Coaching confidence: an exploratory investigation of sources and gender differences.

Marback, Tracy L. ; Short, Sandra E. ; Short, Martin W. 等

Administrators, parents, and athletes sometimes overlook coaches' confidence in their ability to coach. The assumption, especially at the collegiate level, is that coaches are "expelts" in their respective sports and are psychologically prepared for the demands of coaching. Considering the importance of the coach in determining the quality and success of an athlete's sport experience, surprisingly little research was found that identifies optimal coaching behaviors and factors that influence the effectiveness of particular behaviors (Kenow & Williams, 1999).

There is a considerable amount of literature describing the multidimensional and hierarchical nature of self-perceptions in sport. Self-efficacy (defined as one's belief in one's ability to successfully perform a specific behavior or set of behaviors required to obtain a certain outcome; Bandura, 1977) is a specific self-perception, and has been referred to as a situationally specific self-confidence (Feltz, 1988). Perceived competence (defined as the perception that one has the ability to master a task resulting from cumulative interactions with the environment; Nicholls, 1984) is considered a more general self-perception compared to self-efficacy. Although it has been argued that both self-efficacy and perceived competence refer to similar cognitive processes by which people make judgments about their capabilities to accomplish a particular goal in a sport or motor performance context (Feltz & Chase, 1998; Moritz, Feltz, Fahrbach, & Mack, 2000), this study differentiates between these terms because it is focused on comparing these conceptualizations. Confidence (defined as the firmness or strength of one's belief; Bandura, 1997) is used as an overarching concept that encompasses both efficacy and competence. When considered in this manner, one may wonder why two conceptualizations (efficacy and competence) are needed to describe the same construct - that of coaching confidence?

Coaching confidence is a relatively new name for an old construct in sport. In their model, Feltz, Chase, Moritz, and Sullivan (1999) defined coaching efficacy as the extent to which coaches believe they have the capacity to affect the learning and performance of their athletes. The four coaching efficacy dimensions that comprise coaching efficacy include: game strategy, motivation, teaching technique, and character building. Game strategy efficacy is the confidence coaches have in their ability to coach during competition and lead their team to a successful performance. Motivation efficacy involves the confidence coaches have in their ability to affect the psychological skills and states of their athletes. Technique efficacy is defined as the belief coaches have in their instructional and diagnostic skills. Character building efficacy involves the confidence coaches have in their ability to influence the personal development of a positive attitude toward sport in their athletes. Coaching efficacy is based on coaching experience and preparation, prior success, the perceived ability of athletes, and school/community support. In turn, coaching efficacy effects coaching behavior, and the satisfaction, performance, and efficacy beliefs of both players and teams. Support for these relationships was shown in Feltz et al. (1999)

An alternative conceptualization for coaching confidence is Barber's (1998) notion of coaching competence. The components of coaching competence included communication skills, ability to motivate athletes, ability to teach sport skills, knowledge of strategies and tactics, training and conditioning, practice and seasonal planning, and coaching during competition. The sources of coaching competence she considered were performance accomplishments, social comparison, and the use of significant others. Barber examined gender differences in sources of coaching competence and levels of perceived competence in specific coaching areas. The results showed that teaching sport skills" discriminated between female and male coaches; the mean for females was significantly higher than for males. There were also differences in the sources of coaching competence infomation where females placed a greater importance than males on improvement of athletes and improvement of coaching skills. Despite these findings, because of low effect sizes and the similarities in ratings between females and males, Barber concluded that there were not any meaningful gender differences in perceived competence and sources.

Although Barber (1998) understated her results, gender differences have been found for the coaching efficacy construct as well (Lee, Malete, & Feltz, 2002). These findings should not be surprising as sport coaching is often considered to be a male-dominated activity (Coakley, 2001). There are fewer female coaches in competitive sport compared to males, and most female coaches have fewer years experience in coaching than their male counterparts. These differences may have an effect on the confidence levels of female and male coaches. The confidence coaches have in their abilities is an important contributor to coaching involvement (Weiss & Stevens, 1993) and coaching motivation (Sisley, Weiss, Barber, & Ebbeck, 1990; Weiss & Sisley, 1984; Weiss & Stevens, 1993).

