perfonmance must also be assessed. Secondly, the relationship between coaching behaviors and players' perceptions of competence and control might be more accurately assessed if future investigators included more skill-specific measures of relevant variables. The players at this age and skill level tended to partition their self-assessments of over-all ability into more skill-specific components (e.g., "I an a good batter but not a good fielder."). Actually, evaluation of player abilities based on observers' analyses suggests that for many players at this level, this division of ability according to sub-skills may be based on actual performance differences (e.g., designated hitters usually only batted, and pitchers only pitched). In contrast, the psychometric measures of perceived competence and control were assessed in more general ways. Even the softball-specific subscales which were specifically designed for this study assuned that perceptions of control would be similar across all skills. The markedly lower reliability estimates of this set of scales may reflect the inability of players to assess their ability as a softball player in consistent ways when, in actuality, they tend to partition such assessments. 162 Similarly, even the collection of coaches' behavioral data may be more reliably and accurately assessed if skill Specificity is also built into the observational instrument. In this study, both of the trained coders observed differences in the type of instructional feedback which coaches gave to players as a function of the particular skill involved. For example, virtually all of the feedback given to pitchers in response to their performance was supportive or evaluative (reinforcement, mistake-contingent encouragement) rather than technically instructive. In addition, individual coaches also differed considerably in instructional patterns from one subskill to another. A few of the coaches provided very specific feedback in response to players' fielding performance but utilized more evaluative or general instructional behavioral patterns when conducting batting practice. These differences in instructional patterns from one subskill to another may be a reflection of the coaches' own skill-specific expertise. Yerg (1980) found that significant differences in instructional behaviors between individual teachers were a function of their own knowledge or personal skill level. Obviously, if coaches' instructional behaviors are dependent on the particular skill involved, then players may also develOp differential perceptions of their ability in each of the subskills. Therefore, the need for more specific assessment is indicated. To assess coaching effectiveness by dividing data collection according to subskills will certainly complicate the whole observational process but 163 might provide a more accurate assessment of the relationship between instructional behaviors and players' self perceptions. Finally, although this study provides limited support for the hypothesized relationship between coaching behaviors and individual players' psychosocial growth, more specific information might be obtained if individual variation between players was also taken into account. Research from the literature on teaching effectiveness suggests that the correlates of effective instruction are dependent upon situational factors but also on the characteristics of the learners (Brophy, Note 6). In relation to coaching effectiveness, Smith et al. (1979) found that gains in self-esteem and attitude toward participation as a function of coaching behaviors were significantly greater for those Little League athletes who had begun the season with relatively lower levels of self-esteem. In this study, too, it might be hypothesized that the statistical attempt to identify effective coaching behaviors was complicated by such differences between players. Therefore, although the original design of this study did not include comparisons between groups of athletes in relation to coaching effectiveness, additional analyses were conducted to determine if such variation existed. This re-exwmination of the data was conducted primarily to outline some exploratory direction for future research and was not intended to be a definitive assessment of coaching effectiveness. 164 A Re-Examination Of Coach-Athlete Interactions Based on the results Of the Smith et al. (1979) study, it was hypothesized that the players' initial level Of perceived competence might influence their susceptibility to certain coaching behaviors. That is, perhaps the coaching behaviors which would be most facilitative of the psychosocial growth Of these young athletes would be different for low perceived competence players as Opposed to their higher perceived competence teammates. Therefore, it was determined to statistically assess differences between those players who initially scored high on the perceived competence subscale and those who Obtained low perceived competency scores. For this purpose two groups Of athletes were selected from the total number Of 72 available subjects. Approximately 1/3 Of the players on each team were designated as low perceived competency players (3&21) and an additional 1/3 were designated as high competency athletes (gf25). The series Of regression analyses used