Predicting subjective athletic performance from psychological skills after controlling for sex and sport.

Cox, Richard H. ; Shannon, Jennifer K. ; McGuire, Richard T. 等

The sport psychology literature is replete with research that attempts to show a predictive relationship between some sort of psychological variable and athletic performance (Cox, 2007). Initial efforts to predict performance focused upon the relationship between personality characteristics and athletic performance. A comprehensive review of personality/performance related investigations was published by Morgan (1980). Morgan's review and investigations that followed suggested a weak but consistent positive relationship between personality traits and athletic performance. The interest in studying the relationship between personality and performance waned significantly in the 1980s.

The strong interest in relating personality to performance in the 1960s and 1970s was replaced with an interest in mood states in the 1980s and 1990s. The thought was that situation specific mood states might be associated with performance to a greater degree than stable personality traits. The preferred tool for measuring mood states was the Profile of Mood States (POMS; McNair, Lorr & Droppleman, 1992), and Morgan's mental health model was the major theory that was tested (Morgan, 1979). Based on POMS scores, Morgan's model predicted that a mentally healthy athlete would be in a better situation to perform well than a less mentally healthy athlete. A review article published by Beedie, Terry, and Lane (2000) showed a weak but consistent positive association between mood state and positive athletic performance. Similar to personality, mood states failed to show a strong predictive relationship to athletic performance. However, the predictive relationship between mood and performance appears to be moderated by type of sport (Beedie et al.). For example, the relationship is stronger for open skills, individual sports, and short duration sports (compared to the alternatives). Furthermore, the way in which performance is conceptualized also makes a difference in the predictive relationship. The relationship is stronger when performance is measured subjectively as opposed to objectively.

In addition to predicting athletic performance from personality and mood, there currently exists an interest in predicting athletic performance from psychological skill. Psychological skill refers to innate or learned mental characteristics of the athlete that make it possible to succeed in sport (Vealey, 1988). These psychological skills are often referred to as psychological coping skills, because they are believed to assist athletes in coping with the stress associated with athletic performance. Examples of psychological skills that help an athlete cope with adversity include self-confidence, the ability to cope with adversity, attentional focus, intrinsic motivation, and the ability to control anxiety and arousal. One of the first instruments developed for the purpose of measuring psychological skill in sport was the Psychological Skills Inventory for Sports (Mahoney, Gabriel & Perkins, 1987). Other instruments that have been developed to measure psychological skill include the Athletic Coping Skills Inventory-28 (Smith, Schutz, Smoll, & Ptacek, 1995), the Test of Performance Strategies (Thomas, Murphy, & Hardy, 1999), and the Ottawa Mental Skills Assessment Tool (Durand-Bush, Salmela, & Green-Demers, 2001). While each of these instruments has their own strengths, the Athletic Coping Skills Inventory--28 (ACSI-28) was selected in the current investigation because of its ease in administration, its brevity, and for its reported psychometric properties. In their own words, Smith et al. (p. 379) stated that the inventory reflects a "multifaceted psychological skills construct." Once an athlete gains strength in a psychological skill (e.g. arousal control) this strength may then be used to cope with stressful competitive situations.

Utilizing the ACSI-28, Smith et al. (1995) and Smith and Christiansen (1995) demonstrated predictive relationships between certain psychological skills and baseball performance. Smith et al. observed that overachieving high school baseball players scored higher on all subscales of the ACSI-28 compared to normal and underachievers. Similarly, Smith and Christensen reported that confidence and achievement motivation were significant predictors of batting performance, while confidence and peaking under pressure were significant predictors of a pitcher's earned run average. Christensen (2000) correlated objective measures of collegiate golf performance with ACSI-28 measured psychological skills and reported significant relationships with golf performance. Sheldon and Eccles (2005) demonstrated that self-confidence and motivation as measured by the ACSI-28 significantly predict perceived tennis playing ability of male and female adult tennis players. Speiler et al. (2007) reported a significant relationship between the coping with adversity subscale of the ACSI-28 and starting status of Collegiate NCAA Division I football players.