There are several similarities between the constructs of coaching efficacy and coaching competence. First, both conceptualizations suggest that coaching confidence is multidimensional in nature. Most of the subscales in Barber's (1998) conceptualization are apparent in Feltz et al.'s (1999) description of their subscales. For example, while Feltz et al. use the label "technique efficacy" to refer the belief coaches have in their instructional and diagnostic skills, Barber has a subscale called "'teaching sport skills." Second, both Feltz et al. and Barber posit that levels of coaching efficacy and coaching competence will influence coaching behavior. Consistent with Bandura's (1997) theory, they agree that the efficacious and competent coaches are more likely to coach to begin with, to put forth greater effort while coaching, and to continue coaching longer than their less efficacious counterparts. Third, both of these models identify specific sources that are proposed to influence the self-perceptions of confidence. Barber used three broad categories (performance accomplishments, social comparison and the influence of significant others) as the basis for nine subcategories of sources, whereas Feltz et al. had a fewer number of sources (i.e., coaching experience and preparation, prior success, the perceived ability of athletes, and school/community support).

However, there are some differences between the two approaches. The inclusion of "character building efficacy" is unique to Feltz et al.'s (1999) model. Second, their model is firmly entrenched within Bandura's (1997) framework. As a result, their list of sources is relatively restricted. Barber, on the other hand, discusses a wider variety of sources of coaching competence. Third, Barber's research focused on gender differences of coaches' perceptions. While there is a great wealth of literature on gender differences in self-efficacy (Lirgg, 1991), and Feltz et al. mentioned it as a potential factor affecting coaching efficacy, only one study to date has examined gender differences in coaching efficacy. This study was conducted with coaches from Singapore (Lee et al., 2002). Given the cultural differences between Singapore and the United States, it is important to investigate gender differences with American coaches.

The purposes of this study were to clarify the conceptualization and operationalization of coaching confidence. Specifically, this entailed discriminating between coaching efficacy and coaching competence. Second, the sources of coaching confidence were examined. Finally, gender differences in coaching confidence were also considered.

Method

Participants

Participants were 187 intercollegiate coaches (52 female and 135 male) from the Division I (9.1%), II (81.8%), III (1.6%), and NAIA (7.5%) ranks. Of the sample 51.9% were head coaches (n = 97) and 48.1% were assistant coaches (n = 90). The education of the responding coaches varied from having a high school (3.2%), bachelors (29.9%), masters (63.6%), and doctoral (3.2%) degree. The coaches represented the sports of football (24.6%), basketball (21.9%), volleyball (10.7%), hockey (9.6%), softball (8%), track and field/cross-country (6.4%), soccer (4.8%), wrestling (4.3%), baseball (4.3%), swimming (2.7%), golf(1.5%), and tennis (1.1%). They ranged in age from 19 to 72 years (M= 38.8, SD = 11.0). Years of head coaching experience ranged from 0 to 36 years (M = 6.3, SD = 8.7), while years of experience as an assistant coach ranged from 0 to 35 years (M = 5.3, SD = 6.3).

Measures

The background questionnaire included demographic data, such as the primary sport coached, level of sport coached, gender of team coached, age and gender of coach, years of experience as a head and assistant coach, overall and season record (assessed by win/loss), and participation in coaching preparation courses/clinics (i.e., had the coaches ever attended a coaching preparation course/clinic?).

The Coaching Efficacy Scale (CES) developed by Feltz et al. (1999) was used. The CES contains four factors: teaching technique efficacy, motivation efficacy, game strategy efficacy and character building efficacy. Items are rated on a scale of 0 (not at all confident) to 9 (extremely confident). Feltz et al. found the coefficient alphas and test-retest coefficients for each subscale were acceptable; values of .88 and .77 for character building, .89 and .78 for technique, .91 and .83 for motivation, and .88 and .84 for strategy, respectively. The coefficient alpha for the total CES scale was .95, and the test-retest coefficient was .82.

The Perceived Coaching Competence Questionnaire developed by Barber (1998) was also used. This questionnaire consists of seven areas (communication, teaching sport skills, motivation, training and conditioning, sport-specific knowledge of strategies and tactics, coaching during competition, and practice and seasonal planning). Participants are asked to rate how competent they feel in each of these coaching areas using a 5-point Likert-type scale ranging from 1 (extremely competent) to 5 (not at all competent). Barber reported reliability coefficients for the Perceived Coaching Competence Questionnaire ranging from .86 to .94.