earlier to determine the predictive relationship between coaching behaviors and players' ability (predictor set) and changes in players' perceptions of competence and control (criterion sets) was again conducted using only these two groups Of athletes (Mf46). However, a group comparisons test was also added to the analyses for the purpose Of detenmining if the regression planes for the two groups differed. This test for the parallelism of regression planes (Finn, 1974) is designed to assess the homogeneity or equality Of the regression equations for two or more groups. Obtaining a significant difference on this test indicates that the set Of regression weights for one group is statistically different 165 from the set of weights for the other group. In this study, if group differences were found, then separate multivariate multiple regression analyses (exactly as conducted in the previous analyses with all 72 players) were perfonmed for each Of the two groups to Obtain estimates Of these statistically unique regression weights. The results of these parallelism regression tests indicated that three Of the nine tested regression planes were not equal across the two comparison groups, suggesting that the strength and the nature (i.e., the size and sign Of the regression weights) Of the predictive relationship between players' ability, coaching behaviors and players' psychosocial growth differed as a function Of their level Of perceived competence. These parallelism results, presented in more detailed fonm in Appendix E (Table 21) indicated that the majority Of the differences between high and low perceived competence athletes in their response to coaches' feedback occurred when the criterion variables were changes in athletes' perceptions of perfonmance control. That is, the three regression planes which were found to be statistically different for the two groups Of athletes were those which measured the influence of coaching behaviors on changes in players' perceptions Of performance control. The follow-up independent multivariate regression analyses, (detailed in Appendix E, Table 22) conducted for each Of the two groups suggested that low perceived competence players were most strongly influenced by the set of predictor variables (which include coaches' practice behaviors and the ability measure) than were high perceived 166 competence athletes. Standardized regression coefficients (Table 23, Appendix E) and canonical loadings (Table 24, Appendix E) indicated that low competency athletes' beliefs in an unknown source Of control were positively correlated with nonreinforcement (as frequency of nonreinforcement increases, so does the belief in an unknown source Of performance control) but negatively correlated with reinforcement which was accompanied by technical instruction (i.e., not simply an evaluative response to a positive perfonmance). Understandably, Of course, the strength of the low perceived competency athletes' belief in unknown sources increased when ability was rated low in comparison tO teammates. The associative relationship between low competency athletes' ability, their coach's behavior and changes in their perceptions Of perfonmance control was correspondingly not shown to be significant for high competency athletes, indicating that variation in their perceptions Of performance control could not be accounted for by measures Of coaching behaviors or ability level. Although the results Of this series Of regression analyses must be accepted with caution because Of their a posteriori nature and because only a small number Of subjects was assigned tO each group, these exploratory findings demonstrated that variations among individual athletes influenced the identification Of effective coaching behaviors. These results suggest that young athletes who have low perceptions Of competence in relation to physical activity may be more easily influenced by their coaches' behaviors towards them than their teammates with high levels Of perceived competence Smith et al. (1979) 167 also found that the attitudes and self perceptions of young athletes with initially low levels Of self-esteem were more susceptible to coaches' behavior than their high self-esteem peers. The results Of these analyses also suggest that coaches' behavior may be most influential in relation to the athletes' perceptions Of perfonmance control (i.e., their beliefs concerning who or what is responsible for athletic success or failure). Certainly these analyses emphasize the importance Of considering individual variation among athletes as a determinant Of the correlates of coaching effectiveness. 