As portion of a larger survey, 201 participants, consisting 125 work forces ( average age = 29.99 old ages, ( SD= 11.07 ) and 76 adult females ( average age = 26.74 old ages, SD = 11.02 ) took portion in this survey and were actively involved in a broad scope of athleticss across the West Midlands, England. Collectively, participants were recruited via advertisement and direct contact with athleticss nines, bring forthing a convenient, heterogenous sample. The public presentation criterion of participants was classified as either elite ( n = 39 ) , who represented professional, international and national participants, or non-elite ( n = 162 ) , which comprised recreational and nine participants. The choice standard for engagement was wide and included jocks of all abilities who competed ; exclusion merely applied to those below 18 old ages of age. All participants were required to subscribe an informed consent signifier prior to questionnaire completion. DO YOU NEED TO INCLUDE THE BACKGROUND ELEMENTS? !
Background information. The background inquiries solicited demographic, personal and featuring features. Personal information requested included age and gender. Questions related to their athletics gathered information refering their old ages of experience, competitory degree and hours spent preparation and viing. In add-on, participants were asked to see the extent of any anterior athletics psychological science and mental accomplishments developing they had experienced.
Self-regulation in Sport Questionnaire ( SRSQ ) . This fresh questionnaire was developed to measure jocks ‘ usage of self-regulatory procedures in athletics. It is a 36-item stock list with each point tapping a scheme, procedure or scenario comparative to self-management of emotions, knowledges and behavior during featuring public presentation. The SRSQ comprises 5 subscales ; goal-setting ( 12 points ) , regulatory-responses ( 11 points ) , self-talk ( 5-items ) , self-monitoring ( 4 points ) , and self-awareness ( 4 points ) . Participants were asked to see how they would typically react to each point in preparation and competition and rate their response on a 5-point Likert graduated table, widening from 1 ( strongly disagree ) to 5 ( strongly agree ) . The mean mark for each subscale was calculated to bring forth a mean evaluation for each self-regulatory procedure. The Cronbach alpha coefficients for these subscales ranged from.69 to.92, corroborating the internal consistence and dependability of this psychometric instrument ( Tabachnick & A ; Fidell, 1996 ) . The cogency of this questionnaire was established as portion of the current larger survey.
The Revised Competitive Trait Anxiety Inventory-2 ( RCTAI-2 ) . The Competitive State Anxiety Inventory CSAI-2 ( Martens, Burton, Vealey, Bump, & A ; Smith, 1990 ) assesses the multi-dimensional nature of province anxiousness by integrating a measuring of cognitive anxiousness, bodily anxiousness and assurance. Albrecht and Feltz ( 1987 ) adapted the procedural instructions to inquire participants to react based on anxiousness symptoms normally experienced anterior to competition. This trait version is consequently named the Competitive Trait Anxiety Inventory-2 ( CTAI-2 ) , and retains the same subscales and points as the CSAI-2.
A collateral factor analysis ( CFA ) conducted by Lane, Sewell, Terry, Bartram, and Nesti ( 1999 ) revealed a hapless tantrum for the predicted factor construction of the CSAI-2. Furthermore, consequences from the Lagrange Multiplier trial consequence led to the recommendation for alterations to the graduated table ( Lane et al. , 1999 ) . Cox, Martens, and Russell ( 2003 ) revised the graduated table points, and a subsequent CFA provided grounds for the cogency of the Revised Competitive State Anxiety Inventory-2 ( RCSAI-2 ) . The trait version of the RCSAI-2 was used in the present survey and consists of 17 point step organizing three subscales ; cognitive anxiousness, bodily anxiousness and assurance. Participants were asked to rate the strength of the anxiousness symptoms they normally feel before or during a competition on a four-point Likert graduated table ( 1 = non at all to 4 = really much so ) . The marking and computation outlined by Cox et al. , ( 2003 ) was followed for each strength subscale ; points stand foring each subscale was summed so divided by the figure of points associated with the subscale before being multiplied by 10. Therefore, possible mean tonss for all subscales ranged from 10 to 40 ; with a higher value bespeaking greater strength. Cox et Al. ( 2003 ) reported Cronbach alpha coefficients of.81, .81, and.86 for cognitive anxiousness, bodily anxiousness, and assurance severally, corroborating the internal dependability of these subscales in the present survey.