The Sources of Coaching Competence Information Scale (Barber, 1998) was used to assess a coach's preference for sources of competence information. The scale consists of three main categories, which provided the basis for nine subcategories. They are: performance accomplishments (which is defined by athletes' actual and perceived improvement, perceived improvement of coaching skills, and contest outcomes), social comparison (defined by a coach's use of peer comparison as a source of competence information), and influence of significant others (which included feedback from athletic administrators, other coaches, parents, athletes, and the affect expressed by athletes). Items are rated on a 5-point Liken scale ranging from 5 (extremely important source) to 1 (not at all an important source). Barber reported adequate reliability coefficients (Cronbach's Alpha) ranging from .79 to .95.

Procedure

The University Institutional Review Board approved the procedure for this study. The data collection process consisted of mailing a packet containing a personalized cover letter, consent form, background questionnaire, Sources of Coaching Competence Information Scale, Perceived Coaching Competence Questionnaire, and the CES to collegiate coaches in the upper Midwest (North Dakota, Minnesota, South Dakota, Iowa, and Nebraska). A stamped self-addressed return envelope was enclosed to facilitate the return of the materials. Of the 402 questionnaires mailed, 193 were returned for a response rate of 48%. Subsequently, six questionnaires were discarded due to missing data, resulting in 187 collegiate coaches for analyses.

Results

Table 1 gives the means, standard deviations, indicators of skewness and kurtosis and alpha values (Cronbach, 1951) for all factors of coaching efficacy and coaching competence separated according to gender. A MANOVA was conducted to examine gender differences in perceptions of competence and coaching efficacy. Although character-building efficacy was moderately skewed, there was no multicollinearity within either the efficacy or competence sets of variables. The only remaining assumption for the MANOVA concerned equality of variances and covariances (Weinfurt, 1995). The subscales of coaching competence upheld this assumption, and the overall effect of gender was significant, (F (7, 178) = .89, p < .01). There were significant gender differences for coaching during competition (F (1, 184) = 8.24, p < .05), knowledge of skill and tactics (F (1, 184) = 11.03, p < .001), and ability to motivate athletes (F (1, 184) = 5.39,p <.05). Male coaches' perceptions of competence were significantly greater than female coaches for coaching during competition (t (185) - 2.92,p < .01), knowledge of skills and tactics(t(185)=3.38,p<.OO1),and ability to motivate athletes(t(185)=2.21,p < .05).

For coaching efficacy, Box's M test revealed a significant inequality of covariances at p < .01. As this is a fundamental assumption of MANOVA, the analysis could not proceed. Instead, a random selection of male participants was chosen to approximate the female participants. This resulted in sample of 109 coaches (52 female, 57 male). This sample upheld the assumption of homogeneity of covariances. The MANOVA showed a significant overall effect for gender (F (4, 104) - .895,p < .05). The only significant effect was for game strategy efficacy (F(1, 107) = 4.04, p < .05). A t-test revealed that females game strategy efficacy was significantly lower than males (t (107) - 2.01, p < .05). Because of these differences, gender was included as a variable in all other analyses.

Discriminant Validity

Pearson correlations between coaching efficacy and coaching competence were examined to ascertain discriminant validity (see Table 2). To support discrimination, the correlations within each factor should be higher than the correlations between the factors (Bryant, 2000). This was not the case. The correlations between the four factors of coaching efficacy ranged from .39 to .63 and those between the factors of coaching competence ranged from .23 to .71. The intercorrelations between coaching efficacy and coaching competence ranged from .26 to .73. All correlations were significant atp < .01. Several of the correlations between coaching efficacy and coaching competence were above .70, suggesting multicollinearity (Tabachnick & Fidell, 2001). These relationships show no statistical distinction between the constructs of coaching efficacy and coaching competence.

Canonical Correlations

A canonical correlational analysis was run using the four factors of coaching efficacy and the seven factors of coaching competence to analyze the relationships between these two constructs. This data set produced four sets of canonical variates, of which three of the correlations were significant, with Wilk's Lambdas of .08, .30, and .68, respectively, all with p's < .001. The canonical correlation was .86 for the first pair, .75 for the second pair, and .55 for the third. The first two suggest multicollinearity between the sets of variables as indicated by r's > .70.