168 Conclusions and Implications Although the statistical results Of this study provided some important information relative to coaches' perceptions and behaviors and their players' subsequent psychosocial growth over the season, the applicability Of such findings to other athletic situations must certainly be made with caution. The athletes in this study were all females with relatively little previous athletic experience. Their interscholastic coaches were certified and eXperienced teachers (or teachers in training) who had been coaching for a minimum Of two years. Certainly, then, the statistical results as well as the points raised in the ensuing discussion are limited to athletes and coaches from a similar population. As several writers have indicated, the key to determining the correlates Of teaching (coaching) effectiveness is the implementation of field-based, process-product research (with successive replication) across a variety Of contextual situations (Gage, 1979; Locke, 1977; Yinger, Note 10). This study, then, represents only initial work with young athletes, and replication and extension Of such results are needed before generalizable conclusions can be made. Methodological Issues Although the Specific findings from this study may not be generalizable, some methodological information was acquired which 169 certainly may be applicable to future research in coaching behavior. First, and perhaps most Obviously, statistical analyses indicated that the context within which coach-athlete interactions took place influenced the Obtained results. Comparison Of coaching behaviors in game and practice situations revealed very little relationship between the patterns of coaches' responses to players' perfonmance in each Of these two Situations. Additionally, when coach-athlete interactions were examined as a function Of coaches' perceptions of player ability, significant effects were found for game behaviors only. It is obvious, then, that research results may be highly dependent on the context used. Previous researchers who have examined coaching behavior have utilized either game behaviors only (Smith et al., 1979), practice behaviors only (Tharp & Gallimore, 1976), or a combination (i.e., summing together) of practice and game behaviors (Rejeski et al., 1979). The issue Of contextual influences, then, has not previously been examined in the sport psychological literature. A second methodological issue addressed in this study concerns the necessity for imposing some control on inherent differences between groups Of athletes when examining coaching behavior. Brophy (Note 20) has pointed out that high expectancy (ability) students present more Opportunities to the teacher for positive instructional interactions (i.e., higher success rates encourage higher rates Of instructional reinforcement) than do low ability students. Therefore, even if the teacher reacts consistently towards all students, statistical analysis Of rate or percentage scores will indicate higher rates of 170 reinforcement to high ability students. BrOphy strongly recommended that Observational measures Of instructional behaviors must be adjusted for differences in the Opportunities presented to the instructor by the child. The index scores used in this study as a measure Of coaching behaviors were specifically chosen tO control for differences in such success and failure rates. The importance of controlling for player behaviors which may influence coaching behaviors was specifically demonstrated in another way in this study. Significant expectancy effects were found in relation tO the frequency Of coach-athlete interactions. Specifically, the results indicated that high ability athletes experienced more such interactions with their coach in practice situations than did low ability players. However, when the number Of coach-athlete interactions initiated by the athlete was subtracted from the total, no expectancy effects were found. These results suggest that high ability athletes initiated more communications with their coach, but the coach, in turn, initiated more such interactions with low ability players. These findings demonstrate the necessity of controlling for the Opportunities presented to the coach by groups Of athletes. Similiarly, coach-athlete interactions may be influenced by the defensive positions which athletes play. Descriptive statistics from this study revealed that high ability athletes at this level predominantly played infield positions while low expectancy athletes were most Often assigned to the outfield. The small number Of high and low expectancy athletes in this study precluded the possibility Of 171 statistically testing an Ability Group by Position interaction. However, such analyses in future research studies may provide a more accurate assessment of differential patterns Of coaching behavior. In summary, the results Of this study did uncover some methodological issues that were demonstrated to influence the assessment Of coaching behavior. Certainly these issues neeed to be considered in future research designed tO examine the effectiveness Of coaches in facilitating the learning and perfonmance Of young athletes. Theoretical Implications The purposes and design Of this study were specifically based on a proposed coaching effectiveness model which was develOped in Chapter 1 Of this paper following a review Of the related literature. Although only certain aspects of this model were tested in the present investigation, it seems most apprOpriate to conclude this discussion by examining the Obtained results in relation to this model (outlined in Chapter 1). First, the hypothesized relationship between coaches' expectations concerning players' ability and their subsequent behavior towards these individuals was demonstrated. Although, coaches exhibited differential behaviors tO their high and low expectancy athletes, however, nO evidence was provided to support the contention that high expectancy