Finally, utilizing the Psychological Skills Inventory for Sports (PSIS-5), Meyers and LeUnes (1996) were able to discriminate between low and highly skilled rodeo performers as a function of anxiety management, concentration, confidence, and motivation. They concluded that an assessment of psychological skills may enhance prediction of athletic potential in sport. The above review of psychological skill literature shows that while relationships between athletic performance and psychological skills are often significant, the strength of the relationship is typically weak. Self-confidence is a psychological skill that is often observed to be predictive of some measure of athletic performance. This is consistent with the results of a meta-analytic review that showed that self-efficacy (a measure of situation specific confidence) is often observed to significantly predict sport performance (Moritz, Feltz, Fahrbach, & Mack, 2000).

Sex of participant was included in this investigation as a control variable based on past research that has shown that sex of participant moderates the relationship between psychological variables and performance. In a study designed to study the relationship between psychological skill (strategies) and tennis performance, DeFrancesco and Burke (1997) observed that self-confidence influences the performance of the female participants to a greater degree than male participants. Hammereister and Burton (2004) reported that women scored higher in emotion focused coping compared to men. In predicting perceived tennis ability from psychological skill, Sheldon and Eccles (2005) controlled for sex of participant. Finally, Hays, Maynard, Thomas, and Bawden (2007) identified six types of sport confidence and noted that women placed more importance on good personal performances, while men derived confidence from winning.

While studies involving psychological skills typically focus upon a particular sport, three studies are cited to demonstrate the need to include type of sport as a control variable in psychological investigations of this type. The first is a classic sport personality investigation in which Schurr, Ashley, and Joy (1977) demonstrated that personality profile differences exist between team and individual sport athletes. Team sport athletes were found to be more dependent, anxious, and extroverted when compared to individual sport athletes. The second study was the Beedie et al. (2000) article in which sport type was identified as a moderator variable for predicting performance from psychological mood. The third study involved the psychological method of self-talk which has been observed to enhance self-confidence and modify arousal levels in athletes. In this investigation, individual sport athletes reported greater use of self-talk compared to team sport athletes (Hardy, Hall, & Hardy, 2005).

Based on a review of the sport psychology literature, it appears that a weak but fairly consistent relationship exists between athletic performance and the predictor variables of personality, mood state, and psychological skill. The purpose of this investigation was to study the relationship between psychological skill of collegiate athletes and subjective performance, after controlling for sex of participant and type of sport (team/individual). By controlling for sex and type of sport we should be able to observe the effect of psychological skill on perceived athletic performance with the confounding influence of sex and sport removed. This is accomplished through the application of hierarchical multiple regression. It was hypothesized that after controlling for sex and sport, a) psychological skills will predict subjective athletic performance, b) self-confidence will be the single strongest predictor of performance, and c) sex of athlete and type of sport will significantly predict performance. While it is expected that the relationship between psychological skills and athletic performance should be positive in nature (e.g. increased self-confidence should be associated with increased performance), it is conceivable that this will not always be true. For example, studies by Peetoom (1987) and Spieler et al. (2007) provide examples of psychological variables being negatively associated with athletic performance.

Method

Participants

Participants for this investigation were 627 student-athletes from a large Midwest University (Male=326, Female=301). Data were collected over a five-year period, with special care taken to avoid repeated measures of the same student-athlete across years. The majority of the data were collected during the winter semester of 2006 (n=271) with the remainder spread across the other four years (2002-2005). Every intercollegiate sport was represented with the majority coming from the sports of track and field (n = 159), football (n = 115), swimming (n = 57), baseball (n = 53), soccer (n = 50), and softball (n = 41). Of the total sample, 312 student-athletes were categorized as team sport athletes and 311 as individual sport athletes (4 unreported). At time of inventory administration, 199 participants were freshmen, 166 were sophomores, 142 were juniors, 91 were seniors, 19 were fifth-year seniors, nine were graduate students, and one did not report class level. In terms of ethnicity, 497 were Euro-American, 92 were African-American, and 37 were other (1 unreported). The majority of athletes (57%) self-reported semester GPAs of 3.00 or greater. Thirty-five percent reported GPAs between 2.00 and 2.99, with the remainder reporting GPAs of less than 2.00. The research was approved by the Campus Institutional Review Board (CIRB) and all athletes completed and signed a consent form.