The canonical loadings for each of these correlations were interpreted to ascertain the pattern of relationships between coaching efficacy and coaching competence. Loadings greater than .30 were considered significant (see Table 3). On the first canonical correlation, all factors of both scales loaded significantly, and in the same direction. For the second correlation, motivation efficacy loaded with perceptions of ability to motivate and communication skills, whereas teaching technique efficacy loaded with teaching competency. On the third correlation, game strategy efficacy loaded negatively, whereas teaching technique efficacy and the competencies for teaching sport skills, communicating skills, and training and conditioning all loaded positively.

Regression Models

Given the statistical redundancy between the constructs of coaching efficacy and coaching competence, regression models were run to predict only coaching efficacy. From the participant background information, the sources included were gender, participation in a coaching preparation course, years of experience as a head and assistant coach, and won/lost record. For the years of experience variable, values for years of head coaching experience and years of assistant coaching experience were added to create one variable--coaching experience. For won/lost records, the correlation between won/lost record for last season and career won/lost record was .80, so only last season's won/lost record was used (in addition, more coaches knew their last season's win/loss record compared to their overall record).

The models for game strategy efficacy, motivation efficacy, and character building efficacy were significant (see Table 4). The model for motivation (F(4, 170) = 3.80, p < .01, [R.sup.2] = .08) was based on head coaching experience as the sole predictor. For both game strategy (F (4, 170) = 5.96,p < .001, [R.sup.2] =. 12), and character building (F (4, 170) = 4.77,p < .01, [R.sup.2] =. 10), gender and head coaching experience were significant predictors.

A series of multiple regression equations were also run to see if the factors of coaching efficacy could be predicted from the sources of coaching competence. Significant models (p < .05) were estimated for all four factors, although there were no significant individual predictors for teaching technique. For game strategy (F (9, 177) = 3.11, p < .01, [R.sup.2] =. 14), event outcome was the sole predictor. The model for motivation (F (9, 177) = 5.27, p < .001, [R.sup.2] = .21), included event outcome and affect of athletes as predictors, and feedback from athletes, affect of athletes and social comparison predicted character building efficacy (F(9, 177) = 4.58, p < .001, [R.sup.2] =. 19).

Discussion

This study examined the relationships between coaching efficacy and coaching competence, their sources, and considered gender differences. The first conclusion that can be made is that coaching efficacy and coaching competence are statistically redundant and that there is actually only one construct of coaching confidence. The support for this conclusion lay in the lack of discriminant validity between the two models, and the first (and most powerful) canonical correlation. The intercorrelations between the two constructs suggest that there is no operational distinction between coaching efficacy and coaching competence. The first canonical correlation reveals that all of the measured variables are indicators of one higher order model of coaching confidence.

The second two canonical correlations clarified the multidimensional nature of the construct in a way that was consistent with Feltz et al.'s (1999) coaching efficacy construct. More specifically, the second canonical correlation discriminated between the efficacies of motivation (as well as competence in motivating and communicating) and teaching technique (as well as perceptions of competence as a teacher). Two of the four first order factors of the CES were confirmed. Further, the inclusion of teaching competence with teaching efficacy, and motivation and communicating competence with motivation efficacy makes intuitive sense, and is congruent with Feltz et al.'s conceptualization.

The third canonical correlation followed the same pattern. The result discriminated between game strategy and teaching efficacies, with Barber's (1998) notions of teaching and training and conditioning being aligned with teaching efficacy. Again, this distinction is consistent with the factor structure of the CES in that strategy and teaching are two separate aspects of coaching efficacy. The finding that Barber's (1988) subcategory of training and conditioning emerged suggests that efficacy beliefs regarding physical preparation may also be part of the perceptions coaches have in the role to instruct and educate their athletes. This finding is not incompatible with Feltz and colleagues' (1999) definition of teaching technique efficacy, but it does extend the concept somewhat.

In summary, even though the first canonical correlation was congruent with both conceptualizations of coaching confidence, the second two were more consistent with Feltz et al.'s (1999) coaching efficacy construct and the four-factor structure of the CES as opposed to Barber's (1998) nine component model of coaching competence. Further, as opposed to Barber's model, the four-factor structure of the CES has been supported through both exploratory and confirmatory factor analyses (Feltz et al., 1999). The one factor of the coaching efficacy conceptualization that was not specified in these canonical correlations was character-building efficacy, but this factor was included in the first canonical correlation, as being a significant component of a larger factor of generalized coaching confidence. The lesser salience of character building is consistent with Feltz et al.'s findings. They noted that although it was a significant component of coaching efficacy, it loaded least strongly onto the second order factor of general coaching efficacy, that it was comprised of fewer items than the other three factors, and that is was not significantly predicted by their sources.