athletes would receive higher frequencies Of instructional behavior or 172 more positive evaluation Of perfonmance. In actuality, it was the low expectancy players who received more such instruction and reinforcement, and correspondingly, less punitive evaluation Of skill errors. However, the implications Of these differential patterns Of coaches behavior cannot be assessed until more specific infonmation is Obtained relative to coaching effectiveness. The second hypothesized path to be tested was that leading from the players' perfonmance and coaches' evaluation of that performance tO players' perceptions Of competence and control. Results Of these analyses indicated that some Of the variation in players' psychosocial growth over the season could be predicted by a measure Of their actual softball competence apd their coaches' behavior towards them in practice sessions. Although the degree Of success these players had attained over the season was consistently associated with the develOpment Of perceived competence, certain coaching behaviors (e.g., reinforcement, nonreinforcement, punishment) also were identified as important contributors. The demonstrated association Of these coaching behaviors with players' perceptions Of competence was interpreted in light Of the contingency Of these instructional behaviors to perfonmance outcome. It was suggested that both reinforcement and nonreinforcement may be negative contributors to the development Of players' perceived competence because both Of these behaviors did not provide apprOpriate and contingent information concerning perfonmance outcome. Punishment or criticism Of players' perfonmance, in contrast, may be contingently and apprOpriately given, thus providing players 173 with specific infonmation concerning their performance and contributing to the development Of perceived competence. Finally, the results Of the analyses using all players indicated that coaching behaviors and players' attained Skill ranking were significantly associated with changes in players' perceptions Of competence but not with changes in their perceptions Of perfonmance control. However, Similar analyses conducted with two smaller groups Of athletes (high and low perceived competence players). indicated that players who began the season with low levels Of perceived physical competence were more influenced by the events which occurred during the season (i.e., their attained skill and their coaches' behavior towards them) than their teammates with higher levels Of perceived competence. For these low competency athletes, the hypothesized relationship between coaches' evaluation of their performance and their perceptions Of control was found tO be significant. These results suggest that individual variation between athletes will influence their susceptibility to coaches' behavior. Additionally, these findings indicate that the demonstrated relationship between players' perfonmance accomplishments, coaches response towards those accomplishments, and players' develOpment of competence should be examined as a function Of a number Of other factors including the athletes' age, gender, past experience, and competitive skill level. In summary, although the results Of this exploratory investigation did not establish conclusive support for the prOposed model, they indicated that it may be a viable means of 174 develOping and testing theoretically-based hypotheses regarding the influence which participation in competitive athletics exerts on the psychosocial growth Of young athletes. APPENDIX A LETTER TO SCHOOL PERSONNEL 175 February 19 , 19 82 Athletic Personnel Lansing Public Schools The Michigan Youth Sports Institute was established in 1978 for the purpose of assisting parents, coaches, and other sport leaders provide positive and beneficial sport experiences for children. One of the ways in which the Institute has attempted to accomplish this objective is through a continuing research program designed to provide us with informadon daout young athletes and the effects of sports participation on their physical and psychological development. Through my work as a doctoral student with the Youth Sports Institute's research program, I have developed a particular interest in studying the experiences which fell-ale athletes receive through participation in competitive sport program. Since the advent of Title UK, the opportunities for girls to participate in competitive sport program have greatly increased. However, much of the information concerning the influence of sport participation on young athletes has been obtained through research with males. Therefore, my dissertation study, which is being conducted under the supervision of Dr. Dan Gould of the Youth Sports Institute, represents an exploratory investigation designed to identify and assess the achievement behaviors which females commonly exhibit in athletic situations, especially in regard to individual coach- athlete interactions. Data will be collected for this study through observation of coaches and athletes in gene and practice situations as well as through administration of questionaires and surveys to measure the attitudes and motivation of female athletes toward competitive sport participation. Because most females begin participation in canpetitive