KEY WORDS: Competition, skill level, psychological skills.

Participants
Data for the study were collected from 115 male rugby union performers (n = 65 elite, n = 50 nonelite), who ranged in age from 18 to 36 years (M = 20.38, SD = 2.92), all of whom provided written informed consent. Elite participants were sampled from professional competition within the UK while the nonelite players were selected on the basis that they competed at a semi-professional club standard or below (cf. Hanton and Connaughton, 2002). All were in competition or training for competition at the time of data collection.

Instrumentation
Test of Performance Strategies (TOPS). The 64-item TOPS (Thomas et al., 1999) was developed to measure the psychological skills used by athletes in various sporting situations. Specifically, within its 16 subscales, it examines activation, relaxation, imagery, goal setting, self-talk, automaticity, emotional control, and negative thinking/attentional control skills during competition and practice settings. Seven factors are common to both competition and practice contexts, whereas negative thinking is only included in the competition context and attentional control only in the practice context. For the purposes of the current investigation and, in line with the hypotheses, only the competition scale was examined. Examples of items during competition included for relaxation "I am able to relax if I get too nervous at competition" and for goal setting "I set personal performance goals for a competition". Items for imagery included "I visualize competition going exactly the way I want it". Participants rated the frequency of each item on a scale anchored by 1 (never) to 5 (always), with overall psychological skill usage scores ranging from 4 to 20. Initial analyses of the psychometric properties underpinning the TOPS have been encouraging in terms of its construct validity (see Hardy et al., 1997; Thomas et al., 1999), while Thomas et al., 1999 have reported Cronbach alpha coefficients of between 0.78 and 0.80 for the competition subscales. For the current study, values of between 0.72 and 0.83 were reported.

Modified Sport Anxiety Scale (SAS). A modified version of the SAS (Smith et al., 1990) was used to measure the intensity and direction of the trait component of worry and somatic anxiety and comprised 16 of the 21 original items. The scale measuring concentration disruption was removed due to its reported failure to function in accordance with theoretical expectations (Dunn et al., 2000). This left seven items in the worry subscale and nine items in the somatic anxiety subscale. Examples of the worry subscale include "I feel nervous" and "I am concerned about performing poorly", while the somatic scale contains items such as, "I feel tense in my stomach" and "My heart races". For the intensity measure, respondents rated each item on a 4-point Likert scale ranging from 1 (not at all) to 4 (very much so). Intensity subscale scores ranged from 7 to 28 (worry) and 9 to 36 (somatic anxiety). Internal consistencies for the SAS subscales have been reported with Cronbach alpha coefficients ranging from 0.71 to 0.92 for somatic anxiety and 0.70 to 0.86 for worry (Hanton and Connaughton, 2002; Smith et al., 1990; White and Zellner, 1996). For this study, values of between 0.74 and 0.82 were reported for the somatic and worry scales respectively. Satisfactory levels of convergent and discriminant validity have also been observed (Smith et al., 1990).

For the purposes of the present study, the SAS was modified to include Jones and Swain's (1992) direction scale. Participants were required to rate the degree to which the intensity of each symptom experienced was usually interpreted as either facilitative or debilitative to subsequent performance. The direction scale, originally used as a modification of the CSAI-2 (Martens et al., 1990; Jones and Swain, 1992), consisted of a bipolar 7-point Likert scale, ranging from -3 (very debilitative) to +3 (very facilitative), with the midpoint of 0 representing a level of symptom that was interpreted as "unimportant" to performance. The direction subscale scores ranged from -21 to +21 (worry) and -27 to +27 (somatic). High levels of internal consistency have been demonstrated for the direction scale when incorporated into both the CSAI-2 and the SAS. Specifically, Cronbach alpha coefficients for the SAS were 0.87 and 0.88 for worry direction and 0.85 to 0.88 for trait somatic anxiety direction (Hanton et al., 2003). For the current study, values of between 0.85 and 0.91 were reported for the somatic and worry scales respectively.