The gender differences shown in this study can be viewed as support for the construct validity of coaching efficacy. Previous literature within sport has shown consistent gender differences with respect to self-efficacy (Lirgg, 1991), but they have not been investigated with coaching efficacy, at least not with an American sample. This study showed that female university coaches were less confident than males in game strategy efficacy, coaching during competition, knowledge of strategies and tactics, and ability to motivate their athletes. Barber (1998) found that females perceived themselves as more competent than males with respect to teaching sport skills. The gender differences presently found focused on more performance oriented coaching roles. The perceptions reported higher by males refer to those duties required for a successful team performance (e.g., strategizing, motivating). Given that males typically appear more confident in sporting settings (Lenney, 1977; Lirgg, 1991) it is not surprising that males perceive themselves to be more confident in these coaching duties (whether substantiated by superior ability or not).

The second purpose of this study was to examine the sources of coaching confidence. Only coaching efficacy was used because of the statistical redundancy between it and coaching competence. The regression models supported the sources noted by Feltz et al. (1999) and extended their conceptual model of coaching efficacy to include Barber's (1998) list of sources. In their original article, Feltz et al. noted that their list of sources should not be considered exhaustive, and suggested that other sources could be influential, particularly for character building efficacy. The significance of coaching experience is consistent with the conceptualization of coaching efficacy and efficacy in general (Bandura, 1977, 1997). Previous experience has a powerful effect on efficacy beliefs.

The predictive qualities of gender are also consistent with previous research. Gender was a significant predictor for both game strategy efficacy and character building efficacy. Interestingly the beta weight for game strategy efficacy was negative, but it was positive for character building efficacy, implying that females display stronger perceptions of this efficacy than males. While Barber (1998) found that females perceived themselves as stronger teachers compared to males, presently it appears they may perceive themselves as more effective when it comes to instilling (i.e., teaching) mutual respect and good sportspersonship.

The prediction models based on Barber's (1998) sources are congruent with the conceptualization of efficacy in that event outcome (i.e., previous performance) was the most influential source of confidence. As Feltz et al (1999) suggested, there are some notable predictors of character building efficacy outside of their original conceptualization. Specifically, feedback from athletes, affect of athletes, and social comparison were all significant predictors. The results suggest that perceptions of efficacy with respect to the role coaches have in influencing the personal development of and positive attitude toward sport in their athletes is in part based on the social processes of interactions with these athletes. These interactions may involve the exchange of affect or evaluation from their athletes. Also important was the social perception of comparison to other coaches.

These findings can be seen as part of a larger conceptual/research puzzle. Given the rich conceptualization of coaching efficacy and Barber's (1998) significant work on gender differences in coaching competence, we can be very confident of the results of the present study. Namely, these results are (1) that there is one construct of coaching confidence, as operationalized by the CES; (2) that there are a wider variety of sources than given in Feltz et al.'s framework; and (3) that in coaching, as in other areas of sport and physical activity, there are rich and powerful gender differences.

Future research could progress along several paths. Specific areas may include the clarification and addition of more sources for coaching efficacy. All of the studies to date have used quantitative methods; it is likely that a qualitative methodology would compliment this work. Further, researchers should clarify the outcomes of coaching efficacy and more confident coaches. It seems reasonable to expect that factors such as the types of leadership behaviors and styles would interact with coaching efficacy to effect the coach-athlete relationship (see Kent & Sullivan, 2003; Sullivan & Kent, 2003). These outcomes might be specific to the coach's efficacy in certain areas (i.e., a coach who is efficacious in teaching skills might lead to better instruction and ultimately increased learning on the part of the athletes, whereas efficacy in building character might promote moral development in athletes).