athletics during the seventh through ninth grades, I am planning to conduct this study using the junior high softball teams in the Capital Area Conference. Each team will be observed approximately one day a week (three games and four practices) during the course of the entire season. Additionally, all players will be asked to complete a survey form assessing their motivation for and attitudes towards sport participation. This testing 176 session (approximately 20-30 minutes in length) must be completed once at the beginning of the season and once at the end of the playing season. The purposes and procedures of this study have already been briefly explained to you by phone, however this letter provides more specific information concerning the project. A copy of this letter is also enclosed which, contingent on your approval, should be given to the junior high team softball coach as his or her permission is also needed before we can proceed with the study; IMnally. each athlete and her parent(s) or guardian will also be informed by letter of the purposes and procedures. as well as the voluntary nature, of this study and will be asked to give their permission for their daughter to participate. You can be assured that all of the information collected during the course of this study (e.g., recorded behaviors as well as survey responses) will be kept strictly confidential, and the identity of all players, coaches, and school systems will remain anonyggue. Furthermore, the researcher will be making no personal contact with individual players (other than during the administration of the suerY). and the observer(s) will 225,1nterfere with practice activities. The general findings obtained through analyses of the data from all six schools combined will be sent to all interested parties, including athletes, parents, coaches, and athletic administrators. Your school's participation in this project will be greatly appreciated as it will allow us as researchers the opportunity to collect rdalistic and field-based data which we can use to assess the attitudes and motivation of children in sport activity. It is only through the combined efforts of researchers and sport leaders that quality athletic experiences may be made available for all children. we look forward to working with each of you on this project and will send each coach a letter, which should arrive about March 15, detailing the specific data collection procedures. If you have any questions concerning this study. please contact me at 353-6652. Sincerely Thelma Sternberg Born Youth Sports Institute 0205 IM-Sports Circle Michigan State University East Lansing, Michigan 48824 Phone: 353-4652 APPENDIX B PARENT LETTER AND CONSENT FORM 177 Thelma Sternberg Born Youth Sports Institute Michigan State University last Lansing, liichigan 48824 Dear Parent(s) or Guardian: The Michigan Youth Sports Institute was established in 1978 for the purpose of assisting parents, coaches, and other adult leaders pro- vide positive and beneficial sport experiences for children. One of the ways in which the Institute has attempted to accomplish this objective is through a continuing research progr- designed to provide us with information about young athletes and the effects of sport participation on their physical and psychological develOpment. Through my work as a doctoral student with the Youth Sports Institute's research program, I have developed a particular interest in studying the experiences which female athletes receive through partici- pation in competive sports programs. Since the advent of Title IX, the Opportunities for girls to participate in competitive sport programs have greatly increased. However, much of the information concerning the influence of sport participation on young athletes has been ob- tained through research with males. Therefore, my dissertation study, which is being conducted under the supervision of Dr. Dan Gould of the Youth Sports Institute, represents an exploratory study designed to identify and assess the achievement behaviors which females con-only exhibit in athletic situations, especially in regard to individual coach-athlete interactions. , i The data collection procedures for this study include the observa- tion of coaches and athletes in game and practice situations and the administration of a survey to assess the attitudes and motivation of - female athletes towards competitive sport participation. Because most females begin participation in competitive athletics during seventh through ninth grades, I am planning to conduct this study using junior high softball teams in the Capital Area Conference. Each team will be observed approximately one day a week (three games and three practices) during the course of the entire season. Additionally, all players will be asked to complete a survey form assessing their motivation for a- chievement in sport. This testing session (approximately 20-30 minutes in length) must be completed once at the beginning and once at the end of the playing season. 