A directional graduated table ( Jones & A ; Swain, 1992 ) was besides included and required respondents to bespeak whether the strength they encountered was considered by and large good ( facilitative ) or damaging ( enervating ) to their public presentation on a bipolar graduated table widening from +3 ( really helpful ) to -3 ( really unhelpful ) . The way tonss on each subscale were summed and ranged from +15 to -15 for cognitive anxiousness and ego assurance, and +21 to -21 for bodily anxiousness, as advised by Cox et Al. ( 2003 ) . A positive evaluation indicated a facilitative reading of the anxiousness experienced towards, whilst a negative mark reflected a enervating reading. Internal dependability coefficients for the directional subscales have been reported as acceptable by anterior surveies runing from.72 to.90 ( Swain & A ; Jones, 1996 ; Jones & A ; Hanton, 1996 ; Wiggins, 1998 ) .
An initial pilot survey consisting 10 participants, ensured the points composing the above instruments were clearly constructed, suitably phrased and did non arouse any confusion in the participants. Following ethical consent provided from the University of Birmingham Research Ethics Committee ( see appendix 1 ) , participant enlisting was achieved through a assortment of agencies to obtain viing jocks from a broad diverseness of athleticss across differing public presentation degrees. A comprehensive database of different athleticss nine and personal managers across the West Midlands, including The University of Birmingham was generated. Contact was made via electronic mail, missive, or personal visits to inform nines and squads of the nature of the survey, and seek permission to see preparation Sessionss for informations aggregation.
Ads seeking participants were besides displayed in local exercising constitutions to enroll athleticss participants, and class recognition was given to eligible undergraduate pupils analyzing Sport and Exercise Sciences. Additional information sing the survey intent and protocol was offered to prospective persons and inquiries were answered via electronic mail and/or verbal communicating. Once a elaborate information sheet had been read sing the survey standard and consent signifiers signed, the questionnaires were completed by the voluntary participants. The instructions were standardised and followed the recommendations of Martens, Vealey, and Burton ( 1990 ) to foreground the importance of honest responses that would be kept confidential, and that there were no right or incorrect replies. As portion of the broader survey, the questionnaires took about 25-30 proceedingss to finish ; see appendix 2 for several paperss.
In order to rarefy any societal influences and keep consistent environmental impacts on responses, questionnaire completion was conducted either entirely or in little groups, in a non-competitive ambiance, and accompanied by a research worker who would reply inquiries if necessary. Data aggregation was carried out in an array of locations which afforded easy handiness and a relaxed atmosphere. To cut down motivation/fatigue effects, the order of questionnaires presented in the battalion were counter-balanced.
Screening and cleansing of informations. All informations variables were ab initio screened for mistakes in informations entry, losing values and outliers at both univariate and multivariate degrees. Following informations cleaning processs, X instances were deleted.
Internal dependability. An analysis of the internal dependability of the CTAI-2R identified debatable Cronbach alpha coefficients for assurance and cognitive anxiousness strength subscales, with values of.64 and.60 severally. This was rectified by taking one point per subscale ( point 13 for assurance and point 14 for cognitive anxiousness strength ) and resulted in Cronbach alpha coefficients of an equal degree ( .87 and.80 severally ) . Following these changes, all subscales had about acceptable degrees of internal dependability, with alpha coefficients runing from.79 to.87 for the CTAI-2R, and from.69 – .92 for the SRQS. Because the Cronbach alpha for the self-awareness subscale of the SRQS ( .69 ) was merely below the recommended cut off degree of.70 ( Nunnally & A ; Bernstein, 1994 ) , it was decided that this subscale would be retained.
Descriptive statistics. The descriptive statistics for each dependant variable, including agencies, standard divergences and Cronbach Alpha Coefficients for the sample as a whole and by competitory degree are presented in Table 1. This survey considered the competitory degree differences in self-regulation battle and the assurance associated with these behaviors. This was because old research has shown self-confidence strength degrees ( Feltz, 1988 ) , the usage of psychological accomplishments ( Williams & A ; Krane, 2006 ) and anxiousness reading ( Jones et al. , 2004 ) were know aparting factors between these athlete sub-populations. Furthermore, competitory criterion has been shown to act upon the development of assurance ( Vealey, 2001 ) and hence public presentation position needed to be controlled.
