KEY WORDS: Emotion, mood, success, injury, measurement, performance.

It is estimated that in the U.K. alone there are 29 million sports injuries each year (Sperryn, 1994). Therefore, there is clear need for research to identify the antecedents of sports injury. The identification and control of antecedents in training and competition could enable athletes and those responsible for their welfare to develop strategies to reduce the risk of injury occurring. In reviewing the sports injury literature, it is evident there is little research that considers those pleasant and unpleasant psychological states that may be associated with injury, with the majority of research investigating emotional states following injury (Brewer, 1994; 1998).

As injury may occur during competition, it is logical to explore relationships between psychological states assessed before competition and resulting sport performance (for examples of this method see; Beedie et al., 2000; Jones, 1995; Hanin, 2000). Sport performance is typically assessed through an outcome measure with examples including win / loss, or race time depending on the sport under investigation. Hanin (1997; 2000) urged researchers to consider relationships amongst performance states across a range of situations including best, worst, average performance and more recently injury. Given research that shows psychological states assessed before competition relate significantly with performance, it is logical to suggest that such states would relate to a performance that led to injury.

Studies that have investigated the association between psychological states that precede injury tend to emphasize life-stress and daily hassles (Andersen and Williams, 1999). The research completed to date tends to suggest that negative states are associated with injury (Kolt and Kirby, 1994; Fawkner et al., 1999). For example, in a study of 115 gymnasts, Kolt and Kirby (1994) found that feeling anxious and tired preceded injury. Kolt and Kirby argued that the mechanism through which affect led to injury could be attributed to its influence on concentration.

Daily hassles are described as minor daily problems, irritations or changed individual encounters (Andersen and Williams, 1988). Athletes exposed to such stressors are thought to be at an increased risk of injury, again suggesting an association with negative psychological states and injury. Researchers have argued that life stress (a major life event e.g. bereavement, marriage) is cumulative in its effects, enhancing the likelihood of injury by disrupting concentration (Andersen and Williams, 1988). A recent study assessed hassles experienced on a weekly basis over the course of a competitive season (Fawkner et al., 1999). Fawkner et al. found that athletes were more likely to incur an injury when they experienced significant increases in daily hassles the week prior to injury. Comparatively, there were no significant changes in daily hassles for the non-injured athletes.

The relationships between psychological states and risk taking behavior could offer an alternative explanation for the antecedents of a performance induced injury. Evidence suggests that psychological states as assessed by the Profile of Mood States (McNair et al., 1971) are associated with risk-taking behavior (Hockey et al., 2000).

In contrast to the notion that negative psychological state profile are associated with injury, Hanin (2000) emphasized that injury might be associated with a range of positive and negative psychological states. This assumption was based on ample empirical research indicating that success related emotions include not only pleasant but also strong unpleasant experiences; similarly poor performance is often accompanied not only by unpleasant experiences but by pleasant emotions. Moreover, success may result in complacency, producing states such as satisfied, content and pleasant. These emotional states could be associated with a reduction in alertness that might in turn lead to diminished motivation. Following success, the illusion of being "too ready" for a task, may prevent the effective recruitment and utilization of all available resources. It is also possible that over-motivation could result whereby athletes may try harder, ignoring or underestimating ("forgetting") the role of preparation and sound performance routines. Based on anecdotal evidence and numerous observations of athletes, Hanin suggested that the Individual Zones of Optimal Functioning (IZOF) approach could be used to identify those performance states associated with injury, in addition to successful and unsuccessful performance outcomes.

Collectively, there is a great deal of evidence linking emotional states with performance in terms of success/failure, but relatively little research investigating emotional states and performances leading to injuries. Injuries resulting from performances could be associated with negative emotional profiles as suggested by Kolt and Kirby (1994), or associated also with a positive profile as suggested by Hanin (2000).

The aim of the present exploratory study was to explore retrospectively those psychological states experienced by athletes prior injury, and to contrast these experiences with emotional states preceding individually best and worst performance. It should be emphasized that the focus of the present study is pre-injury emotional states evaluated within the framework of individually successful and unsuccessful performances. From a methodological perspective, the strategy was to triangulate the methods used to observe relationships at intra-individual, inter-individual and intra-group levels. Therefore, performance related psychological states were assessed retrospectively using standardized psychometrics and individualized person-oriented methods. Standardized psychometrics focused on scores generated by the POMS that assess anger, confusion, depression, fatigue, tension, and vigor. Research has found this inventory to be an effective predictor of performance (Beedie et al., 2000). The assessment of performance related emotional states using individualized person-oriented methods was accomplished by utilizing five basic dimensions of the IZOF model (Hanin, 1997, 2000). These include:

• Form dimension (selective athlete-generated descriptions of situational emotional experiences)
• Emotion content (idiosyncratic athlete-generated markers within four global categories based on distinctions between hedonic tone (pleasant-unpleasant) and perceived functionality (optimal- dysfunctional)
• Emotion intensity
• Temporal dimension (included pre-performance and partly mid-event situations)
• Context dimension (consisted of three settings: situations when injury occurred; best ever, worst ever performances).