Materials

Demographic Inventory. The Demographic Inventory included 15 items that requested responses related to such things as year in school (class), age range, sex, housing, semester and cumulative GPA, academic major, time spent studying, racial-ethnic identification, sport played, and subjective measures of player performance. In addition to describing the participant sample, the demographic information was used to categorize participants as a function of sex and sport type and to determine subjective athletic performance.

Athlete Coping Skills Inventory--28 (ACSI-28). The ACSI-28 is a 28-item inventory developed by Smith et al. (1995). The ACSI-28 measures seven psychological skill subscales (coping with adversity, coachability, concentration, confidence, goal setting, peaking under pressure, and freedom from worry). The psychometric characteristics of the ACSI-28 were reported by Smith et al. Alpha coefficients (internal consistency) for the seven subscales of the ACSI-28 range from .62 to .78, with the lowest being associated with concentration. The overall alpha coefficient for all subscales combined was .86. For the present investigation, the standardized alpha coefficients for the seven subscales ranged from .64 to .84 for the men (lowest for concentration) and from .64 to .83 for the women (lowest for concentration). Smith et al. also reported on the convergent and discriminant validity of the instrument.

Procedures

After receiving approval from the CIRB, the investigators approached coaches from 15 sports and asked for their permission to solicit the voluntary participation of their athletes. After receiving permission, the investigators presented athletes with a brief overview of the study and obtained written consent for their voluntary participation. No inducement was offered for participation in the study. Accordingly, it was made clear that failure to volunteer to participate in the study would not result in any type of retribution from their coaches or the investigators. Inventories described in the materials section were presented to each participant in either a team meeting or a one-on-one situation. Prior to completing the inventories, participants were informed of the voluntary nature of the research and that it would take approximately 20 minutes to answer all questions.

Statistical Analyses and Design

Prior to subjecting data to statistical analyses, data were thoroughly screened for (a) univariate deviations from normality (skewness & kurtosis), (b) univariate outliers, and (c) multivariate outliers. This process revealed two multivariate outliers in the ACSI-28 data. As a result of the data screening, the total data set was reduced from 627 to 625 participants (male = 325, female = 300). Missing data accounted for an additional reduction of 15 participants, bringing the total sample available for analysis down to 610. Data were analyzed using a two-step hierarchical multiple regression model. In step one, performance was regressed on the dummy coded variables of sex and sport. Sport was divided into team and individual sport athletes. In dummy coding, men and team sport athletes were coded as 0, while women and individual sport athletes were coded as 1. In step two, the seven coping skills of the ACSI-28 were added to the regression model.

The dependent measure for subjective athletic performance was obtained from the demographic inventory. In this inventory, and consistent with Butt, Weinberg, and Horn (2003), the participant was asked to respond to two different subjective measures of athletic performance. Both measures were from the perspective of the athlete, but one reflected the athletes' perception of his/her own performance while the second reflected the athletes' perception of how he/she thought the coach ranked him/her. The two items appeared as follows on the demographic inventory:

1. Indicate your personal assessment of your performance as an athlete relative to your own personal goals.

2. Indicate your perception of how you feel the coaches currently rank your standing on the athletic team you play for.

Item one was set to a five-point Likert scale (1 = I have failed miserably to measure up to my own personal goals, 5 = I have surpassed the personal goals 1 set for myself), while item two was set to a three-point Likert scale (1 = Bottom one-third in terms of contribution to team, 2 = Middle one-third in terms of contribution to team, 3 = Top one-third in terms of contribution to team). In the data analyses, separate hierarchical regression analyses were run for each of the measures of athletic performance as well as one for a combination (sum) of the two.

Results

The total variance accounted for in the full unrestricted model for athletes' perception of own performance, athletes' perception of coaches' raking, and the sum of the two combined was. 158 .108, and. 197 respectively. The correlation between the athletes' perception of own performance and athletes' perception of coaches' ranking was .28, while the correlations between the combined measure with athletes' perception of self and athletes' perception of coaches' perception were .83 and .76 respectively. Both explanation and prediction were important goals in this research, but because prediction was of utmost importance, the model that exhibited the largest [R.sup.2] was selected for further analysis. Relationships between psychological skills with performance were similar in all three regression analyses, but because greater predictive power was observed when the combined performance score was utilized, the remainder of the results section will focus only on the model in which the dependent variable was a single value reflecting the sum of the other two measures. Thus, subjective performance is conceptualized as a combination of both the athletes' perception of own performance and the athletes' perception of the coaches' perception of the athletes' performance level.