Table 1

Descriptive Statistics for Coaching Effficacy and Coaching Competence

Variable Females Males Total

 Mean SD Mean SD Mean SD Skewness

Coaching Efficacy

Game strategy 7.40 .89 7.81 .82 7.70 .86 -0.54
Motivation 7.26 .95 7.55 .88 7.47 .91 -0.61
Technique 7.79 .76 7.93 .81 7.90 .79 -0.76
Character
building 8.13 .77 7.93 .98 7.99 .93 -1.29
Total 7.58 .69 7.79 .70 7.73 .70 -0.43

Coaching Competence

Teaching
sport skills 4.21 .55 4.26 .62 4.24 .60 -0.55
Communication
skills 4.19 .57 4.22 .57 4.21 .57 -0.21
Practice and
season planning 4.18 .49 4.26 .62 4.24 .59 -0.07
Coaching during
competition 3.91 .60 4.17 .55 4.10 .58 -0.51
Knowledge
of strategies
& tactics 3.94 .60 4.24 .52 4.15 .56 -0.21
Ability to
motivate athletes 3.80 .69 4.03 .59 3.97 .62 0.15
Training and
conditioning 4.07 .66 4.07 .67 4.07 .66 -0.25
General coaching
competence 4.08 .49 4.29 .48 4.23 .49 0.05

Variable Kurtosis Alpha

Coaching Efficacy

Game strategy -0.19 .91
Motivation 0.43 .91
Technique 0.61 .84
Character
building 2.32 .88
Total 0.21 .90

Coaching Competence

Teaching
sport skills 0.32 .90
Communication
skills -0.76 .87
Practice and
season planning 0.55 .87
Coaching during
competition -0.54 .90
Knowledge
of strategies
& tactics -0.34 .87
Ability to
motivate athletes -0.79 .91
Training and
conditioning -0.47 .89
General coaching
competence -0.27 .90

Note. CES factors were scored on a scale of 1 to 9 whereas Coaching
Competence factors were scores on a scale of 1 to 5. In all cases,
higher scores indicate greater confidence.

Table 2

Correlation Matrix For Coaching Efficacy and Coaching Competence

Coaching Efficacy 1 2 3 4 5 6

1. Game strategy -- .59 .62 .40 .54 .29
2. Motivation -- .52 .63 .36 .66
3. Technique -- .42 .73 .34
4. Character building -- .36 .41

Coaching Competence

5. Teaching -- .23
6. Communication --
7. Practice and season planning
8. Coaching during competition
9. Knowledge
10. Ability to motivate
11. Training and Conditioning

Coaching Efficacy 7 8 9 10 11

l. Game strategy .53 .77 .74 .35 .62
2. Motivation .42 .60 .51 .29 .56
3. Technique .55 .55 .56 .52 .53
4. Character building .40 .40 .37 .26 .40

Coaching Competence

5. Teaching .64 .52 .56 .28 .57
6. Communication .30 .45 .32 .63 .31
7. Practice and season planning -- .52 .60 .34 .53
8. Coaching during competition -- .71 .48 .38
9. Knowledge -- .44 .45
10. Ability to motivate -- .39
11. Training and Conditioning --

Note. All correlations are significant at p = .01.

Table 3

Canonical Loadings of Coaching Efficacy and Coaching Competence

Variable Function 1 Function 2 Function 3

Coaching Efficacy

Game Strategy -.93 .22 .30
Motivation -.80 -.58 -.12
Teaching -.79 .32 -.51
Character building -.54 -.18 -.25

Coaching Competence

Teaching -.71 .43 -.46
Communication -.56 -.58 -.28
Practice and season planning -.66 .18 -.18
Coaching during competition -.91 .03 .24
Knowledge -.86 .18 .15
Ability to motivate -.65 -.65 -.19
Training and conditioning -.52 .22 -.61

Table 4

Summary of Regression Models Predicting Coaching Efficacy

Model Dependent Predictor B SE B t
 Variable

1 Game Strategy Coaching
 Experience 0.02 0.01 0.23 **
 Gender -0.37 0.14 -0.19 *
2 Motivation Coaching
 Experience 0.02 0.01 0.22 *
3 Character Coaching
 Experience 0.03 0.01 0.30 *
 Building
 Gender 0.33 0.16 0.15 *
4 Game Strategy Event Outcome 0.22 0.08 0.21 *
5 Motivation Event Outcome 0.18 0.08 0.16 *
 Affect of
 athletes 0.32 0.11 0.25 *
6 Character Feedback from
 athletes 0.35 0.14 0.24 *
 Building
 Affect of
 athletes 0.25 0.12 0.19 *
 Social comparison -0.19 0.09 -0.18 *

Note. Models 1-3 are based on sources specified by Feltz et al. (1999).
Models 4-6 are based on Barber's (1998) sources.