Competitive Trait Anxiety Inventory-2 (CTAI-2) Self-Confidence Subscale. Self-confidence was measured using the subscale from Albrecht and Feltz's (1987) trait modification of the CSAI-2 (i.e., CTAI-2), where each item is responded to in terms of how the individual usually feels. The scale comprised 9 items with respondents rating the intensity of each on a 4-point Likert scale ranging from 1 (not at all) to 4 (very much so) with total scores ranging from 9 to 36. Sample items include "I feel self-confident" and "I'm confident I can meet the challenge". A Cronbach alpha value of .83 has been reported for this scale (Perry and Williams, 1998). A value of .85 was reported for the current study.

Procedures
In order to counter for any potential method bias a number of procedural measures were undertaken (cf. Podsakoff et al., 2003). First, to prevent any contextual influences (e.g., audience effects), the TOPS, SAS, and CTAI-2 self-confidence scale, were completed by participants on their own, in random order, and away from the competitive environment. Next, in order to ensure temporal separation of measurement instruments the scales were administered separately within a 24-hour time lag. Finally, prior to completion, each participant was presented with standardized instructions based upon the recommendations of Smith et al., 1990 and Martens et al., 1990 respectively. These emphasized the confidentiality of responses and the need to consider each item on its own merit, thus attempting to minimize social desirability, accentuate honesty, and indicate that there were no right or wrong answers.

Data analysis
Employing a moderate effect size, the sample size used gave a statistical power that exceeded the required value of 0.80 (Cohen, 1988). Data analysis was then divided into two stages. First, data screening procedures were conducted to investigate the accuracy of the data. Elite and nonelite groups were then examined in relation to participants' scores on the modified SAS, CTAI-2 self-confidence, and TOPS subscales using separate Multivariate Analyses of Variance (MANOVA) procedures. Univariate Analyses of Variance (ANOVA) with Bonferroni adjustments (p < 0.01, for SAS and CTAI-2 self-confidence subscales; and p < 0.001 for TOPS subscales) were employed for follow-up analyses.

Preliminary data analysis
Participants' scores on the measures were examined for accuracy of data entry, missing values, and fit between their distribution and the assumptions of multivariate analysis. No missing values were recorded and there were no univariate or multivariate within-cell outliers at p = 0.001. In line with recommendations of Tabachnick and Fidell, 1996, the assumptions of normality, homogeneity of variance-covariance matrices [F (3, 74928) = 1.21, p > 0.05], linearity, and multicollinearity were also observed to be satisfactory.

Modified SAS and self-confidence scores as a function of skill level
A one-way MANOVA was conducted for skill level to determine if any significant differences existed between elite and nonelite groups for the SAS and self-confidence subscales. The MANOVA was significant, Wilks's lambda = 0.93, F(5, 60) = 8.36, p < 0.01, η² = 0.09, with follow-up ANOVAs indicating significance for somatic intensity and worry direction only (Table 1). Specifically, the elite group reported a less debilitative interpretation of symptoms associated with worry (nonelite -7.11; elite -1.26; p < 0.01) and a more facilitative interpretation of somatic responses (nonelite 0.17; elite 7.10; p < 0.01) than the nonelite group. The elite group also reported higher CTAI-2 self-confidence scores (nonelite 23.20; elite 29.35; p < 0.01).

TOPS scores as a function of skill level
One-way MANOVA was conducted for skill level to determine if significant differences existed between elite and nonelite groups for the TOPS subscales in competition. The MANOVA for competition was significant, Wilks's lambda = 0.93, F(10, 55) = 24.81, p < 0.01, η² = 0.13, with follow-up ANOVAs indicating significance for imagery, self-talk, and relaxation usage (Table 2). Specifically, the elite group reported using more imagery (nonelite 13.10; elite 15.29; p < 0.01), and self-talk (nonelite 14.71; elite 16.32; p < 0.01) in competition, while the nonelite performers reported greater usage of relaxation strategies (nonelite 15.10; elite 13.23; p < 0.05).