Displayed in Table 1 are simple correlations among all variables in the full model. Subjective performance correlated best with adversity, concentration, confidence, and freedom from worry. The negative correlation between performance and sport type simply means that team sport athletes scored higher on performance than individual sport athletes (see Table 2). Several relatively strong correlations were observed among psychological skill variables (adversity, confidence, and peaking under pressure correlated above .40 with at least three other ACSI-28 variables).

Step one of the two-step hierarchical multiple regression model revealed a significant predictive relationship between sex and sport with perceived athletic performance, F (2,607) = 4.53, p = .011, [R.sup.2] = .015. Significant regression coefficients were observed for both sex (p = .028, Women > Men) and for sport (p = .029, Team Sport > Individual Sport).

In step two of the hierarchical multiple regression, the seven psychological coping skills measured by the ACSI-28 were added to the model. As can be observed in Table 3, adding the seven coping skills (step 2) to the hierarchical model increased the variance accounted for ([R.sup.2]) from .015 to. 197, a significant increase of. 182, F (7, 598) = 19.32, p < .0001 (see Pedhazur, 1997, p. 108 for calculation formula). The full model was also statistically significant, F(9, 600) = 16.32, p < .0001 (steps 1 and 2 combined). An examination of tolerance and variance inflation factors for the independent variables in the full model failed to show any indication of multicollinearity among these variables. Additionally, an examination of condition indices and proportion of variance indices failed to show any evidence of multicollinearity.

After controlling for sex and sport and all other psychological skills, the coping skills of coachability (p < .0001), confidence (p < .0001), goal setting (p = .025), and freedom from worry (p < .0001) were all statistically significant. Based upon the standardized betas, squared semi-partial correlations ([R.sup.2.sub.sp]) and the structural coefficients ([r.sub.sc]), it appears that confidence and freedom from worry have the strongest predictive relationship with perceived athletic performance. Coachability and goal setting are also significant predictors of perceived performance, but in a negative direction. This is despite the fact that the simple correlations between these two variables and subjective performance were very low (i.e., r = .06 in both).

Discussion

The purpose of this investigation was to study the relationship between psychological skill of collegiate athletes and subjective performance, after controlling for sex of participant and type of sport (team/individual). By controlling for sex and type of sport it was expected that the relationship between psychological skill and athletic performance could be observed without the confounding influence of sex of athlete or type of sport. It was hypothesized that after controlling for sex and sport, a) psychological skill would predict subjective athletic performance, b) self-confidence would be the single strongest predictor of performance, and c) sex of athlete and type of sport would significantly predict performance. While these hypotheses were generally confirmed, it was also observed that some of the significant psychological skills predicted performance in a negative direction (e.g. coachability and goal setting).

In step one of the hierarchical regression analysis, the control variables of sex and sport were entered into the restricted model for predicting subjective athletic performance. This resulted in a small but significant squared multiple correlation ([R.sup.2] = .015). At this stage, sex and sport were observed to be significant but weak predictors of subjective athletic performance. Women and team sport athletes exhibited significantly larger mean performance scores than men and individual sport athletes (see Table 2).

In step two of the hierarchical regression analysis, the seven psychological skills measured by the ACSI-28 were stepped into the model. After controlling for sex and sport type, psychological skills accounted for a significant 18.2% increase in variance. Psychological skills that add to the predictability of subjective performance in a positive direction include the perceived

ability to be confident and to cope effectively with worry and anxiety. Relative to confidence, these finding are consistent with Bandura's model of self-efficacy and other research that shows a positive relationship between self-confidence and self-efficacy with athletic performance (Bandura, 1997; Beattie, Hardy, & Woodman, 2004; Craft, Magyar, Becker, & Feltz, 2003; Hays et al., 2007; Marsh & Perry, 2005; Moritz et al., 2000; Sheldon & Eccles, 2005; Woodman & Hardy, 2003). Relative to coping with worry and stress, these findings are also consistent with previous research that has shown an association between state anxiety and worry with athletic performance (Greenspan & Feltz, 1989; Groslambert, Candau, Grappe, Duque, & Rouillon, 2003; Hardy, Parfitt, & Pates, 1994; Hardy, Woodman, & Carrington, 2004; Zaichkowsky & Fuchs, 1988).