The present study aimed to examine two alternative explanations of retrospective assessments of performance-induced injury. First, if the assumption that injury is linked to negative psychological states (stress-induced) is correct, then emotional states preceding injury would be negative. Second, if the alternative IZOF-based explanation is true then emotional states prior to injury would include both negative (N+ N-) and positive (P+ P-) aspects with different functional interpretations (sources) of the injury outcome.

Participants
Volunteer Sport Studies students (N = 59, Age range 18 - 31 years; Male N = 35, Female N = 24) participated in this study. All participants competed at county level and above, with the majority competing in invasion and combat sports.

Narratives. Given the absence of research into emotional states that precede injury and the exploratory nature of this study, it was decided to use narratives (see Hanin, 2003). It was unclear whether the standard retrospective method of assessing psychological states would be produce meaningful data. Therefore, participants were asked to write a description of the circumstances surrounding a performance that resulted in injury, and produced their best and worst performance.

Recall individualized emotion profiling involves a stepwise procedure using a stimulus list of affect words used to identify the idiosyncratic content and intensity of optimal and dysfunctional emotions (see Hanin 1997, 2000, 2004; Hanin and Syrjä 1995a, 1995b, 1996 for more details). This methodology identifies positive and negative emotions that are subjectively meaningful in terms of the individual's past performance history and significant emotional experiences. For example, nervous is a negative emotion, but it may be considered to help achieve successful task completion (optimal), or prevent successful task completion (dysfunctional). Athletes generate individually relevant emotion words that best describe their optimal (helpful, beneficial) and dysfunctional (harmful, detrimental) positive and negative emotions. To help athletes generate individual items, the positive-negative emotion stimulus list is used. This list includes positive and negative emotions typically experienced in performance. Hanin (1993) compiled the English version of the emotion stimulus list through selection and revision of items from the 10 global affect scales described by Watson and Tellegen (1985). Examples of positive items are "active," and "calm"; negative items include "nervous," and "angry." Hanin and Syrjä (1996), reported reliability of idiosyncratic emotion scales in a sample of high-level soccer players. Mean intraindividual Cronbach alphas of each emotion subscale (P+, N+, P-, and N-) ranged from .54 to .90. Their study also provided evidence of recall and prediction accuracy in athletes. Specifically, significant correspondence between recalled and actual scores, and between predicted and actual scores was found in 76.5% and in 70.6% of the players.

Recall scaling includes several steps. First, optimal emotion patterns are identified. Athletes, using the stimulus list, select 4 or 5 positive and then 4 or 5 negative items that best describe their emotions related to individually successful performances in the past. Following this, dysfunctional emotion patterns are identified by selecting 4 or 5 positive and 4 or 5 negative items that describe their emotions related to individually unsuccessful performances. Finally, athletes use the stimulus list to generate individually relevant positive and negative emotion descriptors related to injured performance. Where it was deemed necessary, athletescould also add emotion words of their own choice. Each athlete generated idiosyncratic emotion descriptors for the four emotion categories: pleasant optimal (P+), unpleasant optimal (N+), pleasant dysfunctional (P-), and (unpleasant dysfunctional (N-).

Emotion intensity. A separate scale assessing intensity was used alongside each of the emotions selected by individual athletes. The intensity scale asked, "Now think about the intensity of your emotion before the (Best, worst or injured performance)". The intensity was measured on the Borg's Category Ratio (CR-10) scale (Borg, 1982) based on the range principle and constructed to avoid the ceiling effect. The CR-10 permits ratio comparisons to be made of intensities as well as determinations of direct intensity levels. Other research (Neely, Ljunggren, Sylven, and Borg, 1992) has shown it to be useful in quantifying stimuli such as exercise capacity and pain. In the present study a standard format of the CR-10 scale (Hanin and Syrjä, 1995a, 1995b) was used with the following verbal anchors: 0 = nothing at all, 0.5 = very, very little, 1 =very little, 2 = little, 3 = moderately, 5 =much, 7 = very much, 10 = very, very much, = maximal possible (no verbal anchors were used for 4, 6, 8, and 9).