The sample as a whole were self-assured ( 27 ) and reported strength evaluations of cognitive anxiousness ( 24 ) which were somewhat above those of bodily anxiousness ( 20 ) ; nevertheless evaluations were comparatively moderate respective to the possible marking scope ( 10 to 40 ) . Taken together, the participants reported anxiousness degrees and assurance as facilitative to public presentation. Jointly, the sample reported regular use of all self-regulatory procedures. Comparative to the other self-regulatory procedures, self-monitoring was used least often by the sample as a whole. More specifically, elect jocks reported lower anxiousness strength ( both cognitive and bodily ) , more positive directional anxiousness evaluations and higher assurance strength degrees, compared to non-elites. In add-on, elect jocks reported greater usage of all self-regulatory procedures.
Competitive degree differences. Two independent one-way multivariate analyses of discrepancy ( MANOVA ) were employed to find whether differences in anxiousness strength and way and assurance degrees between elite and non-elite jocks were important. In the first MANOVA, the strength of cognitive and bodily anxiousness, and assurance were entered as dependent variables and competitory degree was the independent variable. The 2nd MANOVA comprised cognitive and bodily anxiousness way subscales as the dependant variables, and competitory degree was retained as the independent variable. The information conformed to the statistical premise of homogeneousness of variance-covariance matrices, allowing farther statistical analyses to be conducted. However, the premise of equality of discrepancy for bodily anxiousness strength and cognitive anxiousness way was violated, and for this ground, the chance degree for finding significance was reduced to a more conservative alpha of.01 as recommended by Tabachnick and Fidell ( 2007 ) .
Intensity. The MANOVA for jocks ‘ strength degrees of the subscale variables revealed a important multivariate consequence, Pillia ‘s Trace = .16, F3, 197 = 12.42, P & lt ; .001, I·2 = .16. The univariate analyses revealed important effects for cognitive anxiousness, F ( 1, 199 ) = 18.70, P & lt ; .001, I·2= .09, and assurance, F ( 1, 199 ) = 18.70, P a‰¤ .001, I·2 = .05. No important consequence for bodily anxiousness strength was found. Examination of agencies established that elite jocks reported significantly lower cognitive anxiousness and higher assurance degrees respective to their non-elite opposite numbers.
Direction. The MANOVA for anxiousness way provided a similar form of consequences, and revealed a important multivariate consequence, Pillia ‘s Trace = .09, F ( 2,198 ) = 10.2, P & lt ; .001, I·2 = ? . Univariate analyses indicated important differences for both directional subscales: cognitive anxiousness, F ( 1, 199 ) = 20.01, P & lt ; .001, I·2 = .09, and bodily anxiousness, F ( 1, 199 ) = 10.13, P & lt ; .01, I·2 = .05. Comparison of the average tonss highlighted that elect jocks ‘ directional readings of strength degrees were more facilitative, both sing cognitive and bodily anxiousness, to forthcoming public presentation compared to the non-elite jocks. Furthermore, non-elites perceived both cognitive and bodily anxiousness as debilitating towards public presentation.