The psychological skills of coachability and goal setting were also significant predictors of perceived athletic performance, but in a negative direction. That is, as perceived coachability and goal setting scores increase, subjective athletic performance decreases. This seems counter intuitive, as logic would seem to suggest that psychological skills related to coachability and goal setting should be associated with increased athletic performance. In the present sample, the simple correlations between performance with coachability and goal setting were weak but in the positive direction. The test of significance of a regression coefficient in the full model is whether a particular psychological variable predicts performance after the effects of all other predictor variables (including sex and sport type) have been removed and controlled for. Coachability and goal setting correlate poorly with performance by themselves, but each has relatively strong simple correlations with several other psychological variables. It is these relationships with other variables that result in significant regression coefficients in the full model for these two variables. Effects of this variety are discussed in some detail by Pedhazur (1997) under the heading of suppressor variables. Reliance on simple correlations between individual predictor variables and a criterion can be misleading because it implies that relationships are univariate in nature as opposed to multivariate in nature. The simple interactive relationships among independent variables helps to explain why a variable can have a weak but positive simple relationship with the criterion variable (e.g., coachability & goal setting), yet have a significant and negative effect on the criterion (subjective performance) in the regression model.

While logic might suggest that a positive relationship should exist between athletic performance with coachabilty and goal setting, a close look at the related literature might suggest otherwise. In the study by Smith et al. (1995), high school athletes were divided into overachievers, normal achievers, and underachievers. The ACSI-28 scores for goal setting and coachability did not discriminate between the overachievers and the underachievers, although the means for the overachievers were noticeably (but not significantly) larger than the underachievers. In the study by Smith and Christensen (1995), correlations were calculated between minor league baseball performance statistics and the ACSI-28 subscales. In the case of position players, the correlations for the coping skills of coachability and goal setting with batting average (high score good) were an insignificant .01 and .20 respectively. For pitchers, the correlation for the coping skills of coachability and goal setting with earned run average (low score good) was an insignificant. 10 and. 14 respectively. Both of these correlations were actually in the negative direction suggesting an insignificant negative relationship between the two coping skills and earned run average. In conclusion, there is no evidence from these two investigations that high coachability and goal setting coping skills are predictive of superior athletic ability.

An investigation by Spieler et al. (2007) utilized the ACSI-28 in an attempt to discriminate between starters and nonstarters in collegiate football. Their results suggested that only coping with adversity is effective in discriminating between starters and nonstarters in collegiate football. While this appears to be true, the authors failed to point out that mean differences favored the nonstarters instead of the starters, as one would logically predict. Thus, this would be another example using the ACSI-28 in which the lesser skilled athletes scored higher on a psychological skill. The finding in the current investigation that shows a significant negative relationship between perceived athletic performance and the two psychological skills of coachability and goal setting is not particularly inconsistent with the literature.

An investigation by Peetoom (1987) provides some further insight into the counter intuitive relationship between athletic performance and the psychological skill of coachability. In this investigation, Peetoom (1987) attempted to discriminate between professionally drafted and nondraffed baseball players on the basis of personality traits. The four personality traits of coachability, leadership, self-confidence, and mental toughness were observed to reliably discriminate between the two categories of athletes. However, similar to the current study, Peetoom reported that the mean scores for coachability and leadership were larger for the nondrafted players compared to the drafted players. Even though Peetoom measured personality traits instead of psychological skills, many of the coachability items from the respective inventories were very similar.

When attempting to predict athletic performance from psychological mood state, differences emerge as a function of how performance is measured (Beedie et al., 2000). This is true for psychological skills as well. In the case of Smith et al. (1995), performance was measured as a subjective assessment of under and over achieving (coach's assessment). In the case of Smith and Christiansen (1995), performance was conceptualized as an objective measure of hitting and pitching performance. Speiler et al. (2007) measured performance as a function of starting status of Division I football players. In the current investigation, performance was conceptualized as a combination of the athletes' perception of his/her coach's evaluation of their performance and the athletes' perception of their own performance in relation to their goals. It is realistic to surmise that some of the differences in results that emerge from different studies may be related to the manner in which performance is conceptualized and measured.