178 All of the information collected during the course of this study (e.g., recorded behaviors as well as survey responses) will be kept strictll confidential, and the identity of all players, coaches, and school systems will remain anonymous. Individual athletes, coaches, or parents will be free to discontinue participation in this project at 321 time during the course of the study. Furthermore, the researcher will be making no personal contact with the individual players (other than during the admdnistration of the survIY). and the observer will not interfere with practice activities. After the study has been com- pleted, information concerning its findings will be sent to all inter- ested parties including athletes, coaches, athletic administrators and parents. The purposes and procedures of this study have already been ex- plained to your daughter's coach and the school's athletic director, and each of them has agreed to participate in this project. However, the approval of all athletes mid their parents/guardians is also needed; therefore, each of the athletes will also be informed of the purposes of the study and will be asked to volunteer their participa- tion. If an athlete indicates that she does not want to participate, her decision will be respected, and she will not be surveyed or observed. This letter constitutes a request for your permission to allow your daughter to participate in this study. Once again, be assured that all information collected will be totally confidential, and your daughter's name will be replaced with a subject number as soon as the information is collected. If you do approve of the purposes of this study and will allow your child to participate, then please complete the attached form and return it to the address listed at the bottom of this letter or have your daughter return the form to her coach. If you have any questions concerning this project, you can call or write me at the address listed below. Your permission will be greatly appreciated as it will allow us as researchers to collect information concerning the attitudes of female athletes. It is only through studies such as these that more knowledge concerning the values of sport participation for all children can be gained. Thelma Sternberg Horn 1760 Nanoke Trail Haslett, Michigan 48840 Phone: 349-6638 179 PARENTAL CONSENT FORM Youth Sports Institute Michigan State University I have read the information contained in the accompanying letter concerning the proposed project which is being conducted with female athletes in the Lansing (Holt or East Lansing) school districts and I will give permission to let my daughter, participate as a volunteer in the scientific study being conducted by: Thelma Sternberg Horn under the supervision of Dr. Daniel Gould, Assistant Professor of Health and Physical Education at Michigan State University. The study has been explained to me and I understand whatmpy daughter's participation will involve. I understand that I (or my daughter) am free to withdraw my consent and discontinue my child's participation at any time. I understand that the results of the study will be treated in strict confidence and that my daughter's identity will remain anonymous. within these restrictions, results of the study will be made available to me. I understand that my daughter's participation in the study does not guarantee any beneficial results to her or to me. I understand that I can receive additional explanation of the study, at my request, after my daughter's participation is completed. SIGNED DATE APPENDIX C PSYCHOMETRIC TEST BATTERY 180 —.~cvr. ~.o\-e PERCEIVED COMPEI'ENCE SCALE FOR CHILDRDI - »- a . E '.~J—--‘“¢1‘.e. 7....- a WhatIAmI-ike H... ‘ 4. .ar-.. -. -. - «a 1‘- ...rer .. a ...e _. NAME BOY OR GIRL AGE BIRTHDAY___ CLASS 0R GROUP (circle which) SAMPLE SENTENCES REALLY SORT OF SORT OF REALLY TRUE TRUE TRUE TRUE for me for me for use for me a. Some kids would rather play BUT Other kids would rather watch T.V. outdoors in their spare time b. Some kids never worry about BUT Other kids sometimes worry about anything certain things. 1. Some kids feel that they are very BUT Other kids worry about whether good at their school work they can do the school work assimed to them. 2. Some kids find it hard to make BUT For other kids it‘s pretty easy. friends 3. Some kids do very well at all kinds BUT Others don't feel that they are very of sports good when it comes to sports. 4. Some kids feel that there are alot of BUT Other kids would like to stay pretty things about themselves that they much the same. would change if they could 5. Some kids feel like they are just as BUT Other kids aren'tsosure and wonder smart as other kids their age if they are as smart. 6. Some kids have alot of friends BUT Other kids don't have very many friends. REALLY SORT OF TRUE TRUE for me for me 7. 10. 11. 12. 13. 14. 15. 16. Some kids wish they could be alot better at sports Some kids are pretty sure of themselves Some kids are pretty slow in finishing their school work Some kids don't think they are a very important member of their class Some kids think they could do well at just about any new outdoor activity they haven't tried before Some kids feel good about the way they act Some kids often forget what they learn Some kids are always doing things with alot of kids Some kids feel that they are better than others their age at sports Some kids think that maybe they are not a very good person 181 BUT BUT BUT BUT BUT BUT BUT BUT BUT BUT Other kids feel they are good enough. Other kids are not very sure of themselves. Other kids can do their school work quickly. Other kids think they are pretty important to their classmates. Other kids are afraid they might not do well atoutdoor things they haven't ever tried. Other kids wish they acted differently. Other kids can remember things euily. Other kids usually do things by themselves. Other kids don't feel they can play as well. Other kids are pretty sure that they are a good person. SORT OF REALLY TRUE for me TRUE forme REALLY SORT OF 17. 