To find whether competitory degree differences in self-regulatory behavior, existed a 3rd MANOVA was performed. The self-regulatory procedures were entered as the dependent variables with competitory degree as the independent variable. The consequences revealed a important multivariate consequence of competitory degree on the combined usage of self-regulation procedures, Pillai ‘s Trace =.201, F ( 5, 195 ) = 9.22, P & lt ; .001, I·2 = .98, with a strong consequence size harmonizing to Cohen ( 1988 ) . Subsequent important univariate effects were reported for all five dependent variables: end scene, F ( 1, 199 ) = 39.180, P & lt ; .001, I·2 = .16 ; self-monitoring, F ( 1, 199 ) = 30.929, P & lt ; .001, I·2 = .14 ; self-awareness, F ( 1, 199 ) = 18.464, P & lt ; .001, I·2 = .09 ; regulative responses, F ( 1, 199 ) = 30.63, P & lt ; .001, I·2 = .13 ; and self-talk, F ( 1, 199 ) = 28.453, P & lt ; .001, I·2 = .13. Probe of the agencies identified that elect jocks used significantly more self-regulatory procedures than non-elites. The trial besides reinforced that competitory degree was a confounding variable and hence needed to be controlled in subsequent analyses
Bivariate correlativities were examined to find any other potentially confusing variables that needed to be controlled for subsequent analyses. The bivariate correlativities between assurance strength and the undermentioned variables included: old psychological experience ( R = .21, P & lt ; .01 ) , hours of preparation ( r = -.26, P & lt ; .001 ) and old ages of engagement ( r = -.17, P & lt ; .05 ) in the jock ‘s chief athletics. These correlativities suggested little yet important relationships harmonizing to Cohen ( 1988 ) and needed to be accounted for in subsequent analyses
Cognitive ( CA )
Somatic ( SA )
Assurance ( SC )
Cognitive ( CA )
Somatic ( SA )
Assurance ( SC )
Table 1. Descriptive statistics of the survey variables
Significant differences between elect and non-elite jocks: **p a‰¤ .001 ; * P & lt ; .005
Intensity graduated table: Possible scope: 10-40
Direction graduated table: Possible scopes: CA and SC = -15 to +15 ; SA = -21 to +21
Self-regulation: Possible scope 1-5
A Multiple hierarchal arrested development ( MHR ) was run to analyze if self-regulatory procedures contributed alone discrepancy towards the anticipation of jocks ‘ assurance degrees ( hypothesis 1 ) , after confusing variables had been accounted for ( Hoyt, Imel, & A ; Chan, 2008 ) . Using the consequences from the old analysis, and holding established competitory degree was a confusing variable from the 3rd MANOVA, this independent variable was entered in Step 1, together with old ages of experience, psychological experience and hours developing per hebdomad. Conducting an HMR warranted a coincident input of the independent variables in distinguishable groups in line with their prognostic importance to the dependant variable ( Cohen, & A ; Cohen, 1983 ) , which in the current survey was assurance.
Therefore the order of the consecutive input of possible forecasters for Steps 2 and 3 were determined by utilizing anterior research sing assurance beginnings and their association with the separate self-regulatory subscales. Goal puting and self-monitoring were considered to be the most likely self-regulatory dimensions to heighten public presentation ( O’Brien et al. , 2009 ; Anshel & A ; Porter DATE ) and hence serve as public presentation achievements which is the superior beginning of assurance ( Bandura, 1997 ) . Furthermore, regulatory-responses were believed to be a close contemplation of jocks ‘ ability to get by, which is believed positively associated with assurance ( Jones, 1995 ; Carver & A ; Scheier 1998 ; Hays et al. , 2009 ) . Therefore sing this research these three independent variables preceded self-awareness and self-talk in their entry and comprised the Step 2 variables. Self-talk is a signifier of verbal persuasion and harmonizing to Bandura, 1997 is the 3rd most of import beginning of assurance. Small research is presently available sing the influence of self-awareness on assurance, and hence, self-talk was coupled with self-awareness and established Step 3. Consequences are displayed in Table 2.
The first analysis of the hierarchal arrested development revealed that hours of developing per hebdomad, engagement old ages and extent of psychological experience did non significantly predict assurance. Therefore, these variables were removed since if they were retained, the standard mistake of the both I? coefficients may hold been amplified, and generated weaker statistical power ( Becker, 2005 ) .The MHR was repeated once more with merely the inclusion of competitory degree in Step 1.
Table 2. Consequences from MHR ; Self-regulatory forecasters of assurance
95 % CI
Note: *p & lt ; .05, **p a‰¤.001.
Analysis confirmed a important overall theoretical account, F ( 6, 194 ) = 19.48, P & lt ; .001, in which athletics degree and self-regulatory procedures jointly accounted for 37.6 % discrepancy in assurance. Consequences revealed that competitory degree was a important forecaster of assurance, accounting for a little proportion of the discrepancy of assurance strength ( R2 = 0.05 ) . Measure 2, which included regulative responses, end scene and self-monitoring, significantly altered the arrested development equation and accounted for the largest proportion of alone discrepancy ( a?†R2 = 0.32 ) in assurance, after commanding for competitory degree. Within this theoretical account, all three variables were important and positive forecasters, with end puting emerging as the greatest forecaster of assurance ( I? = .29, P a‰¤ .001 ) . Finally, the add-on of Step 3 did non significantly change the arrested development equation, bespeaking that self-talk and self-awareness together did non significantly lend any alone discrepancy in assurance ( a?†R2 = 0.003 ) .