This research suggests that the psychological skills possessed by the athlete may be a reliable predictor of subjective athletic performance. In particular, the psychological skills of confidence and reducing worry are consistent predictors of performance. Athletes who exhibit self-confidence and the ability to manage worry and anxiety are more likely to perform well on the athletic field. While the psychological skills of being coachable and goal setting add to the variance accounted for in performance, they do so in a negative fashion. In combination with other psychological variables, an increase in coachability and goal setting may actually reduce the athlete's subjective evaluation of performance.

The primary weakness of the current investigation is that the coaches' assessment of an athlete's contribution to the team was made by the athlete and not the coach. Performance was partially conceptualized as a function of the athlete's perception of how the coach felt about him/her in terms of performance. Thus, athlete perception was a proxy of coaches' perception of athlete's performance. A second weakness was that in order to get a large sample size, data were collected over several years as opposed to one year. This resulted in repeated measure on some participants. Even though one of the multiple observations were randomly deleted, it is possible that a carry-over effect could bias the research in some way when the retained measure was the second as opposed to the first. Never-the-less one of the strengths of this investigation was the large sample size. For example, many of the investigations cited in this manuscript had a sample size of less than 100 participants.

In summary, the results of the current investigation show that, after controlling for sex and sport type, subjectively measured athletic performance is related to psychological skills as measured by the ACSI-28. In particular, self-confidence and freedom from worry are reliable predictors of performance. That is, freedom from worry and high self-confidence are positively associated with increased subjectively measured athletic performance. Coachability and goal setting were also observed to be reliable predictors of performance, but in a negative direction. As coachability and goal setting increase, actual subjectively measured performance decreases. It is recommended that future research focus on explaining why and under what circumstances some psychological skills have a negative relationship with athletic performance.

References

Bandura, A. (1997). Self-efficacy: The exercise of control. San Francisco, CA: Freeman.

Beattie, S., Hardy, L., & Woodman, T. (2004). Pre-competition self-confidence: The role of self. Journal of Sport & Exercise Psychology, 26, 427-441.

Beedie, C. J., Terry, P. D., & Lane, A. M. (2000). The Profile of Mood States and athletic performance: Two meta-analyses. Journal of Applied Sport Psychology, 12, 49-68.

Butt, J., Weinberg, R., & Horn, T. (2003). The intensity and direction interpretation of anxiety: Fluctuations throughout competition and relationship to performance. The Sport Psychologist, 17, 35-54.

Christensen, D. S. (2000). Self-efficacy, cognitive interference, sport anxiety, and psychological coping skills as predictors of performance in intercollegiate golf. Dissertation Abstracts International: Section B: The Sciences and Engineering, 61(3-B), 1629.

Coppel, D. B. (1980). The relationship of perceived social support and self-efficacy to major and minor stressors. Unpublished doctoral dissertation, University of Washington.

Cox, R. H. (2007). Sport psychology: Concepts and applications (6th Ed.). St. Louis: McGraw-Hill.

Craft, L. L., Magyar, T. M., Becker, B. J., & Feltz, D. L. (2003). The relationship between the Competitive State Anxiety Inventory-2 and sport performance: A meta-analysis. Journal of Sport & Exercise Psychology, 25, 444-465.

DeFrancesco, C., & Burke, K. L. (1997). Performance enhancement strategies used in a professional tennis tournament. International Journal of Sport Psychology, 28, 185-195.

Durand-Bush, N., Salmela, J. H., & Green-Demers, I. (2001). The Ottawa Mental Skills Assessment Tool (OMSAT-3). The Sport Psychologist, 15, 1-19.

Greenspan, M. J., & Feltz, D. L. (1989). Psycholgical interventions with athletes in competitive situations: A review. The Sport Psychologist, 3, 219-236.

Groslambert, A., Candau, R., Groppe, F., Dugue, B., & Rouillon, J. D. (2003). Effects of autogenic and imagery training on the shooting performance in Biathlon. Research Quarterly for Exercise and Sport, 74, 337-341.

Hammereister, J., & Burton, D. (2004). Gender differences in coping with endurance sport stress: Are men from Mars and women from Venus? Journal of Sport Behavior, 27, 148-164.

Hardy, J., Hall, C. R., & Hardy, L. (2005). Quantifying athlete self-talk. Journal of Sports Sciences, 23, 905-917.