19. 21. 22. 23. 24. 25. 26. TRUE for me TRUE forme Some kids like school because they do we" in class Some kids wish that more kids liked them In games and sports some kids usually watch instead of play Some kids are very happy being the way they are Some kids wish it was easier to understand what they read Some kids are popular with others their age Some kids don't do well at new outdoor games Some kids aren't very happy with the way they do alot of things Some kids have trouble figuring out the answers in school Some kids are really easy to like 182 BUT Other kids don't like school because they aren't doing very well. BUT Others feel that most kids do like them. BUT Other kids usually play rather than just watch. BUT Other kids wish they were different. BUT Other kids don’t have any trouble understanding what they read. BUT Other kids are not very popular. BUT Other kids are good at new games right away. BUT Other kids think the way they do things is fine. BUT Other kids almost always can figure out the answers. BUT Other kids are kind of hard to like. SORT OF REALLY TRUE for me TRUE forme 183 REALLY SORT OF SORT OF REALLY TRUE TRUE TRUE TRUE for me for me for me for me 27. Some kids are among the last to be BUT Other kids are usually picked first. chosen for games 28. Some kids are usually sure that what BUT Other kids aren't so sure whether or they are doing is the right thing not they are doing the right thing. © Susan Harter, Ph.D. , University of Denver (Colorado Seminary), 7.978. 184» WMSICNAL MEASURE OP mum's mails OF CONTROL HHY THINGS HAPPEN Directions: Reach each sentence and decide which answer best decribes our feelin s concerning why things happen in sport s tuations. The first 12 sentences describe activity in agntggl‘while the last 12 sentences apply specifically to sanegguzsnous (A) I like chocolate ice crean better than vanilla ice creel. Very True Sort of True Mot Very True Not at all True (8) Host kids really like spinach. Very True Sort of True lot Very True lot at all True I. Hhen I win at a sport, a lot of tiles I can‘t figure out why I won. Very True Sort of True lot Very True lot at all True 2. I can be good at any sport if I try hard enough. Very True Sort of True lot Very True lot at all True 3. Hhen I play an outdoor gale against another kid, and I win. it's probably because the other kid didn't play well. Very True Sort of True lot Very True lot at all True 4. If I try to catch a ball and I eiss it, it's usually because I didn't try hard enough. Very True Sort of True lot Very True lot at all True 5. Vhen I lose an outdoor gale. it is usually because the kid I played against was such better at that gene to begin with. Very True Sort of True llot Very True lot at all True 6. lihen I don't win at an outdoor gene, the person I was playing against was probably a lot better than I was. Very True Sort of True lot Very True lot at all True 8. 9. 10. 11. 12. 13. 14. 15. 1135 when I don't win at an outdoor game. cost of the tine I can't figure out why. Very True Sort of True Not Very True Not at all True If I try a new sport and don't do very well, I wouldn't know why I couldn't do the skill well. Very True Sort of True Not Very True Not at all True If I an not too good at any athletic skill. it's usually because I haven't practiced enough. Very True Sort of True Not Very True Not at all True When I win at a sport. it's usually because the person I was playing against played badly. Very True Sort of True Not Very True Not at all True I can be good at any sport if I work hard enough. Very True Sort of True Not Very T. a Not at all True When I win at an outdoor game, a lot of tines I don‘t know why I won. Very True Sort of True Not Very True Not at all True THE FOLLOHING QUESTIONS APPLY SPECIFICALLY To SOFTBALL when I face a good pitcher and get a hit, it's usually because she wasn't pitching very well. Very True Sort of True Not Very True Not at all True Hhen I play very poorly during softball practice. I really never knon why I played so badly. Very True Sort of True Not Very True Not at all True If I would get five hits in one gene, it would probably happen because we were playing against a really weak teen. Very True Sort of True Not Very True Not at all True 16. 17. 18. 19. 20. 21. 22. 23. 24. 