This study examined multidimensional anxiety and psychological skills usage as a function of skill level in rugby union players. The predictions were based upon previous research that has independently examined skill level (e.g., Jones et al., 1994; Jones and Swain, 1995; Perry and Williams, 1998), psychological skills usage (e.g., Fletcher and Hanton, 2001; Maynard et al., 1995a), and multidimensional anxiety symptoms. Partial support was provided for all the research predictions under investigation. Specifically, in the context of the anxiety responses, while no differences were reported in intensity across skill level groups the elite performers viewed these symptoms as more facilitating to their performance than the nonelite athletes. These findings compare favorably with existing comparisons of competitive anxiety responses as a function of skill level. Interestingly, however, in the present study, although significantly less debilitative than the nonelite sample, the mean values for elite performers' worry anxiety direction were still perceived as debilitating (M = -1.26). This finding would appear to be in line with existing investigations of anxiety intensity and direction that suggest responses differ as function of sport type. Specifically, performers in contact sports, such as rugby union, experience more detrimental effects from cognitive anxiety symptoms due to the increased threat arising from personal confrontation (cf. Mellalieu et al., 2004). Conversely, as somatic state anxiety symptoms are classically conditioned to environmental cues, physical manifestations experienced tend to dissipate at the onset of competition as players become more involved in the activity (i.e., as the game progresses). The presence of somatic symptoms therefore tend to be viewed as facilitating in sports such as rugby union that they signify action, increased effort, or readiness for competition and the forthcoming contact.

Potential explanations for the observed differences in symptom interpretations across skill level may be found in the greater self-confidence reported by elite performers when compared to their nonelite counterparts. Indeed, one of the most consistent findings in the anxiety literature is that 'facilitators' of symptoms associated with the anxiety response report greater levels of self-confidence than debilitators (cf. Hanton et al., 2004). Self-confidence has subsequently been suggested to act as a resiliency factor and protect against the debilitating effects of anxiety (Hardy et al., 1996; Mellalieu, Neil, et al., 2006). The nature by which athletes use self-confidence to manage responses in stressful situations was identified in Hanton et al.'s (2004) qualitative investigation into the role of self-confidence in the competitive anxiety intensity and symptom interpretation relationship. In their study, elite performers reported using cognitive confidence management strategies including mental rehearsal, thought stopping, and positive self-talk to protect against debilitating interpretations of competitive anxiety. In the current study, elite athletes reported greater self-confidence and usage of imagery and self-talk than their nonelite counterparts. These findings are consistent with those of Hanton et al., 2004 and provide further support for the potential protection effects of self-confidence.

Based upon the work of Bandura, 1997, Hanton et al. have suggested that the confidence protection mechanism may take effect via athletes visualizing or recalling forthcoming or past successful skill performances when experiencing symptoms associated with doubts and negative images of performance. Similarly, the use of other cognitive strategies in combination, such as self-talk and cognitive restructuring, are also purported to serve a similar confidence management function by reducing, removing or altering the negative 'doubting' cognitions that athletes' experience. These cognitive strategies are therefore suggested to alter the overall mental experience of athletes from a negative state to a more positive confident outlook towards forthcoming performance.

Hanton et al.'s (2004) suggestions would appear to be congruent with Jones's (1995) model of control, and the proposals of Carver and Scheier (1998; 1999), in relation to how individuals use self-confidence to cope with adversity when attempting to achieve goals. Specifically, when appraising the likelihood of goal attainment individuals retrieve and utilize expectancies in the form of behavioral scenarios that are played through mentally (i.e., imaged). Those individuals that image positive scenarios and positive outcomes are suggested to lead to positive expectancies (i.e., enhanced self-confidence), while negative scenarios are reported to lead to reduced expectances and levels of self-confidence in the ability to reach goal attainment.

Based upon the findings of the current study, together with those of Hanton et al.'s (2004), practitioners should attempt to focus upon developing confidence protection strategies that build robust efficacy expectations in order to influence self-confidence symptoms and protection against anxiety debilitation. In conjunction with the use of mental imagery, individual-specific mental skill packages should therefore be developed that incorporate various forms of efficacy enhancement including enactive mastery or performance accomplishments, and verbal persuasion or positive self-talk.

The findings that nonelite performers experience their anxiety symptoms as debilitative and attempt to reduce these symptoms via the use of relaxation suggest that practitioners should implement relaxation-based programs with this population group. However, while support has been found for the efficacy of psychological relaxation techniques in reducing competitive anxiety intensity and debilitating interpretations of associated symptoms (e.g., Maynard et al., 1995a; 1995b) such methods may not be appropriate for the activation and arousal demands of certain sports such as rugby union. In particular, the reduction of anxiety intensity may decrease the performer's activation state, and subsequent mental and physical readiness to withstand the physical and confrontational nature of the sport. Indeed, it may not be possible, or even desirable, to reduce such symptoms via stress management techniques due to the relative high levels of activation states required for task performance (Hanton et al., 2000; Mellalieu et al., 2004).