To find whether the attendant addition of assurance was a go-between in the relationship between goal-setting and facilitative anxiousness reading ( hypothesis 3 ) , a series of multiple hierarchal arrested developments were conducted. A consecutive analysis followed, adhering to the recommendations made by Baron and Kenny ( 1986 ) , for both cognitive and bodily anxiousness way.
Initially, bivariate correlativities established little to medium important relationships between competitory degree and anxiousness way ( cognitive: R = -.30, P = & lt ; .001 ; bodily R =-.22 P & lt ; .01 ) or with end scene ( r = -.41, P & lt ; .001 ) harmonizing to Cohen ( 1988 ) . However, the first tally of the mediation analysis revealed competitory degree was non a important forecaster of anxiousness way, and hence removed from the arrested development theoretical account. Subsequently, the independent variables were entered in two stairss ; end scene was foremost entered, followed by assurance. The mediation analysis for cognitive anxiousness way ( CAD ) was ab initio run, followed by bodily anxiousness way ( SAD ) . For easiness of reading, consequences for results are reported at each measure and displayed in Table 3 and 4.
Measure 1. This arrested development determined whether anxiousness way was predicted by end puting inclinations. Either cognitive or bodily anxiousness way was the dependent variable and end scene was entered as the independent variable. Analysiss revealed that end puting significantly predicted anxiousness way: CAD, R2 = 0.324, F ( 1, 199 ) = 95.49, P & lt ; .001 ; SAD, R2 = 0.10, F ( 1, 199 ) = 21.99, P & lt ; .001. Both direct effects were in the hypothesised way, in which battle in end scene yielded a facilitative directional reading of anxiousness.
Measure 2. The 2nd arrested development tested whether end puting predicted assurance. Goal puting remained as the independent variable and assurance was treated as the dependant variable. The result and forecaster variables were changeless for both cognitive and bodily anxiousness arrested developments ; therefore the arrested development equation was the same at Step 2. Analysis revealed that end puting significantly contributed to the account of assurance ; R2 = 0.30, F ( 1,199 ) = 86.84 P & lt ; .001. These chief effects were congruous with the hypothesised way, in which end scene was positively associated with increased assurance.
Measure 3. The 3rd arrested development equation tested whether assurance predicted anxiousness way, after commanding for the consequence of end puting. Consequences revealed that assurance predicted anxiousness way when end scene was controlled: Cad, R2 = 0.36, adjusted R2 = 2.35, F ( 2, 198 ) = 54.46, P & lt ; .001 ; SAD, R2 = 0.12, adjusted R2 = 2.11, F ( 2,198 ) = 13.40, P & lt ; .001.Therfore assurance degrees added alone discrepancy to anxiety directional reading.
Measure 4. The concluding measure is the last demand, harmonizing to Baron & A ; Kenny, ( 1986 ) to be fulfilled to corroborate that assurance is a go-between. This measure considered whether the prognostic consequence of goal-setting on anxiousness way was still upheld when the consequence of assurance was included. Anxiety was regressed on goal-setting, together with assurance. The arrested development equation was so compared with the initial arrested development equation in measure 1 to see if the arrested development coefficient was reduced. A decrease in the equation would propose that assurance besides contributed to the anxiousness directional reading, and hence when included at the same time in the theoretical account, the part predicted by end puting decreased to account for this consequence. The I? value is obtained from theoretical account 2 of the arrested development equation 3, in which both independent variables are included.
Consequences indicated a decrease in the arrested development coefficient for end puting foretelling anxiousness way when assurance was included in the theoretical account. The consequence was non reduced to zero ; and hence suggests a partial mediation consequence ( Kenny & A ; Baron, 1986 ) . Regression coefficient of end puting anticipation on cognitive anxiousness was I? = 0.57 ( P & lt ; .001 ) , which had reduced to I? = 0.45 ( P & lt ; .001 ) in measure 4. This consequence was mirrored for bodily anxiousness, in which the arrested development coefficient for end puting gauging the result was reduced from I? = 0.32 ( P & lt ; .001 ) in measure 1, to I? = 0.22 ( P & lt ; .01 ) in measure 4. Therefore assurance besides contributed to the anxiousness directional reading, and was a partial go-between, which was represented by a decreased part predicted by end scene. The way of the I? values showed this indirect consequence was positive, uncovering that higher assurance degrees partly mediated a facilitative directional reading of anxiousness.