* P<.05 ** P<.001

References

Barber, H. (1998). Examining gender differences in sources and levels of perceived competence in interscholastic coaches. The Sport Psychologist, 12, 237-252.

Bandura, A. (1977). Self-efficacy: Toward a unifying theory of behavior change. Psychological Review, 84, 191-215.

Bandura, A. (1997). Self-efficacy: The exercise of control. New York, NY: WH Freeman and Company.

Bryant, F.B. (2000). Assessing the validity of measurement. In L.G Grimm & P.R. Yarnold (Eds.), Reading and understanding more multivariate statistics. Washington, DC: American Psychological Association.

Coakley, J.J. (2001). Sport in society: Issues and controversies (7th ed.). New York, NY: McGraw-Hill College Division.

Cronbach, L.J. (1951). Coefficient alpha and the internal structure of tests. Psychometrika. 16, 297-334.

Feltz, D.L. (1988). Self-confidence and sport performance. In K.B. Pandoff (Ed.), Exercise and sport science reviews (pp. 423-457). New York, NY: Macmillan Publishing Co.

Feltz, D.L., & Chase, M. (1998). The measurement of self-efficacy and confidence in sport. In J. Duda (Ed.), Advancements in sport and exercise psychology measurement (pp. 63-78). Morgantown, WV: Fitness Information Technologies.

Feltz, D.L., Chase, M.A., Moritz, S.E., & Sullivan, EJ. (1999). A conceptual model of coaching efficacy: Preliminary investigation and instrument development. Journal of Educational Psychology; 91, 765-776.

Kenow, L., & Williams, J. (1999). Coach-athlete compatibility and athlete's perceptions of coaching. Journal of Sport Behavior, 22, 251-259.

Kent, A., & Sullivan, P.J. (2003). Coaching efficacy as a predictor of university coaches' commitment. International Sports Journal, 7, 78-88.

Lee, K.S., Malete, L., & Feltz, D.L. (2002). The strength of coaching efficacy between certified and noncertified Singapore coaches. International Journal of Applied Sports Sciences, 14, 55-67.

Lenney, E. (1977). Women's self-confidence in achievement settings. Psychological Bulletin, 84, 1-13.

Lirgg, C.D. (1991). Gender differences in self-confidence in physical activity: A meta-analysis of recent studies. Journal of Sport and Exercise Psychology. 13, 294-310.

Malete, L., & Feltz, D.L. (2001). The effect of a coaching education program on coaching efficacy. The Sport Psychologist, 14, 410-417.

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

Nicholls, J.G. (1984). Achievement motivation: Conceptions of ability, subjective experience, task choice, and performance. Psychological Review, 81, 328-346.

Sisley, B.L., Weiss, M.R., Barber, H., & Ebbeck, V. (1990). Developing competence and confidence in novice women coaches: A study of attitudes, motives and perceptions of ability. Journal of Health. Physical Education, Recreation and Dance, 61, 60-64.

Sullivan, P.J., & Kent, A. (2003). Coaching efficacy as a predictor of leadership style in intercollegiate athletics. Journal o[Applied Sport Psychology, 15, 1-11.

Tabachnick, B.G., & Fidel, L.S. (2001). Using multivariate statistics (4th ed.). New York, NY: Harper Collins College Publishers.

Weinfurt, K.P. (1995). Multivariate analysis of variance. In L.G. Grimm & P.R. Yarnold (Eds.), Reading and understanding multivariate statistics. Washington D.C.: American Psychological Association.

Weiss, M.R., & Sisley, B.L. (1984). Where have all the coaches gone? Sociology of Sport Journal. 1,332-347.

Weiss, M.R., & Stevens, C. (1993). Motivation and attrition of female coaches: An application of social exchange theory. Journal of Sport and Exercise Psychology, 10, 322-333.

Address Correspondence To: Sandra E. Short, Ph.D. Department of Physical Education and Exercise Science, University of North Dakota, Box 8235 Grand Forks, North Dakota, U.S.A. 58202 Phone: (701) 777 4325 Fax: (701) 777 3531 Email: Sandra_Short@und.nodak.edu

Tracy L. Marback, Sandra E. Short and Martin W. Short

University of North Dakota

Philip J. Sullivan

Brock University