In such circumstances, performers may need to reduce symptom intensity, restructure cognitions, and then raise activation states once again to appropriate levels, particularly if individuals possess insufficient self-confidence to manage their symptoms and to protect against negative interpretations. Elite performers who are debilitators may however be better advised to implement some cognitive restructuring techniques using psychological skills and strategies to interpret their anxiety as facilitative to performance including a combination of goal setting, self-talk, and imagery (Hanton and Jones, 1999).

The primary limitation of the study was the cross-sectional nature of the design, which precluded the inference of causality between psychological skill usage and symptom interpretations. However, taken collectively with the existing empirical research that has considered the athletes use of psychological skills and strategies (e.g., Fletcher and Hanton, 2001; Hanton and Connaughton, 2002; Hanton et al., 2004). The findings of the current study provide the basis to indicate that certain psychological skills (i.e., imagery and self-talk) are implicated in helping elite performers maintain robust perceptions of confidence, in order to cope with the stressful demands of high-level competition. Future research should therefore identify which psychological skill, or their combination, most contributes to the affective response in conditions of competitive stress. Indeed, Fletcher and Hanton, 2001 have suggested that any future examination in this area should consider the effectiveness of different interventions in eliciting positive symptom interpretations and performance improvements. In particular, researchers should consider the efficacy of one strategy versus another or the effects of combining different strategies to form a psychological skills package.

An additional limitation with the study rests with the current conceptualization of psychological skills usage through the utilization of the TOPS scale. Specifically, the fact that it only purports to measure the amount an individual utilizes a psychological skill, and does not consider whether the performer perceives he/she is actually using that skill effectively. For example, in the case of our findings regarding nonelite performers, an athlete may continuously attempt to adopt a somatic-based relaxation strategy (e.g., passive stretching) to alleviate precompetition cognitive and somatic symptoms in the hour prior to performance, however, this may be ineffective due to the incorrect technique adopted. In contrast, an elite performer may utilize images of coping successfully in competition for a 30-second period in the dressing room directly prior to running out onto the field for the match, which may be sufficient to maintain his efficacy expectations regarding the upcoming performance. Clearly qualitative methods may be appropriate here to unearth information on these recommendations. Future research into psychological skills and experiences of the competitive anxiety response should therefore attempt to assess not only the frequency of usage but also the perceived effectiveness of the usage of that skill. Initial empirical support for the distinction between these concepts can be found in the coping literature where researchers have identified athletes' perceptions of strategy effectiveness to be as important as coping usage itself (Neil et al., 2004; see also Bolger and Zuckerman, 1995).

The findings of this study suggest that elite and nonelite athletes differ in their use of psychological skills to cope with their experiences of symptoms associated with competitive anxiety. Specifically, nonelite performers primarily use relaxation strategies to reduce anxiety intensity while elite athletes appear to maintain intensity levels and adopt a combination of psychological skills to interpret symptoms as facilitative. Potential mechanisms for this process include the use of imagery and verbal persuasion efficacy-enhancement techniques. Nonelite performers who experience anxiety symptoms as debilitative should implement relaxation-based programs. However, this may be inappropriate for certain sports that require high levels of activation states. Performers may therefore need to reduce symptom intensity, restructure cognitions, and raise activation states once again to appropriate levels. Elite performers who are debilitators are advised to implement cognitive restructuring techniques to interpret their anxiety as facilitative to performance via a combination of goal setting, self-talk, and imagery. Future research into psychological skills and experiences of the competitive anxiety response should attempt to assess not only the frequency of usage but also the perceived effectiveness of the usage of that skill.

• Nonelite performers primarily use relaxation strategies to reduce anxiety intensity.
• Elite athletes maintain intensity levels and adopt a combination of psychological skills to interpret symptoms as facilitative.
• This process occurs through imagery and verbal persuasion efficacy-enhancement techniques.
• Nonelite performers who are debilitators should implement relaxation-based programs. However, in high activation level sports performers should reduce symptom intensity, restructure cognitions, and then raise activation states again to appropriate levels.
• Elite performers who are debilitators should implement cognitive restructuring techniques to interpret their anxiety as facilitative via a combination of goal setting, self-talk, and imagery.