Significance testing of mediation consequence
Hayes ‘ ( 2009 ) recommendation to utilize the Bootstrapping trial was followed as a better option to the flawed Sobel trial ( 1982 ) . The trial determined assurance intervals for the mediation pathways utilizing a SPSS macro made available by Preacher and Hayes ( 2008 ) . The mediation consequence is considered important if the 95 % assurance intervals ( CI ) of this mediation does non bridge nothing. Given that the scope between the upper and lower boundaries of the 95 % assurance intervals for the mediation of assurance on cognitive anxiousness way did non traverse nothing, CAD = .72, 3.33 ; SAM = 0.17, 3.16, the mediation consequence is important.
Table 3: Arrested development equations of the mediation analysis for cognitive anxiousness directionA
Regression equation 1
Regression equation 2
Regression equation 3
Measure 3 ( model 1 )
Measure 4 ( model 2 )
Note: ***p & lt ; .001 ; ** P & lt ; .01 ; *p & lt ; .05
Table 4: Arrested development equations of the mediation analysis for bodily anxiousness way.
Regression equation 1
Regression equation 2
Regression equation 3
Measure 3 ( model 1 )
Measure 4 ( model 2 )
Note: ***p & lt ; .001 ; ** P & lt ; .01 ; *p & lt ; .05
Bandura, A. ( 1997 ) . Self-efficacy: The exercising of control. New York, NY: Freeman.
Baron, R. M. , & A ; Kenny, D. A. ( 1986 ) . The moderator-mediator differentiation in societal psychological research: Conceptual, strategic, and statistical considerations.
Journal of Personality and Social Psychology, 51, 1173-1182.
Becker, T.E. ( 2005 ) . Potential jobs in the statistical control variables in organizational research: A qualitative analysis with recommendations. Organizational Research Methods, 8, 274-289.
Cohen, J. W. ( 1988 ) . Statistical power analysis for the behavioral scientific disciplines ( 2nd edn. ) . Hillsdale, NJ: Lawrence Erlbaum Associates.
Cohen, J. , & A ; Cohen, P. ( 1983 ) . Applied multiple regression~correlation analysis for the behavioralsciences ( 2nd ed. ) . Hillsdale, N J: Erlbaum.
Hayes, A.F. ( 2009 ) Beyond Baron and Kenny: Statistical Mediation Analysis in the New Millennium. Communication Monographs, 76, 408aˆ?420.
Hays, Kate, Thomas, Owen, Maynard, Ian andBawden, Mark ( 2009 ) ‘The function of assurance in worldclass athletics public presentation ‘ , Journal of Sports Sciences, 27: 11, 1185 – 1199
Hoyt, W. T. , Imel, Z.E. , & A ; Chan, F. ( 2008 ) . Multiple arrested development and correlativity techniques: Recent contentions and best patterns. Rehabilitation psychological science, 53, 321-339.
Nunnally J.C. , & A ; I.H. Bernstein. ( 1994 ) . Psychometric theory ( 3rd ed. ) , McGraw-Hill, New York ( 1994 ) .
Preacher, K. J. , & A ; Hayes, A. F. ( 2008 ) . Asymptotic and resampling schemes for measuring and comparing indirect effects in multiple go-between theoretical accounts. Behavior Research Methods, 40, 879-891.
Preacher, K. J. , & A ; Hayes, A. F. ( 2008 ) . Asymptotic and resampling schemes for measuring and comparing indirect effects in multiple go-between theoretical accounts. [ Computer package ] . Retrieved March 18, 2010, from hypertext transfer protocol: //www.quantpsy.org
Sobel, M. E. ( 1982 ) . Asymptotic assurance intervals for indirect effects in structural equation theoretical accounts. In S. Leinhardt ( Ed. ) , Sociological methodological analysis ( pp. 290-312 ) . Washington, DC: American Sociological Association.
Tabachnick, B.G. , & A ; Fidell, L.S. ( 2007 ) . Using multivariate statistics ( 5th edn. ) . Boston: Pearson Education.
Vealey, R. S. ( 2001 ) . Understanding and heightening assurance in jocks. In R. N. Singer, H. A. Hausenblas, & A ; C.M. Janelle ( Eds. ) , Handbook of athletics psychological science ( pp. 550-565 ) . New York: Wiley