Hardy, L., Parfitt, G., & Pates, J. (1994). Performance catastrophes in sport: A test of the hysteresis hypothesis. Journal of Sport Sciences, 12, 327-334.

Hardy, L., Woodman, T., & Carrington, S. (2004). Is self-confidence a bias factor in higher order catastrophe models? Journal of Sport & Exercise Psychology, 26, 359-368.

Hays, K., Maynard, I., Thomas, O., & Bawden, M. (2007). Sources and types of confidence identified by world class sport performers. Journal of Applied Sport Psychology, 19, 434-456.

Mahoney, M. J., Gabriel, T. J., & Perkins, T. S. (1987). Psychological skills and exceptional athletic performance. The Sport Psychologist, 1, 181-199.

Marsh, H. W., & Perry, C. (2005). Self-concept contributes to winning gold medals: Causal ordering of self-concept and elite swimming performance. Journal of Sport & Exercise Psychology, 27, 71-91.

McNair, D. M., Lorr, M., & Droppleman, L. F. (1992). Profile of Mood States manual. San Diego: Educational and Industrial Testing Service.

Meyers, M. C., & LeUnes, A. (1996). Psychological skills assessment and athletic performance in collegiate rodeo athletes. Journal of Sport Behavior, 19, 132-147.

Morgan, W. P. (1979). Prediction of performance in athletics. In P. Klavora & J. V. Daniel (Eds.), Coach, athlete, and the sport psychologist (pp. 172-186). Champaign, IL: Human Kinetics Publishers.

Morgan, W. P. (1980). The trait psychology controversy. Research Quarterly for Exercise and Sport, 51, 50-76.

Moritz, S. E., Feltz, D. L., Fahrbach, K. R., & Mack, D. E. (2000). The relation of self-efficacy measures to sports performance: A meta-analytic review. Research Quarterly for Exercise and Sport, 71, 280-294.

Pedhazur, E. J. (1997). Multiple regression in behavioral research: Explanation and prediction (3rd ed.). New York: Harcourt Brace College Publishers.

Peetoom, G. A. (1987). Predicting success in professional baseball with the Athletic Motivation Inventory. Unpublished Doctoral Dissertation, University of Mississippi.

Rosenbaum, M. (1980). A schedule for assessing self-control behaviors: Preliminary findings. Behavioral Therapy, 11, 109-121.

Sheldon, J. P., & Eccles, J. S. (2005). Physical and psychological predictors of perceived ability in adult male and female tennis players. Journal of Applied Sport Psychology, 17, 48-63.

Schurr, K. T., Ashley, M. A., & Joy, K. L. (1977). A multivariate analysis of male athlete characteristics: Sport type and success. Multivariate Experimental Clinical Research, 3, 53-68.

Smith, R. E., & Christensen, D. S. (1995). Psychological skills as predictors of performance and survival in professional baseball. Journal of Sport and Exercise Psychology, 17, 399-415.

Smith, R. E., Schutz, R. W., Smoll, F. L., & Ptacek, J. T. (1995). Development and validation of a multidimensional measure of sport-specific psychological skills: The Athletic Coping Skills Inventory-28. Journal ot Sport and Exercise Psychology, 17, 379-398.

Spieler, M., Czech, D. R., Joyner, A. B., Munkay, B., Genter, N., & Long, J. (2007). Predicting athletic success: Factors contributing to the success of NCAA Division I AA collegiate football players. Athletic Insight: The Online Journal of Sport Psychology, 9, 1-11.

Thomas, P. R., Murphy, S. M., & Hardy, L. (1999). Test of Performance Strategies: Development and preliminary validation of a comprehensive measure of athletes' psychological skills. Journal of Sport Sciences, 17, 697-713.

Vealey, R. S. (1988). Future directions in psychological skills training. The Sport Psychologist, 2, 318-336.

Woodman, T., & Hardy, L. (2003). The relative impact of cognitive anxiety and self-confidence upon sport performance: A meta-analysis. Journal of Sport Sciences, 21, 443-457.

Zaichkowsky, L. D., & Fuchs, C. (1988). Biofeedback applications in exercise and athletic performance. Exercise and Sport Science Reviews, 16, 381-421.