In summary, our study will examine emotional states using five dimensions proposed in the IZOF model. These included form dimension (emotional functioning), emotion content (idiosyncratic labels within the four emotion categories), emotion intensity dimension, time dimension (pre-event situation), and context dimension (best, worst performance in competitions and injured performance).

Procedure
The institution in which the study was conducted granted ethical approval for this study. Participants signed informed consent forms prior to IZOF profiling. Data was then collected simultaneously from volunteer sport studies students at the start of a scheduled lecture. The first author described the IZOF process, with participants, and was present at all times to resolve any uncertainty regarding this process.

Table 1 contains qualitative descriptions provided by athletes when asked to describe the circumstances surrounding performance relative to three performance conditions. Results indicate that participants were able to provide clear descriptions of factors related to best, worst, and injured performance. Using this information, it was possible to identify a number of key themes. Performance that led to injury was described as playing with high aspirations, playing well, performing in an important competition, enjoyment, poor judgment, injury expectations, and bad luck. Best performance was described as playing without excessive pressure, playing in a competition perceived to be important, playing when well prepared, playing well and experiencing facilitative anxiety. Worst performance was associated with overconfidence, low confidence, excessive pressure, fatigue and distractions. Whilst the qualitative data identified some overlap with the circumstances surrounding best and injured performance, worst performance resulted in themes unique to this performance condition.

In order to illustrate the identification of emotions experienced before best, worst and injured performance an individual's qualitative (emotion content) and quantitative (emotion intensity) data are presented within Figure 1. Presenting data in this way offers a triangulation of qualitative and quantitative data.

A county level netball player indicated that confidence (intensity = 8) and determined (intensity = 7) were amongst the helpful positive affects experienced before the best performance, with the items intense (intensity = 3) and anxious (intensity = 2) being helpful unpleasant emotion. For the injured performance, the participant reported similar positive affects as best performance (confidence, intensity = 7; determined, intensity = 6). Injured negative affects included anxious (intensity = 6) uncertain (intensity = 7) and irritated (intensity = 6), these were seen to be unhelpful, hence different emotional states and functions to those reported to be associated with best performance. For the emotional profiles experienced before worst performance, the participant reported that unhurried (intensity = 5) and exhilarated (intensity = 6) were positive harmful with dissatisfied (intensity = 4) and concerned (intensity = 7) being amongst the harmful negative affects.

Support for the quantitative was provided by the participant's qualitative data. When describing their 'best ever' competition they described how the 'team played really well, confidence was high within the team/squad. I felt I played well because we were all on a high from winning as many games as we did and this helped my confidence in my own performance'. In describing their 'worst ever' performance the participant described how 'we didn't make the final when we all knew we had the potential to win it! I felt my performance wasn't as good as it could have been as my confidence was low'. In describing a performance in which they 'became injured' it becomes apparent that the description is more akin to a best performance as opposed to a poor performance. The player explains that 'I broke my finger in a friendly match with a local rival team. I was playing well up to the incident even though I was playing in a different position to my usual one'.

Table 2 and Table 3 indicate, that when results of the individual emotional profiles were collated, the most consistent emotional profile was associated with best performance and the least consistent profile was that of worst performance. It should be noted that the top 7 most frequently reported emotional states were selected for analysis. This means that frequency counts for all performance outcomes included emotional states that were identified by less than fifty percent of participants. For example, nervous featured in the top 7 emotional states for all three conditions. Participants identified 'Nervous' prior to best, worst and injured performance. Two other emotional states were identified in all three conditions, anxious and determined. The highest intensity for the emotional states of anxious and nervous was in the worst performance condition, with injured performance being in the middle. The intensity of affect for determined was highest in the best performance condition with injured performance again being in the middle. The intensity of the top 7 emotional states identified was relatively similar (for the same emotions) between best and injured performance outcomes, (ranging from 0 for "anxious" to 0.6 "nervous") with the largest difference being 0.6, see Table 2). Spearman correlation between seven top labels selected for injured and best performance situations was also significant (r = .86, p < 0.05). Most selected labels for worst performance had only three items similar to injured and best performances.