1136 If I want to be a good softball player, it's really up to ac to do t. Very True Sort of True Not Very True Not at all True If I hit two hone runs in one gene. I really wouldn't know why I had hit so well. Very True Sort of True Not Very True Not at all True If I an not too good at a particular softball skill. it's probably because I haven't practiced that skill enough. Very True Sort of True Not Very True lot at all True when I play very well in a softball gene, a lot of tines I really don't know why I playes so well. Very True Sort of True Not Very True Not at all True I could be a very good softball player if I would try hard enough. Very True Sort of True Not Very True Not at all True If I an trying to steal second base and I get thrown out by the catcher, it's probably because the catcher has a really good are. Very True Sort of True llot Very True Not at all True If I wouldn't get a single hit during a gale. I wouldn't be able to figure out why I didn't. Very True Sort of True Not Very True Not at all True Hhen I fail to catch a fly ball. it's probably because the batter is a really good hitter. Very True Sort of True Not Very True Not at all True If I an trying to throw a runner out at a base and I don't, it‘s usually qy own fault. Very True Sort of True Not Very True Not at all True DIRECTIONS: 187 GENERALIZED EZPECTANCY OF SPORT SUCCESS SGKUE Each of the lines below contains two adjectives -- one on the right and one on the left. You lust decide whether you are closer to the adjective on the left or to the one on the right. describes you. Put a checknark in the space which best IN SPORTS I HAVE BEEN Example: I LEARN ATHLETIC SKILLS 6. 7. 8. 9. WHEN 10. 11. 12. Happy Active Bad Successful Unnoticed A Winner Easily Fast Poorly Always LEARNING ATHLETIC SKILLS I Persist An Uncoordinated An Successful 0 MY ATHLETIC ABILITY IS 13. 14. 15. 16. 17. 18. 19. 20. Above Average Bad Superior Limited Praised By Others Encouraging Strong Horse Than Host People's Unhappy Inactive Good Unsuccessful Outstanding A Loser Hith Difficulty Slowly Hell Never Give Up An Coordinated An Unsuccessful Below Average Good Inferior Broad Ridiculed By Others Frustrating . leak Better Than host People's APPENDIX D COACHES' DEMOGRAPHIC QUESTIONAIRE COACH: 1883 COACHES' DEMOGRAPHIC QpESTIONAIHE This questionnaire Just asks you to list some demographic information about your coaching and athletic background. All answers will be confidential and my report will include a summary/average of information obtained across all participating coaches. I. COACHING BACKGROUND A. II. A. Are you presently a teacher in the school system where you coach? YES NO ”hat is the highest educational degree you have attained? . (Please include year of graduation) How many years (include this one) have you coached this interscholastic girls' softball team? Are you presently. or have you ever been, coach of another interscholastic athletic team? YES NO If YES, list the team and the number of years you coached that team. TEAM YEARS OF COACHING EXPERIENCE COMPETITIVE ATHLETIC BACKGROUND were you a member of a varsity team in high school? YES NO If YES, indicate what team(s). 189 were you a member of a competitive athletic team in college (varsity or junior varsity level)? YES NO If YES, indicate which team(s). Do you presently play on (or plan to play on) a competitive athletic team? YES NO If YES, indicate the sport and the league in which you play. 190 (for athletes and coaches) CONSENT FORM Youth Sports Institute Michigan State University I have freely consented to take part in a scientific study being conducted by: Thelma Sternberg;Born under the supervision of Dr. Daniel Gould, Aasiiiant Professor of Health and Physical Education and a staff member with the Youth Sports Institute at Michigan State University. The study has been explained to me and I understand the explanation that has been given and what my participation will involve. I understand that I am free to discontinue my participation in the study at any time without penalty. I understand that the results of the study will be treated in strict confidence and that I will remain anonymous. within these restrictions, results of the study will be made available to ms at my request. I understand that my participation in the study does not guarantee any beneficial results to me. I understand that I can receive additional explanation of the study after my participation is completed. Signed Date APPENDIX E STATISTICAL ANALYSES: PARALLELISM REGRESSION RESULTS 191 as. he.— cwwsoo Heownxcd «aouucou nuow>ecnm mcwcomou unmanned Lo ecowunoouod xuwawce gunman n no. mn._ cansoo auowmxcd "aouucou muow>ecom mcmconou Hono— co ucowuaoouod xuqawcn gunman N no. mm.— cameoo Haecucom «Aouucou muow>ecom ocwcoeou aeuop no ncoaunoouud quAAne noxead P uuziu munm<~m<> mu4m<~x<> zo-<=cu zommn—Hmu ma—uuoumd zo—mmuxoux nouoacu< accouonsou cow: cc< so; ”encode cownnououm do xuwocnaoeo: uou uuop och do undone: xumsenm "FN umm<~ 192- .eN one nN madcap cw ocuccuoun nwcncowueaou aceofiuwcmwn our» and nonwoeofi Heowcoceo can nucewowuuooo coumeoumeu noNAnuemceume hoovmse 3.va on.p e¢~P.~ cweeoo aeowexcu "Monacou nuow>ecom mcwcoeou oowuoeud co ucowuaoouud xuwuwce hexane m hm. *eee.— caesoo Heownxce “Houucou nuow>ecom mcwcoeou «euc— uo ncowunoouod >awawce noxead N mm. Nm. caesoo Haecucom «aouucou uuoa>ecom ocwcoeou fiduc— co ecowunooued xuuawce noxean — unn<>iu u=n<>iu mncm<~z<> mucm<fiz<> zo-<=cu muzu—udzou uuzu—uezou zo.xn—~mu «opuuouzd zo~mmucuuz zen: zen noueacu< uocmuoneou om>wuouod cow: mc< rod "monxaec< cownnouaom nauseafiou ”NN unm

umaqn< enamfieu aneooam ouzaueu enooozm confided mooooam crocxcz crocxca muozuo when»: Harcouca Hucuoucq Honduran Hacuezoe mnnm<~m<> zo~mu_umu euoxeae xocouoneou mo>weouod no; can Houucou oo>woouon oc< euow>ncom nowuoaum .uocoeou “mucowowcuoou commandos: oo~quueuceum «nN nnm