Richard H. Cox, Jennifer K. Shannon, Richard T. McGuire and Adrian McBride

University of Missouri-Columbia

Address Correspondence to: Richard H. Cox, Ph.D., Department of Educational & Counseling Psychology, 16 Mill Hall, University of Missouri-Columbia, Columbia, MO 65211. Phone: (573) 882-7602. E-mail: [email protected].

Table 1. Correlations among Variables used in Subjective
Performance Prediction Model (N=610)

Variables               1        2        3        4        5

Performance           .08      -.18      .24     .06 *     .22
1. Sex                          .08     -.12     -19      -.02 *
2. Sport Type                           -.07 *   .11      -.02 *
3. Adversity                                     .28       .57
4. Coachability                                            .31
5. Concentration
6. Confidence
7. Goal Setting
8. Peaking Pressure
9. No Worry

Variables               6        7        8        9

Performance           .34      .05 *     .15      .23
1. Sex                .03 *    .03 *    -.18     -.02 *
2. Sport Type         .04 *    .12      -.10      .07 *
3. Adversity          .46      .31       .44      .30
4. Coachability       .42      .19       .22      .27
5. Concentration      .48      .33       .46      .18
6. Confidence                  .42       .46      .19
7. Goal Setting                          .24     -.02 *
8. Peaking Pressure                               .09
9. No Worry

* Not significant at p = .05

Table 2. Descriptive Statistics Associated with Variables in the
Full Hierarchical Regression Model Psychological Skill Variables

                   Perform       Adversity     Coachab

Sex      Sport      M      SD     M      SD     M      SD

Male     Team      5.6    1.2    7.4    2.0    8.5    2.5
         Individ   5.4    1.4    7.0    2.2    9.2    2.1

Female   Team      5.9    1.2    6.8    2.2    9.6    2.1
         Individ   5.6    1.4    6.7    2.3    9.7    1.9

                   Concentr      Confidence    GoalSet

Sex      Sport      M      SD     M      SD     M      SD

Male     Team      7.5    2.1    8.1    2.2    6.3    2.1
         Individ   7.5    2.1    8.6    2.2    7.0    2.5

Female   Team      7.5    2.1    8.6    2.0    6.5    2.4
         Individ   7.4    2.1    8.4    2.2    7.0    2.5

                   Peaking       Noworry

Sex      Sport      M      SD     M      SD

Male     Team      7.9    2.6    6.4    2.5
         Individ   7.9    3.0    6.6    2.8

Female   Team      7.5    2.7    6.0    3.1
         Individ   6.3    2.7    6.7    2.8

Table 3. Full Model Hierarchical Multiple Regression Analysis
for Predicting Athletic Performance.

Variable             M       SD    beta    Beta      T       p

Sex                0.48 *   0.50    0.32    0.12    3.19    .002
Sport              0.50 *   0.50   -0.22   -0.08   -2.19    .029

Adversity          6.96     2.17    0.04    0.07    1.44    .149
Coachability       9.22     2.24   -0.10   -0.18   -4.15   <.0001
Concentration      7.47     2.11    0.05    0.07    1.50    .135
Confidence         8.42     2.15    0.22    0.36    7.35   <.0001
Goal Setting       6.69     2.39   -0.05   -0.09   -2.24    .025
Peaking/Pressure   7.41     2.79   -0.01   -0.02   -0.41    .680
No Worry           6.43     2.78    0.09    0.18    4.60   <.0001

                   [r.sup.2                [R.sup.2    [R.sup.2
Variable           .sub.sp]   [r.sub.sc]   .sub.tot]   .sub.inc]

Sex                0.014       0.187
Sport              0.006      -0.186       0.015       0.015

Adversity          0.003       0.537
Coachability       0.023       0.132
Concentration      0.003       0.504
Confidence         0.072       0.767
Goal Setting       0.007       0.123
Peaking/Pressure   <.001       0.342
No Worry           0.028       0.521       0.197       0.182

Variable             F       p

Sex
Sport              4.53    0.011

Adversity
Coachability
Concentration
Confidence
Goal Setting
Peaking/Pressure
No Worry           19.39   <.0001

[r.sup.2.sub.sp] = Squared Semi-partial Correlation

[r.sub.sc] = Structural Coefficient

* Mean of dummy coding