The IZOF approach was used to identify and assess participant generated emotion profiles (Hanin, 2000) associated with best, worst and injured performance. Out of the ninety-six emotional states available for selection, participants commonly identified the same 7 emotional states as being experienced prior to best and injured performance. As Table 2 indicates, the frequency with which the top seven emotional states were identified demonstrates a greater degree of consistency for best performance as compared to injured and worst performance. Injured performance provided a more consistent profile than worst performance. The emotion states retrospectively identified prior to worst performance varied considerably.

At the group level, three emotional states were reported across all performance conditions and reflected qualitatively different interaction patterns. Specifically, nervous demonstrated an increase in intensity from best to worst performance, with injured being in the middle (Best < Injured < Worst). The intensity of the item 'anxious' had a similar intensity for best and injured performance, but was scored higher for worst performance (Best = Injured < Worst). Determined was associated with best performance, moderately associated with injured with low determination linked with worst performance (Best > Injured > Worst). Notably absent before worst performance were motivated (Best = Injured > Worst), confident (Best = Injured > Worst), energetic (Best > Injured > Worst) and aggressive (Best < Injured > Worst).

A key aspect of Hanin's work has been the identification of unpleasant states that facilitate performance and pleasant states that are harmful for performance. Findings of the present study indicate that anxious, aggressive and nervous were unpleasant states that were perceived to facilitate good performance. All seven emotional states associated with a best performance describe states of high activation (high intensity and effort). It is notable that the same emotional states in terms of emotional content, but slightly different in intensity, are associated with performances that led to injury. Of the seven emotional states identified prior to best and injured conditions, four were pleasant and three were unpleasant, a finding that lends support to the notion that researchers should assess an equal balance of positive and negative emotional states.

The findings for worst performance show that the balance of pleasant and unpleasant affect was unequal. There were five unpleasant states and two pleasant states. Calmness might be associated with poor performance due to its link with complacency (Hanin, 2000), whereas feeling excited might lead to poor performance due to excessive activation (caused by trying too hard). Qualitative data (see Table 1) illustrates the possible mechanisms responsible for injury such as disrupted focus, more risk taking and unsustainable effort leading to fatigue, thus indicating poor control of those psychological factors regulating performance. The qualitative data produced by this study indicated that the perception of emotional states could influence its functional impact. In the best performance condition, athletes clearly described the potentially facilitative nature of anxiety. In most attributions, there is a strong meta-emotional component - knowledge and preference or rejection of a particular state based on past experiences. One participant described how they were "extremely pumped up prior to the game, it had been on my mind all week, but (usually) I play better in pressure situations" whilst another described how they "felt nervous but had put in lots of hard preparation for the event". The quality of preparation again appeared to be influential prior to worst performances, in this instance participants described poor preparation "I was very nervous about playing abroad and I was not very well prepared"(I need to prepare well not to feel nervous). Participants attributed poor performance to a variety of factors including fatigue, overconfidence, poor confidence, distractions and excessive pressure. Prior to performances in which participants were injured, many described how they were playing well up until the point of the injury "I was playing well and enjoying the game. I got an elbow to the nose which broke my nose", and "It was an important club game with England selectors watching and I was playing well until my injury". However, other factors were identified in describing the circumstances leading up to the injury including poor judgment, high aspirations, enjoyment/ enthusiasm, injury expectations and bad luck.

A number of emotional states were identified in the qualitative data provided by participants prior to best, worst and injured conditions. Prior to best and injured performances these predominantly included pleasant states such as enjoyment, happy, focused, determined, inspired and pumped. However negative states were also identified as being helpful and included aggressive, nervous, and anxious. Conversely when describing worst performance, only unpleasant states were identified within participants qualitative data. These included terms such as lost, fatigued, frustrated, pressured and nervous. The qualitative findings support the quantitative data and offer further support to the notion that researchers should assess an equal balance of positive and negative emotional states when profiling performance.

When viewed collectively, findings of the present study illustrate that using an ideographic approach in the assessment of affect can produce data enriched by the self-generated descriptions of the circumstances surrounding best, worst and injured performance, an aspect of profiling not assessed in standardized scales. Additionally, it was possible to distinguish between perceived (or experienced) states, relatively stable emotion patterns and meta-experiences reflected in athlete's attributions.

• Psychological states experienced by athletes prior to injured, best and worst performances were investigated retrospectively using a mixed methodology.
• Results indicate that retrospective emotional profiles before injured performance are closer to successful performance, than unsuccessful, and confirm differences between successful and unsuccessful performance, a finding that occurred using both methods.
• Future research should further examine the emotional antecedents of injury and that applied sport psychologists recognize the potential risk of injury associated with emotional profiles typically linked with best performance.